Datasets:

mimiklee

/

masterthesis-longt5-sum-8k

like 0

Introduction Mobile phones have become the most ubiquitous information and communication technology (ICT), with an estimated 6.8 billion subscriptions for the world's population of 7.1 billion people (ITU, 2014). In contrast with personal computers, tablets, and the Internet, the adoption of mobile phones has been expansive among all population groups. For example, according to Eurobarometer statistics for 2014, the average percentage of mobile phone users in the 15 and over age group in the 28 EU member states is 92%, with 12 of the 28 EU member states having at least 95% mobile phone users in the population (European Commission, 2014a). Thus, we have witnessed the development of insightful research in the last 15 years on various social aspects of mobile phone use and its integration in social interactions in people's everyday lives (Green and Haddon, 2009). While most of this research has been focused on the use of mobile phones by teenagers, younger adults, and working age peopleplausibly as a consequence of the fact that these user groups have represented the largest proportion of mobile phone early adopters heavily targeted by the industry (Chen et al., 2013)-noticeably less interest has been shown by scholars for other user groups and, in particular, for older adults (Hardhill and Olphert, 2012). This is somewhat surprising since official statistics show that the proportion of mobile phone users in the over-55 age group has been steadily increasing over the last decade. In fact, almost eight out of ten (79.5%) people aged 55 years and over had mobile phone access in the EU at the end of 2013, while this proportion was barely above one out of two persons (52.8%) in 2005 (European Commission, 2014b). Substantial differences, however, are still evident in mobile phone access among older adult age groups-despite the overall increase in the last decade (see Figure 1). For example, whereas in 2013, 90.3% of people aged 55-64 owned a mobile phone, this rate falls to 81.7% for the 65-74 age group and to 55.2% for those aged over 75 in the same year. Consequently, the limited prior literature dealing with older adults and mobile phones has been mainly focused on barriers and incentives for mobile phone use from two perspectives. On the one hand, researchers have shown interest in the predictors of the adoption of mobile devices among older adults. Accordingly, the Technology Acceptance Model (TAM) (Davis, 1989) and its variations (e.g., TAM3, UTAUT) (Venkatesh and Bala, 2008) have been applied to mobile phone adoption (e.g., Mallenius et al., 2007;Van Biljon and Renaud, 2008). Drawing on the suggestion that understanding the limitations associated with the appropriation of ICTs' specific characteristics and constraints of older adults must be considered, TAM-based models have been further extended with other variables, such as demographical, socioeconomic, and personal factors (e.g., cognitive and attitudinal) (e.g., Conci et al., 2009;Van Biljon and Renaud, 2008). Overall, the findings of this research support the suggestion that besides the core TAM variables (e.g., perceived ease of use, perceived usefulness, intention to use, and actual use), these additional characteristics of older adults may importantly condition the intensity and types of use of mobile phones and related services. On the other hand, the second group of studies has addressed the potential usability issues of older adults with the handling of mobile devices. For example, it has been shown that potential or progressive cognitive declines, hearing and visual impairments, and a higher susceptibility for disease render the learning of mobile phone use difficult (Kurniawan, 2008). As a result, older adults are faced with usability problems related to the shape and design of a device and the complexity of mobile phone (interface) handling (Chen et al., 2013;Zajicek, 2004). Commonly, mobile phones were found to be too small and uncomfortable for holding (Nasir et al., 2008). Further, small size implies small buttons and screen size/text on the screen (Kurniawan, 2008). Consequently, usability problems with multilevel user interfaces emerge, constraining older adults to address more easily available functions in the menus (Ziefle and Bay, 2004). These issues seem to become even more pressing toward older adults' approaching smartphones, as the potential benefits associated with their use (Doughty, 2011;Joe and Demiris, 2013;Plaza et al., 2011) could be inhibited by the currently scarce possibility of adaptation of the interface to the individual needs of older users (e.g., Al-Razgan et al., 2012;Zhou et al., 2014). In terms of needs, older adults commonly identify their mobile phone as a kind of safety device (e.g., Hardhill and Olphert, 2012;Oksman, 2006;Plaza et al., 2011), which, on one hand, fosters their independence and, on the other hand, provides inclusion with the potentially perpetual access to their social network. In fact, a consistent finding of prior research has been that older adults perceive and use the mobile phone as a personal communication device to stay in touch with family members and other persons (e.g., friends, neighbors, and caregivers) (see Section 2.1). These members of their social network provide them with emotional support and/or social companionship, which reassures older users that they have someone to rely on when dealing with everyday life occurrences and/or emergency situations. Nonetheless, in spite of such findings, no evidence has been published until now (to our best knowledge) on how different types of social support networks and their characteristics are associated with older adults' use of mobile phones for social support exchange. It is with this limitation in mind that this study aims to theoretically and empirically examine how mobile phones are used by older adults in communication with their providers of social support and how the composition and type of support network, as well as the sociodemographic characteristics of older adults, are related to the frequency of mobile communication. In addition, drawing on previous findings that show that older adults often combine the mobile phone with landline phone communication to maintain family and social relations (Kurniawan, 2008;Fernández-Ardèvol and Arroyo Prieto, 2012;Hardhill and Olphert, 2012), this study also investigates the relation between these two technologies for communication as it relates to emotional support and social companionship networks. The data for this study was collected from a representative sample of retired older adults in Slovenia. In comparison with the EU average, Slovenia has a more accentuated trend towards aging population (Vertot, 2010) and a slightly worse accessibility and utilization of ICTs among older adults (Dolnic<unk>ar and Nagode, 2010). The focus on retired older adults was informed by the social gerontological approach which-in contrast with chronological age definition-argues that old age is a socially constructed category linked to the changes in people's social roles that condition their life courses, with retirement being an important turning point in old age. The paper is structured into six sections. Section 2 presents a review on the uses of mobile phones by older adults for social interaction and supportive communication. The section then elaborates on the characteristics of social support networks of older adults, with the aim of understanding the possible roles of mobile communication in the exchange of different types of social support. Then, the research questions are formulated. Section 3 explains the methodology and describes the nationwide representative survey data used to empirically study the research questions. The results are presented in Sextion 4 and Section 5 presents the discussion of the results and the implications they suggest, including a comment on the limitations of this study and possible issues which might be addressed in future work. Finally, in the last Section 6, the conclusive remarks are drawn. --- Literature review --- Mobile phones and support communication Since the beginning of mobile phone proliferation among the general population, the social uses of mobile phones-being dependent upon the personal and contextual characteristics of their users-have been largely diversified (Leung and Wei, 2000). With the adoption of smartphones, the vast array of their usage patterns has further increased and extended to new social domains and practices (Jihyuk and Sang, 2013;Verkasalo et al., 2010). However, for many older adults, experience with mobile phones has continued to be limited to mobile voice calls for safety and security issues (Martinez-Pecino et al., 2012). Therefore, many older users have adopted the mobile phone for episodic uses (in contrast with ''always-on'' connectivity of young adults and working age groups) while being out of the home or when not having access to other communication means, such as a landline phone (Hardhill and Olphert, 2012;Ling, 2008). This aspect is especially present among mobility-impaired or otherwise health disadvantaged older adults because it has a positive impact on their personal independence (Abascal and Civit, 2001) and self-esteem (Oksman, 2006). Nevertheless, as in the case of the uses of other ICTs, the element of independence is intertwined with other two key elements: socialization and inclusion (Sayago et al., 2011), which in the case of older adults are highlighted in the use of the mobile phone as a device for personal communication with core network members-grandchildren, children, and close friends. Mobile phones can be employed to strengthen relationships across time and space and foster cohesion among social ties in familial networks (Ling, 2008). Oksman (2006) argued that the relative significance of the inclusive aspects of mobile communication for older adults has increased over time. Recent research has found that older adults, compared with the younger generation, experience a better improvement in terms of life and network management with the same amount of increase in mobile use (Lee and Kim, 2014). As network management mainly involves familial relations (Kurniawan, 2008), older adults associate mobile communication with prompt and perpetual access to the coping resources associated with emotional support (Martinez-Pecino et al., 2012). The convenience and importance of the mobile phone for the exchange of emotional support has also been reported in the case of illness, when the multilayered supportive nature of mobile communication seems to have beneficial effects on older adults' mental and physical health and well-being (Pahor et al., 2009). Children, especially, typically provide their older parents with a mobile phone to keep in touch with them because they are worried about their old parents' well-being (Chen et al., 2013;Hardhill and Olphert, 2012;Mallenius et al., 2007;Oksman, 2006;Plaza et al., 2011). However, this communication has been found to be asymmetrical: older adults typically receive calls more often than they make calls with their family members and friends (Chen et al., 2013;Kurniawan, 2008). In part, such findings could be explained by the fact that for many older adults the feeling of balance and reciprocity in the exchange of support is especially important. Namely, they do not want to be a burden to their support network when they are no longer able to provide support to them. On the other hand, their socioeconomic conditions induce them to develop cost-efficient communicative practices to reduce the monetary expenses associated with mobile phones. For instance, socio-economic status (SES) factors, such as education, income, and social class, determine the personal ownership and use of mobile phones. On one hand, especially in low-income groups, financial considerations and the high cost of mobile devices and plans force users to develop costcontrol strategies and employ more selective and less frequent use of mobile phone, which is often combined with other more affordable services (e.g., landline phones) (Fernández-Ardèvol and Arroyo Prieto, 2012;Sawchuk and Crow, 2010). On the other hand, educational attainment is associated with less exposure to technological hindrance and more willingness to learn new digital skills, which have been positively associated with more skillful use of mobile phones (Fernández-Ardèvol and Arroyo Prieto, 2012). Moreover, the frequency of mobile communication with network members is also conditional on specific life stages and personal characteristics of older adults. For example, significant age-cohort effects result in young older adults having better access to mobile phones (Annafari et al., 2014;Hardhill and Olphert, 2012) and being much more active and familiar with the use of mobile phones than old older adults (Ling, 2008). Further, Oksman (2006) associates more intensive mobile contacts with friends and family as a consequence of labor status: retirement, the diminution of paid work activity, and a consequential increase of leisure time relationships generally lead to more intensive mobile communication. His observation corresponds with more general suggestions about retirement as leverage for changes in daily routines, social interactions, and social support networks. For example, the decreasing contacts with co-workers and business partners and the possible termination of mobile phone use at work, together with probable new social roles and activities, may influence mobile phone use. Diversity among older users of mobile phones is also induced by changes in marital status. For example, widowed older adults rely more heavily on mobile phones to share their loneliness, grief, and concerns with other people who are close to them (Oksman, 2006). Widowhood is often associated with older adults living alone. Whereas some studies associate the loss of a spouse and living alone with social isolation (Wenger, 1994), others observe an increase in social contacts and support despite such a loss (Kalmijn, 2012). Living alone brings about specific risks, needs, and opportunities, steering older adults to actively engage in interactions with their network members. In this context, a mobile phone could be beneficial for older adults, as it was found to be associated with larger and more diverse core discussion networks (Hampton et al., 2011). Further, physical and health problems might influence the use of mobile phones for social support (Abascal and Civit, 2001). While many researchers note that aggravated health conditions trigger the adoption of mobile phones, Mallenius et al. (2007) observe that the increased dependency of older adults on external help (i.e., home and health care workers) often implies more frequent use of the mobile phone for handling relationships with caregivers. In addition, Chib and Jiang (2014) underscore how mobile phones provide many benefits to mobility-impaired users. Advantages include not only their better physical mobility but also the sociopsychological aspects of their life associated with a higher sense of empowerment, which emerges from a better management of personal identities and social support networks. As it has been already noted, mobile phones do not represent the only communication device for older adults today in keeping up with their social ties. Older adults have also become involved with the complex media environment (Petri<unk> et al., 2011), where they have access to various communication devices that allow them to establish and sustain mediated supportive communication. For instance, Hardhill and Olphert (2012) suggest that many older adults have started using the mobile phone as a complement to the landline phone in order to keep in touch with people when outside the home. Thus, it seems that mobile phones have been integrated within the existing personal communication system rather than replacing in-person and landline communication (Fernández-Ardèvol and Arroyo Prieto, 2012;Hardhill and Olphert, 2012;Quadrello et al., 2005;Sawchuk and Crow, 2010). Only a small number of older users seem to substitute their landline with mobile phones for sending and receiving phone calls inside the home. Generally, such ubiquitous patterns of use are more common among older adults who have an active social life, have a (part-time) job, an affluent SES, and/or a more positive attitude toward technology (Hardhill and Olphert, 2012;Sawchuk and Crow, 2010). Consequently, for such users, the mobile phone often represents more than a communication tool to keep in touch with their social network. Rather, it has become a multifunctional device and a personal assistant with a range of daily used applications (e.g., alarms and calendars), which are helpful for them in many aspects of their daily life (Chen et al., 2013;Plaza et al., 2011). From the above presented research, we can derive the notion that older adults consider the mobile phone an important tool for supportive communication with their closest network members and that the patterns of its use may vary according to different life-stage circumstances and personal characteristics of older adults. However, the literature survey also highlights the gap in knowledge about how the structural characteristics of social support networks (e.g., size and composition) of the older adults are related to the use of mobile phones. Thus, the next section explores in some detail the specifics of older adults' social support networks, with the aim of identifying some of the characteristics that might be associated with the use of mobile phones. --- The social support networks of older adults Even though the process of creating, maintaining, and losing supportive ties does not stop with people getting older; aging strongly conditions the social support of older adults, affecting the size, composition, and interactions in their personal networks. Most importantly, due to specific life events-such as retirement, settlement in a nursing home, death of a partner, and health problems-which may impede the upkeep of social interactions and balanced social support exchange, many cohort and cross-sectional studies have revealed a visible decrease in network size in old age. For example, using cross-sectional data, Kogov<unk>ek et al. (2003) showed that older adults (67 and older) have smaller networks on average than younger people (18-24), whereas Gurung et al. (2003) detected a decrease in the network size for a cohort of older adults of approximately one person in 2 years. Similar findings can be found by Ajrouch et al. (2001) and Kalmijn (2012). However, changes in the network size are dependent on various factors and particularly on older adults' characteristics. In line with the aged heterogeneity thesis (Nelson and Dannefer, 1992), variability with increasing age has been observed, for instance, by Van Tilburg and Broese van Groenou ( 2002), who suggest age-related variations in network size, whereas Kalmijn (2012) and Shaw et al. (2007) observed noticeable differences in support network size and contacts between men and women. Interestingly, the network size shrinkage often does not affect the amount of support received. In fact, many studies have discovered a sufficient level of support being transferred despite the network contraction in old age (e.g., Kalmijn, 2012;Gurung and Von Dras, 2007;Gurung et al., 2003;Lang and Carstensen, 1994;Shaw et al., 2007;van Tilburg and Broese van Groenou, 2002), supporting the social convoy model (Kahn and Antonucci, 1980) and socioemotional selectivity theory (Carstensen, 1992). This apparent paradox can be reconciled by considering the (specific) composition of social support networks. By and large, older adults' networks are characterized by the loss of parents, a diminishing number of co-workers and friends, and a greater share of relatives and neighbors. Accordingly, with aging, close ties remain and gain in importance due to their multiplexity (the number of different types of social support provided), while the numbers of more distant members decline (Lang and Carstensen, 1994). Shaw et al. (2007) noted that older adults have fewer contacts with friends but nevertheless strong links with close family members, who provide various kinds of stable support. However, the composition of social support networks varies according to the amount and type of social support needed. Gurung et al. (2003) point to Litwak's task-specific model, which suggests that different sources of support typically provide different kinds of support (e.g., family members provide instrumental support more often, while friends are more often engaged in emotional support and socializing), and Weiss's functional-specificity model, which assumes that certain specific kinds of support may be met only in certain relations. Given their limited time and abilities, older adults tend to focus, due to socioemotional needs, primarily on the closest ties, with adult children being their primary source of multiplex social support: they are crucial providers of financial support, substantial instrumental support, and support in the case of illness (Dremelj, 2003). Friends remain predominant for companionship and are also important providers of emotional support. The importance of neighbors as supportive ties also increases with age. The structure of social support network is intertwined with the frequency of social interactions with network members. On the most general level, Ashida and Heaney (2008) corroborated previous studies, with their results indicating that having frequent contacts with more network members is positively associated with higher social support, while Ajrouch et al. (2001) noted that the ego's age 1 has a negative effect on the frequency of contact with network members. However, the magnitude of this effect can vary according to the relationship between the ego and the alter, although older adults have fewer contacts with friends, they remain in stable contacts with family members (Shaw et al., 2007). Social interaction frequencies are also likely to be associated with the size of social support networks. On the one hand, authors report positive correlations between network size and social interactions (e.g., Van den Berg et al., 2012). On the other hand, Boase et al. (2006) found a negative relationship between the size of the support network and social interactions, 1 In ego-centered network approach for measuring personal networks the ego refers to the focal actor in a given network, whereas the alter is the person linked to the ego (Wasserman and Faust, 1994). Since in survey collection of ego-centered network data a respondent provides information about their personal network, we use interchangeably the terms respondent and ego as well as network member and alter in this paper. indicating potential limitations of individuals in sustaining frequent contacts with a large number of network members. Further, a rare study that takes into account different communication modes found out that older adults have fewer social interactions and that smaller network size leads to fewer social interactions on average (Van den Berg et al., 2012). They have also identified age as an important factor in the choice of mobile phone (in comparison with in-person communication) for interactions with network members: as anticipated, older adults were less likely to choose a mobile phone. In addition, the choice of using a mobile phone was associated with a larger proportion of strong and kin-based ties, while the network size was not significant. It is important to understand, however, that none of these studies focused on older adults, nor did any of the studies address the differences between the types of social support networks. --- Research questions Based on the literature survey and the identified gaps in understanding the role of the mobile phone for provision and exchange of social support among older adults, the following explorative research questions were distilled, representing the research framework for the empirical part of this study. <unk> How is the composition of support networks associated with the frequency of mobile phone communication within emotional support networks (RQ1a) and social companionship networks (RQ1b)? <unk> What is the relation between different forms of contact (i.e., in-person, landline phone, and mobile phone communication) in the communication of older adults within their emotional support networks (RQ2a) and social companionship networks (RQ2b)? <unk> How are the sociodemographic characteristics of older adults associated with the frequency of mobile phone communication within their emotional support networks (RQ3a) and social companionship networks (RQ3b)? --- Methods --- Procedure and sample The data for this study was collected with a cross-sectional nationwide telephone survey conducted in Slovenia in November and December 2009. A stratified sample was drawn at random from the database of all landline phone numbers in Slovenia. The questionnaire used in this study was divided into several question modules. The original sample (NT = 1209) was split into two subsamples to which respondents were randomly assigned. Since the data about social support networks were collected only in the first subsample (N1 = 602), the analysis presented below was carried out on the respondents in the first subsample, who reported their labor status as retired (NR = 178). In order to compensate for sample designs and patterns of unit nonresponse, weighting was used in the survey analysis. The raking post-stratification procedure was employed to balance the original sample demographics to match the following population parameters: age, education, gender, region, place of living, and employment status. --- Measures --- Social support networks --- Name network generators The social support networks were measured using the ego-centered network methodology, which is based on a two-step procedure. Firstly, the existing ties (i.e., alters) have to be identified with whom the respondent (i.e., ego) has some sort of (supportive) relationship. Secondly, the characteristics of ties (e.g., duration and multiplexity) have to be assessed. Drawing on the distinction between emotional support and social companionship, the providers of social support were identified with the following two survey questions (also known as name generators): Emotional support: ''From time to time, people discuss important personal matters with other people; for instance, when they have problems with other people or at work, and so on. Who are the people with whom you discuss personal matters that are important to you?'' Social companionship: ''From time to time, people socialize with other people; for instance, they take a trip together or go out for dinner, and so on. Who are the people you usually socialize with?'' The emotional support generator is a modified version of the core discussion network generator developed by Burt (1984). The social companionship network generator was first elaborated by van der Poel (1993) and later adapted and tested in telephone surveys in Slovenia (e.g., Hlebec et al., 2006Hlebec et al.,, 2010)), demonstrating high levels of validity and reliability (Kogov<unk>ek and Ferligoj, 2005). --- Network interpreters In the second step the characteristics of ties (i.e., alters) were identified by a block of subsequent questions, which are also known as network interpreters. Name interpreters were administered to respondents in a question-wise mode after a list of alters for both emotional support and social companionship networks was collected. The respondents were asked a set of name interpreters for each of the first five alters to limit the respondent's burden. If an alter was named by the respondent among the first five alters as providing both emotional support and social companionship, name interpreters referring to that alter were administrated to the respondent only once. Following the theoretical review and the research questions, we focus on six name interpreters in this study. The first group relates to the sociodemographic characteristics of alters and includes three name interpreters. The first is the role-relation interpreter, which was adopted from previous research (Burt, 1984). It was applied to determine the proportion of kin among alters, calculated as the sum of the number of partners, parents (and grandparents), siblings, and other kin divided by the total number of alters for whom the name interpreters were asked. The second interpreter measured the alter's age (in years), whereas the last sociodemographic interpreter assessed the duration of the tie-or the time of knowing each other in years. In order to eliminate the impact of the age of respondents on the alter's age and duration of ties, both interpreters were normalized. The normalized alter's age was calculated by dividing it with the ego's age, whereas the normalized duration of the tie was calculated as the proportion of years of having known the alter in relation to the age of the ego. The higher the values, the older the alter was in comparison with the respondent2 and the longer the respondent had known his or her alter.3 In addition, the second group of interpreters was related to the frequency of respondent's interpersonal communication with the alters, including three questions on how often the ego was used to communicate with each alter via a particular medium (i.e., inperson, landline phone, and mobile phone voice communication). The survey question here was as follows: ''How often do you communicate with (this) person [medium]?'' For each statement, the respondents answered on an 8-point ordinal scale, where 1 = ''never,'' 2 = ''once a year or less,'' 3 = ''several times a year,'' 4 = ''about once a month,'' 5 = ''several times a month,'' 6 = ''several times a week,'' 7 = ''every day or almost every day,'' and 8 = ''several times a day.'' These indicators represent a considerable limitation in terms of the operationalization of the temporal structure of media use, as it only captures the recurrence in the temporal structure of social interactions. Nevertheless, Licoppe (2004) suggested that the frequency of media use appears as one of the most relevant dimensions that distinguishes interpersonal communication via mobile phones. Similar suggestions are to be found in other works that focus also on other ICTs (e.g., Boase, 2008;Boase et al., 2006;Hampton et al., 2009). In addition, such operationalization has been elaborated upon and suggested in other large-scale ego-centered network surveys that measured the use of various media (including mobile and landline phones) and ego-alter communication (e.g., Hlebec et al., 2006); it was also assessed as the most valid self-report measure when asking respondents to estimate their use of mobile phones for personal communication (Boase and Ling, 2013). --- Sociodemographic variables Sociodemographics were measured with a set of standardized Eurostat questions, including gender, age, education, marital status, place of living, household size, and income, as well as subjective class identification. The latter was measured using responses to the following question: ''Among which social class would you place yourself, low, low middle, middle, high middle, or high?'' This question has been identified as a valid single-item measure of social class (Kluegel et al., 1977). In addition, to assess whether a respondent had a health impairment, they were asked whether they had any kind of heath problem that limits them in performing everyday tasks at home or at work so that the amount of work that can be done is limited (0 = ''No''; 1 = ''Yes''). Lastly, a variable measuring the frequency of mobile phone conversations was entered in the analysis, as a potential confounder of sociodemographic variables. Following the procedure of Vyas and Kumaranayake (2006), education, household income, and subjective class identification were used to construct a principal component analysis (PCA)-based SES score. Notably, the three interval variables were standardized and entered into the PCA. 4 Since only the associated eigenvalue (accounting for 58% of the variation in the original data) of the first extracted component was greater than one (<unk> = 1.741), the results indicated that a single component could be extracted with PCA. Accordingly, using the factor scores from the first principal component as weights (see Table 1), a new variable was constructed for each respondent, which was regarded as the respondent's SES score, and the higher the SES score, the higher the implied SES of that respondent. --- Results --- Sample overview As mentioned in Section 3.1, the analyzed subsample consisted of 178 retired respondents, of whom 39.3% were male (see Table 2). The sample contained almost an equal proportion of respondents between 55 and 64 years of age (47.7%) and 65 and 75 years of age (46.1%), while 6.2% were 54 years old or younger (M = 63.9 years, SD = 6.45). Two fifths of the respondents (41.3%) had a lower or middle vocational education, 4.5% had secondary vocational education, while 14.1% of them had some university education. More than six out of ten (61.9%) respondents were married, while the rest were single, either because they were widowed or they had never been married. More than a half (54.8%) of respondents lived in a rural area (settlements with up to 2000 inhabitants), 17.5% lived in semi-urban areas (settlements with 2000 to 10,000 inhabitants), and 27.7% lived in urban centers with more than 10,000 residents. Sixty-five retired respondents (36.9%) reported having a health problem that limited them in performing everyday tasks. With reference to the subjective measure of social class, nearly two-thirds (63.0%) of respondents identified themselves as ''middle,'' 31.8% saw themselves as ''middle low'' or ''low'' class, and 5.2% viewed themselves as ''middle high'' or ''high'' class. In addition, the largest proportion of the subsample lived in two-person households (40.7%), followed by one-person households (27.7%), while 31.6% of respondents were living in households with three or more persons. Finally, 22.4% of respondents answered that the monthly income of their household was below 590 €, 43.6% estimated it to be 591-1440 €, and 34.0% said the members of their household earned overall more than 1441€ per month (M = 1261.9€; SD = 863.7€; Me = 1000€). In terms of mobile phone use, 29 (16.3%) respondents did not use a mobile phone, whereas 149 (83.7%) respondents reported to be mobile phone users. Among them, 27.1% were episodic users (reporting to make a few mobile phone conversations monthly or less often), 54.7% reported to make between one and four conversations per day, while 18.2% of mobile phone users made more than five conversations in a typical working day. Therefore, to analyze the research questions, we could draw on 149 participants who were users of mobile phones. However, because of their specific characteristics-for example, respondents having no person in their support networks and/or giving invalid (i.e., item nonresponse) answers on selected variables (e.g., household income, social class), which were most often caused by explicit refusal or ''don't know'' answers-we were left with 91 eligible mobile phone users who provided valid answers on all variables entered in regression analysis for emotional support and 89 eligible units in regression analysis for social companionship.5 Notes: Percentages may not sum to 100 due to rounding. In cases where the sample size is not 178, this is due to item nonresponse, which was most often caused by explicit refusal or "don't know" answers. a The question was administrated only to the users of mobile phones. --- The composition of social support networks Before addressing the research questions, we outline and compare the characteristics of both types of social support networks (see Table 3). On average, respondents had significantly fewer people to rely on for emotional support than for social companionship (MES = 2.3 vs. MSC = 4.1, t = -4.2, p <unk>.001). Statistically significant differences between the characteristics of the two types of social support networks can also be observed in terms of percentage of kin, frequency of in-person communication, and mobile phone communication with their alters. Notably, for emotional support respondents rely more on kin than for social companionship (MES = 80.2% vs. MSC = 55.7%, t = 3.696, p <unk>.001). Similarly, on average, they were in more frequent in-person (MES = 6.69 vs. MSC = 5.92, t = 4.356, p <unk>.001) and mobile

The present post-print may not replicate the final version published in Elsevier journal. It is not the copy of record and may differ from the final version.

e, which was most often caused by explicit refusal or "don't know" answers. a The question was administrated only to the users of mobile phones. --- The composition of social support networks Before addressing the research questions, we outline and compare the characteristics of both types of social support networks (see Table 3). On average, respondents had significantly fewer people to rely on for emotional support than for social companionship (MES = 2.3 vs. MSC = 4.1, t = -4.2, p <unk>.001). Statistically significant differences between the characteristics of the two types of social support networks can also be observed in terms of percentage of kin, frequency of in-person communication, and mobile phone communication with their alters. Notably, for emotional support respondents rely more on kin than for social companionship (MES = 80.2% vs. MSC = 55.7%, t = 3.696, p <unk>.001). Similarly, on average, they were in more frequent in-person (MES = 6.69 vs. MSC = 5.92, t = 4.356, p <unk>.001) and mobile phone (MES = 4.8 vs. MSC = 4.34, t = 1.761, p = 0.082) contact with emotional support network members than with the people who provided them social companionship. Note: a Paired-sample t-tests were run only on the units entered into both multiple linear regressions (N = 88). Conversely, it is interesting to note that non-significant differences between the two types of support networks were found for the frequency of landline phone communication (MES = 3.64 vs. MSC = 3.8, t = -0.851, ns). In addition, the paired-sample t-test results revealed that there were no significant differences between the two types of social support networks with reference to the average alter's age and tie duration. In other words, no differences could be noted between the emotional support and social companionship networks in terms of the average alter's age (MES = 0.84 vs. MSC = 0.85, t = 0.48, ns) and how long the network members have known the respondent on average (MES = 0.57 vs. MSC = 0.55, t = 1.133, ns). --- The frequency of mobile phone communication within social support networks To answer the research questions, two multiple linear regression models were analyzed. The first model (here after referred as to Model ES) was run with the frequency of respondent's mobile phone voice communication with their network members on whom they rely for emotional support as a dependent variable and the characteristics of the emotional support networks and respondent's sociodemographic characteristics as independent variables. The Model ES was statistically significant [F (12, 90) = 5.280, p <unk>.001], with independent variables explaining 36.5% of the variance of the dependent variable.6 With reference to RQ1a, the results show that after controlling for all variables in the model, none of the analyzed characteristics of the social companionship network was a significant predictor (all betas have p >.1) of the frequency of mobile phone voice communication (see Table 4). Conversely, the results related to RQ2a showed that respondents who were in more frequent in-person contact (<unk> = 0.282, p =.013) or who more often talked with their social companionship network members on the landline phone ( = 0.367, p <unk>.001) also talked more frequently with them via a mobile phone. Moreover, the results for sociodemographic variables in Model ES (RQ3a) showed that the frequency of mobile phone conversations, ego's age, SES score, and living alone are significant predictors. However, while mobile phone conversations (<unk> = 0.381, p <unk>.001), SES score (<unk> = 0.233, p =.030), and living alone (<unk> = 0.192, p =.100) have a positive effect on the frequency of mobile phone communication between respon-dents and their network members, the ego's age has a negative effect (<unk> = -0.245, p =.012). The other two variables comprising gender (<unk> = 0.023, p =.799) and health impairment (<unk> = -0.027, p =.780) were not significantly associated with the frequency of mobile phone communication with network members. The same steps as in the case of the emotional support were followed in the multiple linear regression model (here after referred as to Model SC) for the social companionship network. The Model SC was statistically significant [F (12, 88) = 5.203, p <unk> 0.001], with 36.5% of variance of the frequency of mobile phone voice communication with alters in social companionship network being explained by the entered predictors. With reference to RQ1b, the results showed that among all predictors, only the alter's age marginally significantly predicts the frequency of mobile phone communication with network members (see Table 5). Specifically, the older the alters (in comparison with the ego), the lower the intensity of mobile phone commu-nication (<unk> = -0.191, p =.092). Regarding the RQ2b, the Model SC indicated that both parameters were statistically signifi-cant: respondents who were in more frequent in-person contact (<unk> = 0.264, p =.014) or who more often talked with their social companionship network members on the landline phone (<unk> = 0.174, p =.083) also talked more frequently with them via the mobile phone. In regards to the sociodemographic variables (RQ3b), the analysis revealed that only the frequency of mobile phone conversations and living alone are significant predictors, and notably, that older adults who had more mobile phone conversations were more likely to more frequently talk on the mobile phone with their network members (<unk> = 0.373, p <unk>.001). Moreover, in comparison with individuals who did not live alone, those who lived alone stayed in touch with their alters more often via the mobile phone (b = 0.316, p =.006). Conversely, age (<unk> = -0.135, p =.150), gender (<unk> = 0.036, p =.697), SES (<unk> = 0.148, p =.176), and health impairment (<unk> = -0.065, p =.486) were not statistically significant in the Model SC. 1.244 0.443 0.316.006 Notes: N = 89, F (12, 88) = 5.203 p <unk>.001, and Adj. R 2 = 36.5%. --- Discussion --- Findings and theoretical considerations The empirical investigation of the proposed research questions revealed interesting results with potentially relevant theoretical implications for understanding the role of mobile communication in support networks of older adults. With reference to the first two research questions (RQ1a and RQ1b), we found that with the exception of alters' age, the composition of support networks was not important for the frequency of mobile communication with network members. Notably, only the average age of social companionship providers was negatively related with the dependent variable. This was somewhat expected, as the proportion of mobile phone use (European Commission, 2014a) and the intensity of its use decrease with age (Kubik, 2009). Nevertheless, it is interesting that such a relationship was not observed in both types of support networks, especially because no significant difference between the two networks was found in terms of the average alter's age. A plausible explanation for such a discrepancy might be provided by the differences in the frequency of mobile communica-tion between the two types of networks. In emotional support networks, people are in more frequent contact on average with their providers of support (Section 4.2). Therefore, especially in cases when old adults have health issues, their support providersindependently of their age-rely on mobile phones to manage daily situations. Such a social obligation is less prominent with larger networks of older adults' socializing companions who would prefer in-person over mobile commu-nication. Consequently, with the described agerelated decline in mobile use (Section 1), the effect of alters' age could be higher for the social companionship network. However, it should be noted that network size does not affect the frequency of mobile communication in emotional or in socializing support networks. Even though social interaction frequencies are likely to be influenced by network size (Van den Berg et al., 2012), the absence of such a relationship in our data could stem from the small (absolute) size of both networks. Likewise, the average tie duration and the proportion of kin in the network did not affect the frequency of mobile communication. On one hand, this might be an unexpected result. Following past research we could have anticipated that kin-based and long-lasting networks would imply closer ties, with which people would maintain more regular contacts using a larger number of communication means (Boase et al., 2006). On the other hand, studies have shown that the tie strength is not always related to familial and/or long-lasting ties (Wellman and Wortley, 1990) and that sometimes people who have known each other for a longer time prefer a landline over a mobile phone (Van den Berg et al., 2012). Therefore, these results war-rant further research in the future that would consider additional characteristic of social support networks, such emotional closeness, geographical proximity, and tie multiplexity. In regards to the RQ2a and RQ2b, we should first mention that for in-person and mobile phone communication, the frequency of contacts was significantly higher in emotional support than in social companionship network, whereas no significant differences were observed for landline phone communication. This finding supports prior research, suggesting that as people age, they remain in more frequent contact with kin (who are predominant providers of emotional support) than with friends (who are predominant in socializing ties) (Shaw et al., 2007;Kalmijn, 2012). In addition, the results seem to be in line with what Quadrello et al. (2005) refer to as the accumulation model of personal communication. Notably, in both types of social support networks, if older adults phone or meet in-person often with their network members, they also stay in touch with them via mobile phones more frequently. Thus, as has been the case with landline phones (Wellman and Wortley, 1990), more intensive mobile communication does not replace in-person contact. Such a conclusion is further confirmed by the results showing that the in-person contact remained the most frequent means of communication, independently from the type of social support, followed by the mobile phone and the landline phone (Section 4.2). Moreover, interesting results were unveiled in this study with reference to the socioeconomic characteristics of older adults (RQ3a and RQ3b). As expected, older adults who have more mobile phone conversations are also more likely to be in more frequent contact with the members of both types of support networks. More intriguing is the finding that older adults are more likely, only in an emotional support network, to have less frequent mobile phone contacts than younger older adults. On the one hand, this neatly corresponds with prior research on mobile communication and social support. While with higher-age older adults tend to use a mobile phone less frequently (Kubik, 2009), aging was also found to negatively affect the frequency of social interactions within networks in general (e.g., Ajrouch et al., 2001;Kalmijn, 2012;Shaw et al., 2007;Van den Berg et al., 2012). Nevertheless, it is interesting that the effect of an ego's age disappears in social companionship, which might be related to the different composition of the two network types. While some scholars suggests no differences regarding age-related decrease of mobile contacts with family and friends among older adults (Kubik, 2009), others argue that aging might lead to less mobile-based familial contact (Ling, 2008). Since the emotional support network of older adults is prevalently kin-based (while social companionship includes a larger proportion of non-kin ties, such as neighbors), the detraction of kin ties induced by aging might result in a larger negative effect of age on the frequency of mobile communication within emotional support networks. In addition, this effect might be amplified by more guarded social attitudes of older adults toward any form of mediated personal communication (Sarrica et al., 2014). Thus, when it comes to discussing important personal matters on the mobile phone, old older adults could feel more restraint in terms of socializing communication. Likewise, SES has also been identified as an important antecedent only for the frequency of mobile communication in emotional support networks. It has been already argued that various aspects of SES importantly characterize the differences between mobile phone haves and have-nots among older adults (Annafari et al., 2014;Hardhill and Olphert, 2012). Our findings indicate that SES also has implications on the usage patterns depending on the type of social support. Considering the factors included in our SES variable, it could be suggested that older adults with lower education and lower household income are less likely to be in contact with their network members via the mobile phone. Following previous findings (e.g., Fernández-Ardèvol, 2014;Fernández-Ardèvol and Arroyo Prieto, 2012;Sawchuk and Crow, 2010), we could suppose that due to a lack of monetary resources among low-income and economically dependent older adults (who identify themselves as members of lower social class), restrictions in their use appear, which are observable in a lower frequency of outgoing calls. On the other hand, lower education attainment might be related to higher skill constraints in the use of mobile phones, which has also been shown to be reflected in less versatile and intensive use of mobile phones (Kurniawan, 2008;Fernández-Ardèvol and Arroyo Prieto, 2012). Both aspects of SES might also be relevant to explain its different roles in the analyzed regression models. Since mobile phone communication with providers of social companionship is less frequent than with providers of emotional support (Section 4.2), the effects of SES factors are plausibly less pronounced in the social companionship network. With respect to living alone, the results indicate significant effects for both types of social support networks. In emotional support and social companionship networks, alters are more frequently contacted via mobile phone by egos who live alone. This is consistent with Kalmijn's (2012) findings indicating that separation and widowhood (i.e., living alone) often increase social contacts and support outside the household. Therefore, our results give indirect evidence of a ''mobilization effect'' in accordance with socioemotional selectivity theory (Carstensen, 1992), demonstrating how older adults actively engage in interactions with network members to address the risk of isolation. In addition, if the size of the regression coefficient in both models is observed (bES = 0.192 vs. bSC = 0.316), it seems that the mobile phone might better support the socializing needs of those old adults who live alone. In our case, the mobile phone appears to be further extending the characteristics of the landline phone as a coordinating tool, assisting in the operation of socializing relations and complementing in-person contacts (Wellman and Wortley, 1990) and social activities (Carrasco and Miller, 2009). Finally, the findings indicate no difference in the frequency of mobile communication with network members according to gender and health impairment. Research on senior mobile phone users often underlines gender-specific design require-ments, attitudes, and usage practices (e.g., Kurniawan, 2008). Nevertheless, according to our study, these might not be related to the frequency of mobile phone communication for social support. A similar observation has been already pointed out by Quadrello et al. (2005): the gender of grandparents is not related to more use of mobile voice calls or to short message use (SMS) texting with their grandchildren. Likewise, being affected by a health problem does not condition the intensity of getting in mobile contact with support network members. Such a finding may come as a surprise, particularly because of the potential benefits of mobile phones to enhance social integration, health, and active aging (cf. Gonzales, 2014;Kurniawan, 2008). However, the complexity and novelty of mobile phones may turn away many older adults (particularly those facing health problems) from mobile phone adoption. In a recent Pew Internet report, Smith (2014) identifies physical conditions and health issues as one of the key barriers of older adults in adopting and using new technologies. Mobile phones can help maintain and improve the quality of life of older adults facing health problemsparticularly by keeping them connected with the members of their social support networksif the phones are based on age-friendly designs that focus on good usability and that take into account the decline of different age-related functions. Moreover, some other important considerations arise from the results above. While mobile phones can be a valuable con-duit of social support to older adults, they usually face many difficulties in learning how to use mobile phones. Smith (2014), for example, reported that only 18% of older adults ''would feel comfortable learning to use a new technology device such as a smartphone or tablet on their own,'' while 77% indicate ''they would need someone to help walk them through the process.'' Older adults rely heavily on social networks and/or social support in order to acquire ICT-related skills, which are also needed for the use of mobile phones (Mori and Harada, 2010). Besides studying the structure of social support networks and household size, it is suggested that the family structure (e.g., three-generation households and husband-and-wife-only households) should also be included in the study of the learning and adopting process of mobile phones among older adults. Of course, older adults living alone would benefit most from the use of mobile phones, as it is also indicated by our study that those living alone are more likely to use mobile phones more frequently to get in touch with their social companionship network members. Hence, attention should also be given to exploring possible ways of stimulating intergenerational solidarity in terms of ICT-based skills transfer from younger generations to older adults. --- Limitations and future research Although this study offered many interesting findings, some limitations should be acknowledged. First, although the sample is representative, it includes only older adults aged 75 or less. Expanding the sample with older adults over 75 years of age could potentially change the results, as prior literature shows that they are not only characterized by lower levels of mobile phone use, but they also have different social representations in terms of personal communication (Sarrica et al., 2014). In addition, the structure of the social support network for the oldest age groups is likely to be different: they have smaller networks, with less proximal ties and a higher proportion of kin (Lang and Carstensen, 1994;Ajrouch et al., 2001). Second, the study data was collected with a telephone (CATI) survey, which means that only older adults with a landline phone in their household were included in the sample frame. As the proportion of mobile-only households is increasing in Europe (Vehovar and Slavec, 2012), our findings could suffer from non-coverage bias, especially considering that mobile phone use is a focal point of this study, and non-coverage of older adults who live in mobile-only households could be a substantial problem in interpreting the results. As this group of people has only access to mobile phones, their patterns of use may not resemble the ones of older adults with access to both landline and mobile phones. Third, the study might be limited by possible lower quality data because of the self-assessment of social support networks and mobile phone use. While the methodology used for measuring the size and composition of social support networks has been demonstrated to provide valid and reliable measurements (Kogov<unk>ek and Ferligoj, 2005), the findings about the reliability of selfreported survey data on the frequency of mobile phone use are less conclusive. Recently, Vanden Abeele et al. ( 2013) indicated that there were discrepancies between self-reported and behavioral mobile phone use in terms of the number and duration of mobile calls, with light users overestimating and heavy users underestimating their mobile phone use. Conversely, Boase and Ling (2013) concluded that respondents in general are more likely to overreport than to underreport mobile voice calling. However, they also showed that a survey measure that requires respondents to report how often they used a mobile phone to call others by selecting from a number of predetermined categories (the approach used in this study) demonstrated higher validity when compared to asking respondents about the number of calls made/received yesterday. Even so, survey data could be supplemented with data from diary studies or log data in a potential replication of our study. Another possibly substantial improvement to the study could have been made by adding more predictors to the regression model. On one hand, the models could be extended with additional network interpreters, such as geographical proximity or perceived relationship closeness with the alter (Section 5.1). On the other hand, the analysis could be upgraded with predictors related to the purpose and/or con-tent of the mobile phone conversation as well as other typical predictors of ICTs use. For example, an empirical investigation of the relationship between mobile lifestyles (Hamka et al., 2014;Karnowski and Jandura, 2014;Sell et al., 2014;Vanden Abeele et al., 2014) and the structure of social support networks of older adults would add important insight into their mobile phones' usage patterns. Lastly, an important constraint of our study is its cultural specialty. While Slovenia is a typical, average-level EU country in terms of the frequency of use of ICTs in the general population, the segment of older adults lags behind most EU countries. In addition, the social support networks of older Slovenians have some particular characteristics (e.g., a higher geographical proximity of alters and a higher proportion of kin relationships), which might reduce the generalizability of our findings. Therefore, it would also be desirable to conduct similar research in other sociocultural contexts. --- Conclusion This study confirms that the mobile phone plays an important role when older adults in Slovenia rely on the members of their personal networks for social support. However, the frequency of mobile communication with the providers of social support is scarcely associated with the network characteristics, but rather with sociodemographic characteristics of older adults, although their effects vary considerably depending on the type of social support. In addition, the results also indicate the complementarity of mobile communication with in-person and landline phone communication, which are both firmly incorporated into the communication of older adults with their providers of social support.

The present post-print may not replicate the final version published in Elsevier journal. It is not the copy of record and may differ from the final version.

Background Unpaid (family) carers play a vital role in supporting the health and wellbeing of individuals with a disability or illness. This care often comes at the expense of unpaid carers' own wellbeing, with negative effects for carers' physical and emotional health, finances, and social activity [1][2][3][4][5]. Carers' wellbeing may also be affected by the organisation and delivery of patient services [6]. This includes social care (i.e. practical support with everyday tasks such as personal care, for people who have extra needs due to illness or disability) as well as health care. Clinical research has provided a wealth of information about the way in which treatment and care affects patient outcomes. However, we know very little about the effect of patient services on unpaid carers' outcomes ('carer effects'). One area of research where carer effects are particularly pertinent is economic evaluation. Economic evaluations provide a means of systematically evaluating the costs and benefits of new services. Carer effects are important to consider in order to measure and value the full health and wellbeing impacts of patient services on society [7]. Failure to consider carer effects means the economic evaluation is incomplete and may provide misleading information on the impact of a service on societal health or wellbeing. The importance of including carer effects is explicitly highlighted in influential methodological guidelines for economic evaluation from, for example, National Institute of Health and Care Excellence in the UK [8], the US panel on cost-effectiveness [9], and Zorginstituut in the Netherlands [10]. To date, carer effects are still rarely considered in economic evaluation. A review of the economic evaluation literature to 2010 only identified 20 economic evaluations of patient interventions that considered informal (unpaid) care [11]. A more recent study showed carer effects were still neglected, even in areas where unpaid carers are intrinsically involved, like Parkinson's Disease or Rheumatoid Arthritis [12]. Evidence does suggest that in some areas, such as dementia [12] or paediatric care [13], unpaid care is regularly considered on the cost side when economic evaluations are conducted from a societal perspective. For example, in a UK study of antidepressants in dementia, Romeo and colleagues [14] assess whether the intervention affects the number of hours of unpaid carer time (alongside health and social care resource use). Carer time value is then valued using the opportunity cost method and considered as part of the overall cost, taking a wider payer and carer perspective. However, it is still the exception rather than the rule to value carer quality of life outcomes within an economic evaluation. Indeed a recent review of NICE appraisals found only 16 of 422 appraisals considered carer outcomes [15]. One highlighted area where more information was needed was "...unpaid/carer health outcomes of [National Health Service] interventions [and across] disease areas...". Economic evaluation is intended to be an aid to healthcare decision-makingultimately informing and guiding the services that are provided in society [16]. There is therefore the additional question of whether carer wellbeing should be routinely considered more broadly in decisions about health and care delivery. This includes decisions about the availability of services at a national level and the provision of services to individual patients. Certain government policies, such as the Triangle of Care in the UK [17], and the implementation of mental health services in Australia [18] clearly highlight an important role for unpaid carers in decisions about care provision. However, the degree to which carer wellbeing ought to be considered routinely, alongside patient wellbeing, is open to debate. The aim of the present study was to identify: (i) how different categories of health and social care were perceived to impact on unpaid carers ('carer effects'); and (ii) whether there was consensus about when such effects should be considered in decision-making contexts. --- Methods A Delphi study [19,20] was used to elicit expert judgements about the scope of carer effects and their inclusion in decision-making. The Delphi method provides a framework for transforming individual opinions into a group consensus [20]. Participants are surveyed remotely, quasi-anonymously, and at multiple time points. Delphi studies have been widely used in healthcare research more generally [21][22][23][24] and specifically in health economics to determine quality checklists [25], core resource use items to measure in economic evaluation [26], and evidence needs for public health decisions [26]. Methodological guidance on the Delphi technique has been developed [20,27] but the approach should be seen as pragmatic and flexible, to meet the needs of the study [28,29]. In our study, a modified Delphi approach was used, whereby we planned for up to three online rounds, but could end it earlier if a high degree of consensus was reached. --- Sampling Three types of experts were identified for the Delphi study: 'unpaid carers', 'care professionals'. Unpaid carers were defined as individuals who provided care or support for a family member, friend, or neighbour, due to their illness, old-age, or disability. Care professionals were employed staff working for health and social care organisations that had some experience of the impact of care and treatment on patient and carers' lives. Research professionals were individuals involved in academic interventional research in one of the three clinical areas. Each group brought different insights on unpaid carer wellbeing and healthcare decision-making. Three clinical areas were considered: dementia, mental health, and stroke. In all areas, unpaid care is important, but carers face different challenges in relation to the illness, service availability, and their caring role [30,31]. Our target was to recruit at least 20 unpaid carers, 20 care professionals and 20 researchers across the three clinical areas. This sample size was consistent with previous Delphi studies [28,32] where the aim is to establish group consensus. Unpaid carers and care professionals were identified initially through local charities, NHS trusts, and service contacts from the lay panel supporting the research programme. Participants had been invited to take part in prior qualitative interviews and focus groups to establish the mechanisms by which health and social care delivery affected carer wellbeing [33]. This sample was supplemented with a small number of carers who had taken part in a nationwide survey on quality of life [34]. A purposeful sampling strategy [35] was used to ensure diversity in terms of caring role (carers) and job role (professionals). Researcher participants were identified with the assistance of the project advisory group. In total, 124 individuals were invited in March 2018 to participate. --- Online survey The first round of the Delphi study consisted of an online survey to identify likely carer effects of health and social care. Four broad categories of health and social care were identified based on prior work on mechanisms behind carer wellbeing [33] and input from the lived experience advisory panel [LEAP] [36]. These categories were: Patient treatments (e.g. medication, psychological support). These could affect carer wellbeing, by improving patient outcomes and therefore indirectly reducing the emotional and physical strain on carers. Services to replace or supplement unpaid care (e.g. formal social care). These could affect carer wellbeing by directly reducing the caring load although their use may be linked with guilt or financial expense. Organisational changes to the timing and/or location of care. These can affect carer wellbeing when services become easier (or more difficult) to physically access and combine with daily life. Organisation changes to staffing. These can affect carer wellbeing by changing how well-informed carers feel and their sense of alienation. The first-round survey consisted of two main sections: (i) Part A, which elicited judgements on the likely impact of service changes on carers, based on participants' own experiences; and (ii) Part B, which elicited judgements on whether these impacts should be explicitly considered in decision-making. The LEAP provided input on the survey length, language and content, as well as the use of rating scales to record participants' responses [35]. For Part A, participants were asked whether each of the four categories of health and social care (treatment, replacement, timing/location, staffing) would, on balance, have 'positive', 'negative', 'positive and negative (mixed)' or 'no' impact on each of five domains of carers' lives. The five domains covered mental and physical health effects and resource consequences (personal finances, paid work, free time) highlighted in the literature on carer impact [37,38] and relevant to economic evaluation [16]. For Part B, participants were asked to what extent they agreed or disagreed (on a six-point scale, strongly disagree to strongly agree) with considering carer effects in each of three decision-making contexts (research, funding, patient care). See Table 1 for a summary. The survey had a total of 16 questions and was d e l i v e r e d u s i n g S m a r t S u r v e y s o f t w a r e [ w w w. smartsurvey.co.uk]. Please see Appendix 1 for full survey. In the first round, potential participants were sent a survey link and an information sheet. They were given two weeks to respond and were reminded that their response was voluntary. Non-responders were reminded a day before and three days after the deadline. Data were summarised by participant role (carers, care professionals, researchers) and used to create individual feedback sheets to be used in the second round (see Appendix 2). The feedback sheets reported the proportion in each participant role (and disease area) that agreed/disagreed that carer effects should be considered in the 12 different decisions. Participants' response to the sets of question below were summarised by role and disease area to generate frequencies relating to: the domains of carers' lives affected by service delivery; whether such effects would be positive, negative, mixed, or absent. Consensus was studied in relation to the strength of agreement that carer effects should be considered for different types of service, decision-making context, and disease. Consensus was defined as at least 70 % agreement [20,29,32] in the top third (i.e. strong/moderate agreement) of the scale. In the second round, all responders were sent a follow-up survey two months later by email. Responders were separated into the three health conditions. This survey contained the 'Part B' questions and a ranking question, requiring respondents to identify the most important decision context and category of health and social care for considering carer impacts. This was introduced to encourage respondents to prioritise contexts for collecting and using data on carer effects. Individual feedback sheets were also provided for participants, with their own responses and the aggregate sample responses to the 'agree-disagree' questions. Participants were asked to consider first round responses in their decision (see Table 2 below) and were assured that they did not have to conform to the group view. Quasianonymity [20] of the participants was maintained throughout, with the participants unaware of each other's identities. A third online round of the Delphi was not conducted as consensus was reached. --- Results Round 1 of the Delphi survey was completed by 65 of the invited individuals (52 %), with 59 of these individuals (91 %) completing round 2 of the survey. The characteristics of the baseline sample are shown in Table 3. --- Perceived impacts of service changes on unpaid carers Figure 1 shows the perceived impact of health and social care on domains of carers' lives. Health and social care 'interventions' (i.e. treatment and replacement care) were most often associated with positive effects on carers (see Fig. 1a andb). This was particularly the case for replacement care (Fig. 1b). Here perceived impacts are very positive across all domains, with the notable exception of finances. Conversely 'organisational changes' (i.e. changes in timing/location of service and changes in staffing) were rarely perceived to have positive effects, with effects tending to be either negative or mixed (see Fig. 1c andd). This was particularly the case for staffing changes, where effects were very rarely perceived to be positive on balance for any domains of carers' lives (Fig. 1d). As Table 4 shows, this pattern of positive impacts for 'interventions' and mixed/negative impacts of 'organisational changes' was repeated across all three conditions. The negative impact of organisational changes was most pronounced in mental health. In stroke, impacts on carers were more often perceived to be mixed compared to dementia or mental health. Pooling responses across all categories of health and social care and conditions (Table 5), shows which domains were most likely overall to be affected by health and social care. Overall, carers' 'emotional health' was perceived most likely to be affected, with 94 % of participants responses indicating either positive, negative or mixed impacts. Conversely, finances were least likely to be perceived to be affected, with 63 % of participant responses indicating positive, negative or mixed impacts. When these results are broken down by participant role it is notable that participants who were unpaid carers perceived negative effects with much greater frequency than participants who were care professionals and researchers. Specifically, carers (as compared to non-carers) perceived negative effects more frequently across all domains of life: emotional health (29 % vs. 11 %), physical health (21 % vs. 10 %), finances (29 % vs. 18 %), paid work (21 % vs. 9 %), and free time (29 % vs. 13 %). --- Consensus on including carer effects in decision-making Overall, when pooled across condition and health and social care category, support for considering carer effects in research decisions was 81 %, support for considering carer effects in funding decisions was 81 %, and support for considering carer effects in patient care was 80 %. Across all four clinical areas, all four service changes, and all three decision contexts (a total of 36 cells), a majority of participants agreed that carer effects should be considered in decision-making (Table 6). Consensus was achieved after round 1 for 34/36 cells (all cells for dementia and mental health and 10 of 12 cells relating to stroke). Consensus was achieved after round 2 in the 2 remaining cells (both relating to'staffing changes' in stroke care) when those'mildly agreeing' were included. Agreement tended to increase (towards greater consensus) between round 1 and 2 for stroke participants and was maintained at a similar (high) level in mental health and dementia. In round 2, when pushed to prioritise which category of health and social care was the highest priority for inclusion of carer effects, 'treatment' decisions received the highest priority and considering carer effects in'staff changes' received the lowest priority. For decision contexts, participants prioritised 'patient care' decisions most highly and'research' decisions least highly. --- Discussion This study elicited views on the impact of health and social care on unpaid carers' lives (carer effects) and the relevance of such effects in different decision-making contexts. Expert participants perceived that healthcare and social care would affect a range of domains of (See figure on previous page.) Fig. 1 Perceived impact on carers of different aspects of health and social care (n = 65). Perceived impact on carers of replacement care. Perceived impact on carers of timing or locational changes. Perceived impact on carers of staffing changes. Note: % indicate proportion of the sample indicating that they thought the impact of intervention on the carer was positive, negative, mixed or not present. So for example 54% of the sample thought patient treatment would have a positive impact on family carer emotional health, 37% a mix of positive and negative impacts, 8% no impact and 2% no impact. Note: cells show the modal effect (positive/ negative/mixed/ none) and % of sample giving the modal answer. So, for example, most respondents (71%) felt dementia treatments for patients would have a positive effect on carers' emotional health. Similarly for changes in timing and location of dementia care, the largest sub-group of respondents (33%) felt that this would have a positive effect on carers' emotional health. carers' lives, most commonly their emotional health. Carer effects were not universally positive, particularly for organisational changes, (changes in the timing, location of staffing of services) which were generally perceived to have negative or mixed impacts on carers' lives. Participants viewed carer effects as important to consider in a range of decision-making contexts, most notably in decisions being made about an individual patient's care. This study suggests that positive effects on the carer's emotional and physical health are likely from interventions in dementia, mental health, and stroke. This means that economic evaluations that neglect carer health may be systematically underestimating the health benefits, and therefore value, of new interventions. Conversely detrimental health effects for carers were often perceived to result from organisational changes in timing, location and staffing of health services. In these cases, economic evaluations that neglect carer health, may overestimate the benefit of the service change in the economic evaluation. Neglecting carer effects may therefore lead to 'investment' in organisational changes that cost-shift and are ultimately harmful to carer (and societal) health. The finding that 'emotional health' is perceived to be the most likely domain to be affected is important because it underscores the need to measure carer quality of life effects in addition to time costs in economic evaluation [39,40]. This may require a change in mindset in economic evaluation where unpaid care is not just seen only as a 'cost'. It also highlights the need to use quality of life measures with carers that are sensitive to emotional health effects. This may point to measures of wellbeing (such as ICECAP-A) that have demonstrated sensitivity to mental ill health [41,42] and carers experiences [34]. In this study, participants perceived carers' finances, employment, and time, also likely to be affected by service changes. This complements recent work on the economic burden of caring [43], additionally demonstrating that service delivery may impact on carers' financial burden. In particular, replacement care interventions were invariably perceived to positively affect free time and employment for carers. These domains should be considered on the cost side in an economic evaluation, albeit typically only when employing a societal perspective. This underscores the importance, where feasible, for a societal perspective to be used [9] to ensure the resource implications to society are properly considered. The findings from the Delphi study highlighted the fact that it could be difficult to predict the direction of carer effects. For example, replacement care may free carers up, having positive impacts on employment or free time. However, if carers bear the financial costs of respite care it may also negatively affect their personal finances. This work adds to a body of literature advocating more consideration be given to carer effects within economic evaluation [4,44,45]. The widespread view, held by participants in this study, that carer effects should be considered in decision-making, is perhaps not surprising, given that many participants may have agreed to participate in this study because of an interest in unpaid carers. In view of this, participants were pushed to prioritise the most important contexts for considering carer wellbeing in. When this happened, participants highlighted decisions about 'treatment' services and decisions relating to 'individual patient care'. On the face of it, this was unexpected, as these decisions might be thought of being more 'patient-orientated'. This may reflect the fact that these are the decisions that participants more immediately relate to, compared with funding or research decisions. However, this finding should not take away from the consensus among participants whereby 81 % also moderately or strongly agreed that carer effects should be considered in research studies and in funding decisions. This study has identified four categories of health and social care where collection of carer data may be warranted as well as perceptions of the likely scope and direction of effects. It is worth highlighting some of the limitations and strengths of this study. As noted earlier, a Delphi panel, by its nature, is self-selecting, so we cannot say that views expressed here are representative of care professionals or the research community more generally. In particular the unpaid carers and care professionals had participated in a previous study. This may have shaped their responses and resulted in a sample that had a particular interest in including unpaid carers in decision-making. However, this approach is likely to have reduced the rate of drop out and resulted in a sample with greater insight into carer effects and how they were likely to occur. A further point to reflect on, is that the study is limited to three conditions. Viewpoints on the relevance of including carer effects in decision-making may therefore differ for other contexts, such as end-of-life care or childhood illness where unpaid care is also likely to be important. Some participants may not have clearly understood the decision-making contexts. We briefly explained the meaning of the decision contexts in the study (e.g. for the research decisions: "Research on these interventions should include finding out how they affect carers' lives"). However, a more in-depth explanation of funding decisions and patient care decisions might have been useful for some participants. Finally, we focused on eliciting views on broad categories of health and social care delivery, rather than specific treatments. This was done to focus on the major ways in which health and social care could impact on carers' lives. However, a consequence of having four general categories is that there may be ambiguities about what participants associated with these categories. A strength of this study is that we sampled participants with experience across three major conditions and in different roles. Across all there was a high degree of consistency in terms of the perceived impacts of services on carers and a high degree of consensus that these impacts should be considered in decision-making. It is still the 'norm' in many countries (including the UK) to expect unpaid carers to carry much of the responsibility for patient care. Nevertheless there is much that could be done to better ensure that negative impacts of policies on carers wellbeing are minimised. Further work on this topic could focus on developing practical approaches to consider carer wellbeing in everyday care decisions. This could complement work to include carers in research studies and economic evaluation. From an economic evaluation perspective, utilising a societal perspective, and including outcome measures that encompass emotional health are important to fully capture carer effects. Other challenges exist in incorporating carer outcomes in economic evaluation, for example to include carer outcomes routinely, where primary data cannot easily be collected or in establishing how best to measuring carer outcomes where family caring networks may be complex, and extend beyond a single carer. In conclusion, this study adds to a body of literature that highlights the importance of carer effects in economic evaluation and more generally in healthcare decision-making. It highlights the particular need to measure and value effects on carers' emotional health and the need to use a societal perspective to avoid 'costshifting' to unpaid carers when introducing interventions and making organisational changes. --- Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. --- Supplementary Information The online version contains supplementary material available at https://doi. org/10.1186/s12913-021-06742-4. --- Additional file 1: Additional file 2: Authors' contributions HA conceptualised the study, administered the project, undertook the analysis, and wrote the main manuscript text; HA, NE, CM, MC, JO: developed the methodology and reviewed and edited the manuscript. All authors read and approved the final manuscript. --- Declarations --- Ethics approval and consent to participate The study protocol was reviewed and approved by the University of Birmingham's ethical review committee (14-1444) and the NHS Health Research Authority (IRAS 206161). Informed consent was obtained from all participants. All methods were performed in accordance with the relevant guidelines and regulations. --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background and objective: Health and social care may affect unpaid (family) carers' health and wellbeing in addition to patients' lives. It is recommended that such impacts (carer effects) are considered in decision-making. However, the scope of carer effects and range of decisions where carer effects should be considered is uncertain. This study aimed to identify: (i) how different categories of healthcare and social care were perceived to impact on unpaid carers; and (ii) whether there was consensus about when carer effects should be formally considered in decision-making contexts. Methods: A two round, online Delphi study was conducted with 65 UK-based participants (unpaid carers, care professionals, and researchers) with expertise in dementia, mental health, and stroke. Participants considered two broad forms of 'interventions' (patient treatment and replacement care) and two broad forms of 'organisational change' (staffing and changes in timing/location of care). Participants assessed the likely impacts of these on unpaid carers and whether impacts should be considered in decision-making. Results: Participants predicted interventions and organisational changes would impact on multiple domains of unpaid carers' lives, with 'emotional health' the most likely outcome to be affected. Patient treatment and replacement care services ('interventions') were associated with positive impacts across all domains. Conversely, timing/location changes and staffing changes ('organisational changes') were perceived to have mixed and negative impacts. There was widespread support (80-81 %) for considering carer effects in research studies, funding decisions, and patient decision-making. Conclusions: This study highlights a perception that carer effects are widespread and important to consider in economic evaluation and decision-making. It highlights the particular need to measure and value effects on carers' emotional health and the need to use a societal perspective to avoid cost shifting to unpaid carers when introducing interventions and making organisational changes.

Plain English summary Self-administration of contraception may offer greater privacy and increase access to injectable contraception, particularly for those who face challenges accessing clinic services. In particular, unmarried adolescents often encounter negative stigma when seeking services, and may also lack the financial means to travel to clinics for regular reinjection. Whether self-injection is offered to adolescent women will depend on the willingness of providers to train clients of diverse ages and backgrounds. This study explores the views of providers with regard to self-injection as an option for women and adolescents in Uganda. In-depth qualitative interviews were conducted with family planning providers in Gulu district, to understand their views on injectable self-injection for women, with a specific focus on unmarried adolescents. A total of 40 health care providers were interviewed from public, NGO, and private clinics, pharmacies, and community-based settings. While most providers were receptive to self-injection for adult women, fewer than half were supportive of adolescent self-injection. Their reservations focused on whether young women, women who are unmarried, and women without children should use the injectable, in addition to concerns about the ability of adolescents to self-inject safely. Self-injection presents an opportunity to reduce the enormous burden on the public health system imposed by quarterly reinjection clinic visits. These results reveal cautious support for self-injection among providers when it comes to self-injection by adult women. With respect to adolescent clients, extra attention will likely be needed to reduce provider-imposed restrictions on adolescent access to this mode of delivery for injectable contraception. --- Background The introduction of subcutaneous depot medroxyprogesterone acetate (DMPA-SC) in a number of African countries in 2014 has opened the door to the possibility of self-injection of injectable contraception [1]. Self-injection may offer greater discretion and increase access to the injectable, particularly for those who face challenges accessing clinic services. Many women could benefit in terms of reduced travel and opportunity costs if they could manage their injectable use independently of the health system. In particular, unmarried adolescents often encounter stigma when seeking contraceptives at public sector clinics, and may lack the financial and physical independence to travel to clinics on the periodic basis that injectable contraception requires. Whether self-injection is offered to women on a wide scale basis, and to adolescents specifically, will depend on the willingness of providers as gatekeepers to DMPA-SC. This qualitative study explores the views of providerspublic, NGO, private and community-basedwith regard to self-injection as an option for women and adolescents in Uganda. Like the intramuscular version, DMPA-SC is a three -month, progestin-only product that is stable at room temperature. The subcutaneous version now available in a number of African countries has a lower dose of DMPA (104 mg. vs. 150 mg.), but with comparable safety and efficacy [2]. DMPA-SC is packaged in the UnijectTM injection systema small, prefilled, auto-disable device designed for easy administration after minimal training. The easy-to-use injection system provides opportunities for women to administer themselves through self-injection. DMPA-SC (brand name Sayana Press) was registered for administration by providers by the Ugandan National Drug Authority (NDA) in 2014. While the total fertility rate in Uganda is decliningfrom 6.2 children per woman in 2011 to 5.4 in 2016it remains among the highest in the world [3,4]. Unmet need for contraception is also high, at 28% among currently married women and 32% among sexually active unmarried women, suggesting that a substantial share of women are not using a method though they would like to wait an additional 2 years before their next birth or to limit childbearing altogether [5]. Adolescents aged 15-19 years represent about 11% of the population in Uganda, and one quarter of them have already begun childbearing [5]. The modern contraceptive prevalence rate is 36% among married women ages 15 to 49 years, and the injectable is the most popular method (for married and unmarried women alike), representing nearly half (46%) of the modern method mix [6].Studies of the appeal of DMPA-SC in Uganda found that more than four out of five women preferred it to the intramuscular version (DMPA-IM) [7]. The the country is now undergoing rapid scale up to offer DMPA-SC nationwide, and as of 2018, DMPA-SC represents 11% of the contraceptive method mix among married women [6]. The popularity of injectable contraception, and the growing appeal of DMPA-SC in Uganda creates potentially fertile ground for self-administration. Self-injection of DMPA-SC could overcome barriers for many women, and particularly for adolescents. While women may initially need training from a clinic-or community-based health worker, subsequent self-injection at home will eliminate the need to see a provider every 3 months. Self-injection reduces the financial burden and opportunity costs associated with travelling to the clinic and waiting for services. The ability to store the product at home will make women less vulnerable to stock outs. For women wishing to use the injectable discreetly, self-injection may enhance confidentiality. In short, self-injection may improve injectable continuation by reducing clinic access challenges while simultaneously enhancing women's autonomy and control over contraceptive use [8][9][10][11]. Self-injection may offer solutions to the particular access challenges often faced by unmarried adolescents, including concerns about privacy and confidentiality stemming from the stigma of contraceptive use and premarital sex; the cost of travel and distance to health facilities for adolescents who lack financial autonomy and face challenges travelling independently; and inconvenient clinic hours, which can be particularly difficult for school-attending adolescents [12,13]. Despite efforts to improve services for adolescents, there continues to be a pronounced lack of youth-friendly services, of which non-judgmental, supportive front line health care providers are the critical component [14,15]. Numerous studies have found that health care workers sometimes refuse to provide contraception to unmarried adolescents because of deeply held negative opinions about premarital sex, or limit their contraceptive options due to misperceptions about who can use certain methods safely and effectively [13]. Misplaced fears that some forms of contraception impair fertility lead to parity and/or age restrictions [16,17]. Recent research from Nigeria found that the most common restriction imposed on contraceptive provision was an age requirement [18]. In Tanzania, more than one in three providers impose age restrictions on injectable use, more than one in four impose parity restrictions and one in five impose marriage restrictions [19]. More specific to Uganda, a simulated client survey found that two thirds of providers chose a method on behalf of their clients. The authors observed that 'younger clients seemed to be treated differently than older clients', with more discussion focused on method side effects when the simulated client was younger and unmarried [20]. Another study from Uganda found that one in five providers indicated they would not offer injectable contraception to adolescents, with the authors concluding that most providers had misunderstandings about contraceptives, negative attitudes toward providing contraceptives to young women, and imposed age restrictions and consent requirements on adolescents [21]. While self-injection does not eliminate the challenge of provider bias, it improves the autonomy of clients and may enable unmarried adolescents to minimize the stigma they experience when seeking contraceptives in clinic settings. Research suggests that self-injection is a feasible mode of administration, including for women in low resource settings as well as for adolescent women. Among adult research participants in Uganda, 88% self-injected proficiently three-months after being trained by a nurse [22]. Studies from high resource settings have similarly shown that self-injection of DMPA-SC is feasible [23][24][25]. One study in the United States specifically assessed adolescent competency and acceptability in self-administering DMPA-SC, finding moderate proficiency (63%) after a single training session. Though the study was small, the authors concluded that self-injection was feasible for adolescents with training and support [26]. A qualitative study of Ugandan adolescent interest in self-injection found that approximately half were personally interested in taking up self-injection if that option were available to them [27]. With respect to self-injection in Uganda, the NDA granted approval for self-injection in 2017, and subsequently that year, a pilot self-injection program was launched in four districtsthe first offer of self-injection in sub-Saharan Africa outside of a research setting. However, successful implementation and scale up of self-injection as a delivery strategy in Uganda will require buy-in from family planning providers who are the gatekeepers to self-injection training. To the extent that providers doubt women's ability to self-inject safely and effectively, or feel that self-injection will encourage promiscuity among unmarried adolescents, they will limit the availability of training, or provide training only to particular types of women, such as more educated or married adult women. This qualitative study is designed to better understand Ugandan providers' willingness to endorse and train women, including adolescents, for self-injection. To that end, we first explore providers' views of appropriate contraceptive methods for adolescents, including injectable contraceptive use, as a necessary precondition to self-injection. The study then examines their views regarding self-injection for all women of reproductive age. Lastly it solicits their opinions regarding key program characteristics-such as the training approach and appropriate follow-up-that might facilitate self-injection, and how the program design requirements may be varied to better serve adolescent clients. --- Methods --- Study sites and participants The study was conducted between October -December 2015 in the district of Gulu in Northern Uganda, in collaboration with the Ugandan Ministry of Health. Facilities and establishments from which participants were drawn were identified from a full list of potential sites in the district, and those selected were based on the willingness of facility administrators to have their staff participate, and with consideration for the accessibility of the site. Health care personnel were recruited purposively from private clinics, NGO clinics, public sector clinics, public sector village health teams (VHTs), and pharmacies. Clinic-based family planning providers and community health workers were eligible if they were providing family planning counseling and services as part of their main responsibilities, while pharmacy staff were eligible if their pharmacy was selling injectable contraceptives. Participants were recruited through face to face interaction at facilities and interviews conducted either at the facility or at a location convenient to the provider. Participants were required to speak English or Acholi-the major local language, to provide voluntary informed consent and to agree to being audio-recorded. --- Study design and procedures We conducted qualitative in-depth interviews with family planning providers to understand their views on injectable self-injection for women, with a specific focus on unmarried adolescents ages 15 to 19 years. Participants included a mix of individuals who had previously been trained to administer DMPA-SC, and others who were unfamiliar with DMPA-SC. Health workers who had not been trained to administer DMPA-SC were given basic information, received a demonstration during the interview, and had an opportunity to administer DMPA-SC on a prosthetic. No injections or self-injections were performed during this study. Semi-structured interview guides were developed to facilitate the interviews. Different interview guides were prepared for family planning providers who had previously been trained to administer DMPA-SC, family planning providers not familiar with DMPA-SC, and with pharmacy staff. The data collection team was comprised of both male (2) and female (2) research assistants from Gulu district. They participated in a 5-day training covering recruiting and screening procedures, research ethics and administering informed consent, qualitative interviewing techniques and conducting in-depth interviews, translating and transcribing audio recordings, and data quality assurance. The in-depth interviews lasted up to 60 min and were conducted in English or Acholi, as preferred by the participant. Interviews were audio-recorded, and translated and transcribed simultaneously. Interviews were designed to move from the general to specific, beginning with views on contraceptive use and recommended methods for women in general, and for adolescents specifically, and progressing to views about injectable contraception and self-injection as a delivery modality for adult as well as adolescent women. Interviewers also solicited ideas for program design to facilitate self-injection-such as training, supervision, and reminderswith an eye toward what additional support, if any, might be necessary for adolescent populations. --- Data analysis Data analysis was conducted using the qualitative software program Atlas.ti. The coding scheme was developed and transcripts were coded through an iterative process by two coders, with any discrepancies reviewed and resolved. Coded text was reviewed for each main code and, subsequently, memos were developed to summarize key patterns and themes. Where appropriate, findings were compared by the type of provider (private clinic/public clinic/commercial provider/community based). --- Ethical conduct of the study All research study team members involved in data collection, management, or analysis were trained on research ethics, including confidentiality. This study was approved by the Mulago Hospital institutional review board, the Uganda National Council for Science and Technology, and the PATH Research Ethics Committee. --- Results --- Participant characteristics To capture the views from various types of providers, we interviewed 40 providers in total, 29 of whom were women. The median number of years worked as a health care provider was 6, with a range of one to 30 years. The median age was 32, with a range of 24 to 75 years. To maximize the diversity of experience, providers were drawn equally from public sector health centers, NGO facilities, for profit clinics, pharmacies and the public sector community-based distribution program. Participant characteristics are shown in Table 1. Below we present summaries of themes that emerged from interviews, along with illustrative comments from these 40 health care providers. --- Informed choice for adult women, less so for adolescents When asked what family planning methods they recommend to a client, about three fourths of providers (n = 28) stated that they do not recommend specific methods, but rather, counsel the woman about available family planning methods and allow her to choose. "I do not decide for the mothers any methods. But when they come, I counsel them and then they choose for themselves." -VHT provider The remaining providers who stated that they recommend particular methods were evenly split between those who recommend injectables and those who promote long acting reversible methods and sterilization (intrauterine devices or IUDs, implants and tubal ligation). No one offered that they recommend condoms or oral contraceptives. With regard to unmarried adolescents under age 20 however, far fewer providers (n = 14) expressed that all (non-permanent) methods are potentially appropriate for adolescent women. The most common method actively promoted to adolescents was the condom (n = 18). Two thirds of providers recommending condoms (n = 12) were concerned about exposure to sexually transmitted infections (STIs), and the remaining one third (n = 6) expressed reservations about premarital adolescent sexual activity, recommending the'ABCs' -Abstain, Be faithful but if not, use Condoms. "The method that I feel is appropriate for an adolescent according to me is only condom. Because like for the female adolescent there are side effects of these contraceptives and it can become problematic especially to those who are school going. So condom is okay since it doesn't only prevent them from pregnancy but also from other diseases like the STDs (sexually transmitted diseases), for instance HIV/AIDs (Human immunodeficiency Virus/Acquired Immune Deficiency Syndrome)." -Pharmacy provider "But you know when we talking about adolescents, we are not only looking at preventing pregnancy in them but also other STDs and STIs. Like adolescents who have not yet given birth in their life, as long as they know they are free from getting pregnant, they are very vulnerable and they would forget about HIVs. That would also be my fear that much as we are trying to prevent pregnancy in especially adolescents, we should always not forget to encourage them to use condom on top of preventing pregnancy. Remind them that HIV is still there, so they should protect themselves." -NGO provider "Being an adolescent is really not something easy because we always advocate for abstinence, you see that. And whoever cannot really do that should use condom." (Public provider) Consistent with this tendency to prescribe specific methods to adolescent clients, a substantial share of providers (n = 15) volunteered specific methods they would advise adolescents NOT to use. In particular, they singled out the IUD (n = 4), hormonal methods generally (n = 2), injectable contraception (n = 3), cycle or moon beads (n = 2), long term methods of any type (n = 1), oral contraceptives (n = 1), implants (n = 1) and condoms (n = 1). Their rationales for restricting adolescent choice included concerns about low efficacy (condoms and cycle or moon beads), concerns about possible infertility or cervical cancer due to STI exposure during IUD use, and more commonly, concerns about the impact of hormonal contraception on fertility. --- Providers divided on injectable use by adolescents When asked specifically about injectable contraception, just over half of providers (n = 22) expressed support for adolescent use of this method, with many citing the importance of injectable contraception in preventing unwanted pregnancy and reducing the incidence of school drop due to pregnancy. Other stated benefits of the injectable included the short duration, manageable side effects, and accessibility of the method. --- "I would [offer injectable contraceptive to adolescents] because first of all I know it has no other future dangerous effect to the youth. Secondly it's going to protect them from having unwanted pregnancy. And they will also continue with their studies at school. That is the reason I recommend the injectables to the youth. It's very safe and it has no future effect."-Public provider "My own opinion is generally that it's actually good for adolescents to use injectable contraception because the injectable is a short term method, and the side effects are easily manageable and it can easily be accessed and administered."-NGO provider The remaining providers (n = 18) expressed reservations about adolescent use of the injectable. Their reservations were sometimes tied to marital status, rooted More common however, were reservations related to parity, rooted in the misconception that the injectable causes infertility, which has origins in the delayed return to fertility common among DMPA users. "The injection Depo is not recommended for the adolescent because it may interfere with their fertility since they have not produced in their life. So when it comes time for you to conceive, there may come a problem because the Depo takes a time long to leave the body system." -Private provider "The young ones, we always advise them to use oral contraceptives because inject-a-plan [brand name of the socially marketed injectable] is meant for once you have at least three or four children. It can stop you from getting your normal menstruation period." -Private provider "When they are married and the person has not yet delivered, it (the injectable) is not very advisable." -Pharmacy provider Consistent with their reservations, more than one third of clinic-based providers felt that parental permission should be required (or is advisable) before offering family planning services to adolescents. "Adolescents who are not married, I think parental permission for adolescents to use family planning is appropriate. The parent should be aware because as those people who use family planning methods say, methods have side effects. So if it starts before the parents, when she has no knowledge, the parents could have fears about what is happening." -VHT provider --- Subcutaneous DMPA popular among providers With respect to DMPA-SC specifically, more than three quarters of participants (n = 33) found the device easy to use, with the same number indicating they preferred it to intramuscular DMPA (DMPA-IM). Interviewees noted that the pre-packaged, all-in-one presentation offers advantages to providers, such as avoiding stock-outs of syringes, requiring less skill to administer, saving health workers' time, and minimizing the risk of a needle stick. "It has its needle already attached. I don't have to withdraw the Depo or the medicine. So everything about it is easier or much better or more convenient."-NGO provider "As I said, this issue of we don't have the syringe is not there because everything is connected and you can administer it by yourself. It does not really require much skill in injecting. Actually there is nothing I don't like about that."-NGO provider Providers also identified benefits of DMPA-SC for women, including that the presentation readily lends itself to self-administration (n = 21), injections may be less painful due to the smaller needle (n = 6) and the perception that the lower dose DMPA-SC has fewer side effects (n = 4). "I saw the needle was really very tiny, you don't experience much pain compared to the other injectable, and the (DMPA-IM) needle is a bit big." -NGO provider "I think the product is very good because it is selfadministered by the clients and side effects are not really serious like for other methods." -Public provider "For me what I have liked about this [DMPA-SC], the first thing, it is private. You can use it privately. Secondly, the needle is very small for people who fear injections. Then you can do it even self-injection. Actually there is nothing, for me I feel there is nothing wrong with this."-Public provider --- Most providers receptive to self-injection, perceive advantages for women To gauge receptivity to self-injection, providers were asked whether they consider self-injection to be a good option for Ugandan women. By a margin of just under two to one, providers viewed self-injection favorably (n = 25), citing the benefits of greater convenience and time savings (n = 11), reduction in transport costs (n = 9), and better adherence to the reinjection schedule (n = 5). Two providers noted that offering self-injection may reduce provider workload. Even providers less receptive to self-injection readily identified advantages for women. "Mostly I think the major part of the population in Uganda they are below poverty line so if someone can reduce the cost of transport, and get the injectable contraceptive which can be administered by themselves, it can really give them some [savings]." -Pharmacy provider "In my location, I think it's going to be a good idea to help out women because most of them, during the season for digging (farming) they go really far away; we've had incidences that they miss out on their shots just because they were not close to the clinic. So if they have this, if they have Sayana Press with them, I think they can be able to carry it with them and inject themselves."-NGO provider "Sometimes in the health centers you can go and you find that the health worker is not yet there, but if you are self-injecting yourself and you have got it enough with you at home, you will be using it without again need to look for health worker to help you inject you, all that."-Pharmacy provider A number of providers (n = 11) spontaneously offered that self-injection is likely to enhance discretion for women who are hiding contraception use. When queried specifically about whether self-injection would be more discreet, nearly all the remaining providers (n = 27) thought self-injection would improve privacy. "I think all the women should use [self-injection] because there are some men who don't advise their women to use family planning or to use any method. So Sayana Press can help them. You can even go to your private room and just inject yourself."-Private provider "Somebody who is concerned, maybe a neighbor, would say 'I saw your wife at the facility, is she sick?' You see? So it can bring problems."-Private provider --- Even receptive providers raise concerns For a minority of providers (n = 12) however, their initial reaction to the concept of self-injection was more skeptical. These providers, as well as a number of providers generally receptive to self-injection, cited a number of concerns, including appropriate hygiene (n = 13), storage (n = 11), and disposal (n = 6) practices. "Normally those things have sterile procedures. If maybe the clients are not very clean the site may be infected."-NGO provider "Storage at home. I could be having the kitchen where it is hot throughout the day. I think we have a temperature where we should keep this Sayana Press. When I have only one house, it's my bedroom, it's my storage, it's my kitchen. So it's like we are exposing this Sayana Press at very high temperature which can damage the product."-NGO provider "What I see about this Sayana Press is convenient except the ways of disposing the waste. If it is not properly disposed, it can cause injuries at homes."-Public provider In particular, some providers expressed a lack of confidence in women's ability to administer the injection properly. This included opinions that women would forget the injection steps or not do them correctly (n = 11), forget their reinjection date (n = 4), or choose an inappropriate injection site (n = 5). "[Self-injection] is not so good because women are... they are not capable, intelligent enough to follow all the steps...It is better they go to the clinic or to the health workers to administer, not from home." -Private provider "These women if you put your mind on learning the steps in self-injecting and if the health provider did not write for you your next date of injection, you may forget since you may be too busy doing house work... But if you are to remember it alone, you will forget due to so much house work that women usually do at their homes." -VHT provider "And sometimes they may confuse the injection site. You see the drug has to be injected on the fatty tissue of the body, on the thigh and lower abdomen. So they can forget and end up injecting the wrong site and this may not be good for them."-VHT provider Public sector clinic providers were disproportionately represented among the 12 individuals who were more skeptical about self-injection, with five of the eight public sector clinic providers expressing reservations. Conversely, public sector VHTs were disproportionally represented among those viewing self-injection favorably, with only one of eight VHTs expressing skepticism. --- Mixed feelings about adolescent self-injection Regarding self-injection for adolescents, a slight majority of interviewees (n = 23) opined that self-injection was a good or acceptable option for adolescents-although 7 of these had previously stated that they did not approve of giving the injectable to adolescents. In particular, providers cited the specific benefit of enhanced privacy and ability to conceal contraceptive use (n = 21), which some felt, was particularly important for adolescents who are attending boarding school. "Because I think being in the health facility and also going to a family planning clinic, first of all [adolescents] fear being seen as I told you by the relatives, or the neighbors. But if they are now doing self-injection, they are more confident that nobody is seeing me."-Public provider "I think Sayana Press would be very good for the adolescents since they can even go with it to school and administer without anybody noticing. Because adolescents go to the boarding school far away from home and the drug will cover them even if they come back home for holidays." -Public provider Sixteen interviewees felt that adolescents should not be permitted to do self-injection. Beyond concerns about injectable use by adolescents, the most common reason for opposition was a belief that adolescents do not have the maturity to do the injection on their own. Others worried that self-injection, like use of contraceptives more generally, would lead to sexual promiscuity, and a few were concerned that self-injection would open the door to illegal injectable drug use. "And for adolescents, they have to go to the health facility in order to get the injection because adolescents like taking things for granted. They may forget the procedure to follow when administering..."-Private provider "So someone like that (an adolescent), it is not appropriate for her because she might be given a method and she changes her mind to use another method. So the other method which she already had could expire unused. That is why I think that it (selfinjection) is not very appropriate for adolescents."-VHT provider "The adolescents will misuse the drug. Misusing, I mean like I had live sex (without a condom) and maybe I can get pregnant, so I will also give the injection again."-Pharmacy provider "For the adolescents it would be beneficial but at the same time it would also cause a fear that once they get used to self-injection they may be introduced to some other drug like the drug use injection."-Private provider Despite their general receptivity to self-injection (noted above), more than half of the VHTs (5 of 8) expressed reservations or concerns about self-injection by adolescents. With regard to whether adolescents need parental permission to self-inject, providers were evenly divided with half indicating parental permission is needed and half indicating it is not. The rationales offered were similar to those for family planning use in general by adolescents. "I would think the adolescent should get permission from her parents so that in case of any risks in future the parents would be aware. Because if she starts doing it on her own, if she starts injection on her own, without informing her parents, if the side effects like bleeding starts, who will help her?" -Public provider "Some of the parents they are very tough. Yes, because if you tell your parents I'm injecting myself and obviously the parents will know that you have somebody somewhere. And some of them may even stop paying your school fees when, if you are still at school." -Pharmacy provider --- Clinic providers best for training women The vast majority of providers, regardless of the type of provider, identified clinic-based health workers as the most appropriate personnel to train women for self-injection, due to their medical expertise and contraceptive knowledge (n = 33). "Health workers are the most appropriate to train because there is a bit of medical knowledge involved in the injection and involved in knowing about the drug. Medical personnel may advise in an event that there's anything like swelling or maybe infection at the site which has not been administered well. A medical personnel can also give and can handle that." -Private provider "Midwives who handle matters of family planning in the health center and in the area are appropriate and medical personnel with the knowledge on family planning."-VHT provider That said, quite a few respondents recognized VHTs as capable of training women (n = 15), including nearly one third of those who first proposed clinic-based health workers (n = 10). VHTs are valued as trainers because of their close relationships with the community. "Like you heard me say, first, those people (VHTs) are prior trained and have knowledge on how to administer Sayana Press. Secondly, the VHT is a person closer and known to the community as a person from whom we get these specific things, from whom we go obtain this particular assistance." -VHT provider While all eight pharmacists identified health workers as the most appropriate trainers, five indicated, when queried, that pharmacy staff could also train women for self-injection. One disagreed however, noting that pharmacists don't have the time required to train women individually. "The health workers in the pharmacy should also be trained so that any clients going there, they should also give that knowledge to them... As you are selling the products to the customers, the health workers must know more about the product because you cannot give a product to a client and the client will ask you very many questions and you have nothing to say. The client will not take that unless you have knowledge on that, then you can give and defend with the answers."-Pharmacy provider "And the best people to train the women I would prefer those who work at the family planning section, maternity sites and antenatal care department because they are the ones who always interface with these women more often and they always carryout health education. Because like here at the pharmacy, no one can train these women since we provide a wide range of services and we do not have the time to talk to the clients for long." -Pharmacy provider A few providers (n = 3) suggested that peer-to-peer training would be beneficial as it would provide role models to women. "I would recommend that there should be regular outreaches on Sayana Press and demonstration. And maybe select one participant to come and demonstrate by herself to see how they administer Sayana Press so that women get more knowledge and get acquainted with the method."-NGO provider By about two to one, providers felt that adolescents might prefer or benefit from separate training-away from adults-for reasons such as discomfort and shyness. They opined that, if trained with older women they might not ask questions or receive adequate training. A few providers noted that HIV and STI education should also be a focus of training for adolescents (but did not mention these topics as critical for adult women). "The difference in the training needs maybe is there because there may be adolescents who are still very young like around 15 -16 years, and if you mix them with the older women, they may not ask any questions even if they have them, because of fear and not feeling free or comfortable."-VHT provider "[The training of adolescents should be] a little bit special because these people should be training them not only on family planning methods but also train adolescents on preventing themselves from being exposed to HIV/AIDs and other STIs. So this means your knowledge again will have to go further onto this adolescent. Those one who will be providing the services to the adolescent, they should equip them with more knowledge than these one who train mothers."-Public provider --- Post-training support and follow up important About three quarters of providers (n = 28) mentioned that some form of proactive follow-up would be helpful to make sure clients remember the injection procedures and schedule. The most common form of follow up, identified by 19 interviewees, was home visits or community outreach to villages where women are self-injecting. Some specified that this form of follow-up would be most appropriate and feasible for the VHTs (n = 11). Others advocated that self-injecting clients be asked to return to the clinic periodically to demonstrate their mastery (n = 6). Some felt that phone calls to self-injecting clients would be appropriate support (n = 6) "I think it would be easy if we follow them through the VHTs that are nearer to them because if the VHT near them, this VHT can go to a mother just like he is visiting the home, and will follow up these mothers from home that is one thing. The VHT can also ask them how they are doing it and still remind them."-Public provider "Women, especially those ones in the villages, when you inject them or you tell them to always come back to you after a certain period, they do come. So according to me, I recommend that all providers should tell whoever gets Sayana Press from them to always remember to come back to him/ her after a certain period --maybe after 6 months --so that you can still assess the person to see whether she is doing it rightfully. In that way, you would also be doing refresher training." -Pharmacy provider The remaining participants (n = 10) were confident that, once trained

Background: Self-injection of subcutaneous depot medroxyprogesterone acetate may offer greater discretion and increase access to injectable contraception, particularly for those who face challenges accessing clinic services. In particular, unmarried adolescents often encounter stigma when seeking services, and may also lack the financial means to travel to clinics on the quarterly basis that injectable contraception requires. Whether self-injection is offered to women on a wide scale basis, and to adolescents specifically, will depend in part upon the willingness of providers to train clients of diverse ages and educational backgrounds. This study explores the views of providers with regard to self-injection as an option for women and adolescents in Uganda. Methods: In-depth qualitative interviews were conducted with family planning providers in Gulu district, to understand their views on injectable self-injection for women, with a specific focus on unmarried adolescents ages 15 to 19 years. The in-depth interviews, which lasted up to 60 min were audio-recorded, translated and transcribed simultaneously, and analyzed using Atlas.ti software to identify key themes and common perspectives. Results: A total of 40 health care providers were interviewed with equal numbers of each type (public, NGO, and private clinics, pharmacies, and community-based health workers). While most providers were receptive to selfinjection for adult women, fewer than half were supportive of adolescent self-injection. Their reservations focused on age, marital status and parity concerns around adolescent use of the injectable more broadly, and concerns about the ability of adolescents to self-inject safely. Conclusions: Self-injection presents an opportunity to reduce the enormous burden on the public sector health system in Uganda, which is particularly compounded by the heavy reliance on injectable contraception requiring quarterly clinic visits. The results of this study reveal a level of cautious support for self-injection among providers when it comes to self-injection by adult women. With respect to adolescent clients, family planning policymakers and program implementers should design, implement, and evaluate self-injection interventions with the needs of adolescent clients uppermost in mind, recognizing that extra attention will likely be needed to reduce providerimposed restrictions on adolescent access to this injectable delivery modality.

-injecting clients be asked to return to the clinic periodically to demonstrate their mastery (n = 6). Some felt that phone calls to self-injecting clients would be appropriate support (n = 6) "I think it would be easy if we follow them through the VHTs that are nearer to them because if the VHT near them, this VHT can go to a mother just like he is visiting the home, and will follow up these mothers from home that is one thing. The VHT can also ask them how they are doing it and still remind them."-Public provider "Women, especially those ones in the villages, when you inject them or you tell them to always come back to you after a certain period, they do come. So according to me, I recommend that all providers should tell whoever gets Sayana Press from them to always remember to come back to him/ her after a certain period --maybe after 6 months --so that you can still assess the person to see whether she is doing it rightfully. In that way, you would also be doing refresher training." -Pharmacy provider The remaining participants (n = 10) were confident that, once trained, women would not need additional supervision or follow-up to do self-injection. They advised that follow up should be client-initiated, such that women return to the facility or pharmacy in case of problems. "According to me, I feel if the women are cleared to start injecting from their homes, then they should tell them to always remember to come back to the health center in case they are forgetting the steps to follow in self-administering Sayana Press."-VHT provider A number of providers recommended support in the form of a client instruction job aid and/or calendar be provided to remind women of the procedure and their next injection date (n = 8). "Yeah that one [client support] will be a little bit hard but I think they need to be encouraged to at least have a place where they can record the next date for injection. So they should be provided with either a calendar or a book where they can write their next injection date so that they cannot forget."-Private provider Nine providers were of the opinion that follow-up was particularly fundamental for adolescents because they need more encouragement and supervision to continue with the method. "You know adolescent youth are not like adults. So for them they can change their mind anyhow so they need to be followed and talked with well. --- Advised."-Private provider --- Discussion This study suggests that informed choice is an ideal not yet realized when it comes to contraceptive services for adolescent clients in Uganda. Our findings regarding what methods providers consider appropriate or inappropriate for adolescents are largely consistent with what Chandra-Mouli and colleagues refer to as a 'condoms-only mindset' [28]. Be they motivated by concerns about STIs, promiscuity, or lingering suspicions about the safety of hormonal methods, the attitudes of our participants suggest that providers may impose restrictions on method choice for young, unmarried and/or nulliparous adolescent clients. While providers may feel they have the adolescent's best interests at heart in promoting condoms over other methods, adolescents may have older partners or engage in transactional sex, and may not be able to negotiate condom use in settings where gender norms promote inequality. Suspicion about the safety or appropriateness of high efficacy hormonal methods for adolescent women leaves them vulnerable to higher rates of contraceptive failure and method discontinuation associated with condoms and other short term methods [23,29]. More specific to the main focus of this study, the parity and/or age restrictions on injectable use, which have no medical basis, present an obvious barrier to the offer of self-injection to this demographically-important population. Since fewer than half of providers see self-injection as a good option for adolescents, making self-injection available to adolescent clients will necessitate renewed efforts to overcome provider-imposed method restrictions and ease concerns about the ability of adolescents to self-inject safely. A number of promising interventions to reduce provider bias against adolescents have been proposed by the Beyond Bias Consortium, including: 1) clear directives and clarity from leadership regarding the importance of reproductive health services for adolescents; 2) provider education that involves communications training, values clarification, attitudinal conditioning or the pairing'reluctant' providers with 'champions' in the provision of youth services, and using personalized anecdotes and exercises that encourage providers to see the world through an adolescent lens; and 3) a systems approach rather than one-off provider trainings to address the myriad conditions necessary for Youth-Friendly Services, such as expanded clinic hours, outreach to schools and communities, enhanced privacy, and reduced fees for adolescent clients [30]. With regard to self-injection for adult women, this study suggests that most providers view self-injection favorably, but with some reservations that will need to be addressed if the practice is to become widespread. In particular, provider concerns that women may be unable to self-inject competently should be addressed, with reference to the growing number of studies demonstrating its feasibility and acceptability, including for women in low resource settings and women with limited education [22,31]. Identifying and featuring'self-injection champions' is another strategy that may sway skeptical providers. Offering a client instruction guide and reinjection calendar may reassure health workers that women can self-inject independently. Concerns about safe storage and waste disposal may be assuaged by providing low cost impermeable containers to store the spent units until a convenient time to return the needles to a clinic, community health worker, or pharmacy. With respect to other parameters for the design of a self-injection program, the vast majority of providers (of all types) identify clinic-based health workers as best placed to train women in self-injection. What providers may not recognize is that training women in self-injection can be time-consuming, particularly if training is conducted in a one-on-one fashion. The heavy workload of providersparticularly those in the public sectormay impinge on the availability of self-injection training. While less acceptable to providers in this study, permitting community-based health workers to train women may be a more realistic approach, and one that recent research suggests is feasible [8]. With regard to adolescents, a number of participants proposed that adolescents be trained separately from adult women, to reduce feelings of intimidation and discomfort. Making that proposal a reality likely entails offering self-injection training beyond the clinic, at outreach events or via programs that specifically target adolescent clients. With respect to the private sector, though private clinic providers were not noticeably disinclined to endorse self-injection, it remains to be seen whether they will offer self-injection training, given their financial incentive to encourage repeat clientele in order to collect consultation fees. Training in commercial settings, such as pharmacies and drug shops, may be constrained by lack of a private setting for injections. In terms of post-training support to self-injecting clients, most providers proposed proactive follow up to ensure that clients recall how and when to self-inject. This suggestion presents a number of challenges: First, many clients in Uganda are using methods discreetly, and much of the appeal of self-injection stems from the potential for enhanced confidentiality. In this setting, clients may not welcome a home visit or phone call from a family planning provider. Second, from a practical standpoint, the majority of women in Uganda do not have exclusive access to a cell phone, and providers are not resourced to provide home visits (or make calls to their clientele with phones). Tasking community health workers with follow up may be more practical and cost-effective, but will require strong coordination between clinic and field-based health workers. One option for client-initiated support currently being tested in Uganda is the offer of a toll-free hotline manned by trained self-injection counselors. If successful, this approach may satisfy the World Health Organization recommendation that self-injection be offered "in contexts where mechanisms to provide the woman with appropriate information and training exist, referral linkages to a healthcare provider are strong, and where monitoring and follow-up can be ensured" [32]. --- Study limitations As with all qualitative studies, our findings are not generalizable and may not apply to providers in other settings in Uganda and other countries. While attempts were made to solicit honest and forthcoming opinions, providers may have been subject to some degree of social desirability bias, offering opinions overly favorable to adolescent contraceptive use, injectable use, and/or self-injection. --- Conclusion Self-injection presents an opportunity to reduce the burden on the health system presented by heavy reliance on injectable contraception requiring quarterly clinic visits. For adolescents, the potential discretion and user control inherent in the practice of self-injection align with well-established priorities for improving adolescent access to contraceptive services. While our results reveal a level of cautious support for self-injection among providers in Uganda, their reservations about offering the service to adolescents suggest more needs to be done if self-injection is to be made available to women without regard to age, parity or marital status. In addition to including self-injection in healthcare training curricula, service delivery guidelines will need to be revised and existing providers will require continuous professional development to bring them up to speed with this self-care intervention. With about a dozen countries in sub-Saharan Africa currently poised to introduce self-injection as a delivery option, this study offers insights on provider perspectives that ministries of health may wish to consider, particularly if their goals include improving access to family planning for adolescents. Moving forward, policymakers and program implementers should design, implement, and evaluate self-injection interventions with the needs of adolescent clients uppermost in mind, recognizing that extra effort may be required to shift provider perspectives and assuage their concerns. --- Availability of data and materials The qualitative dataset generated for this study is not publicly available due to concerns that individuals may be recognizable based on the content of their interview transcript. --- Ethics approval and consent to participate This study was approved by the Mulago Hospital IRB, the Uganda National Council for Science and Technology, and the PATH Research Ethics Committee. --- Consent for publication Not applicable --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Self-injection of subcutaneous depot medroxyprogesterone acetate may offer greater discretion and increase access to injectable contraception, particularly for those who face challenges accessing clinic services. In particular, unmarried adolescents often encounter stigma when seeking services, and may also lack the financial means to travel to clinics on the quarterly basis that injectable contraception requires. Whether self-injection is offered to women on a wide scale basis, and to adolescents specifically, will depend in part upon the willingness of providers to train clients of diverse ages and educational backgrounds. This study explores the views of providers with regard to self-injection as an option for women and adolescents in Uganda. Methods: In-depth qualitative interviews were conducted with family planning providers in Gulu district, to understand their views on injectable self-injection for women, with a specific focus on unmarried adolescents ages 15 to 19 years. The in-depth interviews, which lasted up to 60 min were audio-recorded, translated and transcribed simultaneously, and analyzed using Atlas.ti software to identify key themes and common perspectives. Results: A total of 40 health care providers were interviewed with equal numbers of each type (public, NGO, and private clinics, pharmacies, and community-based health workers). While most providers were receptive to selfinjection for adult women, fewer than half were supportive of adolescent self-injection. Their reservations focused on age, marital status and parity concerns around adolescent use of the injectable more broadly, and concerns about the ability of adolescents to self-inject safely. Conclusions: Self-injection presents an opportunity to reduce the enormous burden on the public sector health system in Uganda, which is particularly compounded by the heavy reliance on injectable contraception requiring quarterly clinic visits. The results of this study reveal a level of cautious support for self-injection among providers when it comes to self-injection by adult women. With respect to adolescent clients, family planning policymakers and program implementers should design, implement, and evaluate self-injection interventions with the needs of adolescent clients uppermost in mind, recognizing that extra attention will likely be needed to reduce providerimposed restrictions on adolescent access to this injectable delivery modality.

"Please! No politics this time!" A senior member of staff from the Ugandan Ministry of Health greeted us with these words when we returned to Uganda to do more fieldwork in 2008. He was referring to an article that we had published a few months previously on the control of two neglected tropical diseases (NTDs): schistosomiasis and soil-transmitted helminths (Parker, Allen, and Hastings 2008). The article was based on fieldwork undertaken in Panyimur, north-western Uganda in 2005, and demonstrated the multiple ways in which political, historical, social, and economic processes influenced the rollout of a top-down, biomedical intervention run by the Ministry of Health. In particular, it highlighted that resistance to the take-up of free drugs was linked by many of those with whom we lived to a sense of marginalization, and sometimes outright oppression by the Ugandan government. Our conclusion made the point that while some people benefitted from the treatment program, it was unrealistic to assume that it would lift such impoverished and politically excluded populations out of poverty. It is mostly neglected people who are infected with neglected diseases-and this fact could not just be wished away. The member of staff concerned was not alone in trying to evade public discussion of issues influencing the control program in Uganda. Indeed, his desire to avoid engaging with the political economy of NTDs resonates with those running similar programs in other parts of sub-Saharan Africa and elsewhere. Over the past 10 years, multi-million dollar grants have been made available by international donors in response to claims that it is possible to end diseases of poverty, and maybe even poverty itself, through dispensing medications, free of charge, to adults and children living in areas where NTDs are endemic (see, e.g., Molyneux, Hotez, and Fenwick 2005;World Health Organization [WHO] 2010). This is asserted as a just and moral cause-one that is above criticism and relatively easy to implement in all contexts. In private, however, many of those involved with implementing these programs openly recognize that such top-down approaches can be counterproductive, and that claims about what is happening on the ground are often overly positive. Nevertheless, they continue to act as if successful outcomes are inevitable. How is this cognitive dissonance maintained and managed? Is it so pervasive as to make evidence irrelevant? Or is it strategic? In this article, we suggest that the rise of mass drug administration programs for the treatment and prevention of NTDs, without reference to local political, historical, social, and economic circumstances, are a remarkable example of how global public health remains chronically prone to what James Ferguson has called anti-politics. However, recourse to Ferguson's theoretical model, which suggests a hegemonic discourse that determines thinking and perceiving, misses key issues. In particular, locating the point at which rhetorics are actually believed by those articulating them can be hard to discern. It is, therefore, helpful to analyze how those promoting selective interventions for the control of NTDs respond to alternative perspectives. This throws light on underlying motivations and suggests that anti-politics discourse is sometimes less hegemonic than it seems. Rather, deliberate strategizing is involved, with discourse being used as an instrumental tool to promote vested interests. --- ANTI-POLITICS AND MAGIC BULLET MEDICINE Anthropologists engaging with international development issues, especially students doing so for the first time, often feel compelled to quote James Ferguson's book The Anti-Politics Machine (1990), or to at least make the point associated with his work that the aid industry de-politicizes social realities as a way of asserting control over them. His work is invoked to indicate why social contexts are ignored. The language of 'doing development,' so the argument goes, enables the promotion of technical solutions which can be implemented in a context-free manner with largely international expertise. Power dynamics associated with the notion of development thus allows for a disengagement from the real world and results in ineffective strategies which maintain structural inequalities. The approach can be insightful, but like all theoretical models, it has limitations. It constructs a straw target out of the international development project; and it neither does justice to the complexities of analysis associated with actual development work nor to the intelligence of well-meaning (and less well-meaning) international organizations and their staff. It also tends to exaggerate the effects of aid, and pays curiously little regard both to the long involvement of anthropology with development agendas and to the sophistication and insight of earlier researchers. Perhaps most importantly, it largely sets aside the possibilities of strategic decision-making. Agency is down-played. Part of the problem lies with Ferguson himself. His book is engaging to read, and offers an 'off the peg' critique, which is easy to emulate. He writes so well that flaws in his argument are disguised. Ferguson studied the World Bank in Lesotho between 1975 and 1984, and cited reports that made ludicrous statements given the realities on the ground. However, many other World Bank reports and publications are less open to such ridicule. There are plenty that make realistic and controversial observations about the political dimensions of development programs, referring to corruption, the failure of projects to reach their objectives, and the lack of an evidence base for policymaking (e.g., World Bank 1981, 1997, 2011). Moreover, Ferguson does not discuss the various constraints faced by the World Bank in the 1970s and 1980s. It is an organization comprised of its member governments. It cannot say certain things, or if it does, it has to use a kind of code. World Bank staff may be frank in private, but they have to be diplomatic in public. This does not mean that they cannot work with other groups and individuals who are more explicit; they have always done so. At the time of Ferguson's fieldwork, Lesotho was surrounded by apartheid South Africa. The United Nations, of which the World Bank is formally affiliated, had agreed to implement sanctions. Work in Lesotho required that these sanctions were sidestepped, but such sensitive issues were unlikely to appear in the official reports of bilateral and multilateral organizations working there. This does not make the statements cited by Ferguson any less misleading, but it does beg the question of how seriously they were taken by those writing and reading them. David Mosse's observations are apposite here: "Precisely because such a large proportion of the time and expertise of development personnel is organized with reference to writing and negotiating texts, they cannot be read at face value without reference to the arguments, interests and divergent points of view that they encode and to which they allude " (2005:15). Reports and articles by those involved in development work thus need to be analyzed in relation to the social relationships that produce them, and, in most cases, in relation to how soft funding is being deployed. This raises the issue of how far such texts end up shaping perceptions and policy decisions. A particular sort of language may be essential to 'keep the show on the road.' There is, however, a danger that those deploying arguments and information for strategic purposes end up taking seriously what they write and say. At one level, they may know about local realities and practical difficulties, and will talk about them at length in some settings. They are the staple conversation of restaurants, bars, and compounds frequented by aid workers. Yet, in other circumstances, they may appear to think official assertions, normative perceptions, and institutional ideologies are factually correct, even when they are contradicted by their own day-to-day experiences. Cognitive dissonance of this kind relates to Ferguson's points about the power of language, but it suggests a much more complicated terrain than the one he depicts. Cognitive dissonance is likely to be most pronounced where developers work within closed epistemic communities and networks operating within the aid world, and reinforced by broader belief systems associated with universal codes and principles (Mosse 2005:15). Examples of such codes include socially hegemonic ideas about human rights, poverty, economic liberalism, governance, and biomedicine. In some respects, the latter is the most extreme example. Ideas about biomedicine and related medical hierarchies comprise an almost unquestionable body of knowledge that legitimizes health-related development programs, and helps insulate the statements and motivations of biomedically orientated researchers and policymakers from critical scrutiny. The whole sector is elevated as a kind of moral good, to be assessed in special ways, by carefully trained and initiated analysts. In this sector, perhaps more than any other, anti-politics discourse can shape ways of thinking and seeing. Ferguson draws on the ideas of Michel Foucault (1973), who argued that the clinical or medical gaze takes the human body out of its social context, turning it into an object for amoral and technical investigation. At a social level, it becomes an effective means of control, an aspect of biopower. Ferguson's theoretical contribution was to take Foucault's insights about the history of social policy in Europe and apply them in a distilled and simplified form to interpret a large-scale development scheme in Lesotho. He wrote mainly about mining, so he did not cite Foucault's The Birth of the Clinic (1973). However, he did use Foucault's idea of biopower. In this respect, it is interesting to note the period of Ferguson's fieldwork. Although he makes no mention of health policy, Ferguson was researching at a time when the campaign for a new approach to basic health care was being promoted in Africa and elsewhere. The high watermark of the campaign was the 1978 Primary Health Care Declaration at Alma Ata. He was still doing fieldwork when the declaration was effectively set aside in favor of a more technological and vertical approach, one that might be viewed as illustrative of his core argument. These events in the 1970s and 1980s are germane to the more recent efforts to deal with NTDs. The Alma Ata Declaration was arguably as unrealistic and impractical as the World Bank reports described by Ferguson. It provided a manifesto for mass access to health care that linked poor health with poverty, and proposed responses that moved away from relying on biomedical solutions to lessen the burden of ill-health. While reflecting similar kinds of concerns to those Ferguson raises, but in the health arena, it proposed a radical approach for changing the status quo. It drew on social research to propose a model of basic health care orientated to the needs of deprived populations, rather than the interests of medical establishments (WHO 2005). The social context that produced it included the increased influence of developing counties within the United Nations; recognition of the overall health outcome achievements of certain countriesmostly with socialist governments; a degree of international antipathy toward the United States, particularly in response to its activities in Latin America; skepticism about the political motivations of bilateral development aid and the World Bank; and suspicions about the interests of drug companies. The location of Alma Ata in the former Soviet Union was also significant. Inverting Ferguson, here was 'international development' as a kind of 'politics machine.' The response was to counter the agenda set out in the declaration with a different kind of primary health care, namely'selective primary health care' (Walsh and Warren 1979). Initially focusing on vaccinations, selective primary health care expanded to include the promotion of breast-feeding, growth monitoring of children, and the use of oral rehydration salts to treat diarrheal diseases. Family planning and the provision of nutritional supplements were later added to the list. UNICEF took the lead in promoting the approach, and secured substantial international funding (Cueto 2004). Concerns were voiced from the outset, with critics drawing attention to the way in which biomedical interventions set up unsustainable approaches that not only undermined national health care systems, but also did little to address the fundamental causes of ill-health (Berman 1982;Unger and Killingsworth 1986). These points were subsequently shown to be largely accurate in a number of locations, with some researchers suggesting that published data highlighting the achievements of selective interventions had been exaggerated (Werner and Sanders 1997). However, 'de-politicized strategies' were popular with donors, and restored the authority of biomedical professionals. It was also easier to package such programs for large-scale fundraising than more locally contingent and grounded strategies. In particular, the idea of handing out medications that had the potential to save lives was compelling. It is, therefore, an approach that has been persistently reinvigorated-the adding of vitamin A to the mix in the 1990s being a good example (Ching et al. 2000). In addition, protagonists have become increasingly effective at setting aside critical perspectives with reference to higher values and morally loaded claims. The plight of the impoverished, suffering from curable diseases, is depicted in horrible images and harrowing stories; and dispensing free medicines is presented as a solution. Who could be against these kinds of programs? Human suffering demands social justice and vertical medical interventions appear to offer a form of it without complications. From time to time, evidence continues to emerge that large-scale, top-down, context-free interventions are not as successful as they are reported to be. Occasionally, there are castigating assessments in the medical literature. A 2010 editorial in The Lancet, for example, bemoaned extravagant claims made about UNICEF's current Child Survival and Development program, pointing out that "(w)ithout proper monitoring and accountability, countries and donors-and taxpayers-have no idea whether or how their investments are working.... Evaluation matters.... It's time that the global health community embraced rather than evaded this message" (The Lancet 2010:526). Yet, apolitical, top-down biomedical interventions seem impossible to set aside. Indeed, the very next issue of The Lancet published a series of papers promoting what is in effect a new form of selective primary health care: the rapidly expanding global program of mass drug administration for the control of NTDs (Baker et al. 2010;Conteh, Engels, and Molyneux 2010;Gyapong et al. 2010). --- CONTEXT-FREE MASS DRUG ADMINISTRATION FOR NEGLECTED TROPICAL DISEASES Although mass drug administration is presented as health care implemented through existing systems, in practice it is another top-down, context-free, pathogen-focused strategy, modeled on the mass immunization campaigns of the 1980s. It is open to the same kinds of criticisms, notably with respect to monitoring, but it has successfully revamped the notion of a vaccine against poverty. As Allotey, Reidpath, and Pokhrel observed, under the guise of pro-poor arguments, "it remains a medical and technological fix; the'magic bullet' to combat disease" (2010:e32). The idea of distributing medication for particular diseases in endemic areas has been around for a long time. The current rapid up-scaling is connected to successful lobbying around the Millennium Development Goals (MDGs). The sixth goal aims to combat HIV/AIDS, malaria and 'other diseases.' Tuberculosis is mentioned as one of these 'other diseases,' but no more are named. The category 'other diseases' thus became the object of intense lobbying; and it was successfully captured by those concerned about a range of tropical diseases-the NTDs. The particular diseases included in the group vary, depending on the interests of those involved in control programs. However, the diseases most commonly included are parasitic, and they can potentially be controlled, eliminated, or even eradicated with available medications. The five 'core' NTDs are onchocerciasis, lymphatic filariasis, schistosomiasis, trachoma, and soiltransmitted helminths. A highly effective campaign pooled together the resources and networks of key individuals and institutions involved in combating these afflictions. Five years into the new century, the Millennium Development Project published a set of 'Quick Win' interventions, able to deliver rapid results in attaining the MDGs. Neglected tropical disease control was foregrounded as a quick win. In an influential article published the same year, this was described as "one of the most convincing ways to make poverty history [in sub-Saharan Africa]" (Molyneux et al. 2005:106). Molyneux and colleagues argued that it is possible for public-private partnerships to develop integrated control programs for the core NTDs in Africa with only four drugs: albendazole, ivermectin, praziquantel, and azithromycin. Concerted pressure was placed on pharmaceutical companies to make these medications available on a massive scale. The companies needed to sustain the prices of the drugs in rich countries, and to counter the threat of influxes of generic products in poor countries. These factors, combined with a degree of institutional philanthropy, resulted in a positive response. Millions of tablets were provided free of charge and mass drug administration programs were established in a range of countries, with moves toward an integrated approach whereby multiple infections could be treated simultaneously. In Uganda, for example, an integrated program was introduced in 2007. This program linked together previously separate endeavors to control onchocerciasis and lymphatic filariasis with schistosomiasis and soil-transmitted helminths. In common with many other countries, the aim was to provide tablets to all adults and children living in endemic areas, irrespective of whether people were actually infected. This was considered ethical because the chances of infection were considered to be very high and the drugs themselves were said to be safe and efficacious. Claims made about the success of such operations have been far-reaching. In 2008, for example, the elimination of lymphatic filariasis by 2020 was heralded as a real possibility (BBC 2008); and reported estimates of the numbers of people treated for schistosomiasis are remarkable. In Uganda, for example, more than 13-million received treatment in 4 years (http://ntd.rti.org, accessed on 28th September 2012, cited in Parker and Allen, 2011). In 2010, the WHO published the first ever global report on NTDs. This, too, provided a glowing assessment of the strategies being implemented (WHO 2010). By that point, the GATES Foundation, US Agency for International Development, the UK Department for International Development (DfID), and other donors were already committing themselves to releasing substantial funds; and on January 30, 2012 further pledges were made by donors and pharmaceutical companies at a meeting in London. These included pledges to sustain the supply of free drugs until the threat of infection with the targeted diseases had been controlled and, in some cases, eliminated. The UK DfID alone pledged more than £240 million to the ongoing operations. In May 2012, Margaret Chan, the Director General of WHO, proclaimed in her address to the World Health Assembly that "these Cinderella diseases, long ignored and underappreciated, are a rags-to-riches story" (WHO 2012). Perhaps, predictably, as the enthusiasm of medical establishments, pharmaceutical companies, and donors cohered around the agenda, the context-free qualities of mass drug administration became more pronounced. Statements were made about the need to work closely with target populations, provide good health education, ensure adequate community mobilization, and encourage local ownership of the treatment regimes. However, in practice, as Allotey and colleagues (2010) noted, the tendency has been to rely, ever more, on vertical, externally audited, biomedical interventions. Pointing this out has become increasingly controversial, and it has led to a turning away from evidence highlighting the realities of mass drug administration on the ground. What are these realities? This is a surprisingly difficult question to answer. There is plenty of literature asserting that millions of people have been treated for a variety of NTDs across sub-Saharan Africa. In Uganda and Tanzania, for example, these statements are often based on the number of tablets that are recorded as passing through national and/or district headquarters on their way to villages for distribution (e.g., Fleming et al. 2009;Malecela et al. 2009). However, it is not possible to use this type of data to gauge the proportion of children or adults living in villages who have actually received, let alone consumed, tablets. Village drug distributors are required to, and in some cases keep, detailed registers of the names and ages of villagers receiving tablets, but they rarely record the names of people who have not received treatment, and they rarely mention whether the distributed drugs have been swallowed (Parker andAllen, 2011, 2013). There is also a paucity of epidemiological information documenting changing rates of infection following mass treatment. Available data are mainly based on a small number of sentinel sites and/or cohort studies (e.g., Kabatereine et al. 2007;Mohammed et al. 2006;Simonsen et al. 2010). Ethnographic research highlights the limits of these approaches, and suggests that it is inappropriate to assume that findings from such studies are being replicated in the wider population (Parker et al. 2012(Parker et al., 2013)). Research exploring local responses to mass drug administration was undertaken by the two of us between 2005 and 2011. This research primarily focused on two NTDs: schistosomiasis and lymphatic filariasis in different parts of Tanzania and Uganda (see Parker et al. 2008Parker et al., 2012;;Parker and Allen 2011;Parker and Allen 2013 for further details). Grounded in ethnographic fieldwork (and thus paying close attention to the use of language and the contexts in which it is possible and sometimes impossible to speak openly about the distribution and consumption of drugs), we showed that drug uptake among adults and children varies considerably within and between districts, with many locations falling short of the recommended levels deemed necessary to achieve control. With respect to schistosomiasis, fieldwork undertaken at selected sites in northwestern and southeastern Uganda, Ukererwe Island, and parts of Morogoro and Tanga regions in Tanzania demonstrated that drug uptake among adults rarely reached the World Health Organization's target of treating 75% adults living in endemic areas. Indeed, in Panyimur subcounty, northwestern Uganda, Bugiri district, southeastern Uganda, and Ukererwe Island, Tanzania, it often fell below 45%. By contrast, the uptake of drugs among adults was quite high in Lumino subcounty, southeastern Uganda, and exceeded the recommended 75% target level in a few villages in Moyo and Adjumani districts (both located on the River Nile, bordering South Sudan). Considerable variation was also noted among school children. Drug uptake tended to be higher than that recorded for adults especially in northwestern Uganda, but this was not always the case. In 2008, for example, mass drug administration had to be halted in parts of Tanzania due to the interventions of angry parents and the refusal by some teachers to participate in the program (IRIN Humanitarian News and Analysis 2008; Daily News Online Edition 2008). Research on lymphatic filariasis in Pangani and Muheza districts, northern coastal Tanzania, revealed that drug uptake among adults and children from 2004-2011 was consistently below 50% at the majority of field sites. While higher rates were found in a few places close to district capitals (where the main hospitals are located), they were still mostly below the 70%-90% uptake levels thought to be necessary for sustained control or elimination to occur (Michael et al. 2004). A variety of historical, political, social, and economic factors help explain these findings from East Africa. They include historical experiences of previous endeavors to control tropical diseases, population movement across borders, subjective experiences of side effects from treatment, alternative understandings of affliction, rumors about the'real' purposes of treatment, ineffective communication about the rationale for mass treatment, too great a reliance on volunteers to distribute drugs, and responses to social control measures. Our research shows how vital it is to engage with these issues and to adapt mass drug administration programs to specific local circumstances if there is to be any prospect of controlling schistosomiasis, let alone eliminating lymphatic filariasis. It is striking, however, that little effort has been made to amend current strategies in the light of local level evidence. Why? --- CONTROLLING THE DEBATE, IGNORING THE EVIDENCE Just as economic and political forces shape the way in which policy is formulated, so they also shape responses to data suggesting that the uptake of drugs falls short of the requisite levels to control NTDs. These responses take multiple forms, but collectively reveal an endeavor to set aside discomforting information and, increasingly, to control the terms of the debate and marginalize critics. In so doing, it becomes possible for mass treatment programs to not only be scaled up at speed but also for funds to continue to be concentrated in a small number of institutions promoting context-free, mass treatment as a technical, a-political, and morally appropriate response to aspects of grinding poverty. This section illustrates these points with reference to our research in Uganda and Tanzania, as well as to research undertaken by other scholars who, grounded in the disciplines of epidemiology and parasitology, have questioned the merits of current mass treatment programs. To start with research in Uganda: The reluctance to consume free drugs for the treatment of schistosomiasis in Panyimur was a disappointing finding for those advising and running Uganda's national control program. Initial responses in 2005 included shock, concern, and distress. A junior member of staff, for example, broke down in tears at a meeting when she heard that many villagers were rejecting the drugs, because they were afraid that the treatment might damage their fertility, and in some circumstances, kill them. More senior colleagues thanked us for highlighting difficulties and problems that had gone unnoticed, and told us that this was exactly the sort of local level assessment that was needed to help the national disease control program achieve its objectives. Disturbed by our findings, they subsequently dispatched a member of staff to our field site to track down our research assistants and check with them that our assessment of the situation was 'correct.' When they confirmed our analysis, they were told that they had a duty as Ugandan citizens to report that the drugs were welcomed. If they did not do so, then it was likely that treatment would be withdrawn and other adverse consequences would follow for them personally. Although the attempt to intimidate our research assistants was not authorized from 'above,' the views expressed by the member of staff gave voice to fears and concerns at a district and national level that information which did not suggest the smooth running of the program seeking to control schistosomiasis could easily be construed as 'failure.' This, in turn, could have negative effects on the funding that would be made available by international donors. An attempt was also made to contain and contextualize the results by commissioning a new report (Lubanga 2007). This replicated parts of our research in Panyimur, north-west Uganda, with some of our locally trained research assistants even contributing to focus group discussions about local understandings and responses to mass drug administration! However, this new report made no reference to the fact that the uptake of drugs fell short of the coverage levels recommended by the WHO. Instead, findings from qualitative work in Panyimur, exploring knowledge and attitudes about mass treatment for schistosomiasis, were compared with findings from similar work undertaken in Busia, south-eastern Uganda, where the program was said to be working well. The differing political and historical contexts in which mass treatment was being rolled out was not mentioned nor, therefore, were the various ways in which this was affecting the uptake of drugs. It was hard not to conclude that this was an effort to re-package and de-politicize our findings. Commissioned reports aside, the prevailing view behind the scenes, at least among those responsible for running the mass treatment program, remained the same: it was better to know about difficulties arising from the program than not to know about them. As one senior colleague said, "There is no point spending large sums of money on the control of schistosomiasis if the program is not going to work." It was important, too, not to assume that difficulties with the program in one part of the country were being replicated in other parts of the country. We were thus asked to undertake further research at selected sites in Moyo and Adjumani districts, north-western Uganda and Busia district, south-eastern Uganda in 2008 and 2009. Findings from this research were presented in an article by Parker and Allen (2011). A draft was circulated prior to publication and it was amended in the light of some of the comments received. One colleague, having read the draft, acknowledged that it highlighted some important and complex issues in a helpful way, but expressed concern that it would be interpreted as a 'negative' assessment of mass drug administration programs. With ongoing uncertainty as to whether funding would continue to be forthcoming for programs seeking to control NTDs, he asked us to consider amending the abstract and conclusion so that it sounded more positive. The same colleague also asked us to be careful about the findings we presented at seminars and conferences. To quote one e-mail: "I am worried [by]... what you are saying.... You could endanger our fund raising efforts by negativity." However, as with the paper published in 2008, it was not until the article was published that controversy really broke out. Some of those implementing mass drug administration were furious that it was available in the public arena. One particularly irate communication said: "People who have read this [paper] say 'Are they for you or against you?' On balance I have to say it reads like against. Which considering all the support you have received is disappointing." The latter comment is revealing. Despite the fact that the paper is broadly supportive of mass treatment, and made numerous suggestions about how to improve existing programs (such as using existing public health legislation to foster behavioral change), it did not provoke substantive discussion about how policies could be refined in the light of the available evidence. Research exploring local understandings and responses to the mass distribution of drugs for the treatment and prevention of lymphatic filariasis, schistosomiasis, and soil-transmitted helminths in northern coastal Tanzania generated a different set of responses, although arguably with the same outcome: to set aside evidence suggesting that mass drug administration may not be as effective as that suggested by global policymakers, and to control debate. By way of illustration, it became apparent within the first six weeks of fieldwork that there were substantial difficulties rolling out free drugs for the treatment of lymphatic filariasis. These early findings were reported to senior figures running mass treatment programs at the government's headquarters in Dar es Salaam. Their response led to two rather different courses of action. First, the suggestion that mass treatment was running into difficulties was taken very seriously. It resonated with concerns filtering through from other sources: namely, that mass treatment programs primarily benefit those who are infected, asymptomatic, and likely to develop visible signs of infection (such as swollen limbs and/or enlarged scrotums) in years to come; rather than those who are infected and already display visible signs of infection, for whom treatment can do relatively little. While the clinical benefits of treating those who are infected and asymptomatic are irrefutable (as the drugs kill the filarial worms and thus prevent the emergence of visible signs of infection), it does not follow that those being treated recognize the benefits of treatment. On the contrary, the absence of information about the etiology of lymphatic filariasis and the rationale for mass treatment fostered close questioning about the'real' purposes of handing out free treatment, and fuelled resistance to the consumption of medicines. Indeed, many posed the question: Why should someone who has no visible signs of infection consume tablets for an affliction they do not think they suffer from? Concerned by these findings, a senior figure said that she would look for additional funding to supplement the roll out of mass treatment with other interventions, including hydrocelectomies. This was a hugely positive response. Indeed, in 2008, funding was secured to enable 200 men in Pangani district, northern coastal Tanzania to be given surgery on swollen scrotums (Malecela et al. 2009). However, our findings simultaneously provoked a second less positive response, aimed at limiting the dissemination of findings. Ethnographic fieldwork had been funded as part of a broader program of work monitoring integrated mass treatment of NTDs. In northern coastal Tanzania, it was anticipated that this would involve monitoring the roll out of drugs for the treatment for schistosomiasis and soil-transmitted helminths; and this was stated in our initial permission documents. It transpired, however, that schistosomiasis and soil transmitted helminths were not being treated together. Instead, the treatment for soil-transmitted helminths was being combined with the treatment for lymphatic filariasis. This made good sense. Soil-transmitted helminths and lymphatic filariasis are both endemic in this part of Tanzania; and the drug, albendazole, has the dual purpose of treating both soil-transmitted helminths and, in combination with ivermectin, lymphatic filariasis. The tablets were thus being distributed together, and it proved impossible to investigate the take-up of albendazole without also noting the take-up of ivermectin. Additionally, it became apparent that the low uptake of albendazole was directly connected with local attitudes to mass treatment for lymphatic filariasis. District medical staff and researchers working with Tanzania's National Institute of Medical Research in the region encouraged us to collate and report our results. However, matters were not perceived in the same way at a national level. In fact, the Director of the lymphatic filariasis program in Dar es Salaam took the view that it was unacceptable to record low rates of drug uptake for this disease, even if these were identical to those for soil transmitted helminths. A protocol deviation had to be secured first, and this might take quite a while. In the meantime, research on drug uptake should cease. Without a doubt, bureaucratic procedures were being used to prevent information about local resistance to mass treatment for lymphatic filariasis being more widely known. Having said that, the protocol deviation was accepted almost a year later as part of a broader submission to work on a wide range of NTDs in different parts of Tanzania; and it was back-dated to the time of our 2007 research. Nevertheless, apprehension about the possible consequences of disseminating our results remained. We were explicitly asked by an international figure involved in the promotion of mass treatment not to publish our findings, as he feared it would jeopardize future funding for the program. Sympathetic to the pressures, especially in the wake of the riots that took place in Morogoro and other parts of Tanzania in 2008 (in response to mass drug administration for schistosomiasis), we agreed to do this, and decided to wait until we could carry out follow up research at our field sites. Perhaps naively, we expected to find some significant changes when we returned to northern coastal Tanzania in July-August 2011. That was not the case. Despite making our findings available to those running the programs, there had been no substantive alterations to the way in which the mass treatment program for the elimination of lymphatic filariasis was being administered. There was still much too much reliance on volunteers to distribute the drugs, and the rationale for the program remained poorly understood by both those handing out the tablets and those supposed to be receiving them. Not surprisingly

Large amounts of funding are being allocated to the control of neglected tropical diseases. Strategies primarily rely on the mass distribution of drugs to adults and children living in endemic areas. The approach is presented as morally appropriate, technically effective, and context-free. Drawing on research undertaken in East Africa, we discuss ways in which normative ideas about global health programs are used to set aside social and biological evidence. In particular, there is a tendency to ignore local details, including information about actual drug take up. Ferguson's 'anti-politics' thesis is a useful starting point for analyzing why this happens, but is overly deterministic. Anti-politics discourse about healing the suffering poor may shape thinking and help explain cognitive dissonance. However, use of such discourse is also a means of strategically promoting vested interests and securing funding. Whatever the underlying motivations, rhetoric and realities are conflated, with potentially counterproductive consequences.

our 2007 research. Nevertheless, apprehension about the possible consequences of disseminating our results remained. We were explicitly asked by an international figure involved in the promotion of mass treatment not to publish our findings, as he feared it would jeopardize future funding for the program. Sympathetic to the pressures, especially in the wake of the riots that took place in Morogoro and other parts of Tanzania in 2008 (in response to mass drug administration for schistosomiasis), we agreed to do this, and decided to wait until we could carry out follow up research at our field sites. Perhaps naively, we expected to find some significant changes when we returned to northern coastal Tanzania in July-August 2011. That was not the case. Despite making our findings available to those running the programs, there had been no substantive alterations to the way in which the mass treatment program for the elimination of lymphatic filariasis was being administered. There was still much too much reliance on volunteers to distribute the drugs, and the rationale for the program remained poorly understood by both those handing out the tablets and those supposed to be receiving them. Not surprisingly, fear of treatment remained widespread and the uptake of drugs continued to fall short of the levels required to interrupt transmission. Why did the situation remain unchanged? In contrast to Uganda, there was a reluctance to impart information by those funding the research to their Tanzanian collaborators for fear of disrupting fragile relationships. Thus, a detailed report was not circulated for comment, and requests were made not to place findings in the public arena. This is by no means the only explanation. Within Tanzania, officials working at a district and national level were acutely aware of the challenges of implementing mass treatment, but felt unable to address the difficulties with the resources available. District officials talked of institutional pressures 'to cook' the data in such a way as to suggest that drug uptake levels were in line with national targets; and officials working at a national level spoke of the difficulties of challenging global policy for fear of jeopardizing future funding and the provision of drugs. Indeed, one senior figure in the Ministry of Health, reflecting on the pressures, said "We must swallow the bile to keep the drugs coming." The colleague concerned undoubtedly felt that there was little choice but to accept the drugs on offer as a more holistic approach (involving mass treatment alongside the provision of hydrocelectomies for scrotal swellings, lymphodema management, and vector control of the mosquitoes carrying filarial worms) would not be funded by international donors. Another official, reflecting on the difficulties of speaking out, said "We are black guys with no money, and they are white guys with lots of money." --- Extreme Examples of Exerting Control The self-silencing of dissent discussed previously is understandable, but it is also worrying, especially when reflecting on broader strategies deployed to control information in the public arena. These strategies are discussed in this subsection. Citing examples, it demonstrates how morally imbued discourse is used, with recourse to technical solutions putatively aimed at resolving complex social problems. At one level, the examples illustrate Ferguson's anti-politics machine thesis. Indeed, they could even be viewed as adding to it with reference to Foucault's clinical gaze, as well as his notion of biopower. A striking dynamic in terms of the language deployed is that anti-politics with respect to mass drug administration is often promoted in a highly political way, linked to accusations that those raising concerns are acting immorally. In so doing, there is a drawing on notions about biomedical knowledge having access to some sort of deep reality, above the mundane details of day-to-day events or other kinds of evidence. However, it is misleadingly deterministic to push this kind of analysis too far. It is also a less compelling explanation when dealing with debates between biomedical professionals than debates between biomedical professionals and social scientists. Moreover, in all the instances that follow, it is clear that instrumental choices are also part of the story. Evidence raising concerns with the orientation of programs is purposefully obscured, ignorance is deployed strategically (Mcgoey 2012), and anti-politics discourse is used instrumentally to achieve particular outcomes. These points are underlined by the public treatment of Professor Gryseels, Director of the Institute of Tropical Medicine in Antwerp. An expert on schistosomiasis, Gryseels expressed reservations about mass treatment programs in a letter to the Financial Times (2006). His concerns included the possible emergence of resistance to frontline drugs for the treatment of schistosomiasis, doubts about the efficacy of distributing drugs for very different tropical diseases alongside each other, and the likelihood that scaling up the mass distribution of drugs for NTDs would undermine endeavors to develop comprehensive, sustainable health services and create a top-down,'vertical monster.' The published response, ostensibly authored by African advocates of mass treatment programs, made light of Gryseels' concerns, and rejected them on the basis that it was inappropriate for him to express them. To quote "[Gryseel's] position from an African perspective is unethical and a violation of the fundamental right to health and contrary to the health policy of African countries" (Tchuente et al. 2006). Gryseels is not the only biomedical researcher to be treated in this way. In 2011, David Molyneux, a well-known advocate for the mass distribution of drugs for the treatment and prevention of NTDs, combined forces with a Tanzanian government official to critique an article published a year earlier by Cavalli and colleagues (2010). The data presented in this article suggested that the additional burden of work placed on those rolling out mass treatment programs in Malawi detracted from other essential work that had to be done, and thus undermined already fragile health services. Rather than engage with the methods and analysis, the argument presented by Cavalli and colleagues was set aside as inappropriate and "counter to 'The Right to Health'." A moral statement about the merits of mass treatment was asserted. To quote: "We ask a simple question... would it be acceptable NOT to distribute ivermectin and azithromycin to communities where there was a severe risk of blindness when free drugs and other resources were available, which if they were not given would commit many thousands of people... to a life of misery" (Molyneux and Malecela 2011:e234). Thus, the probity of Cavalli and colleagues was questioned, because they suggested that treatment of those in need should be undertaken in a less vertical and externally driven manner. Molyneux and Malecela similarly took issue with an article written by the two of us (Allen and Parker 2011). In this article, we highlighted the disjuncture between global rhetorics (suggesting that mass treatment for NTDs could alleviate the suffering of 'the bottom billion' and 'end poverty') and local realities (suggesting that it was unlikely that mass drug administration would be able to alleviate poverty in the way anticipated). At one point, we briefly illustrated our argument with reference to the low drug uptake rates in Tanzania mentioned previously. The response was to misrepresent key points that we made, dismiss them as inaccurate, and to question our integrity. For example, Molyneux and Malecela (2011:5) incorrectly claimed we believed that mass treatment has no beneficial clinical effects for those suffering from lymphatic filariasis. A series of articles are then cited to discredit such a position, while our points about monitoring, running, and understanding the program in Tanzania were simply ignored. With respect to drug uptake data, they again misrepresented our findings, and set aside information that the uptake of drugs in Pangani and Muheza districts, northern coastal Tanzania, was too low to enable effective and sustained control. They even cited a paper (Simonsen et al. 2010) that makes similar points to us as if it contradicted our findings. Echoing the response to Gryseels, they then asserted that raising problems with mass treatment is "disrespectful to endemic countries... unethical and grossly negligent." The publication of Molyneux and Malecela's article in Parasites and Vectors subsequently prompted discussion in The Lancet, with many of the same authors participating and largely restating their positions (see Allen and Parker 2012a;Meheus et al 2012;Molyneux et al. 2012). In those exchanges, leading advocates of context-free treatment made their underlying motivations explicit, articulating a fear that funding would be withdrawn and international agreements to provide free drugs abandoned, if it became known that there were difficulties rolling out treatment. Highlighting difficulties was said to be "cynical" and something that would do "the poor a disservice" if it ended up leading to the withdrawal of drugs (Molyneux et al. 2012(Molyneux et al. :1099)). Mindful of Murray Last's influential paper, "The importance of knowing about not knowing" (1988) and Wenzel Geissler's more recent paper "Public secrets in public health" (2012), it is helpful, at this juncture, to reflect on what is not said as well as what is said in the public sphere. Just as government officials may privately acknowledge fundamental problems with strategies relying on mass treatment for the control and elimination of NTDs, but publicly feel compelled to support the programs, so academics find themselves in similarly difficult situations. One parasitologist in Europe, for example, let it be known that while he recognized that the mass distribution of drugs was unlikely to eliminate lymphatic filariasis at certain highly endemic locations, the political economy of research made it little short of impossible to speak out. To quote from an e-mail: Another leading parasitologist wrote in a comparable vein, stating that he had already had some grants cut, and feared losing others, if he spoke out. Meanwhile, another academic specializing in global health emailed us. She, too, expressed support. Nevertheless she wondered, while noting difficulties with current approaches to the control of NTDs, if there was any value in directly challenging dominant voices. It was, she argued, a battle that could not be won, because there was too much riding on the status quo and it was not worth fighting. --- CONCLUSION How should anthropologists working in the arena of NTDs respond to endeavors to set aside biological and social data suggesting current programs to control these diseases are unlikely to achieve their objectives at some locations? One approach is to follow the advice of the academic working in public health, mentioned previously, and to write scholarly articles with the intention of gradually refining and developing existing approaches. This may work in time, but it assumes that accumulated evidence will eventually prevail over entrenched rhetoric. At present, there is little sign of that happening with the global campaign of mass drug administration for NTDs. While local level data identifying the complexities and challenges of rolling out free treatment may privately be recognized as providing useful information that could be drawn upon to improve existing strategies, the reality is that vertical, context-free policies have been expanded. Moreover, a wide range of strategies are increasingly being deployed to limit the kind of information gathered and to contain debate. If anything, antipathy to assessing alternative strategies or refining existing ones is increasing over time. This is discouraging, but for medical anthropologists there is an alternative project. While we may share normative values with those running medical projects about the imperative to heal the sick and to alleviate suffering, we also have the option to stand back and assess the projects themselves as part of our ethnography. We have found Ferguson's anti-politics thesis to be a useful starting point, not least because it helps us understand the way in which technical and moral rhetorics are used to de-politicize the highly politicized arena of controlling the control of NTDs. It is, however, not without its limitations. Despite its postmodern theoretical status, the anti-politics gaze lends itself to functionalist interpretations. It is like an old anthropological study of African linage shrines that explains the underlying structures of ritual action, and omits the intense competition between elders for authority. In the world of NTDs, there are doubtless times when cognitive dissonance is so pervasive that rhetoric and realities cannot be separated by those committed to mass drug administration. However, analyzing responses to alternative points of view is revealing of ways in which conscious strategic decisions are also being made. It shows that hegemonic rhetoric is sometimes a tool, used instrumentally to secure grants, impose hierarchies, and promote institutional interests. Incentives to do this have been apparent in global health programs at least since the selective primary health care campaigns of the 1980s. The current renewed emphasis on impacts and targets can only be expected to increase the trend. As we noted in our discussion of Ferguson's book, The Anti-Politics Machine, there is a danger in using his model in ways that actually ignore such political realities. It cannot be assumed that manifestly misleading statements are not recognized as such by those articulating them. Similar problems arise with the tropes of'mobile' or'scientific' sovereignty. The latter shares associations with Foucault's biopower thesis, ascribing to highly qualified, medical-scientific technocrats, decisions ordinarily reserved for state organs. Samsky (2012) used the idea in his interesting study of onchocerciasis in Tanzania; but this, too, can end up suggesting a web of thinking that shapes behaviors in specific and consistent ways. In some respects, an alternative, if again related, theoretical notion is potentially more insightful in assessing the global control programs for NTDs. Examining who is infected with NTDs lends itself to analysis drawing on ideas about structural violence (Farmer 1999). It is overwhelmingly the poor and marginal who are infected. In this respect, the issue for those involved in NTD policy is whether they are working to challenge such structures or whether they are perhaps inadvertently involved in sustaining them. The approach is predictably controversial and can be impossible for researchers dependent on soft funding (Allen and Parker 2012b). Medical anthropologists, however, are often less constrained in this respect than epidemiologists and parasitologists. Our research is not so costly, and we are not so restricted by clinical protocols and biomedical hierarchies. Potentially, this gives us greater academic freedom. It can place us in a position to present evidence in ways that others cannot, and it can force debates into the public arena. This is what we have endeavored to do with respect to context-free mass drug administration for NTDs. It is what we can contribute as critically engaged medical anthropologists. Perhaps naively, we have done so in the hope that there will be a greater emphasis on monitoring and an adjustment of existing policy, leading to the more effective delivery of treatment to the millions who certainly need them. While acknowledging the strength of the anti-politics perspective, to take the implications too far is misleading. If there is deliberation involved in the use of hegemonic discourse, there is the possibility of different choices being made. Small shifts in the incentives provided by donors can have big effects. We remain optimistic that such changes are possible.

Large amounts of funding are being allocated to the control of neglected tropical diseases. Strategies primarily rely on the mass distribution of drugs to adults and children living in endemic areas. The approach is presented as morally appropriate, technically effective, and context-free. Drawing on research undertaken in East Africa, we discuss ways in which normative ideas about global health programs are used to set aside social and biological evidence. In particular, there is a tendency to ignore local details, including information about actual drug take up. Ferguson's 'anti-politics' thesis is a useful starting point for analyzing why this happens, but is overly deterministic. Anti-politics discourse about healing the suffering poor may shape thinking and help explain cognitive dissonance. However, use of such discourse is also a means of strategically promoting vested interests and securing funding. Whatever the underlying motivations, rhetoric and realities are conflated, with potentially counterproductive consequences.

Introduction The World Health Organization's Commission on the Social Determinants of Health(CSDH) [1] drew attention to the marked inequalities in health that exist between and within populations, and focussed attention on health inequities. One of the approaches taken by the CSDH was to examine 'priority public health conditions', such as tuberculosis, neglected tropical diseases, mental health, cardiovascular diseases and diabetes [2]. The overall aim of these condition specific analyses was to understand the impact of social determinants on their occurrence and consequences and the potential for public health programmes to reduce health inequities. Diabetes (DM) is estimated to affect between 10% -15% of the adult population in the Caribbean region [3], and to be a major contributor to premature mortality [3,4].Where studies have been done, DM is responsible for high rates of complications, such as lower limb amputation [5]. In other parts of the world it is well known that the incidence and prevalence of Type 2 DM, which accounts for over 90% of DM in most populations, tends to be unevenly distributed by some markers of socio-economic position (SEP), including education and income [6]. In high income countries, for example, prevalence and incidence of type 2 DM is higher in those with lower education and income [6]. The relationship between type 2 DM and gender is less clear. Globally, for example, the prevalence in men and women is very similar [3,7]. More recently, however, it has been reported that once major risk factors for type 2 DM are taken into account, in particular obesity, men have a higher incidence than women [8]. Understanding how diabetes is distributed by gender and what factors underlie that distribution is important for planning interventions to control and prevent this common condition. In the work described here we sought to determine what is currently known about health inequities and their determinants for DM in the Caribbean with a view to informing health policy and further research. We used the theoretical framework from the CSDH analyses of Priority Public Health Conditions [2], which has five levels and three dimensions, as shown in the figure in the protocol (S1 Appendix). Our specific aim was to determine what is currently known about the social distribution of DM (type 1 and type 2); the social distribution of the risk factors for type 2 DM ('vulnerabilities' in the framework); and the social distribution of its major complications ('consequences'), including DM related mortality, in the Caribbean. As described in the methods section,'social distribution' was interpreted to mean the distribution by the following markers of socio-economic position: gender, race/ethnicity, education, occupation and income. Available data from studies conducted in the Caribbean on all of these, with the exception of gender, was very limited. In this paper, therefore, we present the findings on the distribution by gender of risk factors for type 2 diabetes, diabetes, and diabetes control and complications. --- Methods --- Protocol and eligibility criteria A study protocol was developed in advance (Supporting Information: S1 Appendix). The final protocol for this review was informed by an initial scoping exercise undertaken in Pubmed and Embase to gain insight into the volume, timeliness and range of published literature available to meet the aim. We sought literature from studies conducted in the Caribbean that described the distribution by one or more markers of socio-economic position on one or more of the following outcomes: e. Metabolic syndrome 2. The incidence or prevalence of type 1 and type 2 DM, including reported diagnosis by a health professional and/or based on glucose or glycated haemoglobin measurement 3. In those with diabetes, assessment of disease 'control' based on one or more of glycated haemoglobin, fasting glucose and blood pressure. 4. In those with diabetes, mortality and/or the incidence or prevalence of retinopathy, nephropathy, neuropathy, diabetic foot, lower extremity amputation. We were guided in our choice of markers of socio-economic position by the Progress-plus check list [9] and by the markers predominantly used by the Commission on the Social Determinants of Health [1]. We focussed on identifying data on distribution of risk factors, disease, control or complications by: gender, ethnicity/race, education, income, occupation. We defined 'the Caribbean' as consisting of 28 countries and territories [10]. For cultural and historical reasons three mainland countries are considered part of the Caribbean. These are Belize, Guyana and Suriname, and each is a member of Caribbean Community (CARI-COM). Bermuda, which is not geographically part of the Caribbean, was also included as it is an associate member of CARICOM, sharing cultural and historical ties with the English-speaking Caribbean. We did not apply language limits to our search. We deliberately kept eligibility criteria broad, wishing to identify any study that presented data on the distribution of one or more of the outcomes of interest by at least one of the markers of socio-economic position. Inclusion Criteria: • Persons living in the Caribbean; • Age of study participants: for diabetes outcomes, only studies in adults (18 and older) were considered, whereas for risk factors studies that included children/adolescents (12 years and older) were also included; • Described the distribution of at least one of the outcomes of interest by one of the markers of socio-economic position; • Quantitative study design (observation or intervention) with minimum sample size of 50. Exclusion Criteria: • Narrative review papers, commentaries, case series, qualitative studies and single case reports were not included. • Studies that only included the Caribbean diaspora (as opposed to populations living within the Caribbean) were not included. We limited our final search to studies published between January 1 st 2007 and December 31 st 2013, with the intention of identifying data that are timely enough to be able to inform policy and further research around the control of DM and related NCDs within the region. --- Information sources, search strategy and study selection We searched Medline (through Pubmed) and Embase (through Ovid), plus an additional five data bases through the Virtual Health Library (VHL) [11]. The databases searched through the VHL are: LILACS (the database of the Latin American and Caribbean of Health Sciences Information System); MedCarib (mainly health sciences data from the English Speaking Caribbean); IBECS (Biographic Index on Health Sciences from Spain, a potential source of Spanish language publications from the Caribbean); and the evidence databases of the Pan-American Health Organization and the World Health Organization. The detailed search strategy for Pubmed is given in S1 Appendix. This was adapted as appropriate for the other databases. The citations identified by the search were downloaded into Endnote (version X4) and the titles and abstracts independently reviewed by two individuals (AN and NU) for potential relevance. Citations were only excluded when both reviewers agreed that they were not relevant. Where doubt existed the citation was retained. --- Data collection, summary measures, and assessment of risk of bias The full text of all retained articles was sought. Data from articles were abstracted independently by two reviewers in two stages into a Microsoft Access database. The first stage determined whether or not the study met the inclusion criteria, and in the next stage relevant data were abstracted. Differences between first and second data abstractions were resolved at meetings of all the authors. The data abstraction form was designed to extract key study characteristics and findings relevant to the objective and to enable an assessment of the risk of bias in the study. The content of the data abstraction form was guided by the STROBE statement [12], on reporting observational epidemiology, and by the PRISMA statement on systematic reviews concerning health equity [9]. The reporting of the findings of this systematic review was also guided by the PRISMA statement (S1 Checklist). All available summary measures by category of each marker of socio-economic position were abstracted from each study, including prevalence, incidence, mean and median, and odds ratios. When available, confidence intervals, p values and measures of dispersion were also collected. Risk of bias was assessed for each individual relationship. For a relationship to be classified as 'low risk of bias' both the study as a whole and the individual relationship had to meet certain pragmatic criteria, as described in the protocol (S1 Appendix). As an example, for a relationship in a cross sectional study to have a low risk of bias, the study itself should be population based and have a response rate of more than 75% with at least 500 participants. In addition, the outcome (e.g. risk factor or diabetes) should be based on objective measures, the relationship assessed controlling for important potential confounders, such as age, and the number of individuals excluded from the analysis must be less than 15%. Three levels of risk of bias were ascribed: low, medium, high/unclear. --- Data Synthesis and Meta-analysis There was no a-priori plan to undertake meta-analyses, given the anticipated heterogeneity of the studies and settings. In addition, as is shown in the results section, many of the studies were considered to have a high risk of bias. However, on reviewing the findings meta-analyses were considered applicable to summarising the relationships on gender and generalised obesity (I 2 -93%), gender and smoking (I 2 -91%) and gender and diabetes prevalence (I 2 -20%). In these instances there were sufficient studies of low or medium risk of bias to facilitate synthesis. Only population-based cross-sectional studies which measured the outcome as a dichotomous variable were used in the meta-analyses. In the case of diabetes prevalence, only type 2 DM was considered; while for obesity only papers that used an objective measure of body mass index were used. However, smoking was most often measured by self-report. Meta-analysis was undertaken using RevMan 5.2 [13]. RevMan 5.2 was also used to create and examine funnel plots, qualitative examination of which revealed no evidence of publication bias. As no funnel plot included 10 or more studies, formal statistical tests for asymmetry were not applied [14]. --- Results --- Study selection From our search of seven databases we found 2796 unique records published between the beginning of 2007 and the end of 2013, of which 2709 were excluded after titles and abstracts were screened by two independent assessors (Fig 1) leaving 87 articles for review. Of the 81 articles for which we were able to obtain the full text, a further 25 did not meet the inclusion criteria, leaving 56 articles for review. Three of the remaining six abstracts were based on conference presentations, thus no full text was available and the final unavailable three each originated from an English, Spanish and French speaking territory. For this report the 50 articles which described the distribution of diabetes, risk factors or diabetes control or complications by gender were used in the qualitative synthesis of our results. In addition to producing a narrative synthesis, we performed meta-analyses examining the relationships between gender and diabetes prevalence (n = 7), gender and obesity (n = 8), and gender and smoking (n = 6). In each meta-analysis, Barbosa et al contributed two effect measures from the Study of Health and Well-being in the elderly (SABE) conducted in two Caribbean countries. --- Study Characteristics Within the 50 articles reviewed there were 27 unique studies, and within these studies we found 114 relationships between the distribution of diabetes mellitus prevalence/incidence or diabetes control or diabetes complications or a related diabetes risk factor and gender (Table 1). The studies were conducted in the English-speaking (Bahamas [15], Barbados [16][17][18][19][20][21][22][23][24][25][26], Grenada [27], Jamaica [28][29][30][31][32][33][34][35][36][37][38], Trinidad& Tobago [39][40][41] and the US virgin islands [42]); French-speaking (Guadeloupe [43][44][45][46][47][48]); Dutch-speaking (Suriname [49]and Saba [50]) and the Spanish-speaking (Cuba [16,18,19,24,26,[51][52][53][54][55][56]and Puerto Rico [57][58][59][60][61][62][63]) countries and territories across the Caribbean. As shown in Table 1 the relationships examined provided information either on the distribution of diabetes prevalence/incidence and gender(n = 19) and/or a diabetes-related risk factors by gender: generalized obesity (n = 29), abdominal obesity (n = 19), smoking (n = 15), physical activity (n = 15), aspect of diet (n = 6) and metabolic syndrome (n = 3).There was a scarcity of information(only 3 studies) examining a measure of diabetes control and gender and only one study examining the distribution of a complication of diabetes(mortality). Forty-four of the 50 articles were based on studies with a cross-sectional study design while the remainder utilized a cohort design. Some studies contributed more than one article for example the SABE --- Risk of Bias within Studies Over half of the relationships studied (74) were found to have a high risk of bias (Table 1). In some cases, the high risk of bias was due to the lack of a population based study design, with this being the case in 17 of the 50 articles identified. In these 17 articles the studies were based in health care facilities (n = 7), schools (n = 7), workplaces (n = 2) and a church (n = 1). We also considered cross-sectional studies with fewer than 500 persons and/or low (<unk>50%)) or unclear response rates to have high risk of bias (n = 11). There were several articles where the study as a whole was judged to have a low or medium risk of bias but where the relationship with gender was not adjusted for age. In some studies the health outcomes and risk factors were measured using subjective measures such as self-report where an objective measure is the accepted standard. Of the 19 articles that reported gender and diabetes prevalence, 9 used self-reported diagnosis only i.e. did not measure glucose. For physical activity assessment all studies used a questionnaire. The questionnaires included the International Physical Activity Questionnaire (IPAQ), 24 hour recall of 3 consecutive days, PAHO STEPS survey and single questions such as "How many times have you been active for 3 or more times per week in the past 12 months?" Smoking status was also exclusively assessed by self-report. With respect to obesity, it was based on measured BMI in 28 articles, and in two articles self-reported height and weight were used to calculate BMI. --- Results of Individuals Studies Gender and Diabetes prevalence. There were 19 articles from 12 unique studies which examined the relationship between diabetes prevalence and gender, 9 of these were based on population-based studies (Table 2).All 9 studies found a higher prevalence of DM in women than in men although in 2 studies this was not statistically significant (p>0.05), while in another 3 there was no assessment of type 1 error. In the meta-analysis, we included population-based studies that used objective and self-reported measures of diabetes prevalence and were considered low to medium risk of bias. The summary result of these studies indicates that women were more likely to have diabetes than men (OR 1.75, 95% CI 1.42, 2.14)(Fig 2 ), with subgroup analysis including only those population-based studies which used objective measures of diabetes mellitus finding a similar, if slightly weaker, relationship(OR 1.65, 95% CI 1.43, 1.91) (Fig 3). Four of these studies were from Jamaica, while the largest was from Cuba where women were 1.58 times more likely to have diabetes than men. (OR 1.58, 95% CI 1.31, 1.89).There was significant heterogeneity in combining the studies with I 2 values of 71% and 20% for the combined and subgroup analysis respectively. Gender and risk factors. We examined the relationships between gender and various diabetes risk factors. There were 30 papers describing the relationship between gender and obesity as assessed by body mass index (BMI) from Jamaica (n = 7), Barbados (n = 6), Cuba (n = 6), Guadeloupe (n = 5), Suriname (n = 2), Puerto Rico (n = 2) Bahamas, Grenada, Saba and Trinidad (1 each)-3 papers covered both Cuba and Barbados (S2 Table ). Studies that were population-based, measured body mass index by an objective standard protocol, adjusted for age or considered subjects in a narrow age range were assessed as having a low risk of bias. Men tended to have lower mean BMI values than women and this was true of all studies reporting this statistic in the adult population. Block et al [27] demonstrated that women in Grenada had higher mean BMI than men in each age group. Studies from Guadelope [47,48]and Suriname [49] examined mean BMI in adolescent populations and found similar mean BMI values in girls and boys. Of the 16 articles from 13 unique studies which provided prevalence rates for obesity, 8 were population-based and measured obesity objectively (Table 3). These eight studies all found higher rates of obesity in women compared to men and when meta-analysed, provided an odds ratio in favour of women of 3.10 (95% CI 2.43, 3.94)(Fig 4). There were 18 articles that examined gender and abdominal obesity as measured using waist circumference and/or waist-to-hip ratio. Eight of these articles were assessed as having high risk of bias, while the remaining ten consistently reported higher rates of abdominal obesity in women compared to men. Fifteen articles described the relationship between physical activity and gender(S1 Table ), all of which were assessed as having a high risk of bias as samples were either not populationbased (6) or obtained by convenience sampling (1) or measured physical activity by self-report via a questionnaire (15). All studies except one [46]reported higher levels of physical activity among men and/or higher levels of sedentary time among women. Men were more likely to smoke than women. There were 15 articles drawn from 9 studies which examined the relationship between gender and smoking. Six of these studies were population-based and were used in a meta-analysis using a random effects model to summarize the relationship between gender and smoking (Table 4). There was significant heterogeneity (p<unk>0.001) between studies (I 2 = 91%), and a random effects model was used. The odds of smoking among women was 0.24 times that of smoking among men, 95% CI (0.17, 0.34) (Fig 5). There were five articles in which one or more aspects of diet were assessed and described by gender. The most robust of these studies was conducted in Puerto Rico where researchers found a slightly higher percentage of men consumed at least 5 servings of fruit and vegetables per day than women (8.2% vs. 6.3%) but this difference was not significant (p = 0.23). In a population based study from Jamaica, Anderson [34] found that men had a higher total energy intake than women and similar percentage energy intakes from carbohydrates, while women 32<unk>8% (95% CI 32<unk>1, 33<unk>4) had slightly higher energy percentages from fat than men (31<unk>8% 95% CI 31<unk>0, 32<unk>7), and men had significantly higher percentages from protein. Block et al [27], who also conducted a population-based study, found that younger women in Grenada were less likely to eat 2 or more servings of fish per week than younger men but this gender difference was not seen in older ages. Men and women had similar patterns of fruit and vegetable consumption. A small health facility-based study conducted in Guadeloupe [46] found that men and women had similar percentage energy intakes from protein, lipid and carbohydrates. A university-based study in Puerto Rico [58]evaluated the adequacy of dietary elements that must be supplied in sufficient amounts to guarantee a 'healthy diet'. They found very little difference between men and women in the aspects of diet they assessed. We found 3 population-based cross-sectional studies that examined the relationship between gender and metabolic syndrome. Two of these from Grenada and Jamaica [27,36] showed that women had higher rates of metabolic syndrome while the third was done in Puerto Rico and found men 42.1% (36.5-47.7), to have slightly higher rates than women 36.4% (32.7-40.2). For only one of these studies (Grenada) [27] was the difference statistically significant (women had a metabolic syndrome prevalence of 36% versus 17% for men (P <unk>. 001)). The Jamaican study defined the metabolic syndrome using the International Consensus Criteria [64]while in both Grenada and Puerto Rico they used the NCEP-ATP III criteria [65]. Gender and Diabetes control and complications. There was a paucity of studies describing the distribution of diabetes control or complications by gender(S3 Table ). We found three articles [31,39,42] in which the outcome was diabetes control and all of which were classified as having a high risk of bias. Two of these studies used HbA1c while the third used fasting glucose as a measure of control. They each showed that women were more likely to have poor control or be uncontrolled compared to men. Only in one was statistical significance attained where men had significantly tighter glycaemic control than women(p = 0.0271) [42]. Only one study described the relationship between mortality and gender and found similar mortality rates for the two sexes for persons with type 1 diabetes [51]. --- Discussion --- Summary of the evidence We undertook a systematic review of the distribution by gender of diabetes, risk factors for type 2 diabetes, and diabetes control and complications in the Caribbean, based on studies published between 2007 and the end of 2013. The choice of 2007 as a start point was based on a major political initiative within the Caribbean to take action on chronic diseases [66], and this work is in part designed to help inform policy and research in support of this initiative. We found a statistically significant association between gender and diabetes in the Caribbean region, with women being at more than one and a half times greater risk than men. The prevalence and incidence data we examined did not distinguish between type 1 and type 2 diabetes, but as type 2 diabetes represents 90 to 95% of all diabetes [7,67] it is reasonable to conclude that the gender difference is largely based on differences in type 2 diabetes. Obesity is by far the strongest modifiable risk factor for type 2 diabetes [68], and this was three times higher in women. In addition, the data we found overwhelmingly suggested they were less physically active. Of the risk factors for type 2 diabetes that we found data on, the only one that was more common in men was tobacco smoking, where men were at four times greater risk. We found only three studies describing glycemic control by gender. In each of these studies a high risk of bias was found, thus we regarded the better control found in men as inconclusive., Overall, our findings suggest greater risk of diabetes, greater prevalence of diabetes, and potentially poorer diabetes control in women compared to men. The relationships we found between gender and diabetes prevalence, and between gender and obesity, were consistent across different settings in the Caribbean, older and younger adults, and based on studies with a low to medium risk of bias. The relationships were found in population based studies, with good response rates, and with objective measures e.g. fasting glucose, and measured height and weight. Our finding that women are at significantly greater risk of diabetes is in marked contrast to what is described in most populations, where the prevalence is either similar between the sexes or higher in men. Indeed evidence suggests that for a given level of obesity adult men are at greater risk of type 2 diabetes than women [8]. While the reasons for this are not fully understood they are likely to include a greater tendency to abdominal and hepatic fat deposition in men. This is likely to be part of the explanation as to why in many populations the prevalence of diabetes is similar between men and women despite higher levels of obesity in women [8]. Another factor that may increase a tendency in many populations to type 2 diabetes in men compared to women is a much higher prevalence of tobacco smoking in men [69]. Tobacco smoking is associated with a 40 to 50% increase in the risk of type 2 diabetes [70]. While smoking in the Caribbean is on average much higher in men than women, its prevalence in men is typically between 10 and 20%, lower than most world regions [69]. The only other world region in which an excess of diabetes in adult women compared to men has been described, as far as we are aware, is Southern Africa. A systematic review and meta-analysis of diabetes prevalence studies in sub-Saharan Africa found a male excess in diabetes prevalence in Central and Eastern Africa but a female excess in Southern Africa (OR 1.25, 95% CIs 1.09, 1.45) [71]. Estimates of the prevalence of obesity from the Global Burden of Disease study [72] suggest a modest excess (1.5% to 4.4%) of obesity in females compared to males in Central and Eastern Africa, but a much larger excess 26% (11.7% men vs 37.7% women), in Southern Africa, which is similar to what we have found in the Caribbean. We postulate therefore that this degree of excess is in part responsible for the higher prevalence of type 2 diabetes in women compared to men which we found. Finally, it is relevant to note that there is evidence that the strength of the relationship between increased BMI and the risk of type 2 diabetes differs between men and women. Thus, while the incidence (absolute risk) of type 2 diabetes tends to be higher in men than in women at all BMI levels, the relative risk of a raised BMI compared to a normal BMI tends to be greater in women than in men. For example, a systematic review and meta-analysis describes a pooled relative risk of 6.4 (95% CIs 5.6-8.2) in men for BMI of 30 and above compared to less than 25, and pooled relative risk of 12.4 (9.0-12.1) in women [73]. This seems partly to reflect the fact that at normal BMI levels (e.g. <unk> 25) the incidence of diabetes is substantially lower in women than in men [74]. While overweight/obesity is the strongest single risk factor for type 2 diabetes, other factors may also play an important role in the female excess we have found. In particular, physical inactivity, independent of its role in obesity [75], is a risk factor for type 2 diabetes, and we found evidence (albeit not suitable for meta-analysis) of higher levels of physical inactivity in women compared to men across several settings in the Caribbean. Our findings have two broad implications. The first is around understanding the gender distribution of type 2 diabetes within populations. We suggest that our findings provide some insight into the relative distribution of risk factors between genders, particular obesity, needed to produce a population level female excess of type 2 diabetes. In both the Caribbean and Southern Africa women are about three times more likely to be obese, with an absolute difference of 20 to 40% in prevalence. It may be speculated that the gender disparity in both regions is a manifestation of a shared cultural heritage as well as dietary and lifestyle practices. There has been a long held tradition in the Caribbean that obesity is healthy and that larger, more full bodied, women are preferred [76]. Some have speculated that this may be related to the former custom amongst some groups in Africa of hosting brides-to-be in fattening huts to please their future partner [76]. Studies conducted in Barbados on obesity attitudes and preferences indicated a change in male preference towards a normal or near-normal female figure but a continuing increase in prevalence and acceptance by females of obesity [77]. It is conceivable that with economic development other regions of Africa, and indeed other developing regions outside Africa, may also see a marked female obesity excess and that in these regions type 2 diabetes will be significantly more common in women than men. The other broad implication is around research and policy for the prevention of type 2 diabetes. In particular, research is needed to better understand why in some populations, such as in the Caribbean, women are at so much greater risk of obesity, and what can be done to reduce that risk. The mechanisms when elucidated are bound to be complex and may revolve around women's roles and responsibilities in society. Health promotion campaigns that are based on the research undertaken and mechanisms elucidated can then be appropriately conceptualized. Similar considerations apply to physical inactivity. Policy makers may need to consider measures that specifically target the physical activity needs of women, and further research is required to understand properly how different these needs are relative to those of men. Our findings suggest that the determinants of obesity and physical inactivity differ substantially between men and women in the Caribbean and that preventive policies must separately target and monitor their impact on men and women. --- Limitations The relationships we abstracted on diabetes prevalence, obesity, physical inactivity and smoking by gender are consistent in direction across studies and settings. Nonetheless the number of studies of low to medium risk of bias we identified was small. In addition, none of the studies were concerned specifically with examining the distribution of their findings by gender thus we are unable to comment on whether this finding of diabetes excess in women differed by ethnic group, education occupation or income within the Caribbean. For example, none of the studies we found adjusted for differences in risk factors or other social determinants between men and women in describing diabetes prevalence, or investigated potential underlying determinants for gender differences in obesity. While our search strategy was broad, and included through the Virtual Health Library databases that record some 'grey literature', such as government reports and PhD theses, we did not have the resources to undertake more extensive grey literature searching. --- Conclusions We have described in the Caribbean region a 50% excess risk of diabetes in women, associated with an approximate threefold excess risk of obesity, and consistently lower levels of physical activity. As far as we are aware no other world region has such an extreme female excess of diabetes, with the only other region showing a similar pattern being Southern Africa. It is reasonable to speculate that with economic development other developing world regions, such as other parts of Africa, may see similar differences between men and women. Our findings have major implications for research to understand the reasons for gendered differences in the risk of diabetes, and for policy aimed at its prevention. --- All data used in the manuscript of this systematic review has been obtained from already published data and as such is available.

Diabetes (DM) is estimated to affect 10-15% of the adult population in the Caribbean. Preventive efforts require population wide measures to address its social determinants. We undertook a systematic review to determine current knowledge about the social distribution of diabetes, its risk factors and major complications in the Caribbean. This paper describes our findings on the distribution by gender.We searched Medline, Embase and five databases through the Virtual Health Library, for Caribbean studies published between 2007 and 2013 that described the distribution by gender for: known risk factors for Type 2 DM, prevalence of DM, and DM control or complications. PRISMA guidance on reporting systematic reviews on health equity was followed. Only quantitative studies (n>50) were included; each was assessed for risk of bias. Metaanalyses were performed, where appropriate, on studies with a low or medium risk of bias, using random effects models.We found 50 articles from 27 studies, yielding 118 relationships between gender and the outcomes. Women were more likely to have DM, obesity, be less physically active but less likely to smoke. In meta-analyses of good quality population-based studies odds ratios for women vs. men for DM, obesity and smoking were: 1.65 (95% CI 1.43, 1.91), 3.10 (2.43, 3.94), and 0.24 (0.17, 0.34). Three studies found men more likely to have better glycaemic control but only one achieved statistical significance.

Introduction Poverty governance is a critical challenge faced by numerous developing countries worldwide. The United Nations' Sustainable Development Goals have prioritized "No Poverty" as the first goal, aiming to "End poverty in all its forms everywhere" by 2030 [1]. China is also committed to this goal. Over the last four decades, China has successfully implemented the "Targeted Poverty Alleviation Strategy," a campaign that has yielded remarkable results in poverty governance and made significant contributions to global poverty reduction efforts. This strategy has effectively counteracted the adverse impact of economic growth decline and income inequality on poverty. As per current standards, an impressive 98.99 million rural individuals have been lifted out of poverty in China. Additionally, 832 poverty-stricken counties and 128,000 impoverished villages have been successfully uplifted. China has effectively tackled overall regional poverty, and this accomplishment has substantially improved people's capacity to work and thrive. Nevertheless, even with the objective poverty issue addressed, subjective poverty has emerged as a prominent social problem in China's pursuit of constructing a comprehensive socialist modern state [2]. Therefore, it becomes crucial to devise follow-up plans and measures to sustain the achievements made in poverty elimination. In a similar context, Mahmood et al. discovered discrepancies between objective and subjective poverty in Pakistan, attributable to various influencing factors [3]. Understanding and addressing such nuances are pivotal for crafting effective poverty governance strategies. Subjective poverty refers to individuals' self-perception of poverty, representing their personal assessment of living conditions rather than an external determination. The concept of subjective poverty arose as a rethinking of the objective poverty line in academic history, with Dutch scholars pioneering subjective poverty line measurement in the 1970s. It is important to note that subjective poverty and objective poverty are not equivalent. Koczan proposed a dual definition of poverty: one is objective economic poverty, where income is insufficient to cover daily expenses, and the other is psychological subjective poverty [4]. The subjective evaluation of poverty takes into account various factors such as property and income, access to education, pension and medical security, and participation in decision-making processes. In contrast to objective poverty, subjective poverty offers a more comprehensive assessment, considering an individual's overall attributes and values. This includes residents' perceptions of income distribution, the allocation of educational resources, and the effectiveness of social security systems. Subjective poverty also encompasses the overall evaluation of policy implementation. In China, the notion of equal rights in income distribution holds significant value. The Chinese government has taken significant steps toward improving the welfare of its citizens. Their efforts have focused on promoting equal educational opportunities, providing comprehensive social security coverage, and enhancing the overall sense of acquisition, security, and happiness among residents. As pointed out by Koczan, addressing both residents' objective economic poverty and psychologically subjective poverty is crucial to minimizing resistance to reform and development [4]. Consequently, beyond solving the issue of objective poverty, eradicating subjective poverty has become a major political agenda in China after 2020. To achieve this, we utilized data from the Chinese General Social Survey in 2019 to study the influencing factors of subjective poverty. This research is instrumental in assisting the government in timely policy adjustments and facilitating social and economic reforms. --- Literature review According to the existing research results, most of the literature discusses the impact of differences in household micro-features on subjective poverty from multiple aspects, such as population, economy, and policy. Scholars have found that demographic characteristics, including family consumption, household size, marital status, gender differences, educational background, and ancestry, are significantly related to subjective poverty [2,3,5]. Na <unk>ndori [6] analyzes subjective poverty in Hungary and finds that Roma descent, entitlement to social support, and unemployment affect subjective poverty. Alem et al. [7] find that families with a history of poverty will continue to think they are in poverty even if their material consumption increases. Mahmood et al. [3] use Pakistan Panel Household Survey 2010 data, and they find that the determinants of subjective poverty are not limited to household consumption but include physical health, food security, sanitation facility, household size, household demographic structure, and agriculture land ownership. The economic aspect mainly involves the relationship between subjective poverty and factors such as socioeconomic background, income level, household disposable income, and employment status [8]. It has been found that all of these factors significantly affect the occurrence of subjective poverty directly or indirectly [1,3]. Joyce and Ziliak [9] compare the causes of relative poverty in Britain and the United States and find that relative poverty in Britain is related to working households, while low-skilled is the main cause of relative poverty in the United States. Some scholars have also studied the relationship between policy and poverty. Li and Li [10] find that the anti-poverty relocation and settlement program affects subjective well-being by affecting the probability that people will be exposed to risks due to policy. Agasisti et al. [11] collected relevant data from European countries between 2006 and 2015 and found a significant correlation between education and poverty. Zheng and Wang [12] discovered that social pension insurance can effectively reduce poverty among 95% of rural elderly individuals. In summary, the papers and opinions discussed in this context offer valuable insights into understanding subjective poverty. However, certain research gaps still exist. Many studies primarily focus on the impact of individual factors on subjective poverty, overlooking the interconnectedness between various factors and their combined effect on subjective poverty. Additionally, there is a dearth of research exploring the influence of livelihood capital on subjective poverty. As subjective poverty is more comprehensive than objective poverty, it eventually encompasses multidimensional aspects of poverty. To contribute to the theoretical research in this area, our study aims to analyze subjective perceptions of poverty in China. Specifically, we employ logistic regression and the "self-rating scale" method, using public services supply, subsistence capital, and livelihood strategies as the measurement dimensions to investigate the impact of subjective poverty in this region. The results of this research could assist the government in making timely policy adjustments and promoting social and economic reforms. Moreover, it serves as a valuable addition to empirical research on subjective poverty, offering a more comprehensive understanding of the complexities involved in assessing and addressing poverty from a subjective perspective. --- Theoretical basis and research hypothesis The concept of "sustainable livelihoods" has its origins in the expansion and reflections of antipoverty research by eminent scholars such as Amartya Sen [13]. Towards the end of the 20th century, Chambers and Conway introduced the term "livelihood" for the first time, defining it as a comprehensive concept encompassing an individual's ability to earn a living, the assets they possess, the external support available to them, and the actions they undertake to generate income for long-term improvements in their production and living conditions [14]. This concept highlights that at its core, livelihood involves "the action and process of converting livelihood capital through the exchange of various forms of capital" [3]. Capital conversion aims to foster sustainable and continuous livelihood expansion. Having abundant livelihood capital can lead to improved personal production and living standards, bolstering one's ability to cope with risks and breaking the cycle of intergenerational poverty transmission. Furthermore, different types of livelihood capital can be transformed into each other, giving rise to various welfare states depending on how they are utilized and exchanged. This interplay of livelihood capital forms the basis for creating resilient and thriving livelihoods. In the report "Our Common Future", Brundtland first introduced the concept of sustainable livelihoods during the 1987 United Nations Conference on Environment and Development. In the 1990s, this concept underwent further development and clarification by Western scholars, leading to the emergence of the sustainable livelihoods approach. International agencies such as the United Nations Development Programme, the Department for International Development, and the Food and Agriculture Organization of the United Nations adopted this approach as a method to comprehensively understand livelihoods, assess the influence of livelihood strategies on poverty, and identify appropriate interventions to address povertyrelated challenges. Among the various frameworks proposed for sustainable livelihoods, the one put forth by the Department for International Development has gained widespread acceptance and is most commonly used in the Sustainable Livelihoods Guide [15]. We will utilize this framework to explore the interactions between four types of public services supply, livelihood capital, and livelihood strategies, which include physical capital, human capital, social capital, and financial capital. By employing this theory, we aim to shed light on how individual livelihood systems impact subjective poverty, providing valuable insights for addressing poverty-related issues and promoting sustainable development 1. --- Public services supply and subjective poverty Appropriate public service supply can play a crucial role in addressing the issue of subjective poverty to a certain extent. In the 1970s and 1980s, Western social sciences underwent a "rediscovery" of the significance and role of institutional analysis in explaining practical problems, leading to the development of the new institutionalist analysis paradigm. At the heart of new institutionalism lies the concept of a "system," which encompasses the various rules that govern human social life, including both formal and informal rules, as well as operational mechanisms. In the context of poverty governance, the system serves as a transformative mechanism through which the public sector designs and integrates various resources to achieve rational poverty alleviation. Public services supply is an integral part of this institutional arrangement and policy practice. It is directly linked to the redistribution process of public resources. However, when the public services supply is insufficient and unbalanced, it results in a relative lack of public welfare rights. This, in turn, becomes a significant obstacle to promoting the flow of resources, enhancing development capabilities, and ultimately eradicating poverty. Addressing the challenges in public services supply and ensuring a more equitable distribution of resources is essential to effectively combat subjective poverty and facilitate overall socioeconomic development. By strengthening public services and making them more accessible to all segments of society, governments can play a critical role in reducing poverty and enhancing the well-being of their citizens. Basic public services play a crucial role in empowering the poor to overcome internal and external challenges. One of the significant ways in which these services help is by assisting the poor in finding employment opportunities, thereby expanding their capabilities and possibilities. By providing essential support, public services can greatly encourage and inspire the impoverished population, enabling them to break free from the shackles of poverty. Hypothesis 1: The more perfect the government's public services supply is, the better able it is to reduce the occurrence of subjective poverty. --- Livelihood capital and subjective poverty The main manifestation of subjective poverty is the lack of livelihood capital, which encompasses a full or partial shortage of physical capital, financial capital, human capital, and social capital [16]. Currently, poverty measurement primarily focuses on the amount of physical and financial capital, as they form the foundation of livelihood capital and influence the acquisition of other forms of capital. The scarcity of human capital is evident at both regional and individual levels. The rapid pace of industrialization and urbanization has led to the migration of many skilled laborers to big cities, leaving rural areas or underdeveloped cities with a limited number of low-skilled laborers. Combined with various factors like environmental conditions and resource availability, this situation hinders fundamental internal motivation for economic development and makes certain groups susceptible to subjective poverty. Furthermore, social capital also plays a role in perpetuating subjective poverty across generations. Studies have indicated that individual social capital is significantly and negatively correlated with the intergenerational transmission of poverty. As a result, the stock of livelihood capital serves as an observable indication of residents' subjective poverty. Families with higher levels of livelihood capital tend to exhibit greater livelihood adaptability and are less prone to subjective poverty. Conversely, families with lower livelihood capital have reduced livelihood adaptability and are more susceptible to subjective poverty. Livelihood strategies play a crucial role in achieving the family's livelihood goals by effectively combining and utilizing various forms of capital. Scholars generally agree that livelihood strategies, especially non-agricultural ones, significantly contribute to breaking the cycle of poverty and fostering sustainable development. Physical capital and subjective poverty. The interpretation of poverty from both physical and economic perspectives has been a topic of enduring interest among researchers. Physical capital, encompassing the infrastructure and production tools necessary for sustaining livelihoods, constitutes a vital aspect of multidimensional poverty. It is predominantly reflected in financial deficiencies concerning income, consumption, access to food, clothing, housing, and transportation [17,18]. Poverty is a comprehensive term that encompasses economic, social, and cultural backwardness resulting from low income, a lack of essential physical goods and services required for life, and limited opportunities and resources for development. Economic factors are the fundamental drivers of poverty, making economic growth a powerful force in reducing both income and non-income poverty. From this viewpoint, physical poverty denotes a state of "survival poverty," where individuals lack the income and resources necessary to meet basic life needs. Hypothesis 2: The more abundant physical capital is, the more subjective poverty can be reduced. Human capital and subjective poverty. Human capital refers to the skills, knowledge, employability, and health that individuals possess when pursuing various livelihood strategies and striving to achieve their livelihood goals. From a macro perspective, labor mobility facilitates the efficient spatial and regional allocation of labor and its associated components, thereby reducing the relative likelihood of falling into poverty [19]. At the micro level, human capital significantly influences individual employment, income, and career advancement, directly impacting their overall welfare. The absence of human capital, which includes education level, labor skills, and health status, not only indicates relative poverty but also serves as a fundamental cause of such poverty. Enhancing the stock of human capital can play a pivotal role in helping the poor escape poverty [20,21]. It can be asserted that human capital serves as an intrinsic strength for the sustainable development of impoverished families, fostering improvement in various aspects of livelihood capital. Amartya Sen has aptly pointed out that improved education and healthcare can directly enhance the quality of life and increase individuals' capacity to earn income, ultimately breaking the cycle of income poverty [22]. The provision of universal education and healthcare can significantly benefit the poor by offering them greater opportunities to overcome poverty. Hypothesis 3: More sufficient human capital can reduce the occurrence of subjective poverty. Social capital and subjective poverty. Social capital serves as an essential complement to traditional forms of capital, as it involves the characteristics of social organization that promote cooperative behavior, thereby enhancing societal efficiency [23]. At the individual level, social capital can occupy strategic network positions, allowing actors to acquire and utilize resources embedded in their social relations, particularly those obtained from significant organizational positions within the social network. Social capital can be categorized into two types: structural social capital, which relates to social organizations, networks, and informal institutional arrangements, and cognitive social capital, encompassing various norms, trust, and cultural models prevalent in society. Numerous studies have underscored the significance of social capital, including social networks, trust, and norms, in poverty reduction. Social capital plays a crucial role in influencing family poverty and significantly impacts individuals' subjective perceptions of poverty [24,25]. It can partially compensate for the limitations of human capital and financial capital. Moreover, there exists a mutually beneficial interactive relationship between different forms of capital. Physical and human capital aid individuals in expanding their social network, altering the nature of their relationships, and building a new type of social capital that facilitates the achievement of their goals. In turn, social capital acts as a lubricant, providing a conducive environment for optimal performance of physical and human capital. Financial capital and subjective poverty. Financial capital refers to the financial resources necessary for farmers to achieve their livelihood goals [26,27]. It encompasses various wealth management products, including personal deposits in domestic and foreign currencies, treasury bonds, funds, securities, collective wealth, bank wealth management products, third-party depository deposits, insurance, gold and gold deposits, as well as collective fund trust plans and other wealth management products. Financial poverty, an essential aspect of multidimensional poverty, primarily manifests as a lack of sufficient funds. In response to the challenge of poverty, the Chinese government has introduced policies centered around the "trinity" big poverty-alleviation pattern, which emphasizes the integration of multiple strengths, measures, and support mechanisms. This approach involves special poverty alleviation initiatives, industrial poverty alleviation efforts, and social poverty alleviation programs. As such, the financial industry must play a pivotal role in fostering precise poverty alleviation, charting a hopeful path from poverty to prosperity, and contributing its financial strength to society's poverty alleviation endeavors. Poverty alleviation work extends beyond merely addressing basic needs like food and clothing; it also involves nurturing aspirations and wisdom. Financial poverty alleviation plays a significant role in comprehensive poverty alleviation efforts, focusing on cultivating the financial investment concepts of impoverished households to ensure multiple sources of economic income. Hypothesis 5: The increase of sufficient financial capital can reduce the occurrence of subjective poverty. --- Livelihood strategies and subjective poverty To achieve livelihood goals and attain positive livelihood outcomes, individuals employ livelihood strategies that involve combining and utilizing their owned livelihood assets. These strategies encompass a range of activities, such as production endeavors, investment plans, and fertility arrangements. The structure and quality of livelihood capital play a crucial role in determining the capacity of impoverished groups to adopt and implement livelihood strategies effectively. The tangible results of these strategies are evident in the benefits accrued from livelihood capital and the level of poverty risk faced by individuals. When individuals possess a higher level of diverse livelihood capital and employ various livelihood strategies, they can engage in both agricultural and non-agricultural activities, such as being chefs, welders, cleaners, or pursuing other non-agricultural vocations. This increased diversity in livelihood activities strengthens their ability to withstand risks [28]. As a result, individuals with varying levels of livelihood capital and involved in diverse livelihood activities exhibit diverse outcomes, leading to the formation of differentiated livelihood strategies. The adoption of such diversified strategies is instrumental in reducing relative poverty. In essence, when an individual possesses a low total amount of livelihood capital and limited options for livelihood strategies, they may face a higher likelihood of subjective poverty. The ability to choose freely between multiple strategies becomes constrained in such circumstances, which may hinder their efforts to improve their living conditions and overall well-being. Hypothesis 6: More diverse livelihood strategies can reduce the occurrence of subjective poverty. --- Data, variables and methods --- Data source This paper employs data from the Chinese General Social Survey (CGSS), which the Renmin University of China implemented. The questionnaire covers basic personal information, family production and life, quality of life, social security, social participation, and other items. It adopts multistage stratified sampling, which is currently recognized as representative data with scientific research value in academia. To ensure the quality of the data and the accuracy of the statistical analysis, we eliminated samples with incomplete variables and some questions in the original answer, for example, "know," "not applicable," and "refuse to answer" were deemed invalid and then deleted. Finally, a valid sample size of 2,554 copies was obtained. The remaining sample size ensures that 30 provinces in China have a certain number of samples. --- Variables Dependent variable. Various studies have explored different methods for measuring subjective poverty, aiming to bridge the gap between objective poverty measurements and the subjective experiences of residents [3] Since individuals have the most insightful information about their circumstances, they are best suited to judge whether they are in a state of poverty [1]. Researchers have identified two primary approaches to measuring subjective poverty based on its definition. The first approach involves collecting residents' subjective information and quantifying the data using specific questionnaires. Respondents are asked about their minimum living standards, which helps establish a subjective income poverty line for defining subjective poverty. However, this methodology often limits poverty measurement to income indicators, leading to a one-sided assessment [21]. The second approach is based on the "selfrating scale," inspired by the work of psychologist Cantril. This method gauges residents' satisfaction with their economic life using a graded measurement system [26]. Residents' self-determined evaluation results are then used to assess poverty levels among individuals. The results not only measure poverty and subjective well-being but also reflect residents' overall feelings about their living conditions and happiness. This second approach has become the mainstream method in empirical analysis. In this paper, we adopt the second approach for measuring subjective poverty. Specifically, we use the question "Which level do you think your socioeconomic status belongs to in the local area?" as the basis for measurement. The questionnaire offers a range of responses, including "up," "middle up," "middle," "middle down," and "down," which are graded with values of 1 to 5, respectively. Among these, "down" is identified as the indicator of subjective poverty and is assigned a value of 1, while the other responses are assigned a value of 2. Independent variables. The core of public services supply lies in the authoritative and legal allocation of resources. The unequal distribution of resources and elements among various regions has exacerbated the occurrence of poverty among individuals. Evaluating the effectiveness of public services supply can be observed through individual behavioral choices under subjective consciousness, such as institutional trust and policy cognition. In this study, the public service satisfaction indicator is used to measure the status of the system's supply. It is important to note that the CGSS questionnaire measures individuals' subjective evaluation of the government's overall satisfaction with providing public services across 14 different aspects. These aspects include medical and health services, social security services, environmental protection, safeguarding political rights, maintaining social security, combating corruption, ensuring fairness in law enforcement, promoting economic development, expanding employment opportunities, providing open government information, improving service consciousness, safeguarding education equity, ensuring food and drug safety, and overall work and services. To ensure the reliability of the measurements, this paper conducts a reliability analysis. According to Table 1, the results show that the alpha value was 0.9325, and the alpha coefficient was greater than 0.6. It can be seen that this group of items has high reliability. We use the average satisfaction of the 14 aspects to measure public services supply status. Physical capital plays a crucial role as a fundamental basis for poverty alleviation among the poor. It includes essential components such as infrastructure and production tools that fulfill their livelihood needs. However, due to the lack of viable investment channels and the limitations of their own capabilities, converting funds into physical capital poses significant challenges for the poor. In light of this, this article employs the question "Do you own real estate?" as a measure of the state of physical capital. Financial capital represents the financial resources that individuals possess to pursue their life goals. As the economic foundation determines the overall superstructure, financial capital wields a decisive influence on residents' ability to thrive. Hence, it becomes imperative to proactively explore the involvement of financial funds in poverty alleviation and development, leveraging the potential impact of financial capital. In this study, "family income" is utilized as a measure of the endowment of financial capital. Among various forms of livelihood capital, human capital holds a more central and crucial role. The human capital perspective primarily emphasizes the influence of economically valuable knowledge, skills, and physical abilities possessed by individuals in poverty. In this study, "education degree" is employed as the measure of human capital. Social capital serves as a crucial complement to physical capital, encompassing the social resources and network of relationships that individuals can utilize in their livelihood activities. In this study, the question "What online social groups or circles have you joined in the past two years?" is employed as an indicator to measure social capital. It is important to note that the CGSS questionnaire includes the measurement of trust in various types of subjects. To ensure the reliability of the measurement, this article conducts a reliability analysis. The results from Table 2 reveal an alpha value of 0.6737, with the alpha coefficient exceeding 0.6, indicating high reliability for this group of items. The livelihood strategy encompasses an individual's integrated utilization of livelihood capital, serving as a means of converting and leveraging the potential benefits of livelihood capital. The effectiveness of livelihood capital is closely tied to the adoption and implementation of livelihood strategies. When individuals have a wider range of diverse livelihood strategies at their disposal, it is more likely that their livelihood situation will improve. In this study, the "working way" is employed as an indicator to measure the choice of livelihood strategy and a detailed description of the variables can be found in Table 3. --- Model Logistic regression analysis is mainly used to form a multiple linear regression relationship between a dependent variable and multiple independent variables and to predict the probability of an event. The advantage of logistic regression is that the independent variables can be continuous or discrete, and there is no need to satisfy a normal distribution, because in the general multivariate statistical analysis model, the variables must satisfy the normal distribution. In logistic regression analysis, the dependent variable Y is a binary variable with values Y = 1 and Y = 0, representing subjective poverty and non-subjective poverty, respectively. The n independent variables that affect the value of Y are |X 1, X 2,, and X n, and the conditional probability of subjective poverty occurrence under the action of n independent variables is P 1<unk>4 P<unk>Y 1<unk>4 1jX 1 ; X 2 ; ; X n <unk>. The logistic regression model can be expressed as follows: Z i 1<unk>4 a o <unk> a 1 X i1 <unk>a 2 X i2 <unk> <unk> a n X in<unk>1<unk> P i 1<unk>4 1 1 <unk> exp<unk> z i <unk>2<unk> Z i refers to the intermediate variable parameters, a 0 refers to the regression constants, a 1 refers the regression coefficient of the j variable (i,j,,n), X ij express the value of the i unit, and the value of the j variable, if subjective poverty exists a value of 1, otherwise a value of 0, and P i refers the regression-prediction value of subjective poverty's probability of occurrence in the i variable, i = (1,2,,n). --- Empirical results and analysis --- Analysis of spatial results Before conducting the empirical analysis, we aim to visually present the provincial spatial characteristics of subjective poverty incidence, public service supply, and livelihood capital in China for the year 2019. To achieve this, we employ ArcMap to create a spatial distribution map that covers all 30 provinces in the country. To classify the variable values, we utilize the natural breakpoint hierarchy, which divides each variable into three categories: high-value area, medium-value area, and low-value area. For the incidence of subjective poverty, areas with a higher incidence represent a higher probability of subjective poverty occurrence. Regarding public service supply, higher values indicate that residents are less satisfied with the public services provided by the government. For physical capital, a higher value represents more house ownership. For financial capital, a higher value signifies higher household income. For human capital, a higher value denotes higher educational attainment. Finally, for social capital, a higher value indicates stronger social connections. Based on the findings presented in Fig 1, it is evident that the high-value areas of subjective poverty incidence in China in 2019 were primarily concentrated in Heilongjiang, Jilin, and Shanxi. Meanwhile, Hebei, Chongqing, and Yunnan were categorized as medium-value areas in terms of subjective poverty incidence. On the other hand, the low-value areas, indicating a lower incidence of subjective poverty, were primarily observed in Fujian, Guangdong, Hubei, and Shanghai. Based on the information presented in Fig 2, it is evident that the high-value areas of public service supply in China in 2019 were predominantly found in Liaoning, Jilin, Inner Mongolia, and Shanxi. Meanwhile, Hebei, Henan, Chongqing, and Yunnan were categorized as mediumvalue areas in terms of public service supply. On the other hand, the low-value areas, indicating lower levels of public service supply, were observed in Tibet and Qinghai. Based on Fig 3, we discover that Jiangsu, Zhejiang, and Inner Mongolia were the high-value areas of physical capital in China in 2019. The medium-value areas mainly occurred in Henan, Hebei, Chongqing, and Yunnan. The low-value areas primarily occurred in Heilongjiang, Liaoning, and Ningxia. From Fig 4, we know that the high-value areas of financial capital in China in 2019 mainly occurred in Jiangsu, Zhejiang, Anhui, and Shanghai. The medium-value areas occurred in Henan, Hebei, Chongqing, and Yunnan. Jilin, Gansu, and Tibet were low-value areas. Based on the data presented in Fig 5, it is evident that Fujian, Zhejiang, and Shanghai were identified as the high-value areas of human capital in China in 2019. Meanwhile, the medium-value areas were primarily observed in Henan, Hebei, Chongqing, and Yunnan. On the other hand, the low-value areas, indicating lower levels of human capital, were predominantly found in Liaoning, Qinghai, and Tibet. Based on the data presented in Fig 6, it is evident that the high-value areas of social capital in China in 2019 were primarily concentrated in Jiangsu, Zhejiang, Fujian, and Shanghai. Meanwhile, the medium-value areas were observed in Henan, Hebei, Heilongjiang, and Shandong. On the other hand, the low-value areas, indicating lower levels of social capital, were predominantly found in regions such as Tibet. Upon comparing Figs 1-6, we have observed a significant overlap between the high, medium, and low-value areas of subjective poverty incidence and the high, medium, and lowvalue areas of public service supply and the four types of capital. Provinces and cities with higher subjective poverty incidence also tend to exhibit higher levels of public service dissatisfaction among residents and lower ownership of all four types of capital. These findings suggest a provincial spatial connection between subjective poverty incidence and public service supply, as well as physical capital, financial capital, human capital, and social capital. --- Correlation analysis Table 4 shows the correlation analysis of the Pearson correlation analysis results between the variables. The variable of subjective poverty is found to be significantly correlated with the variable of public services supply, as well as with physical capital, financial capital, human capital, social capital, and livelihood strategies, all at the 1% significance level. This preliminary analysis suggests that higher levels of public services supply and more diverse livelihood strategies are associated with a lower likelihood of subjective poverty. However, it should be noted that the variables of physical capital, financial capital, and social capital did not show significant correlations with the other variables. Apart from physical capital, financial capital, and social capital, all other variables demonstrated either positive or negative correlations at the 1% or 5% significance level. This indicates that they may have a strong influence on the occurrence of subjective poverty. To gain a deeper understanding of these relationships, regression analysis will be conducted on each variable. This analysis aims to explore the specific impacts of each factor on subjective poverty and identify potential areas for policy interventions and poverty alleviation efforts. By investigating these correlations, we can better comprehend the complex dynamics --- Benchmark regression To examine the presence of collinearity in the multiple linear regression model variables, we conducted a Variance Inflation Factor test. The test results revealed that the average VIF value was 1.33, which is below the critical value of 10. This indicates that there is no collinearity issue among the independent variables. Column (1) of Table 5 examines the relationship between subjective poverty and public services supply. The results indicate a significant negative effect of public services supply on the occurrence of subjective poverty at the 1% significance level, confirming the validity of Hypothesis 1. This suggests that a well-functioning and comprehensive government public services supply can effectively reduce the occurrence of subjective poverty. In Column (2), there is a negative correlation between livelihood strategies and the occurrence of subjective poverty, supporting the validity of Hypothesis 6. The findings demonstrate that diverse livelihood strategies, encompassing both agricultural and non-agricultural activities, can lead to increased income and a reduction in subjective poverty. Column (3) investigates the relationship between capital conversion and subjective poverty. It reveals that physical capital, financial capital, human capital, and social capital all have significant positive effects on subjective poverty at the 1% significance level, thus confirming the establishment of Hypotheses 2-5. The greater the abundance of physical capital, financial capital, human capital, and social capital, the more subjective poverty occurrence can be reduced. Furthermore, Column (4) presents the regression results after accounting for the control variables, and it verifies the robustness of the research findings. The conclusions drawn in the previous columns remain valid, indicating the consistency and reliability of the results. As presented in Table 5, Columns ( 5) and ( 6) present the regression results after controlling for urban-rural attributes. The findings reveal that public services supply, physical capital, financial capital, and human capital all have varying degrees of impact on the occurrence of subjective poverty among urban and rural residents at different levels of significance. These results are consistent with previous research findings by other scholars [20]. Specifically, public services supply, human capital, and physical capital have a more pronounced effect on the subjective poverty of urban residents compared to rural residents. On the other hand, financial capital has a greater impact on the subjective poverty of rural residents than urban residents. Additionally, social capital solely affects the occurrence of subjective poverty among urban residents. These observations suggest that the factors influencing subjective poverty among urban residents are more diverse and multifaceted, whereas rural communities are still heavily influenced by economic factors. It indicates that rural groups possess significantly lower economic capacity compared to their urban counterparts and face challenges related to weak social capital. This discrepancy may be attributed to the urban-rural divide in terms of economic development, access to resources, and opportunities for social connectivity. --- Robustness test Replaced the dependent variable. To ensure the robustness of our empirical findings, we employed a different variable as the explained variable in our analysis. We used the question "How to evaluate your family's economic situation?" from the questionnaire, with respondents providing a value on a scale from 1 to 10 to grade their family's economic situation. We then processed the data on subjective poverty accordingly and reassigned values to create a new variable. We categorized respondents who rated their family's economic situation as less than 5 points as experiencing subjective poverty, assigning them a value of 1 while assigning a value of 2 to the rest. The results obtained from this new analysis, as shown in Table 6, were found to be consistent with those in Table 5. Public services supply, physical capital, financial capital, human capital, social capital, and livelihood strategies still demonstrated significant effects on the occurrence of subjective poverty at varying levels of significance. This provides further evidence to support the validity and reliability of our model. Change model. To further validate the robustness of our empirical findings, we employed the Generalized Linear Model (GLM) as an alternative statistical approach, and the results are presented in Table 7. We are pleased to report that the outcomes obtained through the GLM model are consistent with those derived from the logistic model. Once again, we observe that public services supply, physical capital, financial capital, human capital, social capital, and livelihood strategy continue to exert significant effects on the occurrence of subjective poverty. The consistency between the results obtained from both the logistic and GLM models reinforces the credibility and reliability of our findings. It indicates that the impact of public services supply and various livelihood capitals on subjective poverty is robust and reliable across different statistical techniques. This strengthens the validity of our research and bolsters confidence in the relationships identified. --- Research conclusions and policy implications --- Research conclusions This study highlights the multidimensional nature of poverty, where economic poverty is just one aspect. Addressing poverty requires not only tackling economic challenges but also addressing the psychological and subjective aspects of poverty. Understanding the influencing factors of subjective poverty is crucial for the government to adjust policies effectively and promote social and economic reforms. The results demonstrate that public services supply has a significant negative effect on the occurrence of subjective poverty at the 1% significance level. Additionally, physical capital, financial capital, human capital, social capital, and livelihood strategy all have significant positive effects on the occurrence of subjective poverty at the 1% significance level. Specifically, for each one-unit change in public services supply, the probability of subjective poverty occurrence decreases by 0.0398. For each one-unit change in physical capital and financial capital, the probability of subjective poverty occurrence decreases by 0.149 and 0.0325, respectively. Similarly, for each one-unit change in human capital and social capital, the probability of subjective poverty occurrence decreases by 0.0244 and 0.107, respectively. The formation of subjective poverty is influenced by

Poverty is not only an economic problem but also a social problem, and there are certain limitations of objective poverty based on the population's income. It does not reflect the residents' true feelings regarding education opportunities, pension and medical security, and participation in decision-making. Researchers have studied it intensively in different objective dimensions of Chinese poverty, and little attention has been paid to subjective poverty. This study analyzes how public services supply, livelihood capital, and livelihood strategies affect subjective perceptions of poverty. The results show that public services supply, livelihood capital, and livelihood strategies significantly correlate with subjective poverty. Physical capital and social capital have the greatest effects on the occurrence of subjective poverty. The probability of subjective poverty decreases by 0.149 and 0.107 for each unit change in physical and social capital, respectively. What's more, public services supply, physical capital, financial capital, and human capital affect the subjective poverty of urban and rural residents at different significance levels. It means that the formation of subjective poverty results from the superposition of multiple factors.

of our research and bolsters confidence in the relationships identified. --- Research conclusions and policy implications --- Research conclusions This study highlights the multidimensional nature of poverty, where economic poverty is just one aspect. Addressing poverty requires not only tackling economic challenges but also addressing the psychological and subjective aspects of poverty. Understanding the influencing factors of subjective poverty is crucial for the government to adjust policies effectively and promote social and economic reforms. The results demonstrate that public services supply has a significant negative effect on the occurrence of subjective poverty at the 1% significance level. Additionally, physical capital, financial capital, human capital, social capital, and livelihood strategy all have significant positive effects on the occurrence of subjective poverty at the 1% significance level. Specifically, for each one-unit change in public services supply, the probability of subjective poverty occurrence decreases by 0.0398. For each one-unit change in physical capital and financial capital, the probability of subjective poverty occurrence decreases by 0.149 and 0.0325, respectively. Similarly, for each one-unit change in human capital and social capital, the probability of subjective poverty occurrence decreases by 0.0244 and 0.107, respectively. The formation of subjective poverty is influenced by a combination of multiple factors, --- Policy implications Governments should prioritize supply-side reforms and enhance the provision of public services to tackle poverty effectively. A key aspect is to reform the income distribution system and the fiscal and taxation system, ensuring a fair distribution of income. This entails strengthening income distribution controls and implementing policies to create a redistributive adjustment mechanism that utilizes taxation and social security measures, with a greater focus on directing benefits toward the poor. Fiscal policy should play a crucial role in supporting impoverished regions by increasing financial assistance and encouraging financial institutions to provide more credit to these areas, addressing the issue of limited capital accumulation. Coordination between social security and poverty reduction policies is paramount. Social security measures should be leveraged to alleviate subjective poverty, and a sustainable mechanism should be established to consolidate the achievements of poverty alleviation efforts. Additionally, policymakers must prioritize the elimination of subjective poverty and enhancing residents' well-being as key objectives of social security policies. To achieve these objectives, the government should adopt a rational time frame and spatial planning to effectively address subjective poverty in different regions and stages. Implementing pilot programs can help identify successful models, which can then be expanded gradually for broader impact. Governments must prioritize capital conversion as a key strategy in their poverty reduction efforts. Ensuring equal access to education should be a top priority, and this can be achieved by strengthening educational opportunities in impoverished regions and low-income families. To do so, the government can implement financial discounts and offer guarantees to encourage financial institutions to support education and increase targeted enrollment in underprivileged areas. Additionally, improving educational infrastructure and providing high-quality resources in areas with educational deficits will be crucial to leveling the playing field. Furthermore, vocational skills training should be normalized in impoverished areas, and the government should offer free vocational skills education to the poor. By equipping individuals with the necessary skills and knowledge, the government can empower them to become self-sufficient and break free from the cycle of poverty. Regulating the real estate market to boost investment in affordable housing is also essential. This will allow citizens to access education, employment, medical care, and housing, ultimately improving their overall quality of life. By enhancing the living conditions of the poor and providing them with access to essential services, governments can effectively reduce poverty and promote the well-being of all their citizens. These efforts will create a more inclusive and equitable society where everyone has the opportunity to thrive. Based on our findings, it is evident that the nature of work may not directly impact subjective poverty. However, employment remains a critical factor affecting income, which, in turn, influences subjective poverty. To address this issue, governments should prioritize the improvement of labor market institutions to enhance employment opportunities and participation rates. Moreover, governments should explore the potential of "Internet + livelihood strategies" to leverage modern information technology and the development of the tertiary industry. By doing so, they can expand livelihood strategies and options, stimulating economic opportunities for groups vulnerable to subjective poverty. Adopting a localized approach is also crucial for governments. By capitalizing on the unique strengths and resources of each area, they can develop characteristic industries and promote employment in those regions. --- All data files are available from the Chinese General Social Survey database (http://cgss.ruc.edu.cn/index.htm). --- Supporting information --- S1 Data. (ZIP) --- Author Contributions Data curation: Li Chen, Yuan Meng.

Poverty is not only an economic problem but also a social problem, and there are certain limitations of objective poverty based on the population's income. It does not reflect the residents' true feelings regarding education opportunities, pension and medical security, and participation in decision-making. Researchers have studied it intensively in different objective dimensions of Chinese poverty, and little attention has been paid to subjective poverty. This study analyzes how public services supply, livelihood capital, and livelihood strategies affect subjective perceptions of poverty. The results show that public services supply, livelihood capital, and livelihood strategies significantly correlate with subjective poverty. Physical capital and social capital have the greatest effects on the occurrence of subjective poverty. The probability of subjective poverty decreases by 0.149 and 0.107 for each unit change in physical and social capital, respectively. What's more, public services supply, physical capital, financial capital, and human capital affect the subjective poverty of urban and rural residents at different significance levels. It means that the formation of subjective poverty results from the superposition of multiple factors.

Background Early detection of emergent influenza strains is a global health priority. Environments where wildlife is in close contact with domestic animals and humans are considered hotspots for the emergence of potentially pathogenic influenza strains [1][2][3][4][5]. The 2009 H1N1 virus responsible for the most recent pandemic contained genetic material derived from both avian and swine viruses and zoonotic transmission of the highly pathogenic avian H5N1 to humans resulted in an abnormally high case fatality rate [6,7]. Notably, the H5N1 global spread was linked to movement of birds, both domestic poultry and migratory waterfowl [8][9][10][11]. Correspondingly, an increase in surveillance, especially in more vulnerable countries and populations where multiple animal species interact in close proximity to humans has been called for and, to varying degrees of success, implemented [12][13][14]. However maintaining active surveillance, especially in low income countries and communities where poultry and livestock are predominately raised in the absence of biosecurity and in close proximity to households, is both economically and logistically challenging given other pressing health and non-health priorities. An alternative approach is a community-based surveillance program in which the community itself is responsible for detecting changes in disease patterns in their livestock and poultry, understanding the risk of zoonotic transmission, and empowered to alert regional and national health authorities [15][16][17]. Community participation is essential for the design of efficient epidemiological surveillance programs. The systematic use of research methods applied to epidemiological programs facilitates the documentation of community understandings of disease and options for disease control. This type of research emphasizes community involvement in defining and prioritizing healthrelated problems, with the goal of improving detection, reporting, and response [18][19][20]. Participatory epidemiology [21] has been used to address zoonotic disease emergence, transmission, and control [22][23][24][25][26][27][28], including for avian influenza [16,[29][30][31]. As part of this, understanding how a community conceptualizes disease in their animals, assesses prevalence and categorizes specific diseases and patterns, and seeks both animal and human health guidance and treatment are essential. Communities along Guatemala's southern Pacific coast represent populations that are socioeconomically and epidemiologically vulnerable. Candelaria and Monterrico are the two main towns in the coast of the municipality of Taxisco, where 66 % of residents live in poverty and 18 % in extreme poverty [32], 44 % do not have access to safe sources of water or sanitary services [33], 25 % have no formal education, and 60 % only have some primary education [33,34]. Predominately rural, families raise domestic animals for household consumption or as an income source. Poultry (predominately chickens and ducks) and swine are free ranging with frequent interactions among these species and with family members. Even if maintained with informal enclosures, poultry and pigs are closely located to the house and linked to the household through free-ranging dogs and cats and a contaminated environment. These free range and high contact rate conditions provide the opportunity to introduce new pathogens or pathogen strains into different animal hosts and humans that would not occur in other ecosystems where potential hosts are physically separated. Specifically free ranging household swine may have a critical role as they have frequently served as intermediate hosts for genetic reassortment of avian influenza viruses, resulting in strains transmissible to and pathogenic for humans [1,3]. In addition to the role of free ranging household livestock, the Taxisco coast is a major overwintering zone for migratory aquatic birds: two natural reserves, Monterrico (2,800 ha) and Hawaii (3,657 ha) are large mangrove wetlands [35,36]. Due to the close and frequent interactions between aquatic migratory fowl, domestic poultry and swine, and humans, these communities are potential hot spots for emergence and spread of influenza. The Centro de Estudios en Salud de la Universidad del Valle de Guatemala, in collaboration with the U.S. Centers for Disease Control and Prevention, has conducted active surveillance for influenza in Taxisco, focused on these two communities directly bordering the Pacific Ocean, Monterrico and Candelaria [35]. Concomitantly, we initiated a study to identify opportunities for and barriers to community engagement in sustainable community-based surveillance. Using an ethnographic approach followed by a descriptive cross-sectional survey, we documented the characteristics of the main animal illnesses as defined by the community, as well as their signs, perceived causes, and consequent healthseeking behaviors. We report the findings of the study and present them in context of how community-based understanding of disease can be linked to improved public health reporting. --- Methods --- Research design The study consisted of two phases. The first phase was exploratory and aimed at identifying the repertoire of local terms, concepts and practices relevant to animal health. The ethnographic approach used in this phase included informal interviews and observations, followed by qualitative analysis [37,38]. The second phase was descriptive and aimed at documenting the extent to which the local terms, concepts and practices identified in the first phase were used by a representative sample of households. This phase used a cross-sectional survey and descriptive statistics [38]. Study design, data collection, and data analysis were all conducted in Spanish, the local language in the two villages. All authors and a research assistant are native speakers of Spanish, with the exception of GP, who is also fluent. --- Population and sample The study was carried out in two villages, Monterrico and Candelaria, surrounding the Chiquimulilla Channel wetlands in the Pacific shore of Guatemala. This region was selected because it is an overwintering destination for migratory birds and household livestock include pigs, ducks, and chickens. There are twelve villages in this region, where domestic animals are in contact with the migratory fowl. The two selected villages are the ones with health posts staffed by the Ministry of Human Health, therefore considered to have greater potential for the future implementation of an epidemiological surveillance pilot program. The ethnographic portion of the study used purposive sampling for selecting individuals with knowledge about animal illnesses and care (i.e. animal owners, healers), as well as places where animal care was delivered (i.e. backyards, kitchens). The cross-sectional survey was conducted in a sample of 90 households (42 in Monterrico and 48 in Candelaria) previously selected to randomly sample swine and duck populations as part of an active surveillance program. Monterrico and Candelaria have a total combined population of 961 inhabitants, and a total of 341 households according to a February 2013 census. --- Data collection An initial phase of ethnographic research was conducted between February and April 2013. This phase consisted of observations and informal interviews conducted by a single interviewer (MRO) oriented by an observation and interview guide [37] aimed at identifying the local terms for household animal illnesses, as well as understanding the local beliefs and practices related to each of those illnesses as associated with care seeking behaviors. Species of interest for this ethnographic phase were chickens, ducks, and swine. Data from observations and informal interviews were documented in notebooks by the interviewer and transcribed into Microsoft Word. MRO and AC cross-checked transcriptions for accuracy and qualitatively analyzed them to identify relevant local terms and concepts.. A subsequent phase of cross-sectional survey was conducted in June 2013. The questionnaire design was informed by the findings of the ethnographic phase, and one questionnaire was administered per household. The questionnaire consisted of close-ended questions organized in six sections. An initial section was aimed at establishing prevalence of illness in the previous 15 days. If illness was reported, subsequent sections investigated name of illness, signs, perceived causes, treatments, and help sought by animal caregivers. Species of interest for this survey were chickens, ducks and swine. The questionnaire was created as an electronic form installed in a tablet computer, checked for accuracy by a field supervisor, and uploaded to a database that generated an Excel spreadsheet used for analysis. The questionnaire was validated by interviewing 10 households with the aim of detecting comprehension problems and to assess if the questions responded to the research aims. --- Data analysis For the ethnographic phase, MRO and AC analyzed the original field notes and synthesized the findings into profile matrices for each of the reported illnesses. These matrices allowed organization of the information according to signs, perceived causes, treatments, and people/ places sought for help. Matrices identified knowledge gaps to be addressed through additional interviews, and provided means for better understanding each of the illnesses as well as making comparisons between illnesses. Survey data displayed in a Microsoft Excel spreadsheet were analyzed using the pivot tables feature for generating tables of frequencies/percentages and cross tabulation analysis [38]. --- Ethics, consent and permissions The study protocol was approved by the Universidad del Valle de Guatemala's Ethical Review Committee (protocol number 079-04-2013). All participants gave written informed consent. To preserve participants' anonymity, neither the fieldnotes nor the survey documented household or individual identifiers. --- Results --- Animal ownership A total of 55 informal interviews and 30 observations were documented through the ethnographic phase of the study, and a total of 90 households took part in the cross-sectional survey, with an average of five inhabitants per household (range 1-15). Ownership of animals, transportation assets, and education levels are provided in Table 1. --- Described illnesses, perceived causes, and treatments as identified by ethnographic research Illnesses of chickens, ducks, and swine were specifically described (Table 2). Chicken illnesses include two with predominantly respiratory signs, (coriza and soco, the latter of which was also described as pujo or moco); two with gastrointestinal signs (accidente por abajo and cólera), two with problems of locomotion (polio or apolio and calambres), one with cutaneous signs (viruela or buba), and one with non-specific signs (accidente por arriba). Participants reported that common illness signs include sadness (tristeza), droopy beak (pico ca<unk>do), inappetence (no come), and loss of feathers (plumas ca<unk>das). Illnesses are predominantly associated with changes in weather or climatic conditions, as well as with the presence of dust or dead animals. Treatments vary according to the predominant signs, with most illnesses treated with antibiotics and over-the-counter medications, some illnesses treated with local plants, and skin illnesses treated with topical procedures. Medications were procured primarily in local pharmacies and supply stores. Illnesses in ducks are described very similar to those of chickens (Table 2), with the exception of the mal de ojo (evil eye), which is described in very similar terms as the syndrome amply described in humans [39]. Swine illnesses also include the mal de ojo, together with additional illnesses that present a combination of systemic, cutaneous, and gastrointestinal signs (Table 2). Perceived causes of illness, treatments, and help sought are similar to those described for chickens and ducks. --- Described illnesses, perceived causes, and treatments as identified by cross-sectional survey The most common illness reported to have occurred in the 15 days prior to the day of the household survey was soco, found in 56 % of the households that raised chickens. Accidente was the most prevalent described illness in ducks (24 %) as was jiote in swine (50 %). Table 3 shows reported illness prevalence by species. Nonspecific signs such as sadness (tristeza), inappetence (no come) and prostration (echado) were the most prevalent reported for each species, accounting for between 50 and 75 % of the households. Importantly, these general signs were the ones reported as having being first noticed by survey respondents. This pattern remains when analyzing respiratory illnesses (soco, coriza) and gastrointestinal illnesses (colera, accidente por abajo), where non-specific signs were reported as being noticed initially and more frequently than syndrome defining signs. In contrast, for diseases of the skin, feather or cutaneous lesions were the first identified signs. Reported signs of illness are shown in Table 4. Perceived causes of illness were predominantly (60-80 % of reported illnesses) related to the physical environment and weather, such as temperatures warmer or cooler than normal, sudden changes in temperature, rain, and dust (Table 5). This pattern is even stronger for respiratory and gastrointestinal illnesses, where these climatic changes were perceived as the cause of up to 95 % of reported cases. Unburied dead animals were also reported as a perceived cause, but only in 20-35 % of cases. Skin illnesses were more often associated with non-weather related environmental contamination than were other illnesses but were also associated with dust due to changes in weather. Diseases in both chickens and ducks were overwhelmingly treated with antibiotics obtained without prescription at local shops or pharmacies. Tetracycline and amoxicillin were the most commonly used antibiotics (Table 6). This was particularly true for respiratory and gastrointestinal illnesses such as soco, where 83 % of cases were treated with tetracycline. The fewer number of reported disease cases in swine limited assessment of specific treatment frequency, however both antibiotics and changing the animal's diet were reported. The use of home remedies or medicinal plants was markedly less common. Treatment decisions were made in 90 % of cases by the household itself and in most of the remaining 10 % with the help of neighbors or relatives. Only two households reported having gone to a pharmacy or to a store specialized in animal products in the nearest city (Taxisco, 19 km from Monterrico, 22 km from Candelaria). --- Discussion Local terms and understandings of animal health problems pose challenges to the implementation of community-based surveillance programs because they highlight the knowledge and language gaps between institutional and community actors. The scientific underpinnings of infectious disease transmission and, especially, drivers for emergence of new zoonotic pathogens are unlikely to be well understood in communities with a low level of formal education unless translated into locally recognizable terms. This is reflected in the two rural communities involved in this study-over 75 % of the population lacks secondary or higher education. This is exacerbated by the very limited public health and animal health resources available to the communities. This gulf in scientific understanding of infectious disease concepts and the language used to express these concepts between the formal public health sector and the communities most likely to first encounter an emerging zoonotic pathogen is an impediment to implementing community-based surveillance. However, this impediment can be markedly reduced by building bridges between what the community observes and how these observations can be translated into actionable public health information. This way, the challenge posed by local understandings of animal health problems is transformed into an opportunity for public health practitioners implementing epidemiological surveillance. --- Early detection The community members clearly associate disease with changes in seasons and local weather patterns. This association is compatible with patterns of certain infectious diseases, including seasonal influenza. This is similarly relevant to the spread of new influenza viruses associated with seasonal migrations of wild waterfowl. Furthermore, the community members are observant in reporting initial, more general signs of disease, which strengthens their ability to detect early events. Understanding the local terminology and perceptions of animal health problems can aid public health practitioners in bridging the gaps that make it difficult to implement community-based surveillance programs. --- Diagnosis The public health community needs to understand diseases of epidemiologic priority in the context of endemic diseases as understood by local communities. Descriptive names may be linked to specific pathogens or, alternatively, be more generalized and encompass multiple pathogens unified only by initial presentation. Rabies for example, derived from the Latin for rage, has evolved from a syndromic description to represent a specific etiology that is commonly understood at both community and public health institutional levels. In contrast, "flu" denotes a specific etiology, influenza virus, at the public health level but may represent a variety of etiologies at the community level that have similar elements of clinical presentation. Providing diagnostic services focused on common illnesses in rural communities can define specific etiologies or sets of etiologies linked to the syndrome that would improve translation across educational divides, essentially focusing on education of public health professionals in the meaning of local terminologies rather than the reverse. By focusing diagnostic services on endemic diseases, the ability to detect a shift in disease signs, patterns, or affected species will be enhanced, a requirement for detection of a newly emergent pathogen. --- Treatment and prevention Importantly, diagnosis of disease must be accompanied by action, whether vaccination, treatment or changes in animal husbandry, in order to provide an incentive for reporting in community-based surveillance. Decisions about whether or not to treat an ill animal and with what were heavily influenced by neighbors rather than by animal health professionals. This in itself is unsurprising and is not necessarily linked to either level of education or socioeconomic status. For example, even in wealthy countries with available maternal health care services, expectant and new mothers seek and act on advice from experienced mothers among family and friends. However, there is also a need and opportunity for evidence-based information to be introduced into those familial and social networks [40]. This also applies to introducing the concept of infectious agents as a cause of disease. Attribution of disease causation only to climatic events, clearly not under the control of individuals or the community, inhibits the implementation of effective preventive measures, such as vaccination and animal husbandry. Use of the community engagement initiated during this study offers the opportunity to provide this information, especially to leaders within the community, and therefore take advantage of the existing community structure for dissemination. Limitations of this study include those known for cross-sectional studies, and reliance on self-reported knowledge and behaviors. Results are only representative of the two studied villages. --- Conclusions Community-based surveillance of diseases of high epidemiologic priority is necessary and feasible. An estimated 60 % of new human infectious diseases emerge from animals and the role of animals, especially swine, chickens, ducks, and migratory waterfowl, is well established for emergence and spread of new influenza viruses [5]. Although many zoonotic pathogens emerge in rural areas with a scarcity of either animal or human health services, they remain undetected until either affecting large numbers of people or animals or spreading to urban areas where detection capacity and capability exists. Enhanced community-based surveillance provides the opportunity to enable earlier detection and response by community-based recognition of a change in disease pattern and prompt notification of public health authorities. Understanding how communities view and express disease concepts, and engaging with them on their endemic disease problems, provides the opportunity to link their continuous observations and local knowledge to effect enhanced surveillance. Areas for future research include those aimed at designing participatory surveillance systems and measuring their effectiveness in disease control. --- Competing interests The author(s) declare that they have no competing interests. Authors' contributions AC, MRO, DA, and CC participated in study design. MRO and AC conducted data collection and data analysis for the ethnographic portion. AC conducted quantitative data analysis. AC, MRO, DA, GP, and CC participated in results interpretation. AC drafted this manuscript while GP contributed specific sections and provided critical feedback on multiple drafts. MRO, DA, and CC also provided feedback on multiple drafts. AC, MRO, DA, GP, and CC have all reviewed and approved of the final manuscript.

Background: Early detection of emergent influenza strains is a global health priority. However, maintaining active surveillance is economically and logistically challenging. While community-based surveillance is an attractive alternative, design and operation of an effective epidemiological surveillance program requires community engagement that can be linked to public health reporting and response. We report the results of a study in rural Guatemalan communities aimed at identifying opportunities for and barriers to community engagement in disease surveillance. Methods: Using an ethnographic approach followed by a descriptive cross-sectional survey, we documented local terms and ideas about animal illnesses, including the possibility of animal-human transmission. Results: The community perceived disease causation principally in terms of changes in the physical environment and weather and categorized illnesses using local terminology based on observable clinical signs. Knowledge about prevention and treatment was derived predominantly from local networks of family and friends without evidence of professionally-based knowledge being regularly introduced into the community. Conclusions: Bridging the divide between professional and community-based descriptive disease terminology, incorporating animal and human health responsiveness to common illnesses, and providing professional knowledge into the community-based networks were identified as addressable challenges to effective implementation of community-based surveillance.

Introduction Modern social networks allow users to organize their friends into groups -social circles on Google+, community pages on Facebook, and lists on Twitter. This functionality allows users to selectively interact with posted content, and more easily browse relevant information. However, grouping users manually is a time-consuming task, and it is thus desirable to automate the group-creation task. The task is formally known as the social-circle discovery problem (McAuley and Leskovec 2012). In this problem, the unit of analysis is the egonet, which is the local network formed by a user's friends and the friendship links between them. The goal is to discover the most meaningful clusters of users in the egonet. Performance is typically measured by comparing discovered clusters with ground-truth groupings that users have already created in their egonets. In this work, we study the social circle discovery problem in twitter, in which groups are known as lists. Any twitter Copyright c 2015, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. user can create a list by manually selecting a subset of twitter users and designating a list title. If a user makes a list public, then other users can follow that list, as well. It has been shown that lists are used to group users who share common expertise, have common interests, belong to the same group of friends, or are located in the same geographical region. Lists are also used to aggregate updates on breaking news stories (Morrison and Hayes 2012;Greene et al. 2012). Previous work on creating twitter lists automatically has achieved moderate success, with one method achieving F 1 score of 0.33, on a dataset collected for their study (Yang, McAuley, and Leskovec 2013), and another method achieving normalized mutual information of 0.27, on a different dataset (Morrison and Hayes 2012). Both of these methods rely on features derived from tweet content and from the structure of the follower graph. Our premise is that these methods ignore an important source of information: the millions of lists already created by twitter users. The information value of these lists is multifaceted; co-listing of two users suggests similarity between them, while list titles have been shown to indicate user interests (Kim et al. 2010). At the same time, existing lists cover a large portion of twitter users; in our dataset 87% of the users belongs to at least one list. On the other hand, note that the problem we consider is different than the listrecommendation problem (Rakesh et al. 2014), since our goal is to group the friends of the egonet of a user into meaningful lists, and not to recommend any relevant list to a user. Contributions. We introduce the use of existing twitter lists for "crowdsourcing" the task of automatically organizing the friends of a user into lists. We use the existing lists that overlap with each egonet as candidate lists. Since the number of such lists can exceed 1 million, we define an optimization problem for selecting k high-quality candidate lists based on an empirical analysis of list and egonet properties. We prove that the select-k problem is NP-hard, but the objective function is submodular, enabling a greedy algorithm that provides a good approximation of the optimal solution. To evaluate the performance of our method, we collect a dataset of 24 egonets with all necessary information. We compare our method with CESNA, a state-ofthe-art overlapping-clustering method (Yang, McAuley, and Leskovec 2013). We find that our method selects candidates with titles similar to the ground-truth and that results are onpar with the comparison method, while there is also room for improvement. --- Background Terminology: In Twitter, if user a follows user b, then a is a follower of b, and b is a friend of a. We work with the extended egonet, defined as the friends of the ego user, the members of the lists that the ego subscribes to or owns, and the follower relationships between them. We include such list members as the ego user indirectly follows those users via interaction with the lists. --- Data collection: We collect 24 extended egonets using the public twitter API. Ego users are selected using a random walk starting from Stephen Colbert, a prominent comedian with almost 7 million followers. A visited user is selected if (i) they subscribe to or own at least 3 lists and (ii) the size of the union of their friends and followers is less than 3 500 users. To help avoid collection of spam and business accounts we also ensure that the ratio of friends to followers, and vice versa, does not exceed 3:1. We manually check egonet pages in twitter to ensure that they are personal accounts. For every user in the egonet, we collect (i) list memberships and titles; (ii) hashtags and mentions from the 200 most recent tweets; and (iii) follower relationships between users in the egonet. In total the dataset includes 42 946 users, 4 181 698 follower relationships, and 25 699 928 list memberships. To evaluate the quality of our methods, we use all 135 lists that our ego users subscribe to or own that contain at least three members, henceforward referred to as ground-truth lists. --- Distinguishing features As we will see below, current methods based on tweet content and network structure do not achieve high scores on retrieving the ground-truth lists. Before introducing another algorithm to solve the task, we first study in detail the features that intuitively provide useful information. Density of lists: Previous work has found that users within lists are densely connected by follower relationships (Morrison and Hayes 2012). Besides density in the follower graph, we expect that being co-listed is a strong indicator of user relatedness. To test this hypothesis, we form the co-listed graph, where there is an edge between users if they already appear together in a list (excluding the considered list). We then examine the density of lists in the follower graph and the co-listed graph, relative to the density of the entire extended egonet. We measure density by |E| |V ||V -1|. The results are showed in Figure 2 (left and middle). We observe that most lists have high density in the follower graph, compared to the extended egonet. The same is true for the co-listed graph. We also observe that there are many ground-truth lists whose density is substantially lower than the density of the egonet. Obviously, an algorithm based only on density cannot identify such lists. The rightmost plot of Figure 2 shows follower and colisted density divided by the density of the egonet. We observe that they are correlated, but also that for some lists both ratios are below 1, indicating that neither is informative, as well as that there exist lists for which follower density is informative but not co-listed density and vice versa. Thus, existing lists made by other users provide information that cannot be found by looking only at the network structure. Topical vs. social lists: It has been argued that communities in social media platforms primarily form either around topical or around social relationships (Grabowicz et al. 2013). To gain further insight in the informativeness of network structure, tweet content, and existing lists, we compute the following features for pairs of users, both in lists and in egonets: Jaccard similarity of: (1) friends ( 2) followers (3) list memberships (4) list titles, and (5) hashtags and mentions in their tweets. Figure 3 presents ratios of pairs within lists vs. within egonets for each of these features. Each is informative, while we find that having many of the same friends indicates high relatedness, and being co-listed is informative even if both are part of many lists (Jaccard small but > 0). Finally, we analyze these features also for all ground-truth lists individually, some examples of which are given in Figure 4. --- Method Our method is a greedy algorithm that, given an egonet and a set of available lists, selects k of those lists that provide a good coverage of the egonet; "good" will be defined shortly. Motivated by the high density of the ground-truth lists, we apply a density-based quality metric. The method allows for list overlap, as it happens with ground-truth lists. Problem formulation: We are given a set of users V = <unk>v 1,..., v n <unk> in a target egonet, and a collection of subsets of users, or candidate lists, L = <unk>L 1,..., L m <unk>, L j <unk> V. These candidate lists are obtained by all existing lists that include at least three users in the egonet. Given a function s(•) that assigns a score to each v i, our goal is to select k lists that maximize the summed score over all users. The objective can be formally defined as follows: maximize f (L) = n v<unk>V s(v) such that |L| = k. We also define a quality function over the lists q : L <unk> [0, 1], such that q(L) measures to what degree L exhibits desirable properties. The score of an item s(v) is then defined as the quality of the best list in which v is a member, i.e., s(v) = max L v|L<unk>L q(L). (1) We experiment with four different q functions. Motivated by the analysis presented in the previous section, we define the quality of the list as the density of its members in the friendship or co-listed graph. For density we use both average degree q(L) = |E|/|V |, and normalized density q(L) = |E|/(|V ||V -1|). Algorithm: We can show that the problem defined above is NP-hard, while the objective function f is nondecreasing and submodular, as long as the function s is also nondecreasing and submodular. The latter fact is a consequence of the non-negativity of the density functions we consider and the max function in Equation (1). The proof follows the lines of the proof of submodularity for the uncapacitated facility location problem (Cornuejols, Fisher, and Nemhauser 1977). We omit the details for brevity. As a result of the monotonicity and the submodularity of f, the greedy algorithm can be shown to achieve a e-1 e approximation guarantee. --- Experiments Our problem definition takes as input an egonet, a set of candidate lists, and a number k, defined as the number of lists to be selected. We experiment with k <unk> <unk>3, 20<unk>, where 3 reflects a use-case scenario in which we want to select the three best lists for a user, and 20 is the maximum number of lists in any egonet in our dataset. As list candidates we input the existing lists to which at least 3 egonet users belong. Our method also takes as input the function q that specifies the quality of each candidate list. We experiment with four different settings, which we compactly denote by G/X/Y. G stands for Greedy. X <unk> <unk>F, C<unk> indicates whether we are using the friendship graph or the co-listed graph. Y <unk> <unk>N, A<unk> indicates whether we are using the normalised density or the average-degree density. To evaluate the predicted lists, we first find the linear assignment of ground truth lists to predicted lists that maximizes the summed F 1 score in each egonet. That is, given ground truth lists L GT and predicted lists L P, we find a map- f : L GT <unk> L P that maximizes L<unk>LGT F 1 (L, f (L)). We also calculate an upper bound on the F 1 score that can be achieved by using only existing lists as selection candidates. When k = 20 we select the existing list that maximizes the F 1 score averaged over ground truth lists. When k = 3 we select the 3 pairs of existing lists and ground truth lists that yield the highest F 1 score. This method is denoted as BEST in Table 1. For comparing to CESNA, we use the default parameters provided in the SNAP package. 1 Results: Our results are shown in Table 1 andFigure 5. The proposed method achieves a highest F 1 score of.22 when the greedy method is applied in conjunction with the normalized density of lists in the colisted graph. Normalized density achieves higher results than average degree in the co-listed graph, and vice versa in the follower graph. Results are on-par with the.21 achieved by CESNA. As seen in the far-right plot in Figure 5, CESNA dominates G/C/N on large lists. A qualitative comparison of the lists found with our method, against ground truth lists, is shown in Table 2, in terms of titles, sizes, and F 1 scores. --- Discussion We introduce a method for using existing lists on twitter for recommending informative groupings in a user's local network. The method is shown to be on-par with the state-ofthe-art overlapping clustering algorithm, while our analysis 1 http://snap.stanford.edu/snap/ demonstrates that there is significant room for improvement on the task. One shortcoming of the approach is its inability to create lists for users who do not already belong to a list outside of the ground-truth. A possible solution to this would be the application of a seed set expansion method.

Twitter and other social media provide the functionality of manually grouping users into lists. The goal is to enable selective viewing of content and easier information acquisition. However, creating lists manually requires significant time and effort. To mitigate this effort, a number of recent methods attempt to create lists automatically using content and/or network structure, but results are far from perfect. In this work, we study the power of the millions of lists that are already created by other twitter users in order to "crowdsource" the task of list creation. We find that in a large dataset, collected specifically for this study, an optimal matching of existing lists from other twitter users to the ground-truth lists in egonets gives an F1 score of 0.43, while the best existing method achieves only 0.21. We explore the informativeness of features derived from network structure, existing lists, and posted content. We observe that different types of features are informative for different lists, and introduce a simple algorithm for ranking candidate lists. The proposed algorithm outperforms existing methods, but still falls short of the optimal selection of existing lists.

Background The government health system in Vietnam Coverage of government primary health care (PHC) facilities has greatly improved in the past two decades in Vietnam. By the end of 2008, 99% of all communes (the local administrative unit in Vietnam) in the country had a communal health station (CHS), including the northern mountainous communes inhabited by ethnic minorities [1]. Communal health stations provide primary health care including free immunization for children, treatment for minor diseases and common drugs (including treatment of child diarrhoea and respiratory infections), assistance during deliveries and take part in health promotion activities in the communes including mother and child health education, and hygiene and sanitation education. This study was conducted in the Northern mountainous regions of Vietnam where 32% of CHSs having doctors and 86% having midwives or assistant doctors specialized in paediatric and obstetric care [2]. CHSs are small health stations situated in each commune, which refer seriously ill patients to larger inter-communal clinics and district hospitals, which provide facilities for surgeries and in-patient treatment as well as more diagnostic services (e.g. simple lab-tests etc.). Most (97%) villages in the Northern regions of Vietnam also have active Village Health Worker (VHWs) [2], who form a crucial part of the PHC system. They are laymen with short-course training in health and human diseases, typically three months of training plus occasional upgrade courses. They work part-time and have five official tasks; perform health education, disease prevention including hygiene promotion, mother and child health care including advocacy of family planning, first aid and disease treatment, and to participate in health programs with authorities and CHSs (e.g. inform caregivers about child immunization and give education to caregivers about child nutrition, care and hygiene) [3]. Private health providers are also a crucial part of the wider health system in Vietnam including a large numbers of traditional medical and herbal experts, spiritualists, and private bio-medical practitioners [4]. It has also become normal practice among Vietnamese to purchase and self-medicate with over-the-counter drugs, many sold without prescriptions, from the large number of private drugstores and pharmacies [5][6][7]. --- The health situation of ethnic minority children Improving child health indicators is a major priority for the Vietnamese government, who have initiated several child health programs and policies under the National Target Programs on Health since the early 1990'ties. Prevention of communicable diseases, improving coverage of the expanded child immunization program and improving child nutrition status are some of the main child health objectives. These programs all aim at providing basic child health services and promotion to all parts of the country. To ensure affordability and increase coverage of services, the National Health Care Fund for the Poor was established in 2002 to provide insurance for ethnic minorities and poor communities, making all treatment and drugs at CHSs and district hospitals free of charge for those insured [8]. However, child health inequalities still exist between rural and urban, and between majority and minority population groups. Diarrhoeal diseases are the number two leading cause of morbidity and is endemic in the northern mountainous regions of the country inhabited mainly by EMGs [9], as are parasitic infections of children [10]. It was recently estimated that 12% of children in a mountainous rural district of Vietnam had experienced diarrhoea within two weeks [11]. Children living in the northern and central highlands also experience higher risks of malnutrition compared with children living in urban and rural lowland [11,12]. A recent analysis of official statistics also highlights that far more ethnic minorities and children from the northern regions live in poverty, have poor shelter and limited coverage of water, sanitation and immunization compared with urban regions and Kinh children [13]. Caregivers' use of health care services and drugs for child illness management In principle, all caregivers in Vietnam have access to a well-established and free PHC system for children. And several studies have shown that self-medication of child illnesses is also common in urban as well as rural Vietnam [14] and that antibiotics are commonly prescribed, and overprescribed, to children [14][15][16]. One study in a Kinh populated rural district found that only 54% of caregivers used antibiotics and 36% used analgesic and antipyretics to treat the child's diarrhoea, while only 9.7% gave the child ORS [15]. Compared with the majority of Vietnamese (Kinh) parents, quantitative studies have found that ethnic minority caregivers are even less likely to give ORS and seek treatment at CHSs when children have diarrhoea [17,18]. One of these studies also found that ethnic minority caregivers perceived diarrhoea episodes as less severe compared with Kinh caregivers [18]. Indirect costs of treatment, user fees and long distances to health facilities have also been mentioned as important factors that hinder the use of health facilities among poor and the EMGs living in highland areas [19,20]. However, the perspectives of Vietnamese ethnic minority caregivers on child illness management and treatment preferences and perceptions are not well described in the literature. This study explores treatment seeking strategies among ethnic minority caregivers of children with diarrhoea in Northern Vietnam, in order to suggest improved health services for EMGs and other vulnerable groups in Vietnam. --- Methods --- Study area, population and health services Six months of fieldwork was conducted in two adjacent rural communes in the Northern Province of Lao Cai in Vietnam in 2008, investigating the general living and working conditions, perceptions and practices of health, hygiene and sanitation among ethnic minorities [21]. For the current study, and building on previous findings, two additional months of fieldwork was conducted in 2009 focusing on caregivers' management and perceptions of child diarrhoea. The total population of the two communes was 10,000 with 85% being ethnic minorities. Fieldwork was conducted among the four most populated minority groups in the area; Giáy, Tày, Xa Phó (living in lowland villages) and Red Dao (highland people). About 40% of households were below the government poverty line of 2006 -2010 for rural areas (defined by a monthly income below 200,000 Vietnamese Dong per person, approximating 10.5 USD), with a higher concentration of poor households among the Xa Phó (living on the edge of the forested hills) and the Red Dao in the highland [22]. All ethnic minorities depend on subsistence rice farming, animal husbandry and forestry with all adults doing hard physical fieldwork daily. All four ethnic groups have traditionally lived in multigenerational households and are all patrilineal, with men being the heads of households and owners of all property, and patrilocal with daughters moving to their groom's families when married. Further details on the study area and the minority groups have been reported earlier [21]. Two CHSs, one also hosting the Preventive Medicine Office in charge of conducting preventive activities in both communes, were located in the lowland part of each commune. Both CHSs were run by assistant doctors and had nurses, pharmacists and nurse midwives employed. Both CHSs had a selection of common drugs and remedies available including ORS, antibiotics, other anti-diarrhoeals and allopathic medicines. Village Health Workers had been appointed by the CHSs for almost all of the 39 villages in the communes. Government district and private hospitals and drug stores were located in the district town 3-5 kilometres away from the clinics. No private bio-medical practitioners, clinics or drug stores were found in the two communes. But several spiritualists (Thày Mo's) were identified as important health providers in highland villages, performing fortune telling and warding rituals to strengthen health. The lowland villages (Tày and Giáy) were located closest and within five kilometres to the CHSs. The Xa Phó villages were located within eight kilometres, while highland villages were scattered and located on steep dirt roads up in the mountains 10-15 kilometres away from the CHS. No public transportation exists and people walk, bike or drive motorbikes to the CHSs and district towns. --- Data Collection The main source of data was 43 semi-structured indepth interviews with caregivers of pre-school children (six years and below) who had experienced a case of diarrhoea in the past month. In addition, two weeks of observations of patient-staff interactions were performed at the two CHSs and three focus group discussions (FGDs) were conducted with men. Data from participant observations in four different study communities conducted in 2008 was included as general background information of the living conditions in highland and lowland settings (methodology has been described in details elsewhere [23]). Information about diarrhoea cases in 10 lowland and two highland villages (Table 1) of the four ethnic groups were collected by asking caregivers if their child had experienced an incidence of loose or watery stools at least three times per 24 hours during the past month, using the diarrhoea definition of WHO [24]. Several sampling methods were used to ensure that cases included mild and severe cases, and cases treated at home and by various health providers. These included two weeks of observations at the two CHSs (eight cases identified), a project community diarrhoea survey covering both communes (seven cases), by references from VHW's (nine cases), and house-to-house visits by the first author and assistants in villages (19 cases). Interviews included questions on perceived causes of diarrhoea, decision making about treatment, treatment choices, and use of health services (See additional file 1: Caregiver Interview Guide). The main informant was the adult caregiver who had attended to the child during sickness, but during seven of the interviews more than one caregiver was present and answering questions for their children. Attempts were made to avoid having grandparents in the same interviews as mothers to allow mothers to talk openly about family decision making. To gain in-sight into influences on treatment seeking and choices of fathers and grandfathers, three FGD's were conducted among men from two lowland villages (Tày and Xa Phó groups) and a highland village (Red Dao). Characteristics of participants are shown in Table 1. Interviews and FGDs were conducted in Vietnamese by two Vietnamese assistants with the first author being present during half of the interviews and all of the FGDs to ask elaborating and clarifying questions. Young local women assisted with translation during interviews in the highland and Xa Phó villages if the interviewees did not speak Vietnamese. Caregivers with sick children encountered at the CHSs and in households were not interviewed, but asked if they were willing to be interviewed at a later time once the child had recovered. Two caregivers with sick children met at the CHS declined to be contacted again. One caregiver with a sick child undergoing treatment was met in a Xa Phó village and interviewed at a later stage. All interviews were audio recorded with informed verbal consent and promised full anonymity and confidentiality. Recordings were transcribed, cross-checked and translated ad verbatim into English by the two assistants. Data was coded for thematic content analysis [25] by the first author using pre-set and emerging empiric and theoretical categories by the NVivo 8 software program [26]. The study was approved by the National Institute of Hygiene and Epidemiology in Hanoi, The Department of Health in Lao Cai province, and Peoples' Committees (government authorities) in the two study communes. --- Results Two ethnographic descriptions of two diarrhoea cases drawing on observations and interview data from various sources, illustrate the identified similarities and differences in perceptions of disease and health choices in a highland and lowland setting (Tables 2 and3). These cases serve as a point of departure for the analysis and discussion of the underlying logic for treatment seeking in ethnic minority families. The two cases show that the non-professional 'popular medical sector' constituted by health care taking place within the spheres of family and community [27], plays a crucial role in the treatment of sick children among ethnic minorities of northern Vietnam. It is within this popular sector that all treatment is initiated and where families negotiate whether to contact the professional sector, such as government health facilities, or the 'folk sector' with other health specialists, such as spiritualists or herbal experts [27]. Secondly, the cases depict several of the social, economical and structural obstacles identified for using the free government health services. These obstacles are investigated below. --- Table 2 A Highland case of child diarrhoea management Discovering the disease One evening when a 23-year old mother returns from field work in the hills, the children who have been attending her 12-months old baby say that the child has a 'hard belly' and hasn't defecated during the day. During the night the child defecates liquid and greenish stools thrice, cries, and develops a bellyache and slight fever. The mother describes the child as very sleepy, weak, and not wanting to eat anything except for taking breast milk. --- Treatment in the home The diarrhoea goes on for three days. During this period the parents-in-law allow the mother to remain home to attend to the child and cook for the family. When asked about the cause of the diarrhoea the mother is uncertain: "Maybe because of the weather -but I don't know [...] I thought it was maybe because the child ate something unsuitable, drank dirty water or played dirtily". She explains 'unsuitable foods' as something which 'couldn't be digested' by the child, including sweet milk or toxic forest fruits. For the first two days the mother carries the child around, all the while breastfeeding and urging the child to drink boiled water and eat some rice in the hope that the disease will cease. On the third day the parents-inlaw decide to call the village 'th<unk>y mo' (spiritualist). He explains how to cure diarrhoea "First we have to do 'bói' (fortune telling) and if there is a ghost, we can see it. Then we can do the'mo' (warding ritual)" (Spiritualist, Red Dao). He talks in tongues and counts his fingers and says that the disease does not take a comprehensive ritual or large offerings to chase away the disturbing ghost. He chants and burns incense to contact the ghost while offering small gifts of rice, meat, and home-brewed wine. But the child keeps defecating and crying during the evening and night. The mother is now unsure of what she should do since she is "inexperienced as a mother". She has not attended village meetings where health advocacy takes place and there is no women's group in her village and she doesn't understand what is communicated on the radio and TV in Vietnamese. She places her trust in her mother-in-law who decides to prepare a bitter drink of boiled guava buds for the child. During the next five days the mother-in-law attends to the child and feeds it with the drink thrice daily, while the mother is away working in the fields. The child gradually stops defecating liquid stools and regains some strength. --- Using drugs and government health services The mother did not go to the CHS for medication, but preferred drugs from the pharmacy in town, which she perceived were of a higher quality. After another six days at home the diarrhoea resumes and the mother now perceives it as a'serious' (nang) disease. She wants to go to the CHS, but: "I don't know how to ride a motorbike and my husband was not at home". The local custom prohibits her to sit on a motorbike with another man, and the CHS is too far away for her to walk to and she feels unable to communicate with any of the health workers at the CHS in Vietnamese. The parents-in-law decide to send her brother-in-law with the sick child to the CHS. He returns with some sachets of powder; a fever reducing drug and ORS, and written instructions on correct dosages of the medicaments in the medical notebook of the child. He was told to ask the VHW for further explanation if needed. The mother is illiterate and says that she never meets the VHW, who lives on the other side of the hill. Further, he never comes around to advice on child health she says. So she decides to give both powders the same way; two -three times a day in a teaspoon of boiled water or tea until the child stops defecating liquid stools. She is very afraid that the diarrhoea would not stop before the medicines run out, she says. But after three days of medication the child recovers and the mother is relieved that she would not have to ask her family-in-laws to take the child back to the clinic or to the district hospital, which would have been very time consuming and costly for the family. --- Economic and logistical constraints to accessing health services Some clear differences in treatment seeking and health care choices were found between the poorer ethnic groups living and working in the highland (Dao and Xa Phó) compared with richer groups living in the lowland setting (Tày and Giáy). As described in case 1 (Table 2), greater difficulties in assessing treatment at health facilities were observed for highland caregivers living furthest away from CHSs and hospitals, compared with the richer and more centrally located communities in the lowland. Even though health insurance would cover the direct costs of treatment and admissions, highland caregivers (e.g. case 1, Table 2) in particular worried a lot about the substantial indirect costs such as transport, buying food, arranging home-sitting, and paying for extra drugs and services at the CHS or hospitals. Distance and time was also a constraint for highland caregivers from poor families without means of transportation, who would have to spend half or a full day if walking to the CHS. Caregivers acknowledged that this was too time consuming if a child was seriously sick. In focus group discussion and interviews, caregivers all said that they would borrow or spend their money to pay for transport to get to the CHS in time, despite having very low incomes. The cost of petrol or renting a motorbike driver for a round-trip from a highland village and the CHS was 20-25,000 Vietnamese Dong, equalling several days of work for a highland farmer. In comparison, lowland caregivers (Giáy and Tày) lived a maximum of five kilometres away from the CHS and all had bicycles or motorbikes accessible within their households. --- Gender roles as a constraint for accessing health services Observations in households (conducted during fieldwork in 2008) and interviews with caregivers confirmed that all four groups of ethnic minorities live according to patriarchal gender roles, with the oldest men taking all major decisions and the youngest women being the main domestic labour force. This was most strongly expressed in highland families, where young mothers were also the main agricultural labour force of the families, and who expressed low decision taking powers about their children's health, which complicated seeking the government health facilities (Table 2). In the highland, six out of 19 caregivers of children during sickness were grandmothers, and mothers and grandmothers indicated that elders were the primary persons in the household in charge of deciding, preparing and Discovering the disease One evening a 25-year old mother from a Tày lowland village discovers that her 17-month old child is slightly feverish and coughing. Without any objections from her family, she stays at home the next day to attend the child. In the morning the child starts defecating very watery and yellowgreenish stools. The mother describes the child as weak, tired, crying, thirsty, having stomach pains and only wanting breast milk. --- Treatment in the home The mother immediately initiates a range of treatments at home: To ward off the fever caused by the "cold winds hitting the child", she performs the traditional '<unk>ánh gió' (hitting the wind); with a silver coin, an egg and some ginger mixed in a piece of cloth she strokes the head and body of the child. At the same time she gives the child a dose of a fever reducing drug, a left-over from the previous time the child was sick. Maybe the diarrhoea was caused by 'incompatible foods' she says. She therefore restricts the child's diet from such 'harmful foods': "Avoid sour and fishy things such as eggs and we didn't have any vegetable soup in 3-4 days. I didn't dare to let him eat it!". She urges the child to drink water and eat rice mixed with water, salt and minced meat. She also restricts her own diet from green vegetables, being afraid of passing any harmful substances on to the child in the breast milk. On the second day the diarrhoea stops and the mother is happy to have handled this 'light' (nhe) diarrhoea at home. Apparently this treatment was 'compatible' to the disease of the child, she says. When asked how she knew what to do, she says that she always asks her mother-in-law and other elder women in the village: "They know everything. They have been through it all". She has also noticed some campaigns during epidemics and some advice sanitation when watching TV at the neighbours. At the regular Women's Union meetings she also learned about child care and personal hygiene. Finally, she prepares a herbal drink for the child. The old people in the village told her to do so: "The old women said that they brought up many children in the past. They all collected the herbal medicines for drinking". She finds the herbal medicines more 'compatible' for a body of a child since they are not harmful as drugs can be she says. She also finds it more practical and effective compared to drugs: "The clinic medicine is drunk just twice per day. About the herbal medicine, he can drink it anytime he's thirsty. I just let him drink whenever he asks for drinking water. I boil it and leave it in the pot". Furthermore, she can drink it herself and pass the healthy substances to the child in the breast milk -unlike any drugs given directly to the child. After four days of this combination of treatments the child recovers and the mother is proud to have successfully handled this disease of her first child. --- Using drugs and government health services But after another four days the diarrhoea begins again. This time the mother immediately goes to the CHS on her bike without involving her husband or parents-in-law. She now describes the diarrhoea as serious and therefore not treatable at home. When asked why she did not firstly consult the VHW she says: "She doesn't have medicine to provide the people. And she doesn't care about the child's diarrhoea". At the CHS she does not tell the assisting doctor about the treatment she carried out at home or what might have caused the disease. She returns home with four kinds of drugs, which she knows as: antibiotics, fever reducing drugs, digestion enzymes and ORS. Without consulting anybody, she decides to only give the fever reducing drugs and the enzymes. She finds the antibiotics 'incompatible' with a sick child's stomach. It is 'too sweet' and 'too hard' for the child which can cause even more diarrhoea. She does not know what the ORS is for and did not dare ask the CHS staff: "They are often cross with me. They think I come too often to ask for medicines", she says. administering the treatment for a sick child. Mothers were not allowed to travel unaccompanied or with male non-family members to health facilities and needed permission from elders to stay at home from field work to care for the sick child (Table 2). A group of highland men explained: "when the child gets fever, falls ill, belly pain or something similar, the grandparents have to take care of the child and prepare the medicines". The men continued: "Everything must be discussed with the grandparents [...]. After discussing, if all agree, we let the child go (to the CHS)" (FGD, Dao men). Highland women also had very few opportunities of accessing information in the community on health care options and preventive measures compared with women in the lowland. TVs and radios were only accessible in few households and not broadcasted in local languages, making it difficult for most women to understand the provided information since they had short or no education and therefore did not speak much Kinh. Village meetings were mostly attended by male heads of households and women's groups were not established in the highland communities. Furthermore, VHWs were not consulted much (elaborated later). In comparison, many lowland family units were smaller and less influenced by the older generations. A group of Xa Phó fathers said: "The men here all live separately (from their parents). They do not depend on the grandparents; if our child is sick, the husband and wife are in-dependent" (FGD, Xa Phó fathers). Most lowland mothers (Giáy and Tày) also had basic education and spoke fluently Kinh. They all stressed that they took decisions on their own about staying home from work to tend the child, going to the CHSs or buying drugs in town. Most lowland women also attended Women's Union meetings in their communities regularly, watched TV on daily basis, and could understand the radio and loudspeaker announcements broadcasted in Kinh language. --- Reluctance towards government health services Despite the fact that all caregivers, from highland as well as lowland communities, highly appreciated access to Government health services and drugs free of charge, many caregivers sought these services with some apprehension. None of the interviewed caregivers sought the advice or were visited regularly by the VHWs working in their villages. Caregivers said they normally only had contact with the VHWs when attending child vaccinations or child nutrition surveillance on a monthly or bimonthly basis, which were also perceived to be the main tasks of a VHW; "He just weighs the small children once per month. I only saw him doing that"(Xa Phó mother). As expressed in both cases (Tables 2 and3), caregivers also said that VHWs did not have important health services to offer sick children such as drugs, first aid, ORS, or good medical knowledge. All caregivers had therefore experienced receiving a standard advice of going to the CHS from the VHW. This made them go directly to the CHS without consulting the VHW first. Health Station staff, on the other hand, were perceived to be qualified to diagnose diseases and prescribe drugs. But as exemplified in case 2 (Table 3), the majority of caregivers had bad experiences when consulting the CHS. Their illiteracy and 'backward' lifestyles were often commented on negatively by different types of health staff. Thirteen caregivers told stories of being scolded, ignored, or blamed for seeking treatment too late or too often. A caregiver from the highland explained; "Most of the people here are scolded by the doctors. Doctors often blame people for not taking good care of their children [...] They should try to understand the situation here. It is really hard (to live) in a place like this" (Dao woman). A young lowland mother added: "I just want the doctors to be more enthusiastic with patients [...] Only when people fall sick, they come to the clinic, so the doctors have to behave softly with them so people can be satisfied" (Tày mother). Language barriers were also mentioned by caregivers during observations and in all interviews with highland women. They felt they could not communicate accurately in Kinh language with staff at the CHS. Health staff also expressed frustration about not being able to communicate with patients. They therefore used very simple language or, in rare cases, used other waiting patients as translators or asked patients to have someone in their own community read and explain the prescriptions to them (from a prescription book). To avoid being misunderstood or perceived as backwards, caregivers uniformly said that they never shared ideas about causes of diseases, asked clarifying questions about the prescribed drugs or told any health staff including VHWs, about the home-made treatments they had used before coming to the CHS. This was verified during observations at the CHS. Here, no caregivers were observed to share any information on the child disease or ask about prescriptions, diagnose, or the guidelines given by health staff. Staffs were observed to only ask few direct questions on disease symptoms and personal details. --- Explanatory model for treatment seeking: 'Praying in four directions' and perceived severity of disease As illustrated in the two cases (Table 2 and3), caregivers sought and chose treatment according to a pragmatic logic of medical pluralism, applying and combining a range of treatments from different medical systems to increase chances of effect and recovery. This strategy was described with the Vietnamese proverb 'praying in four directions' (Vái t<unk> Ph<unk>ng): "We call it 'praying in four directions'; when we get sick; we do whatever we have to" (FGD, men, Highland). A highland mother added: "It is safer when doing all. Because sometimes doing'mo' will (make him) recover, sometimes going to ask for drugs at the health clinic will" (highland mother). As described in the two cases (Table 2 and3), caregivers commonly shifted between folk, popular, and professional bio-medical treatment systems as a disease developed and choices were often based on the perceived severity of the disease. Home-administered remedies, including herbal mixtures, concoctions and massage were perceived sufficient for 'light' or'minor' (nh<unk>) episodes lasting two -four days with irregular defecation of liquid stool, belly pains and general uneasiness. This type of diarrhoea was also described as a 'normal disease' (b<unk>nh b<unk>nh th<unk>ng) common during childhood, easy to treat and not severely affecting the child: "The old people said that the children have diarrhoea when they grow up [...] It's not a special thing if a child gets diarrhoea" (Xa Phó mother, lowland). A lowland mother added: "I saw the diarrhoea as normal; he had diarrhoea for three-four days and then recovered. [...] He didn't lie down and he went out to play all day as usual" (Tày mother). Among all caregivers, diarrhoea was perceived 'heavy' or'serious' (n<unk>ng) when lasting more than 2-4 days, re-emerged shortly after recovery or included high fevers, sudden onset of very frequent defecation, and the child becoming very lethargic, or losing weight. In these cases all caregivers mentioned consulting the CHS and taking 'western medicines' (thu<unk>c tây) as a necessary treatment. Referral to a hospital was also generally agreed as a necessary action if treatment with western drugs did not stop the diarrhoea. This was also acknowledged among older caregivers from highland communities with longer distances to the CHS; "When he kept being very sick, we couldn't let him stay at home. We went down to the health clinic to ask for drugs... He was sick at home for two-three days and it was getting too serious... The longer we stay at home, the more serious it gets" (Dao Grandparents). Nevertheless, a total of six children (four from the highland) had been sick with diarrhoea for several weeks and were seriously weakened before being taken to the CHS and admitted up to 12 days at the inter-communal clinic or the district hospitals receiving frequent intravenous sodium glucose solution and drugs. --- Explanatory model for treatment seeking; 'Compatibility' as a guiding principle Respondents frequently used the expression of a treatment or method being 'compatible' (h<unk>p) or'suitable' (phù h<unk>p) for a child with diarrhoea. Compatibility turned out to be a central and complex concept, functioning at various levels and having implications on all aspects of diarrhoea management, including the perceived cause, the chosen treatment and evaluations of its outcome. --- Compatibility and causal factors Reflections about the possible causes of a diarrhoeal episode were important for the choice of treatment. In all four groups of EMGs, a local cosmology of keeping the body of mother and child well-balanced and un-exposed to certain 'un-suitable' elements in the environments or 'in-compatible foods and drugs', dominated the explanations of diarrhoea. 'Unsuitable' environmental factors for children included exposure to sudden changes in weather e.g. from hot to cold or rainy weather or being exposed to something dirty, e.g. dirty playing environment or eating dirty foods. 'Un-suitable' items for children's stomachs' causing diarrhoea mainly included too hot, sour or sweet items, including sweet drugs, foods with 'fishy' (tanh) taste or smell (e.g. eggs), sour vegetables or fruits, fatty foods and 'hot breast milk' (from a mother who had worked in the sun). As described in case 2, the suitable treatment for exposure to these factors was to re-balance the body by restricting the behaviour (not going out in the sun etc.) or diet of the child or the breastfeeding mother (Table 3). In all four EMGs, a diet for a sick child often included rehydration measures such as giving rice porridge (cháo) with salt and sugar, green teas or herbal concoctions. All breastfeeding mothers also said they had continued breastfeeding during diarrhoea unless they were forced to go to work. --- Compatibility and treatment testing The process of testing and establishing the 'compatibility' of a treatment was crucial for disease management. In general, caregivers perceived a treatment to be compatible, if they saw the diarrhoea lessening or stopping within a short time. As in case 1 (Table 2), this was expressed clearly by highland caregivers who always sought spiritual treatment before or parallel with CHS based treatment. This was perceived a compatible treatment for children affected by angry ghosts, a spell or a discontent ancestor, since such factors had to be eliminated first in order to make the child susceptible to biomedical treatment: "The CHS checks if the child is having a disease. [...] But at home the 'th<unk>y mo' (spiritualist) keeps doing 'bói' and'mo' (fortune telling and warding) to make the child recover or not. When finishing 'bói' and water transmissions (intravenous sodium glucose solutions) and injections we can see if the child can recover" (Dao men, FGD). --- Compatibility and choice of medication All types of medication were also evaluated according to whether it was perceived compatible with the specific person or the specific disease. A lowland father explained the choice of herbal medicines: "If the child is compatible (h<unk>p) to that herbal plant, he/she will recover soon [...]. If that medicine was not compatible, then we go to the health clinic"(

Background: In Vietnam, primary government health services are now accessible for the whole population including ethnic minority groups (EMGs) living in rural and mountainous areas. However, little is known about EMGs' own perspectives on illness treatment and use of health services. This study investigates treatment seeking strategies for child diarrhoea among ethnic minority caregivers in Northern Vietnam in order to suggest improvements to health services for EMGs and other vulnerable groups. Methods: The study obtained qualitative data from eight months of field work among four EMGs in lowland and highland villages in the Northern Lao Cai province. Triangulation of methods included in-depth interviews with 43 caregivers of pre-school children (six years and below) who had a case of diarrhoea during the past month, three focus group discussions (FGDs) with men, and two weeks of observations at two Communal Health Stations (CHGs). Data was content-analyzed by ordering data into empirically and theoretically inspired themes and subcategories assisted by the software NVivo8. Results: This study identified several obstacles for EMG caregivers seeking health services, including: gender roles, long travelling distances for highland villagers, concerns about the indirect costs of treatment and a reluctance to use government health facilities due to feelings of being treated disrespectfully by health staff. However, ethnic minority caregivers all recognized the danger signs of child diarrhoea and actively sought simultaneous treatment in different health care systems and home-based care. Treatments were selected by matching the perceived cause and severity of the disease with the 'compatibility' of different treatments to the child. Conclusions: In order to improve EMGs' use of government health services it is necessary to improve the communication skills of health staff and to acknowledge both EMGs' explanatory disease models and the significant socio-economic constraints they experience. Broader health promotion programs should address the significant gender roles preventing highland mothers from seeking health services and include family elders and fathers in future health promotion programs. Encouraging existing child health care practices, including continued breastfeeding during illness and the use of home-made rehydration solutions, also present important opportunities for future child health promotion.

sought spiritual treatment before or parallel with CHS based treatment. This was perceived a compatible treatment for children affected by angry ghosts, a spell or a discontent ancestor, since such factors had to be eliminated first in order to make the child susceptible to biomedical treatment: "The CHS checks if the child is having a disease. [...] But at home the 'th<unk>y mo' (spiritualist) keeps doing 'bói' and'mo' (fortune telling and warding) to make the child recover or not. When finishing 'bói' and water transmissions (intravenous sodium glucose solutions) and injections we can see if the child can recover" (Dao men, FGD). --- Compatibility and choice of medication All types of medication were also evaluated according to whether it was perceived compatible with the specific person or the specific disease. A lowland father explained the choice of herbal medicines: "If the child is compatible (h<unk>p) to that herbal plant, he/she will recover soon [...]. If that medicine was not compatible, then we go to the health clinic"(Giáy father). A lowland mother similarly explained the effect of western medicines: "Depending on which medicine is compatible to him, he will recover after drinking it 1-2 times" (Giáy mother). A group of highland fathers explained how they also had to consider the compatibility of a disease with a drug:" If the disease is suitable to the foreign medicine, you take a little bit to recover [...] (FGD, Dao fathers). As already mentioned, shifts from herbal medicines to western drugs were related to perceptions of severity of disease. As the quote indicates and as described in case 2 (Table 3), these perceptions were linked to general perceptions of 'western drugs' being very powerful and therefore potentially harmful and 'incompatible with the body' compared with natural medicines. Antibiotics were perceived as particularly powerful. Six caregivers described them as 'too hard', 'too strong', and 'too sweet' for a sick child. Unfortunately, as exemplified in case 2 (Table 3) and confirmed in interviews and by observations, it was common to receive 2-4 different kinds of prescribed drugs for diarrhoea at the CHS, with antibiotics and fever reducing powders the most frequently mentioned. In order to limit the intake and harm from western drugs, many caregivers said they gave smaller daily doses of the drugs than prescribed. Or they decided to give only some of the drugs and shifted from one drug to another if recovery did not take place quickly. These perceptions also had implications for the use of ORS. Only one Xa Phó mother knew how ORS differed from drugs. Others perceived it as any other drug, did not understand the principle of rehydration and did not believe that the child could tolerate the large quantities of powder mixed with water or the unpleasant taste of it. A highland mother explained: "They (CHS staff) said: mix it with a bowl of water but the child couldn't drink it all. I had to mix with a little bit of water for him to drink it. Then it was sweet and the child liked to drink. If the water is too acrid (chát), the child doesn't like to drink"(Dao mother). Some caregivers also expressed suspicion and doubts about the quality and effectiveness of the free western drugs handed out at the CHS:: "In the health clinic they are afraid that we do not know how to drink it. They give us the diluted (lo<unk>ng) medicine"(Highland father). Therefore, despite having very low incomes or living in remote areas, some caregivers occasionally bought drugs at private pharmacies in town instead expecting higher quality drugs: "When we buy, we pay the money and they will give us a good drug so we can get well soon"(-Highland mother). Hence, the expectations to the effectiveness of western medicines and the easy access to drugs provided by commercial providers seemed to sometimes override concerns of cost, distance and harmfulness of drugs. --- Summing up Findings suggest that logistical and social barriers exist for accessing the CHSs. In addition, two guiding principles constitute a local explanatory model for managing childhood diarrhoea among ethnic minority caregivers in this area. Simultaneous resort to multiple treatments or "praying in all four directions" is the leading strategy, relying on the two core concepts of'severity' and 'compatibility'. Reflections of compatibility were drawing on local cosmologies of balancing the body, as well as testing and determining effectiveness of various therapies and drugs. --- Discussion This study describes how logistic constraints, gender roles, local treatment seeking strategies and reluctance towards government health workers together constitute a suboptimal use of free government health services among ethnic minority caregivers in two rural communes in northern rural Vietnam. These constraints are discussed in the following to suggest ways of improving the quality and use of government health services. --- Economic and logistical constraints when choosing health provider Low usage of government health providers among EMGs compared to Kinh have previously been ascribed to failures in insurance coverage [19] and logistical constraints to access health services from the highlands [19,28]. But as pointed out by London [8], health service fee exemptions only cover one component of health service costs, while indirect costs may present bigger challenges for the poor. This was also expressed by highland caregivers in this study, who worried a lot about the substantial related costs of admittance, which potentially delayed seeking treatment for seriously sick children. Interestingly, despite having insurance with access to free treatment for their children at government CHSs, higher costs and longer distances, this and one other study [19] have identified a willingness among ethnic minority caregivers to buy drugs from private drug stores. The common practice among the rural population of Vietnam of seeking private health services and buying over-the counter drugs for self-medication [5,7] apparently extend to poor EMGs. The study also showed that VHWs, who are the closest health providers, were not considered qualified by caregivers, who instead bypassed VHWs and travelled long distances to access private or public health facilities. Problems of distance to health services might therefore be lessened by upgrading services at community level, either by increasing skills and competences of VHWs in remote communities to perform effective health promotion and basic treatment, or by increasing the frequency of outreach visits by CHS staff to remote communities. --- Balancing between a local explanatory model of disease management and government health services The study also identified a local explanatory model of disease management among ethnic minority caregivers, clearly differing from a bio-medical treatment system. Simultaneous resort, local medical cosmologies of obtaining bodily balances, concepts of compatibility of treatment and humoral qualities of medicines (e.g. hot/ cold and heavy/light) as identified in this study are all commonly found in the Asian context of medicine and well described in minority as well as majority populations in Asia and Vietnam [7,[29][30][31]. However, our study also found that health staff ascribed such health seeking patterns as 'ethnic' and 'backwards', and that government health services are not always the first and most well-liked choice of health service for ethnic minority caregivers. Health is a key objective of state policies and governance in Vietnam and health sciences in modern Vietnam have become symbolically associated with socialist modernity, rationality, and progress [8,32]. Further, ethnic minorities have commonly been described as'marginal' and 'developmental backwards' and therefore targeted by the socialist state to 'assimilate' to mainstream developmental standards [33,34] -while also being encouraged by the state to preserve cultural traditions not posing any threats to progress of the socialist state [33,35]. 'Superstitious beliefs and practices' such as shamanistic rituals and animal sacrifice have been deemed as 'backward' health practices and a constraint for progress and modernity [32,35]. Hence our study provide further evidence that ethnic minorities seem to balance between practicing, to them, meaningful health rituals, while also interacting with a modern government health system. This might explain why ethnic minority caregivers do not reveal local explanatory models of disease to health staff and opt for drugs from the private sector, where they are not met with demands to change health behaviours. --- Improving health services and health staff competencies for ethnic minorities This study agrees with previous findings on health service use in Vietnam, that low trust and dissatisfaction with medical staff [17], and discouraging communication with health personnel affects disadvantaged women's and ethnic minorities' treatment seeking behaviours and use of the public health sector [28,36]. This is underlined in a recent World Bank report stressing that female ethnic minority caregivers in Vietnam are particular vulnerable and in special need of appropriate health services [34]. The public health system of Vietnam has been described as particularly 'providercentred', not anticipating patients to speak up their needs or questioning health providers [37,38]. This is possibly worsened for illiterate EMGs who also live in areas with poor health service coverage and where it is difficult to attract and retain qualified health staff [2]. Hence, this study argues for a need for the Vietnamese health system to focus on service levels in remote areas of the country and to change the approach to marginal population groups in Vietnam to increase their trust in and use of government health services. Cross-cultural competence models [39,40] have been developed to address communication challenges in the health sector. They stress that health staff working in ethnically diverse populations must understand patients' actions and perceptions of disease as expressions of their social realities, including constraining socio-economic factors and local explanatory models of disease. For health staff in daily contact with EMGs in Vietnam this clearly implies acknowledging the flexible 'praying in four directions' approach to choosing treatment and local cosmologies of disease management, as well as avoiding the persisting stereotyping of EMGs as 'culturally backwards'. To address these needs, Vietnamese Medical Universities have recently strengthened community health training of medical students [41], while other staff groups remain to be considered. This study primarily focused on understanding the perspective of ethnic minority caregivers' on use of child health services. Further studies into the constraints of delivering effective government health care in mountainous areas of the country, would add important information to understanding the full complexity of providing effective health care for all children in Vietnam. Our findings also corroborate other studies highlighting the potential problem of self-medication and overprescription of drugs to children in Vietnam. Training of health staff should therefore focus on improved prescription practices and guidance to patients on how to administer drugs and ORS. They should encourage open discussions with patients on when to expect effects from drugs, and when to shift treatment and health care system. Small-scale interventions have managed to improve drug prescription and purchasing practices in urban sites in Vietnam [42][43][44]. It is strongly recommended that such interventions be scaled up to include rural and mountainous areas, where ethnic minorities are in need of good guidance. --- Opportunities for improved health promotion and trust in government health systems Empowering and encouraging Vietnamese caregivers with good care practices can enhance and sustain better childcare and health care practices in rural communities [45][46][47]. The current study highlights that most caregivers from all four EMGs recognize symptoms and evaluate the severity of diarrhoea within few days of onset and respond caringly and rapidly by applying various treatments. However, caregivers felt that hardly any encouragements or preventive advice were given to support these care giving efforts during contact with VHWs or CHS staffs. We highly recommend that all staff at government health facilities use every opportunity to praise and encourage such good care giving behaviours. This can form a fruitful starting point for further integrating preventive messages and child health promotion into the contact with caregivers. This may strengthen caregivers' experience of being in contact with a caring and trustworthy health system. In the present study we also saw strong indications of patriarchal gender roles creating difficulties for young highland mothers in particular to follow health messages and seek CHSs, depending on male and older family members' permissions. These findings are in agreement with other studies in the rural population of Vietnam [30,48], and other ethnic minorities in Asia [29], which show that older community members and grandmothers, in particular, play core roles in child disease management and care. However, current child health promotion programs in Vietnam and health promoters mainly target and educate mothers as sole caregivers. It is therefore strongly recommended that health promotion in Vietnam include other family members as key people in children's 'care groups' and their resources in child care. These people can be reached through different grassroots organisations (e.g. Farmers' Unions and Women's Union) or community stakeholders in daily contact with men and elders, e.g. village heads, diviners and CHS staff. Finally, we acknowledge the beneficial practices of rehydrating children with diarrhoea using continued breastfeeding and home-made oral rehydration solutions of green teas, water or rice gruels with added salt and meats. These practices should be known to all health providers including spiritualists, pharmacists and VHWs as beneficial and included as advice during contact with caregivers. This might be more effective than prescribing ORS to less severe cases, since it is perceived 'compatible' with children, while ORS is perceived a noncompatible western drug and also administered incorrectly by many caregivers. --- Study strengths and limitations This study has provided new information on the qualitative aspects of why and how ethnic minority caregivers in Vietnam use and choose health services for child diarrhoea. However, despite triangulation of methods and a thorough sampling strategy the study faces some limitations: Firstly, as also pointed out by Toan et al [19], the geographical inequity in access and use of government and private health services, as well as language barriers and poor economical conditions, are anticipated to be stronger for EMGs living in less developed areas of the country e.g. in high mountains with longer distances to health facilities and urban environments. Secondly, difficulties of gaining access to 'the insider perspective' as outsiders to a research community can present serious limitations to qualitative research [49]. Since some caregivers expressed being marginalised and disrespected by Kinh people due to their ethnicity, the researchers' and assistants' backgrounds (a white westerner and of Kinh origin from urban developed environments) might have caused reluctance from respondents to disclose actions which could be perceived as 'wrong', 'traditional' or 'backward'. This problem was anticipated by the first author having good prior knowledge of the living conditions and having socialized with many community members and their children during a previous six month long fieldwork in the same study area. Being outsiders to the area was also mentioned as a benefit by some respondents, who felt safe to disclose bad experiences with health staffs and family members since the researcher and assistants were not affiliated with any local authorities or organisations. However, it did constitute a special challenge to gain the trust of young highland mothers living in patriarchal families and making them speak up about sensitive issues of gender and decision making. Research assistants therefore aimed at interviewing mothers without the presence of elders and men and encouraged trusted friends and sisters to take part. Thirdly, some information is likely to be lost or transformed during translations with the first author not speaking Kinh, the research assistants not speaking ethnic minority languages and the need for double translation in Xa Phó and Dao communities where some women did not speak Kinh. The quality of translation was validated as research assistant's cross-checked and discussed recordings and each other's transcriptions of all interviews and FGDs with the first author. --- Conclusions This study investigated treatment seeking strategies for child diarrhoea among four groups of ethnic minority caregivers in northern Vietnam. The study identified several socio-economic and logistical constraints for highland caregivers in particular to seek treatment, while also identifying a local explanatory model which guided disease management among caregivers in all four ethnic groups. Suggestions for future child health promotion include increased quality of government health services, including improved communication skills and drug prescription practices of government health staff. Village health workers who are important primary health providers, but not sought by caregivers, should also be upgraded. Broader health promotion programs should address the strong patriarchal gender norms limiting highland mothers to seek treatments. Encouraging existing good child care practices and including elder family members as important caregivers, also present important opportunities for effective future child health promotion. --- Additional material Additional file 1: Caregiver Interview Guide. The interview guide which was used during the collection of the main data source; 43 semistructured in-depth interviews with caregivers of pre-school children (six years and below) who had experienced a case of diarrhoea in the past month. --- Authors' contributions TR was responsible for and carried out the data collection, analysed the data material and drafted the manuscript. FK and HS participated in designing the study and FK, HS, and AD provided comments on the analysis and revised the manuscript together with TR. All authors read and approved the final manuscript. --- Competing interests The authors declare that they have no competing interests.

Background: In Vietnam, primary government health services are now accessible for the whole population including ethnic minority groups (EMGs) living in rural and mountainous areas. However, little is known about EMGs' own perspectives on illness treatment and use of health services. This study investigates treatment seeking strategies for child diarrhoea among ethnic minority caregivers in Northern Vietnam in order to suggest improvements to health services for EMGs and other vulnerable groups. Methods: The study obtained qualitative data from eight months of field work among four EMGs in lowland and highland villages in the Northern Lao Cai province. Triangulation of methods included in-depth interviews with 43 caregivers of pre-school children (six years and below) who had a case of diarrhoea during the past month, three focus group discussions (FGDs) with men, and two weeks of observations at two Communal Health Stations (CHGs). Data was content-analyzed by ordering data into empirically and theoretically inspired themes and subcategories assisted by the software NVivo8. Results: This study identified several obstacles for EMG caregivers seeking health services, including: gender roles, long travelling distances for highland villagers, concerns about the indirect costs of treatment and a reluctance to use government health facilities due to feelings of being treated disrespectfully by health staff. However, ethnic minority caregivers all recognized the danger signs of child diarrhoea and actively sought simultaneous treatment in different health care systems and home-based care. Treatments were selected by matching the perceived cause and severity of the disease with the 'compatibility' of different treatments to the child. Conclusions: In order to improve EMGs' use of government health services it is necessary to improve the communication skills of health staff and to acknowledge both EMGs' explanatory disease models and the significant socio-economic constraints they experience. Broader health promotion programs should address the significant gender roles preventing highland mothers from seeking health services and include family elders and fathers in future health promotion programs. Encouraging existing child health care practices, including continued breastfeeding during illness and the use of home-made rehydration solutions, also present important opportunities for future child health promotion.

Introduction Adolescence is a developmental period during which major biological and psychosocial changes take place including puberty, social reorientation toward peers, transition to autonomy, and a developmental peak in risk-taking behaviour (Adams and Berzonsky, 2003;Steinberg, 2008;Duell et al., 2018;Nelson et al., 2005). These adolescent changes coincide with a developmentally heightened risk of antisocial and criminal behaviour (Landsheer and Dijkum, 2005;Grigorenko, 2011). Notably, the rate and severity of crime committed during adolescence is predictive of future offences (Overbeek et al., 2001). We need better understanding of the processes and individual decision pathways by which adolescents become involved in antisocial and criminal behaviour, in order to develop better intervention strategies. What are the factors that determine a young person's risk and degree of involvement in antisocial/criminal behaviour? Research has proposed a multitude of factors operating at the level of society, community, immediate social contacts, and the individual (Mak, 1990). In the current paper, we focus on individual factors and processes, and their interaction with the immediate peer group, taking into account community-level context in the West Midlands region of the United Kingdom. --- Youth involvement in gangs Recent statistics indicate high reported rates of violent crime amongst young people in the United Kingdom. In 2021, West Midlands Police reported the highest rate of knife offences nationally (156 per 100,000 population), of which 18% were committed by adolescents (aged 10-17) (Allen and Harding, 2021). A substantial proportion of this violent criminal behaviour takes place in a context of gangs, characterised as "relatively durable, predominantly street-based groups of young people who see themselves (and are seen by others) as a discernible group, engage in criminal activity and violence, lay claim over geographical and/or illegal economic territory, have some form of identifying structural feature, and are in conflict with other gangs" (Home Office, 2015, p. 2). The presence of gangs in a local community poses substantial risks to young people, including a higher likelihood of criminal and sexual exploitation, violence, and substance abuse (Krohn et al., 2011;Moyle, 2019;HM Government, 2022). According to a recent report, more than 300,000 10-17-year-olds in the United Kingdom know a gang member (Children's Commissioner, 2019). Typically, the most vulnerable children (e.g., looked-after, poorer socioeconomic backgrounds) are at most risk from gangs (Home Office, 2018), which deliberately recruit from within this age range (Children's Society, 2022). Exposure to the risks associated with gang involvement as a young person are associated with mental health problems, victimization and future criminality (Wood and Dennard, 2017;Moyle, 2019;Frisby-Osman and Wood, 2020). This provides a strong rationale for conducting research on this topic. In addition to the above-mentioned community-level processes and associated risk factors, an array of individual cognitive and social developmental mechanisms are thought to contribute to a young person's trajectory of involvement in antisocial and criminal behaviour. Here, we focus on factors including the formation of individual and social identity, susceptibility to peer influence and morality/moral development. --- Identity formation and peer groups The formation of individual and social identity is an important component of adolescent development (Grigorenko, 2011). Whilst the development of personal identity involves individuation and independence, social identity development requires a sense of belonging within social groups (Tajfel, 1982;Benish-Weisman et al., 2015). The tension between personal and social identity is said to present a developmental challenge for adolescents (Kroger, 2005). During adolescence, peer groups are especially salient for social identity and fulfilment of belongingness (Newman and Newman, 1976). Individuals form positive evaluations of groups to which they belong (Tajfel, 1979), which may influence behaviour within social groups, potentially via peer socialization and social influence processes (Brechwald and Prinstein, 2011). Susceptibility to peer influence peaks in mid-adolescence across risk-taking, prosocial and neutral contexts (Steinberg and Monahan, 2007;Knoll et al., 2015;Foulkes et al., 2018). This susceptibility may contribute to adolescents' greater tendency to take risks and commit crime in groups (rather than alone) (Zimring, 1998), and drive additional associations between peer-related factors (e.g., attitudes favourable to antisocial behaviour, interactions with deviant peers) and adolescent antisocial behaviour (Forsyth et al., 2018). As such, a young person's perceptions of their identity in relation to that of a salient peer group, and processes of individuation vs. social influence, are potentially relevant for understanding adolescent decisions around risk-taking, gangs and antisocial/criminal behaviour. Another facet of identity, and one which plays a role in individual decision-making, is morality (Tangney et al., 2007). Individuals hold moral standards comprising universal and culture-specific moral rules. These translate to moral behaviour in varying degrees depending on individual and social contextual factors. In a peer group setting, interpersonal negotiation and diffusion of responsibility can limit the capacity to act in accordance with one's moral standards. Conversely, experiencing consequential or anticipatory moral emotions such as guilt strengthens the link between moral standards and moral behaviour (Tangney et al., 2007). During adolescence, moral emotions continue to develop (Malti, 2013;Malti et al., 2013;Krettenauer et al., 2014). In one adolescent study, experiencing guilt (as opposed to shame) was associated with lower aggression, greater empathy, and increased propensity to take responsibility (Stuewig et al., 2010). Furthermore, low levels of morality regarding attitudes towards violence and criminality in adolescence are associated with antisocial behaviour and delinquent peer associations (Tarry and Emler, 2010;Chrysoulakis, 2020). However, this can also be explained by moral disengagement theory, which suggests that to commit offending behaviour (e.g., violence) in a gang, individuals detach the moral element from the act itself to rationalise their actions (Bandura, 2006). This is a particular problem within street gangs, where the context can increase the likelihood of moral disengagement (Niebieszczanski et al., 2015). According to social contagion theory, this is due to the contagious nature of gang violence from one gang member to another which sustains gang membership (Tsvetkova and Macy, 2015;Brantingham et al., 2021). To summarise, converging evidence indicates that cognitive and social developmental factors, such as individual and social identity, peer influence, and moral emotions, may play a role in adolescent decision pathways to antisocial and criminal behaviour. As such, this is relevant within the context of the current research, in which we explore young people's decisionmaking in a virtual gang scenario. --- Current interventions for gang involvement The current study explores a novel addition to existing schoolbased interventions around antisocial/violent behaviour in the context of gangs. Previous preventive interventions targeting young people at risk of gang membership tend to comprise educational programmes in which young people can have open conversations with peers and trusted adults (e.g., charity workers, police officers), and be challenged on their existing beliefs about gang membership (HM Government, 2015). An example of this is the G.R.E.A.T. programme (Gang Resistance Education and Training Program) which was implemented and evaluated in the US with two key goals: a) preventing gang membership and other forms of violence, and b) building positive relationships with the police (Esbensen et al., 2011). The programme documented significant improvements in reducing gang membership in the US (Esbensen et al., 2011), and was later adapted and evaluated in the United Kingdom as the Growing Against Gangs and Violence (GAGV) programme (Densley et al., 2017). However, no significant programme effect was found in the United Kingdom (Densley et al., 2017). Issues around young people's involvement in gangs and associated antisocial/violent behaviour remains a pressing issue within the United Kingdom, highlighting a need for research to support and contribute to the evidence base around novel interventions (e.g., Wood, 2019). The current study investigated a virtual reality (VR) experience designed to address issues around risk-taking, peer pressure, gang involvement and violent/antisocial behaviour among young people, developed via a theatre-in-education (TiE) approach. --- Theatre-in-education approach Theatre-in-education (TiE) uses performances, workshops and role play to explore challenging, sensitive topics with children and adolescents (Jackson and Vine, 2013;Wooster, 2016). As an interactive method of education, it allows individuals to engage in their own autonomous learning, visualising and responding to the experience and forming emotional connections to characters (Krahe and Knappert, 2009;Jackson and Vine, 2013). Arguably, TiE's most effective feature is interactively involving young people in the intervention (e.g., making decisions, interacting with characters; Jackson and Vine, 2013). A body of research including randomized controlled trials (RCTs) and systematic reviews indicates effectiveness of TiE interventions on topics including child sexual abuse and mental health, demonstrating increases in risk perception and knowledge gain for adolescents (Fryda and Hulme, 2015;Krahe and Knappert, 2009;Rousseau et al., 2014). Moreover, recent studies reported positive findings following a school-based TiE programme focusing on two similarly complex topics, specifically child sexual exploitation and abuse (CSEA) and child criminal exploitation (CCE; May et al., 2021;Swancott et al., 2023). This prior TiE programme included a live theatre performance by actors followed by an interactive workshop. Adolescent participants reported that the performance was impactful for them, and that they felt more aware of exploitation and had an increased understanding of the topic as a result of attending the programme (May et al., 2021). As such, the TiE approach has documented potential for actively engaging young people on potentially difficult, challenging topics. --- Current research A key TiE method that is used increasingly in recent years and incorporated into the novel 'tech-in-education' sector is virtual reality (VR). VR is defined as "the use of computer modelling and simulation that enables a person to interact with an artificial three-dimensional visual or other sensory environment" Frontiers in Virtual Reality frontiersin.org (Halldorsson et al., 2021, p.585). Interactive VR in particular lends itself well to the interactive nature of TiE, as participants make decisions which may change the course of the experience (cf. Boal, 1993;P. Hyde, personal communication, February 2022). Research highlights a wealth of advantages to using VR, including positive effects on learning, motivation, and enjoyment (Kavanagh et al., 2017). VR can be used to enable participants to experience life-like situations that would be too dangerous to undertake in real life (Abdul Rahim et al., 2012), with resultant learning and insight that translates to the real world (Xie et al., 2021). Furthermore, compared with classic TiE methods, VR offers increased capacity for flexible, consistent and widespread delivery without the need for live actors. Moreover, studies have supported VR use in sophisticated social environments (Kozlov and Johansen, 2010;Haddad et al., 2014), showing promising findings regarding individuals' understanding of their decisions. Although VR use is less well documented in adolescent compared to adult populations, evidence indicates impressive completion rates and ability to evoke authentic emotions, including empathy and the moral emotion guilt (Barreda-Angeles et al., 2021;Halldorsson et al., 2021). The current study investigated adolescents' decision-making and emotional responses to an immersive VR experience of gang involvement and antisocial behaviour. To our knowledge, this is the first virtual reality intervention used in schools to address gangs and violence, and the first to be evaluated within the academic literature. In conducting the current study, we aimed to address some limitations documented in prior VR research. It has been noted that VR studies need clearer theoretical grounding and provision of methodological detail (Halldorsson et al., 2021). Other studies highlight software usability issues, including poor interaction quality and readability (e.g., Hseih et al., 2010); although this limitation may predominantly apply to non-immersive VR, with immersive VR using headsets resulting in a more compelling experience (e.g., Halldorsson et al., 2021). The current VR programme is informed by psychological theory of adolescent risk-taking and was developed in a research-informed, participatory manner to overcome the previously mentioned limitation (see Method section for a more detailed description). There is scope to expand research in this area, particularly in the context of methodology, experiences of VR and adolescent populations. The aim of the current study was to explore adolescents' experiences of risk-taking and decision-making during a virtual encounter with a potential gang developed using a TiE approach (Round Midnight Ltd, 2019). Live-action footage was used to simulate a naturalistic environment and characters, while VR headsets increased immersion (Sütfeld et al., 2017). The VR format was interactive, such that participants were able to make decisions and interact with the virtual characters, applying the most effective features of TiE (Jackson and Vine, 2013). We explored the experiences of adolescents interacting with this immersive VR. A qualitative approach to data collection was used to understand participants' thoughts, feelings and experiences in depth. Data were gathered via focus group discussions, a format that allows participants to build upon and reflect on each other's contributions, increasing data richness (Braun and Clarke, 2013). Finally, the study sought to overcome some prior limitations in VR, including the lack of sufficient methodological detail on the VR experience. In order to answer the overall research question of "What are young peoples' experiences of a VR encounter with a gang, in terms of decisions, emotions and consequences?", the study aimed: 1) To elicit young people's accounts of their decisions in a virtual context of risk-taking, gang involvement and antisocial behaviour, including motivational, social and emotional factors and consequences, using a qualitative (focus group) design; 2) To explore the VR experience as an educational intervention tool to promote young people's discussion and reflection on issues with real-world relevance, i.e., risk-taking, gang involvement and antisocial behaviour; and 3) To investigate young people's experiences of interacting with the VR, including featured characters and events, and the virtual as opposed to real-world setting. 2 Materials and methods --- Participants and recruitment The present study recruited 24 adolescents aged 13-15 years (11 female, 13 male) from four state secondary schools in the West Midlands. Schools were selected based on current contact with our TiE partner Round Midnight Ltd. and local relevance of issues featured in the VR experience. The resultant sample was drawn from predominantly mixed or low socioeconomic central and suburban areas within the West Midlands conurbation. Participating schools distributed information sheets and parent/guardian consent forms to pupils. Parent/guardian consent and participant assent were obtained. The study was granted ethical approval by the University of Birmingham Science, Technology, Engineering and Mathematics Ethical Review Committee (ERN 19-1099) and researchers adhered to the British Psychological Society's (2021) Code of Ethics and Conduct. --- Materials --- The virtual reality experience "Virtual_Decisions: GANGS" is an interactive, immersive VR experience developed by West Midlands-based creative arts and --- FIGURE 3 Steps undertaken during the process of thematic analysis (based on the six phases of thematic analysis by Braun and Clarke (2006); Braun and Clarke (2013). tech-in-education company Round Midnight Ltd. It is used to facilitate school workshops discussing personal, social and health education (PSHE) topics, such as youth violence/gang behaviour, risk-taking and peer pressure. The VR experience was created using Unity, a cross platform game engine developed by Unity Technologies (https://unity.com/). Through the VR experience, participants are placed in a live-action scenario involving a simulated encounter with a potential gang (see Figure 1). This takes place via an 8-min interactive live-action audio-visual experience delivered via Oculus Go and Oculus Quest 2 headsets providing an immersive 180-degree interactive experience. Sound input is received via headphones and participants navigate through the experience, making decisions by moving their head to navigation targets shown on-screen. Specifically, in the VR scenario, participants make decisions about who to speak to, what to say and their level of involvement in two escalating acts of antisocial behaviour enacted toward passers-by. Main events, decision points and decision options are depicted in Figure 2. Some decisions impact the events experienced (e.g., which group members to approach, see Figure 2), but the overall structure of events remains the same. The semi-structured interview guide focused on epochs 1-4 depicted in Figure 2, with participants free to discuss any part of the experience. In line with the aims of the current study, we did not record participant decisions on the VR headsets. In terms of the acts of antisocial behaviour depicted in the VR, in the first of these, the participant witnesses the group intimidating a passer-by (by not allowing them to pass, taking an item out of their shopping bag and throwing it between the group, then returning it and letting the passer-by leave). In the second, more serious, act of antisocial behaviour, the group intimidates two passers-by (a young couple) by shouting at them, approaching them, getting physically close to the young male and continuing to shout and jostle him, and subsequently the young male passer-by ends up on the ground being assaulted; however, what exactly this entails (i.e., assault via kicking/ punching and/or use of weapons) is ambiguous as the action is partially obscured from view by the group members' bodies. Before starting the VR experience, the participant is told that they will be introduced to a 'group' that two of their former primary school friends now hang out with. This group (totalling 12 individuals) is not defined as a gang to participants, but typically, including in the current study, young people spontaneously use the term 'gang' to describe the group. The word 'gang' has more than one common usage. First, young people sometimes use the term to refer to a group of people hanging out together, without any criminal features. Second, the term gang (e.g., street gang, criminal gang) is used and defined as a group of three or more individuals where illegal activity is a core part of their group identity (Klein et al., 2001). In Round Midnight's "Virtual_Decisions: GANGS" educational programme, the VR experience is delivered within workshops and 12-part interactive curriculum created in conjunction with Manchester Violence Reduction Unit, led by experienced TiE facilitators, typically as a primary prevention strategy to discuss complex issues in a trauma-informed manner. However, for the purposes of conducting the current research study, we administered the VR experience and backstory, without the accompanying curriculum. The involvement of Round Midnight in the research is summarised as follows. Round Midnight applied for and received funding from United Kingdom Research and Innovation (UKRI) to develop a VR experience featuring a youth-relevant issue. They conducted extensive, iterative grassroots consultations with >1,000 young people via surveys, workshops, and theatre improvisations in order to select the topic (risk- Frontiers in Virtual Reality frontiersin.org taking behaviours) and develop the scenario and script. Partway through this process, the last author was approached to provide research consultancy input based on her expertise on the neuroscience/psychology of adolescent risk-taking. This input was incorporated as part of Round Midnight's research and development process, specifically during script and character development. Round Midnight own and license the VR experience, and they made it available to the research team free of charge (along with a number of VR headsets) for the purposes of conducting this research. Finally, Round Midnight facilitated participant recruitment for the current study via their school contacts. --- Procedure 2.3.1 Data collection A total of six focus groups were conducted in school classrooms. Focus Groups 1, 2 and 5 (FG1, FG2, FG5) consisted of five participants and Focus Groups 3, 4 and 6 (FG3, FG4, FG6) consisted of three participants. In accordance with the preferences of each of the four schools, FG1 had one teacher present (seated at the back of the classroom); FG2, FG3 and FG4 had no member of school staff present; and FG5 and FG6 had a member of school staff present (seated at the back of the classroom). All focus groups were 45-60 minutes in duration, in accordance with school timetabling constraints. Participants sat in a circular formation around a cluster of desks with the two researchers. Due to changing COVID-19 restrictions, the first two sets of focus group participants sat two metres apart with face masks on, two metres away from the researchers. Subsequent focus groups sat within one metre of each other with no face masks, two metres away from the researchers, who wore face masks. The researchers began each focus group by outlining the 'backstory' and context of the VR scenario which participants would enter virtually (an introduction to the group of people who two former childhood friends hang out with). Participants underwent the eight-minute VR experience individually. Immediately after, participants spent five minutes completing a short anonymous questionnaire which asked them to circle the decisions they made at each of the four key decision points shown in Figure 2, and to note an emotion they felt at the point of decision-making, intended as a memory and reflection aid. Participants were asked to pick a pseudonym out of a choice of name labels. The researchers explained the ground rules of the ensuing focus group discussions which followed a semi-structured topic guide (see Supplementary Material). The focus group discussions were led by one researcher and supported by the second researcher. Questions focused on participants' decision-making during the VR experience as a whole and at the four key epochs shown in Figure 2, including motivations, thoughts, feelings, consequences, reflections, VR vs. real-world comparisons, and user experiences. The discussions were recorded using an encrypted Dictaphone. --- Transcription A professional transcription service transcribed the audio recordings verbatim and the transcripts were imported into a qualitative data analysis software program, NVivo20, to facilitate coding and analysis. --- Analysis The researchers used a thematic analysis approach to analyse the focus group data. Thematic Analysis (TA) was chosen for its datadriven, flexible nature and ability to give a voice to participants, which is important for topics that have not yet been explored (Braun and Clarke, 2013). TA identifies and analyses patterns across the data, allowing themes to be developed which capture important aspects of the data in relation to the research question (Braun and Clarke, 2013). The researchers followed the six-phase process recommended by Braun and Clarke (2013) which is represented in Figure 3. The researchers kept a reflective diary during the process of collecting and analysing the data which encompassed their thoughts, feelings and other notable observations regarding each focus group. The two researchers began the process of analysis independently by allocating and reading three transcripts each and coding them line-by-line using descriptive codes. Next, the two researchers independently identified relevant themes across their allocated transcripts. Following this, the two researchers met together to discuss the themes representing the whole dataset. They subsequently met with the supervisory team to discuss the themes in more depth, resolve any disagreement or discrepancies, develop overarching themes that represented the full dataset, and discuss interpretations of the data in relation to the theme development. This iterative process was repeated three times to identify themes and subthemes, discuss possible interpretations and organization of the data, and further develop themes relevant to the research questions. Finally, through this process, the final set of themes and subthemes were developed. --- Epistemological position TA is theoretically flexible which allows for a data-driven approach to analysis (Braun and Clarke, 2013). The researchers in the current study adopted a realist position, aiming to describe young people's experiences and decision-making in the VR experience independent of pre-existing perceptions or theory. Furthermore, the researchers' characteristics, role and influence in the focus group discussions, and later interpretation, was also reflected upon. Particularly, the researchers were both white, young, educated females. Some of these characteristics may have contributed to creating an informal, non-threatening and respectful environment in which participants felt able to share their experiences. On the other hand, this same set of characteristics may have limited participants' ability to relate to the researchers or feel understood, reducing the information shared, or encouraging a set of desirable, rather than true, responses. The latter was also felt to be the case in the two focus group discussions in which a member of teaching staff was present. These possible influences have been considered and taken into account as part of making sense of the study's findings. The present study considers young people's accounts of the decisions they made, what motivated and influenced these decisions, how they felt, consequences, and VR vs. real-world comparisons, in a context of virtual risk-taking and peer pressure. The following findings are framed around individuals' decisions to go with or against the group, and the emotions, perceptions and risks associated with these decisions. Most participants reported that they were willing to oppose the group in order to do the "right thing". These decisions were primarily influenced by participants' perceptions of themselves, the group and the situation. Three themes were developed encapsulating these influences on virtual decision-making: See Figure 4 1) "The Gang: Perceptions, expectations and peer influence" centres around participants' perceptions of the group and its dynamics; 2) "The Individual: Morality, autonomy and resistance to peers" reflects the process of constructing an individual identity through decision-making; and finally 3) "The Environment: Making decisions in VR versus real-life" considers the characteristics of the VR experience and relationship to real-life decisions. Direct quotes from participants are presented to illustrate the three themes. --- Theme 1. The gang: perceptions, expectations and peer influence Descriptions about the group naturally emerged in discussions, highlighting its importance throughout the experience. This theme represents participants' perceptions of the group and its influence on their decisions. Three main aspects were discussed: a) "The Unknown Group: Forming first impressions" considers the group's features, b) "Leaders and followers: Understanding group dynamics and developing expectations" reflects on group dynamics, and 3) "Conformity, the strength of peer pressure and social influence" mainly centres around peer influence. --- The unknown group: forming first impressions All six focus groups reflected upon and contributed to explaining the features of the group. Since no prior information was provided about them, participants' perceptions originated in initial impressions and developed during the experience. Initially, some participants considered the group to be "normal people [...] they just looked like everybody that's round here" (FG2), as well as thinking "they would probably be nice" (FG5). Instead, others assumed that "they were all like roadmen, women, people" (FG4) and that they would be bad from the start: "When I saw them, the first reaction that I had, I was like oh no, they're bad people-they don't have the good mindsets and stuff, they're not good human beings." (FG6). These initial differences across focus groups seemed to emerge as a result of participants relying on their own understanding and past experiences with similar people, groups, and situations. As participants advanced through the experience, they formed impressions and expectations about the group: "I did expect something bad to happen, at the end, because of just like where they were, what they were doing and how big of a group it was" (FG2). By the end of the experience, all groups seemed to agree that "obviously, they're bad people" (FG4) who were "horrible to anyone just because they feel like it, belittling others because they're bored and feel like they're better than everyone else" (FG3). Additionally, some hinted at the possibility that the group was a gang whose aim was "to recruit new people" (FG5): "I think, by the body language and the way they acted within the groups, and I could hear what they were saying as well, I had an insight that they would probably be in some sort of (gang) subculture and they would probably encourage me to go into county lines and do criminal activities" (FG6). In general, the group was perceived to be bad people, however, this label was not applied uniformly across the group members. Two characters in the story were introduced as participants' former friends. Participants initially reported perceiving them as good and trustworthy people who were different from the group: "They didn't seem like they'd associate with the rest of them (the group), but I thought that I could trust them, they seemed like decent" (FG4). In turn, their behaviour seemed to be rationalized "They might have felt pressured themselves, so they were just sticking with the majority and trying to blend in." (FG1). At its core, the mere notion of friendship, even if hypothetical, contributed to shaping participants' perceptions and expectations differently. While their former friends were mainly considered good people going "with the wrong crowd" (FG1), the group was seen as bad people who "had lost their morals" (FG3). --- Leaders and followers: understanding group dynamics and developing expectations Participants' descriptions of the group demonstrated their perceptions of gangs: the way they function, their structure and group dynamics. For them, the gang consisted of a leader whose role was primarily to control and intimidate the group: "By the way he acts, people are probably usually scared of him" (FG3), and members who "just followed their leader" (FG4) and "will do anything he says" (FG3). This group dynamic meant that rather than being friends, the group was seen as unrelated people only coming together for violence: "you can just tell by the people that none of them are actually good mates. They're just there for violence, something to laugh at." (FG4). Therefore, there was no expectation that group members would support each other, but that each individual would protect their own interests: "I don't even think they were really caring for each other, to be honest, the way that they were, em, talking to each other and acting [...] With groups like this, they are together, but then they tend to stab each other in the back quite quickly if the chance comes." (FG2). Due to the awareness of the type of group they might be, participants felt a sense of uncertainty and unease in terms of their behaviour. In particular, they expressed concern for how the group would react to them and how they would be treated: "I don't know if they would use violence on you. I mean, I've seen it with other people, but I don't know if they'd do it to someone else who was in the group with them" (FG4): "I did feel slightly afraid, but that's because [...] I didn't know how they would act. I did get a general idea of their personalities and what group they're part of, but, how specifically they would act, I wouldn't have known." (FG6). Given the possibility that the group might turn on them, the general agreement was that participants were expected to "just be a lapdog, just listen to whatever they say and do whatever they asked you to do" (FG1). For that reason, most participants believed they were expected to follow along and act like the rest of the group: "If they were fighting people, they'd want to see you fight as well" (FG5). Overall, awareness of group dynamics and behaviours shaped participants' expectations, which then played a role in and affected the way they made decisions. --- Conformity, the strength of peer pressure and social influence Altogether, participants across the focus groups reported that perceptions and expectations of the group led them to feel pressured to conform "I laughed at them (passer-by #1) just because the rest of the group were laughing" (FG4). Most participants shared that they "did feel under pressure, but that's mostly because, as an outsider from the group, they wouldn't really let you in easily." (FG6), or that they felt "under pressure because there was a lot going on and, you know, you could have made one wrong move and they might have targeted you." (FG2). Some participants, albeit in the minority, reported that they ended up going with the group. The motivations behind their decisions shifted throughout the scenario. At the beginning, participants who decided to follow the group were curious to "see what kind of friends they've (i.e., the former friends) made over the years" (FG6) since "if you like keep yourself aside from people, you're just boring" (FG4), or they wanted to "fit in" (FG5). As the experience progressed, the main reasons to follow the group were self-protection and need to not be "the odd one out" (FG4) or be targeted by the group: "I just... just said what he wanted to hear, so that, he wouldn't do anything to me" (FG3). This appeared to be informed by their observations of group dynamics and expectations of how they should behave in the group. By the end, some participants felt scared they would get into trouble, either with the group itself or the Police, given their involvement in the situation, and therefore decided to stick with the group: "Then, well I was going to go home, but then I went with the group because I don't know, I didn't really..." (FG1), "I didn't want them to feel like I was [a] snitch so... and I didn't want to go home, so just went with them" (FG5). Other participants who had initially wanted to know the group changed their minds and decided to separate from them at the end: "I kind of wanted to go with them, at first, because I was very curious of like what it would actually be like to be part of like the group and the way it is. But in the end, I was just like, nah!" (FG2). Altogether, this theme emphasises participants' perceptions of the group's features and dynamics, how this developed throughout the experience, and its influence on virtual decisions. Most participants acknowledged and recognised the role of social pressure. --- Theme 2. The individual: morality, autonomy and resistance to peers Despite participants' awareness of peer pressure, the majority reported opposing the group in order to do the "right thing" and help the passers-by. This theme discusses the individual motivations behind participants' decisions and consists of two subthemes, "Morally good people make morally good decisions" and "The struggle between individual and collective responsibility". --- Morally good people make morally good decisions All participants expressed concern for their own and the group's actions and how they were viewed by others. Participants'

Introduction: Gang involvement poses serious risks to young people, including antisocial and criminal behaviour, sexual and criminal exploitation, and mental health problems. There is a need for research-informed development of preventive interventions. To this end, we conducted a qualitative study of young people's responses to an educational virtual reality (VR) experience of an encounter with a gang, to understand young people's decisions, emotions and consequences. Methods: Young people (N = 24 aged 13-15, 11 female, 13 male) underwent the VR experience followed by semi-structured focus group discussions. Questions focused on virtual decision-making (motivations, thoughts, feelings, consequences) and user experiences of taking part. Data were analysed using Thematic Analysis. Results: Three themes were developed to represent how participants' perceptions of the gang, themselves, and the context influenced virtual decisions. Social pressure from the gang competed with participants' wish to stand by their morals and establish individual identity. The VR setting, through its escalating events and plausible characters, created an "illusion of reality" and sense of authentic decisions and emotions, yielding insights for real-life in a safe, virtual environment. Discussion: Findings shed light on processes influencing adolescent decisionmaking in a virtual context of risk-taking, peer pressure and contact with a gang. Particularly, they highlight the potential for using VR in interventions with young people, given its engaging and realistic nature.

" (FG5). Other participants who had initially wanted to know the group changed their minds and decided to separate from them at the end: "I kind of wanted to go with them, at first, because I was very curious of like what it would actually be like to be part of like the group and the way it is. But in the end, I was just like, nah!" (FG2). Altogether, this theme emphasises participants' perceptions of the group's features and dynamics, how this developed throughout the experience, and its influence on virtual decisions. Most participants acknowledged and recognised the role of social pressure. --- Theme 2. The individual: morality, autonomy and resistance to peers Despite participants' awareness of peer pressure, the majority reported opposing the group in order to do the "right thing" and help the passers-by. This theme discusses the individual motivations behind participants' decisions and consists of two subthemes, "Morally good people make morally good decisions" and "The struggle between individual and collective responsibility". --- Morally good people make morally good decisions All participants expressed concern for their own and the group's actions and how they were viewed by others. Participants' own moral values were reported as one of the main driving forces behind their decisions: "At the end of the day, I still don't know them, so I'm not going to go and agree with random people just to fit in. Obviously, I have my morals, and I intend to stand by them" (FG1). Particularly, most participants valued doing the right thing, being a good person and being perceived as such: "Like, you'd feel like more bad of yourself because you weren't like helping the people [...]. And then (by helping) you won't seem like as a bad person, basically" (FG5). For them, morality was not only seen as having the potential to influence their decisions, but also a reflection of "what type of person someone is, based on their actions and how they decide to... or what they decide to do in that situation" (FG3): "It could also reflect the type of person you are because if you decide to, for example, record on your phone, I feel like you'd only do that just so you could fit in with the group, even if you know they are bad people, because everyone knows what's right and wrong" (FG4). As a result, the majority of participants constructed their own and the group's identity around the morality of their decisions: "I just felt like saying it (saying sorry) because that's just who I am" (FG2). This became the primary difference between them and the group: "You (participants) are good people, like they're bad people" (FG5). For that reason, participants avoided following the group since that could "kind of corrupt" (FG3) them and make them "worse" (FG3). That is, they could lose their morals and identity in favour of those of the group: "I was a bit conscious about joining in the group because they could brainwash me against my morals and they could do things that I morally do not want to do" (FG6). These discussions demonstrated participants' awareness of the moral dilemmas they were presented with throughout the VR experience. For some, the moral complexity of the situation and the fact that "there's not a wrong or right decision" (FG4) was highlighted. Others considered it a black-and-white distinction and saw themselves as either purely good or bad: "I'm still a bad person for being there" (FG4). Overall, for the majority of participants, morality represented a fundamental separation between them and the group as well as being the major driving force to oppose them. --- The struggle between collective and individual responsibility Throughout all focus groups, participants struggled to make sense of their position within the group and their role in the experience. On the one hand, participants emphasised that they were different from the group and that they were "just an outsider" (FG6); on the other hand, they felt part of the group and, to an extent, responsible for the group's actions. This had implications for the decisions they made. Given the fundamental differences participants perceived to exist between themselves and the group, the majority felt that they "just didn't want to be involved" (FG3) from the start. Some participants reported acting in a way that would highlight these differences ("I was really out of place, so I wanted to make myself more out of place", FG2), in order to separate themselves from the group: "They didn't really allow me to join the group, and this is what I wanted to happen" (FG6): "I knew straightaway that I didn't want to be associated with them, so I knew that I would act in the way that I was...isolating myself from the group so they didn't want to associate with me and that I didn't want to associate with them. So, I wanted the first impression that I wasn't a part of the group anyway and I wasn't supposed to fit in there." (FG4). Equally, by opposing the group, participants wanted to reinforce that they had "the ability to choose (in the VR), even if it does feel like a bad idea" (FG6), and that they were making their "own decisions, not going along with what they (the group) are doing" (FG2). Despite wanting to assert their individuality, participants still struggled to reconcile the idea of not fitting in with the group with the fact that they were, on some level, involved with them. This can be seen throughout the focus groups, with participants reporting that they "felt guilty" (FG3) for being with the group, even if they did not partake in or agree with their actions: "I would feel bad for what they did to him, even though I am with the group in a way" (FG4). Some also reflected on the possibility of being perceived as guilty by the Police "because if they find your phone and search you, they have the evidence that you were there, and then you could be convicted for something" (FG1). Participants appeared to attempt to resolve their internal conflict and alleviate their feelings of guilt by justifying their role and responsibility in the experience. Some felt that they had to take responsibility for and remain with the group "to say sorry on their behalf" (FG5), both to protect themselves and the passers-by: "If I did leave as soon as I felt they were being aggressive and had the wrong attitude, then everything would have happened, but maybe it could have been worse because I wasn't there to make the decisions that I made and I wouldn't have distracted the group or told them to stop or said sorry to the man" (FG3). Instead, others seemed to try to diminish their responsibility for and involvement in the situation: "I didn't actually do anything wrong. I didn't really help that much, like I wasn't the one like standing in front" (FG2). Overall, this theme highlights how decisions were shaped by participants' perceptions of themselves and their morality. In turn, through the choices they made and reflections on their role in the experience, participants constructed their identity and affirmed their autonomy and individuality. --- Theme 3. The environment: making decisions in VR versus real-life This theme discusses the influence of the setting in decisionmaking, both in terms of the particular scenario, characters and events, as well as the virtual as opposed to real-world setting. Two subthemes were developed to represent how the realism of the experience influenced decisions: "Real-life decisions through the "Illusion of Reality" and "Real-life decisions have real-life consequences". --- Real-life decisions through the "illusion of reality" The VR experience presented an overarching scenario and series of events which for the majority of participants emulated real-life situations and elicited realistic emotions. Given the 'illusion of reality' created by the VR, participants were, to a certain extent, able to make decisions they felt would correspond to decisions in real life. Most conversations centred around the gang and the events. Specifically, participants talked about how the group reminded them of real-life gangs and how "some people do pressure you like that" (FG5): "There are gangs and groups that do similar stuff to what's happening in the VR. So it's relatable and personal as well because I know that stuff like this happens" (FG4). In turn, the escalating nature of the events and consequences were noted: "It just gets worse and worse as it goes on because you become more involved with the gang, making it worse for yourself" (FG1). This was seen as a realistic element that participants "could imagine [...] happening in real life" (FG2), which in turn contributed to their immersion in the experience: "As you keep going, yeah, so you start thinking and it becomes more and more realistic with the decisions and stuff" (FG1). Additionally, the time-limited nature of making decisions was noted: "like real life-obviously you're not going to have all the time in the world to make a decision" (FG1). In some cases, a sense of familiarity with the events allowed participants to make decisions by drawing from their own experiences: "I based it off real-life experience with people like this, and I did what I would have actually done if I was in that situation" (FG2). Similarly, some participants considered how the VR element made the experience feel real: "You could still feel the emotions of what that person would be going through if they weren't in the VR" (FG3). Therefore, by eliciting realistic feelings, the process of decision-making resembled how participants would act in real life: "Because it felt so real, I actually was able to answer with what I would have done in real life" (FG3). Overall, participants felt the VR experience had real-world authenticity. As one participant noted, this gave them "like an experience of what could happen if I was in that... a group like that" (FG3), which in turn permitted reflection on how they might react in comparable real-world situations. --- Real-life decisions have real-life consequences Participants discussed aspects of the VR experience that did not feel realistic, including aspects of the group and lack of real-life consequences. One of the elements noted to differ from real life was the group itself. A few participants suggested that some members of the group displayed unrealistic reactions and behaviours, as "no one would do that" (FG5), and that the characters seemed to "just act all like... hard" (FG4). Participants further commented that the decisions they made in the VR experience did not result in real-world consequences, and that they "just didn't feel anything could happen really, because, at the end of the day it's just virtual reality" (FG2): "To be honest, in virtual reality, there isn't that, do you know, that danger that's in real life, so you feel a lot more, if you like, free with your decisions because there's not going to be as heavy consequences. But in real life, obviously that danger is there, so when you make a decision, you've got to actually think about what could happen afterwards" (FG1). Given the lack of real-life consequences, some participants reported that their decisions would have differed in real life: "I'd have done different things in real life" (FG5). Reflecting on this difference, the majority of participants noted that in real life they "wouldn't have been in that group in the first place" (FG3) or that they "wouldn't be there at all" (FG2). Overall, participants recognized that the difference in real-life consequences between VR and a comparable real-world scenario could result in different choices. --- Discussion The current study was conducted to investigate adolescent decisionmaking and risk-taking in a VR setting. Specifically, we considered the processes and elements shaping decisions in a context of risky and antisocial behaviour, gang involvement and peer pressure. Participants' decisions and reflections, as well as group processes in the virtual gang and focus group discussion, will now be considered. Focus group discussions highlighted the complexity of the decision-making process and variety of responses among individuals. Often, discussions revolved around whether or not young people opposed the group in order to 'do the right thing' and help the passers-by. Participants discussed reasons and motivations for their decisions, including their perceptions of the group, themselves and the situation. Three themes were developed reflecting these aspects: 1) "The Gang: Perceptions, expectations and peer influence"; 2) "The Individual: Morality, autonomy and resistance to peers"; and 3) "The Environment: Making decisions in VR versus real-life". --- Morality Participants' wish to assist the passers-by (i.e., the victims of the group's antisocial behaviour) was fundamentally motivated by morality. That is, participants wanted to do what was morally 'right'. This is consistent with research highlighting adolescents' higher propensity for morally vs. personally oriented decisions (Sommer et al., 2014). In the majority of cases, this was a product of participants' moral reasoning and perspective-taking. Participants justified their decisions based on justice and the fairness of the group's actions (Sullivan et al., 2008), potentially indicative of mature moral judgment (Myyrya et al., 2010), which is linked to less delinquent behaviour in adolescents (Lardén et al., 2006). Evidence indicates the developmental emergence of moral judgement is linked to cognitive development of perspective-taking (Myyrya et al., 2010). Participants showed preoccupation for how the victims would feel, and how they themselves would feel if they were in the victim's situation, suggestive of empathy and perspective-taking. Specifically, perspective-taking ability is related to moral emotions, such as empathetic concern for the victims (Eisenberg, 2000) and guilt, which is experienced as the result of a mismatch between an individual's moral ideals and their actions (Stets and Carter, 2011). Both empathy and guilt seemed to be key drivers influencing participants' moral decisions in the VR experience. Interestingly, this pathway from perspectivetaking to moral-emotions to decision-making has been previously theorised as a deterrent of crime (Martinez et al., 2014). It was apparent that participants reflected upon and recognised the moral complexity and moral dilemmas they were presented with during the VR experience. These reflections may have been especially facilitated by the approach utilised in the present study. First, the VR experience gave participants the possibility to experience different roles based on their decisions (e.g., be involved in antisocial behaviour, or risk becoming a target by speaking up for the passer-by) which may have facilitated their perspective-taking. Second, the focus group discussions exposed participants to different perspectives, experiences and decisions from fellow focus group members with, plausibly, varying levels of empathy, attitudes towards antisocial behaviour, and experience with gangs. Altogether, this may have encouraged a deeper level of reflection, emphasising perspective-taking and moral emotions, which ultimately should support participants to act in accordance with their moral values. --- Moral identity Another driving force behind participants' decision to oppose the group and do the morally 'right thing' was that they considered it a direct reflection of the 'type of person' they were (Moshman, 2011). Hence, by acting morally, they asserted that they were a 'good person'. This reflects how morality acts as a foundation for individual identity (Malti and Ongley, 2013). In effect, moral decisions in the VR experience served as selfverification that participants were 'good people' and became an opportunity to enhance their self-esteem since they acted according to their goals and values (Sullivan et al., 2008). Given the design of the study, it remains unclear whether moral identity was an already central and important consideration for participants when entering the VR. Yet, we still observed that discussions around morality naturally emerged within the focus group discussions. This suggests that the VR setting may have created the conditions to highlight moral considerations, with these becoming the basis for participants' evaluations of themselves and the group, and ultimately influencing their decisions. --- Perceptions of the group The way participants perceived themselves also reveals a set of assumptions about the group. Upon completing the experience, most participants agreed that the group was a gang, who beyond getting involved in criminal activity was also characterised as seeking trouble. This coincides with the negative views about gangs and their behaviour expressed by young offenders, observed in other studies (Ashton and Bussu, 2020). For this reason, most participants ultimately preferred to avoid conflict with the group and tried to stick to their morals, thus reflecting their awareness that they could potentially be drawn into the gang if they became too involved (Kelly and Anderson, 2012). Overall, this alludes to young people's ability to recognise and understand group dynamics, gang norms and how peer pressure may present a pathway to gang involvement (Swetnam and Pope, 2001). Interestingly, as observed in the present and other focus groups, young people still understood that whilst they may have friends in a gang, that does not mean they also have to join (Annan et al., 2022). Beyond peer pressure, individual factors were recognised as important aspects contributing to gang involvement. --- Autonomy vs. peer influence Making decisions in the VR presented an opportunity for young people to assert their individuality and autonomy. By following their moral values and making morally motivated decisions, participants highlighted their separation from the group, who were seen as amoral. Since participants did not identify with the group, they were at pains to express that their decisions were independent, autonomous, and free from peer influence. Increased resistance to peer influence marks a key developmental transition in mid-to lateadolescence (Steinberg and Monahan, 2007), and is linked to lower engagement in externalising, delinquent and criminal behaviour (Allen et al., 2006;Walters, 2018). Another driver of autonomy was the absence of virtual group membership or in-group identification. With the exception of the two 'former school friends', participants were informed that they were not previously acquainted with the virtual group and therefore did not (yet) belong to it. Subsequently, fundamental differences in moral values between the participant and group further reinforced their wish to remain aloof. Participants rejected the group's identity and did not develop trust towards them, decreasing its potential to influence risk-taking behaviour (Vasquez et al., 2015;Cruwys et al., 2021). Despite rejecting the group's norms, social pressure was acknowledged by participants. Most reported feeling under pressure from the group, and were concerned by how the group would perceive and react to them. This is especially observed in participants' wish not to be seen as a "snitch", suggesting a subtle and potentially unacknowledged source of peer influence, albeit one that could arise in reference to group norms of the virtual and/or focus group (Clayman and Skinns, 2012). Considering participants' attitudes and past experiences with gangs may shed light on the strength of peer vs. individual effects. --- Cost/benefit analysis The way adolescents reflected upon the risks presented during the VR experience played a role in their decisions. One theoretical approach by Furby and Beyth-Marom (1992) proposes that risk-taking involves choosing one of various alternatives, with all alternatives associated with some form of risk or potential loss. Decisions in the VR experience incorporated distinct types and severity of virtual risk, on a variety of timescales. For example, participants could choose between morally positive, societally-acceptable virtual decisions vs. amoral, dangerous actions which nevertheless safeguarded short-term status and personal safety (Duell et al., 2018). Participants' decision-making involved evaluating the potential severity as well as the cost and benefits of each decision, which reflected their riskperception (Graham et al., 2017). In the virtual setting, it appeared that participants were willing to oppose the group and incur personal risks to help the victims. Participants felt it was possible for them to disagree with the group with relatively few consequences, or leave the group altogether without repercussions. Therefore, the risks they could face for opposing the group were perceived as less severe than the risks of following the group. Partly, this may be related to adolescents' lower aversion to ambiguity and uncertainty (i.e., adolescents are more willing to take risks if the outcome is unclear; Blankenstein et al., 2016). It may also reflect the VR setting, with minimal real-world consequences of opposing the virtual group. --- Virtual vs. real world peer influence The VR setting and its features can also explain how, despite peer pressure being recognised by participants, it did not eventually influence them to go with the group, but rather encouraged them to go against it. Particularly, the types of peers and peer relationships presented within the VR may have interacted with individual factors to influence participants' decisions. First, research indicates that peer similarity, such as similar levels of delinquency, may account for peer selection and influence over time (Laninga-Wijnen and Veenstra, 2021). As participants were introduced in the VR scenario, they did not initially have enough information about the group to establish the presence of similarities. Instead, most seemed to highlight their dissimilarities, potentially accounting for participants' decision not to befriend the group and limiting its influence. Furthermore, young people felt able to oppose the group given that there were no real-world consequences; and equally, there was no potential to gain real-life rewards, social or otherwise, from alignment with the virtual group (Chein et al., 2010). Instead, the researchers did observe group processes and peer influence during the focus group discussions, particularly in the form of group polarization and social desirability. The group norms appeared to mainly be of low tolerance to violence, which may have influenced participants to report similar decisions (de Boer and Harakeh, 2017). That is, some participants may have masked their true virtual decisions in order to conform to the focus group majority. This is a well-documented limitation of adopting a focus group format in research with adolescent participants (Adler et al., 2019). However, it also represents a strength and potential advantage of the group format of TiE workshops making use of the VR experience (Round Midnight Ltd, 2019): First, it may enable identification of individuals who are less resistant to peer influence (both positive and negative; i.e., a highly susceptible subgroup), by comparing focus group responses with decisions recorded on the VR headsets. Second, the group format of the TiE workshops may facilitate 'positive' peer influence via peer discouragement (Cavalca et al., 2013), with potential for real-world benefit. --- Limitations and suggestions for future research It is important to recognise that risk-taking and antisocial behaviour is often not a simple product of morality and peer influence. Indeed, it depends on additional individual, psychological and socio-cultural factors; for instance, a person's willingness and predisposition to engage in antisocial behaviour, their position in the relevant social network (DeLay et al., 2022), and the presence of specific risk factors or neighbourhood characteristics. The present study focused on understanding individual reasons and motivations underlying virtual decision-making during the VR. These relied primarily on participants' views and experiences. However, certain individual dimensions (e.g., participants' levels of empathy or antisocial behaviour, past experiences with gangs, decisions made in the VR) were not objectively recorded during the study. These and other additional risk factors and moderators should be explored further in subsequent research, and when translating research findings into real-world interventions. This could serve two main purposes, namely, understanding the factors and mechanisms through which the intervention may be tackling antisocial behaviour and gang involvement, as well as identifying subgroups who may benefit more from the intervention. --- Conclusion and practical implications The VR experience provided an immersive, compelling environment for adolescents to gain first-hand 'knowledge through experience' of a virtual encounter with a gang. Young people reported authentic emotional responses, and many perceived the group, setting and events to be realistic. This provided an opportunity for young people to explore decisionmaking in a safe, virtual environment, an advantage that has been highlighted elsewhere (Sütfeld et al., 2017). Participants commented on the experience of events snowballing out of control, an insight with potential real-world value. A range of consequences was experienced, both virtual and external (i.e., threat of being targeted by the gang, getting in trouble with the police), and real and internal (i.e., experience of decisions shaping one's identity, impact on self-esteem, experience of guilt by association, reflections on behaviour in a comparable real-world situation). Aspects of the VR experience promoted the 'illusion of reality', including context-realism and perspectival-fidelity (Ramirez and LaBarge, 2018;Slater, 2019). We propose that VR experiences such as this, delivered in an appropriately sensitive TiE workshop setting, provide a valuable learning experience for young people, stimulating discussion on sensitive, important topics. This is a promising avenue for research, primary prevention and early intervention which should be further explored in the future. tremendously grateful for the contributions of participating schools, teachers, parents/guardians and pupils, without whom this research would not have been possible. We hope this research contributes, in some small way, to addressing issues affecting your local communities and young people. We are grateful to Dr. Michael Larkin for helpful discussions around developing the semi-structured topic guide. --- Data availability statement The datasets presented in this article are not readily available because the focus group discussions represent personal perspectives and experiences of participants. It is therefore not deemed appropriate for the data to be made publicly available. Requests to access the datasets should be directed to the corresponding author, SBH [email protected]. --- Ethics statement The study involving human participants was reviewed and approved by the Science, Technology, Engineering and Mathematics Ethical Review Committee at the University of Birmingham. The participants' parents/guardians provided their written informed consent for them to participate in this study. Written informed consent was obtained from the individual(s) and and minor(s)' legal guardian/next of kin for the publication of any identifiable images or data included in this article. --- Author contributions SBH and JAK conceived the study. All authors contributed to study design. DB and LS collected, coded and analysed the data with JAK providing guidance. All authors contributed to the article and approved the submitted version. --- 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. --- Supplementary material The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/frvir.2023.1142241/ full#supplementary-material

Introduction: Gang involvement poses serious risks to young people, including antisocial and criminal behaviour, sexual and criminal exploitation, and mental health problems. There is a need for research-informed development of preventive interventions. To this end, we conducted a qualitative study of young people's responses to an educational virtual reality (VR) experience of an encounter with a gang, to understand young people's decisions, emotions and consequences. Methods: Young people (N = 24 aged 13-15, 11 female, 13 male) underwent the VR experience followed by semi-structured focus group discussions. Questions focused on virtual decision-making (motivations, thoughts, feelings, consequences) and user experiences of taking part. Data were analysed using Thematic Analysis. Results: Three themes were developed to represent how participants' perceptions of the gang, themselves, and the context influenced virtual decisions. Social pressure from the gang competed with participants' wish to stand by their morals and establish individual identity. The VR setting, through its escalating events and plausible characters, created an "illusion of reality" and sense of authentic decisions and emotions, yielding insights for real-life in a safe, virtual environment. Discussion: Findings shed light on processes influencing adolescent decisionmaking in a virtual context of risk-taking, peer pressure and contact with a gang. Particularly, they highlight the potential for using VR in interventions with young people, given its engaging and realistic nature.

Background National trends show that most African Americans (AAs) are Christian, express a high degree of spirituality, and regularly attend church services. 1 The black church is the spiritual and psychosocial staple for binding together the middle class and poor, culturally and religiously. 2,3 Church-based institutions are prime venues for health promotion trials because many promote healthy lifestyles through health-care ministries. 4,5 Such trials can reach a broad AA population and have great potential for reducing health disparities related to end-of-life (EOL) care. However, historical and current mistrust in researchers by AAs decreases access to these faith-based institutions. For example, medical experimentation on AA slaves during the antebellum period, the Tuskegee Syphilis Study, unethical practices by researchers in the Henrietta Lacks story, and other clinical trials are historical tragedies that created mistrust. [6][7][8][9][10] Current mistrust is exacerbated with the murder of 9 AA church leaders by a white male in a South Carolina church and the national divisive sociopolitical atmosphere. 11,12 Similar etiologies exemplify ethnic differences in underutilization of EOL palliative and hospice care by AAs compared to white Americans. 13. Empirical data document the mistrust of physicians caused by historical and current medical mistreatment of AAs. 14 Poor communication between dyads is also problematic. Serious care decision regrets and conflicts decrease when AA family members engage in quality EOL communications with health-care providers prior to the death of loved ones. 15 Comfort-focused care versus lifeprolonging treatments is more often chosen for EOL care under similar circumstances. 16 Another cause for underutilization is the lack of knowledge about advance care planning and serious EOL treatment options. 17,18 Additional causes are ethnic differences in family decision-making practices. African Americans are more likely to make serious EOL decisions through family consensus as opposed to white Americans who more often make similar decisions unilaterally. [19][20][21] Faith-based hope describes the spiritual needs of AAs when faced with medically predicted death. 22,23 Spiritual hope is expressed through liberation for strength against a variety of insurmountable pressures created from despair within their lives. 23 When AAs receive respect from medical teams based on the need for spiritual guidance, they are more likely to use hospice care over aggressive EOL interventions. Mutual support between religious communities and medical teams further enhances the use of comfort care by this minority group. 24 A church-based dementia education randomized controlled trial (RCT) is currently being conducted in 4 mega urban AA churches. The purpose is to determine the efficacy of a culturally tailored education intervention developed to improve the quality of advance care planning and informed EOL treatment decisions made by AA family caregivers of patients without decisional capacity. Cardiopulmonary resuscitation, mechanical ventilation, and tube feeding are the EOL treatments of interest. The investigation embraces communitybased participatory research (CBPR) concepts and culturally appropriate and spiritually sensitive methods throughout the study. The sample (n 1<unk>4 304 planned) is being recruited from the 4 churches: 2 intervention (n 1<unk>4 152) and 2 control (n 1<unk>4 152) sites. A leader from each church serves as a liaison to the RCT and is a collaborative member of the research team to enhance recruitment strategies and other related initiatives associated with the investigation. During the initial phase of the study, church leaders voiced mistrust concerns of researchers who conduct investigations in the faith-based institutions. As part of the startup work to initiate the trial, the investigators launched several in-depth interviews with the church leaders to explore strategies to ameliorate trust for the successful implementation of the study that is now underway. Specifically, as a collaborative team building process, interview sessions were held to identify trust rebuilding elements for the research team that was following others who violated trust of the church leaders. --- Methods --- Design Individual, face-to-face, in-depth, debriefing discussions were held with the 4 church leaders. One interview was held with each informant that lasted approximately 1.5 hours and was tape-recorded and transcribed. All sessions were held at the churches to promote candor and comfort in revealing sensitive information about trust/mistrust. Discussion guides led to the highly interactive format. The RCT, with its collaborative implementation process with the church leaders as research team members, was approved by the institutional review board at the University of Illinois at Chicago. --- Sample and Setting The sample (n 1<unk>4 4) consisted of 2 ministers and 2 deacons each re their respective church. The 2 ministers presenting were men and church pastors. The deacons were women, administrative assistants to the pastors, and directors of very active health-care ministries. Church denominations were the Church of God in Christ (1), nondenominational (1) and Pentecostal (2). They are among the largest religious institutions in the metropolitan area with attendance that ranged from 2000 to 7000 congregants with a median age of 45 years (see Table 1). --- Measures Demographic characteristics of the informants included age, gender, education, employment status, and years served in the ministry. The 6-item discussion guide included open-ended questions about prior personal experiences with researchers. --- • Tell me about your experiences with researchers? • Do you know any other pastors or church leaders who have had experiences with researchers? • Tell me about those experiences? • What did the researchers do that fostered your trust? • What did the researchers do that violated your trust? • What did they do to rebuild the trust? --- Procedures The informants were solicited to receive detailed information about the RCT and discuss strategies for the recruitment of study individuals. Mistrust concerns of researchers were initiated and described by 2 informants during the meetings. The researcher expressed dismay and a need to explore the possibility of similar concerns felt by the other informants. A second face-to-face meeting was requested and subsequently held with each informant to ask permission for an interview to address possible experiences with previous researchers that generated mistrust. Details of the discussion guide were described including in-depth open-ended dialogue and audiotaping of the sessions. Respect was stressed including confidentiality, privacy, and anonymity of information shared by the informants to the researcher. Permission was granted and interviews were later held at respective church sites. --- Data Analysis The audio-recorded discussions were transcribed verbatim. A content analysis framework was used to analyze the data. 25 Comparative thematic processes were applied for data encoding. 26 Information from each informant was compared and contrasted until the researcher was satisfied that no new issues emerged. Transcriptions were combined with interview notes recorded by the researcher and discussed in detail among the analysis group. Elements identified were then used to create themes. The themes represented a pattern of responses that shaped the content of the underlying data described by the 4 informants. Trust worthiness of the analysis process was assured with credibility of findings through representativeness of details described across informants, verbatim descriptions of mistrust experiences, and triangulation techniques, comparing consistencies of information provided against related empirical data. 27 --- Results --- Demographics The pastors were older than 43 years with doctorates in divinity from the established wellknown theological institutions. They served in the ministry over 20 years and pastored over mega-churches from 10 to 34 years. Directors of health ministry leaders were deacons, older than 55 years, and church members for more than 18 years. Both had a Bachelor of Science degree and 1 also had a Master of Science degree in nursing. --- Content Analyses Three categories emerged during analysis: experience with researchers, violation of trust, and trust building strategies. Within each of these 3 categories, themes also emerged. Experience with researchers-Positive and extremely negative themes emerged from the transcripts. Positive experience consisted of information about researchers, which was filtered to the pastors through trusted church leaders known as gate-keepers. They were deacons, elders, trustees, or ministry directors. Decisions made by gatekeepers heavily influenced the pastor's choice for confirming appointments with research team members and authorizing studies for implementation in the church. One pastor stated, "I have a pretty good experience with researchers but only because I have trusted church members to first connect with them." Another positive experience occurred when researchers took initiatives to learn about the church culture prior to requesting permission to conduct studies at the church. They obtained flyers or other public notices about health initiatives sponsored by the church and attended the events. In addition, they volunteered for service at future events. Over time, positive relationships formed among researchers and church leaders, which opened discussions for researchers to describe their studies and solicit collaborative partnerships with church leaders. Negative experiences with researchers were attributed to vague research aims and confusion over reciprocal roles between church leaders and the research team. Some unclear aims were inconsistent with the church mission when researchers clarified study goals. A pastor reported, "We would not have let them in (conduct study in the church) if we had known the goals of the study before it started. We said no to them a second time when they wanted to return (with another study)." Relationships improved with researchers when church leaders acquired the knowledge to ask appropriate questions to make informed decisions prior to study authorization. Violation of trust-Mistrust themes included paternalistic attitudes, failed promises, invisibility of principal investigators, and unethical study design. Paternalistic attitudes were exhibited by researchers who assumed complete authority over the project. Church leaders preferred collaborative processes. One participant stated, "They took data from us but did not involve us with interpretation of findings nor results from the study." Failed promises were experienced when researchers violated a contract in a blood drive study. Church leaders reported donated blood thought to be given to sickle cell babies were administered to others and not to the designated chronically ill infants. In another instance, researchers promised to train church members after completion of the study, but no follow-up training was provided. Invisibility of principal investigators was a common complaint as well. Participants stated, "They (researchers) would send AA staff members to meet with us. We never saw leaders of the research team or we saw study leaders only at the beginning of the project without any additional contact." Violation of trust was further described in an unethical study design implemented to screen church parishioners for HIV. The study purpose was to identify positive cases but failed to have protocols for follow-up treatment or referrals. When church leaders discovered no follow-up treatment plans existed, they intervened and refused access for researchers to the study site until requirements for followup were established and implemented. Within the mistrust themes were subcategories of anger, confusion, and disappointment expressed by informants toward researchers. One reported anger at researchers who failed to refer for follow-up the parishioners who were tested positive for HIV. A second expressed confusion with researchers who were unable to clearly define the purpose of their study but expressed a dire need to recruit individuals on behalf of service to the black community. Several described anger and disappointment at behaviors of researchers who continually referred to them and parishioners as "you people" as opposed to addressing them by their given name. The church leaders felt violated and betrayed stating, "They kicked us to the curb." Such negative feelings were generated from perceptions that trust was violated through negative actions exhibited by some investigators. Trust building strategies-Informants discussed successful trust building strategies that are consistent with frameworks derived from principles of CBPR and National Culturally and Linguistically Appropriate Services (CLAS). 28,29 The themes of this category include careful attention to partnership development, efforts to understand the cultural context of the church community, and reciprocity that includes plans to ensure program sustainability. Researchers were favorably received by church leaders who conducted health promotion lectures at the church site on topics entirely different from the aims of the investigation under study. The activity enhanced trust and facilitated collaboration with the church leaders. Other researchers met periodically with church leaders giving ongoing updates on progress of the study. One principal investigator attended church services and events to enhance their own cultural understanding of the black church. Some collaborated with church leaders to write grants for seed money to fund needed church-sponsored projects. Such partnerships helped build genuine relationships. One informant stated, "They (Researchers) must understand we are not like regular community-based groups, we are faith based and our goals are driven by scripture. Therefore, project goals must be aligned and consistent with the pastor's theological mission." --- Discussion The black church remains a historical and foundational structure within the AA community, and high levels of religiosity are still documented for many AAs. 30 Based on the centrality of the church and also findings that AAs use religion and connections to pastors to cope with health, it is clear why researchers desire to partner with the church to recruit minority participants and to improve health outcomes. However, a question that researchers are now forced to answer is why the church would want to partner with them. Consequently, it is critical that researchers consider how a research partnership will benefit the church-its leaders, congregation, and the broader church community. Researchers have found that pastors and lay leaders remain willing to engage with researchers to address health issues within their churches. 31,32 The positive experience theme that emerged from our findings and the willingness of the churches to participate in the larger study support that churches are indeed willing to participate in research studies. However, church leaders emphasize the need for researchers to be sensitive to the church environment and to take time to develop relationships. Unfortunately, findings from this study related to negative experience with researchers support other evidence that church leaders and members still view researchers as "taking but not giving." Fortunately, findings from this study also provide guidance related to how researchers and universities can be better stewards with black churches to build and sustain partnerships. Our findings confirm that when working with churches, similar to working with many others, it is the "relationship" that has to be built and sustained. Relationship building requires respect, consistent contact, and reciprocity. Respect, an underlying ethical principle as well as an underlying principle of CBPR and CLAS remains critical to relationship building. 28,29,33 Despite this, church leaders and community members still perceive paternalistic attitudes rather than respect from their academic partners. In addition, they state that the "faith-based" foundation of the church is not respected and considered. Throughout the process of working with churches, researchers must continually ensure that their research goals and the goals of the church are clear and mutually accepted. In a context where researchers want to discuss advance care planning and EOL with church members, they must consider how this fits with the church's mission to help and support its members. African American communities deal with death often. Writings from scholars can help researchers understand and appreciate that the community's relationship with death that is disproportionately violent death and/or untimely death. 34 Furthermore, researchers may want to explore and identify any implicit or explicit biases toward individuals who hold religious "hope" and belief in miracles as foundational beliefs. Contributing to the central tenet of respect for the church and its members and community is consistent face-to-face contact with principal investigators and leaders of the research study. Findings still show that church leaders want to meet and engage with the principal investigator and other leaders of the study. 35,36 The importance of this to church leaders was noted within our violation of trust category. Inconsistent contact with leaders of the research study not only can adversely affect the research relationship but can also affect recruitment efforts and have a negative effect on willingness of churches and congregation to engage in other research partnerships. Church gatekeepers, including pastors and others, are increasingly more likely to insist that they as researchers give back to the church and/or community. Importantly, if they are to indeed create partnerships, it is critical that they collaborate from the beginning and that the partnership is beneficial for all. Specifically, noted within our mistrust theme was the desire to be involved in the interpretation of findings from studies that involved the church. Efforts to help the church with applying for other grants or leaving resources in place for the church to continue to help its congregants and communities to reduce risk and/or overcome obstacles can be helpful. 37 Such trust building strategies contributed to sustainability of church programs. Other researchers have found that expanded opportunities for professional development, exposure, and networking also work. 38 In the language of many black churches, researchers and universities must be good stewards. --- Conclusions Church-based institutions in the AA communities are prime venues for health promotion trials because many churches have health-care ministries that promote healthy lifestyles. Although such trials can reach a broad AA population and have great potential for reducing health disparities, historical mistrust in researchers decreases access to these institutions. Trust rebuilding strategies were identified for researchers to follow, which may facilitate amelioration of relationships with AA church leaders. Utilizing principles of CBPR, CLAS, and general moral practice is required for successful trust building. If the principles are violated, threat to future hospice and palliative care research within AA church-based institutions may prevail. Thus, potential benefits are thwarted for the church members, AA community, and advancement of the EOL care scholarship. --- Author Manuscript Bonner et al. Page 11

Background-An initial and vital important step in recruiting participants for church-based hospice and palliative care research is the establishment of trust and credibility within the church community. Mistrust of medical research is an extremely important barrier hindering recruitment in African American (AA) communities. A church-based EOL dementia education project is currently being conducted at four large urban AA churches. Church leaders voiced mistrust concerns of previous researchers who conducted investigations in their faith-based institutions. We explored strategies to ameliorate the mistrust concerns. Specific aim-To identify trust-rebuilding elements for researchers following others who violated trust of AA church leaders. Methods-Face-to-face, in-depth interviews were conducted from a convenient sample of four established AA church leaders. Interviews were held in the informants' churches to promote candor and comfort in revealing sensitive information about trust/mistrust. Content analysis framework was used to analyze the data. Elements identified from the analysis were then used to create themes.

cultural practices and constrained financial resources, some ethnic minority families in China still lack toilets [4]. Furthermore, the relationship between household toilet accessibility and the health of adolescents in China's ethnic minority areas has been underexamined in the existing literature. Using two waves of data from the China Education Panel Survey (CEPS), this study explored the effect of household toilet accessibility on Chinese ethnic minority adolescents' physical health. There is a wealth of research on the effects of toilet sanitation facilities on the health of adolescents worldwide, and many have found that toilet accessibility is closely linked to adolescents' health outcomes [5][6][7][8][9][10]. For instance, a case study from Bengaluru, India, revealed that adolescent girls who lack access to toilets have to use open defecation, which makes them prone to reproductive tract infections, urinary tract infections, and snake bites [5]. In addition, a study conducted in India using school-going adolescents as a sample group indicated that access to toilets could significantly reduce incidences of diarrheal [6]. According to empirical research conducted in Venezuela, a lack of toilet accessibility could increase sexual violence and negatively affect the physical and psychological health of young female adolescents [7]. A qualitative study conducted in the United States found that restricted toilet use might significantly affect bladder health among adolescents [8]. Furthermore, some studies have revealed that the quality and dignity of toilets also have an important effect on adolescents' health [9,10]. Specifically, a study of 13-to-17-year-old students in Bahia, Brazil, using semi-structured interviews, argued that toilet acceptability, quality, and safety were also essential for adolescents to avoid the risk of diarrhea [9]. According to a Danish empirical study that recruited adolescents aged 11.2 years on average, poor-quality toilets were significantly correlated with students' toilet refusal and bladder and bowel dysfunction symptoms [10]. However, studies on the effect of household toilet accessibility on the physical health of adolescents in China's ethnic minority groups are relatively sparse. The toilet facilities of Chinese families are undergoing changes and development. In the 1960s, the Chinese government initiated a patriotic public health campaign in an effort to minimize infectious diseases and improve health conditions. This campaign involved encouraging people to use toilets and wash their hands afterwards [4]. In the 1980s, the Chinese government promoted the "Three Improvements" project, which included improvements to toilet facilities, water quality, and health education in rural areas [4]. Moreover, the "Toilet Revolution" has become a hot phrase since 2015 [11]. Since then, it has been substantially implemented in China as a campaign to ensure the hygienic separation of human excreta and environmental pollution from human contact, and to realize resource recycling [12]. However, implementing toilet improvement projects and the "Toilet Revolution" in China's ethnic minority areas has been difficult. For example, in 2017, the prevalence of sanitary toilets in the rural areas of the Ningxia Hui and Xinjiang Uyghur Autonomous Regions was 74.3% and 65.6%, respectively, which was much lower than the average prevalence of sanitary toilets in the country (81.8%) [13]. The local natural environment, economic conditions, and people's opinion on toilets all contribute to the low prevalence of sanitary toilets in ethnic minority communities. The inhabited areas of China's ethnic minorities are typically remote both environmentally and economically. The natural environment is harsh, and economic development is relatively slow [14,15]. In China's ethnic minority areas, implementing toilet improvement projects faces not only the technical challenges of climate (e.g., freezing temperatures or drought), but also the financial constraints of maintaining the normal functioning of toilet sanitation facilities [4]. From the perspective of public opinion, each ethnic group has its own toilet civilization with a distinct and lengthy history of fecal disposal and toilet use. Consequently, cultural clashes among ethnic groups frequently occur during toilet improvement projects. For instance, the Yi ethnic people, who respect fire, tend to oppose the promotion of household toilets with biogas digesters [16]. They regard biogas fires that are produced using human and animal feces as filthy, and they will not use these for heat and cooking [16]. In addition, Tibetans clearly distinguish between inside and outside items [4]. Outside items are filthy, and, to a certain extent, feces from the human body belong outside [17]. Therefore, they are hesitant to build toilets in their houses and generally lack them [4]. Overall, toilet improvement in China's ethnic minority areas is not only a vital public health issue, but also a profound social and cultural movement that embodies the contradiction between modern public health concepts and traditional perspectives [18]. In the existing literature, however, few studies have attempted to explore the effect of household toilet accessibility on the health of adolescents in China's ethnic minority areas. Thus, this study's contributions are as follows: First, using two waves of CEPS data (2013 and 2014), we investigated the effect of household toilet accessibility on the physical health of ethnic minority adolescents in China (including but not limited to Mongolian, Tibetan, Yi, Manchu, Hui, Zhuang, and Uyghurs), which constitutes a group worthy of attention but relatively ignored in previous studies. Second, we determined the heterogeneous health effects of different toilets (including soil cesspit, cement cesspit, squat toilet, and flush toilet) on ethnic minority adolescents. Third, we examined the moderating role played by different family living arrangements (i.e., adolescents who live with parents and those who live without parents) in the effect of household toilet accessibility on ethnic minority adolescents' physical health. The findings of this study have policy implications for the promotion of the "Toilet Revolution" and provide guidelines for the healthy development of adolescents in China's ethnic minority areas and other developing countries and regions. --- Methods --- Data Data of this study were obtained from the CEPS, which was conducted by the National Survey Research Center at the Renmin University of China. The CEPS is a nationally representative longitudinal survey with two waves (i.e., CEPS2013 and CEPS2014) that examines the influences of family, school, community, and society on the development of Chinese adolescents. The first wave of CEPS2013 was conducted anonymously and confidentially during 2013-2014, with a follow-up investigation during 2014-2015 [19]. Using a stratified, multistage sampling design with a probability proportionate to size, the CEPS randomly selected a school-based, nationally representative sample of seventh to ninth graders and their parents nested in 28 county-level units in China [20]. In the Chinese educational regime, the seventh to ninth graders are generally secondary school adolescents between 13 and 15 years of age. Thus, the CEPS provides an appropriate dataset to explore the effect of household toilet accessibility on adolescents' physical health. We cleaned the data by excluding missing items, outliers, and other abnormal values. Furthermore, since our focus was the effect of household toilet accessibility on the physical health of ethnic minority adolescents, respondents of the Han ethnic group were excluded, as Han is China's majority nationality. Finally, 576 ethnic minority seventh to ninth graders were matched across the CEPS2013 and CEPS2014. Approximately 46% of the ethnic minority students in our sample were male, and the mean age of the samples was 13.477 years old. --- Measures --- Dependent variables In this study, the dependent variable was the physical health of ethnic minority adolescents. This was measured by asking the parents of these adolescents, "How would you rate your child's physical health?" On a fivepoint Likert scale, the response of "very unhealthy" was coded as "1", "somewhat unhealthy" as "2, " "normal" as "3, " "somewhat healthy" as "4, " and "very healthy" as "5. " To test the robustness of the empirical results, we used the adolescents' self-rated physical health and height as alternative dependent variables. The variable of adolescents' self-rated physical health was acquired by asking these adolescents, "How would you rate your current physical health?" On a five-point Likert scale, the answer of "very unhealthy" was coded as "1, " "somewhat unhealthy" as "2, " "normal" as "3, " "somewhat healthy" as "4, " and "very healthy" as "5. " In general, adolescents' body height is closely linked with their physical health [21]. Thus, adolescents' height during the survey year was also employed as an alternative dependent variable. --- Independent variables The toilet accessibility was a binary independent variable in this study. These ethnic minority adolescents were asked, "Does your house have a toilet?" The answer of "Yes" was coded as "1"; otherwise, "0" was used. To capture the heterogeneous health effects of different forms of toilet accessibility on ethnic minority adolescents, we obtained information on toilet type by further asking these adolescents whose houses have toilets: "What kind of toilets do you have in your house?" The responses to the question were categorized as follows: (1) soil cesspit, (2) cement cesspit, (3) squat toilet, and (4) flush toilet. In the empirical analysis, the toilet type was treated as a categorical variable. --- Moderating variable The family living arrangement is a vital factor in the improvement of family facilities and adolescents' physical health [22,23]. In some economically underdeveloped ethnic minority areas, the young labor force largely flows out to seek jobs, resulting in intergenerational co-residence and left-behind children [24]. Compared to adolescents who reside with their parents, adolescents who do not reside with their parents may receive less family health education and healthcare intervention, and may therefore have worse physical health. Thus, to test the moderating role of the family living arrangement in the effect of household toilet accessibility on the health of ethnic minority adolescents, we classified the family living arrangement into two categories: (1) living with parents and (2) living without parents (including living alone and living with grandparents, relatives, and others). Living with parents was coded as "1"; otherwise, it was coded as "0. " --- Control variables To precisely determine the effect of household toilet accessibility on ethnic minority adolescents' physical health, we controlled for the variables that could affect their physical health. First, family economic conditions may be prerequisites for adolescents to acquire healthy foods and better medical services [25][26][27][28][29][30][31]. Thus, we controlled the three variables of family financial condition, family income level, and parents' occupations. Second, the living environment is an important external factor that may affect the physical health of ethnic minority adolescents [32][33][34]. Thus, we controlled for environmental sanitation and environmental pollution around adolescents' living houses. In addition, we considered the health effects of safe water and controlled for the variable of tap water. Third, adolescents' lifestyle behaviors were considered [35]. Although we did not find questions about adolescents' lifestyle behaviors in the CEPS, we found a variable that may indirectly reflect adolescents' lifestyle behaviors-close friends' smoking and drinking behaviors. Due to the influence of peers, adolescents' lifestyle choices may be impacted by close friends' smoking and drinking habits to a large extent [35,36]. We thus controlled for adolescents' close friends' behaviors of smoking and drinking. Table 1 presents the definitions and descriptive analyses of these variables. --- Statistical analyses --- Fixed effects ordinary least squares model As the key dependent variable of this study was ethnic minority adolescents' physical health, measured on a five-point Likert scale, adolescents' physical health can be considered continuous data. The results of the Hausman test (Chi 2 = 32.010, p <unk> 0.01) rejected the null hypothesis of a random effect. Consequently, we investigated the effect of household toilet accessibility on ethnic minority adolescents' physical health by estimating the following fixed effects ordinary least squares (OLS) model: where the dependent variable is Adolescents ′ physicalhealth it, it represents the physical health of ethnic minority adolescents. Toilet accessibility is a binary indicator and represents the independent variable. <unk> 0 is the intercept, and <unk> 1 is the coefficient for hav- ing a household toilet. Z jit indicates the j time-varying control variable that may affect ethnic minority adolescents' physical health. <unk> i and year t are the fixed effects of individual and year, respectively. <unk> c is for the fixed effect of counties where the adolescents lived, and <unk> it is the error term. Equivalent to a dummy variable for each ethnic minority adolescent, <unk> i captures all time-constant (1) characteristics of an adolescent, even when these characteristics are unobservant. With year t and <unk> c as the equiv- alent of the dummy variables for each survey year and the living county of adolescents, the fixed effects model further accounts for any year-to-year and county-to-county changes that may affect adolescents' physical health [37]. Adolescents physicalhealth it = 0 + 1 Toiletaccessibility it + <unk> y j Z jit + i + year t + c + it --- Difference-in-differences combined with propensity score matching method Using a fixed effects model, we obtained the effect of household toilet accessibility on the physical health of ethnic minority adolescents without the influences of individual time-invariant factors and year-to-year and county-to-county shifts to some extent. To capture the "true" health effect of household toilet accessibility on adolescents, we need to compare the health outcomes of adolescents whose houses have toilets with those of adolescents whose houses do not have toilets. However, there may be systematic differences among these adolescents. If we compare the two groups of respondents directly, the estimation results may be biased owing to systematic differences in the samples. Consequently, we face the challenge of a lack of counterfactual information. The difference-in-differences (DID) combined with propensity score matching (PSM) method is the preferred method for constructing a counterfactual framework and precisely obtaining the health effect of household toilet accessibility on ethnic minority adolescents [38,39]. To construct an appropriate control group (respondent's house does not have a toilet) and compare outcomes to those of the treatment group (respondent's house has a toilet), we used the PSM estimation based on the propensity scores. With the PSM estimation, we can identify individuals in the control group with similar characteristics as those in the treatment group and exclude observations that were not on common support [39]. In the analysis, the physical health of adolescents was considered a continuous variable. Consequently, based on the observations of common support of the PSM estimation, the following equation of the DID model was derived for the continuous outcomes of adolescents' physical health: where Y it denotes the physical health status of ethnic minority adolescents i in year t. Toilet accessibility it is a binary indicator and represents a dummy variable. Toilet accessibility it = 1 represents the treatment group in which the house of ethnic minority adolescents had toilets, whereas Toilet accessibility it = 0 indicates the control group in which the house of ethnic minority (2) Y it = 1 + 2 Toiletaccessibility it + 3 After t + Toiletaccessibility it <unk> After t + yX jit + year t + it adolescents lacked toilets. After t represents a time dummy variable. After t = 0 means 2013, and After t = 1 represents 2014. Toiletaccessibility it <unk> After t. is the inter- action between groups and time. X jit is the j covariate of ethnic minority adolescent i at time t. year t indicates the fixed effect of the survey year. <unk> it means error term. <unk> denotes the average treatment effect, also known as the DID value. <unk> is the coefficient in the DID model in which we are most interested. --- Results --- Results of fixed effects ordinary least squares model The results of the fixed effects OLS model are presented in Table 2. As shown in Table 2, toilet accessibility was significantly and positively associated with ethnic minority adolescents' physical health (<unk> = 0.306, p <unk> 0.01) when control variables were omitted from the model. Furthermore, after controlling for control variables and the fixed effects of individual, year, and county, toilet accessibility remained significantly and positively correlated with ethnic minority adolescents' physical health (<unk> = 0.337, p <unk> 0.01). These results indicated that com- pared with adolescents whose families did not have toilets, adolescents whose families had toilets had better physical health. Table 2 shows that the control variables of parents' occupations y = 1.051, p <unk> 0.01, environmental sanita- tion y = 1.093, p <unk> 0.05 and environmental pollution y = 0.090, p <unk> 0.1 around adolescents' place of resi- dence, and family living arrangement y = 0.278, p <unk> 0.01 had significant and positive influences on ethnic minority adolescents' physical health when utilizing the fixed effects OLS model. --- Heterogeneous effect of different toilet types Table 3 presents, using the fixed effects OLS model, the heterogeneous effects of different toilet types (including soil cesspit, cement cesspit, squat toilet, and flush toilet) on the physical health of ethnic minority adolescents. Accounting for the control variables and the fixed effects of individual, year, and county, the coefficients of cement cesspit (<unk> = 0.552, p <unk> 0.01), squat toilet (<unk> = 0.833, p <unk> 0.01), and flush toilet (<unk> = 1.162, p <unk> 0.01) were positive and significant. Among the three toilet types, the coefficient of the flush toilet was the highest, whereas that of the cement cesspit was the lowest. Overall, in comparison to the soil cesspit, the three types of toilets (i.e., cement cesspit, squat toilet, and flush toilet) can improve ethnic minority adolescents' physical health more efficiently. Moreover, the flush toilet had the most positive effect on health, followed by the squat toilet, and the soil cesspit had the worst effect. --- Moderating effect of family living arrangement Figure 1 shows the average marginal effect of household toilet accessibility on the physical health of adolescents from different family living arrangements. Figure 1 shows that the 95% confidential interval (CI) of adolescents who live with their parents was [0.029 -0.443], and the average marginal effect was 0.236. Moreover, the 95% CI of the adolescents who live without parents was [0.334 -1.008], and the average marginal effect was 0.673. These results reveal that the average marginal effects of household toilet accessibility on the physical health of ethnic minority adolescents who lived with grandparents, relatives, and others were more evident and significant than those who lived with their parents. --- Results of difference-in-differences combined with propensity score matching The estimation results for DID combined with PSM are presented in Table 4. Using the three matching algorithms: nearest neighbor matching <unk> = 0.335, p <unk> 0.01, radius matching <unk> = 0.335, p <unk> 0.01, and kernel match- ing <unk> = 0.335, p <unk> 0.01, the results of DID combined with PSM implied that the household toilet accessibility could significantly improve ethnic minority adolescents' physical health. These results are consistent with those estimated using the fixed effects model. Table 5 provides a comparison of the samples after PSM. Except for the key outcome variable of adolescents' physical health, the p values of other variables' mean differences between the control and treatment groups were greater than 10% after applying the PSM estimation. Hence, after matching the control and treatment groups, the findings indicated that, except for the key outcome variable, there were no significant differences among the variables, and the DID combined with the PSM model was correctly specified. --- Robustness checks To test the robustness of the results, subgroup regressions were conducted, the estimation method was changed, and the dependent variable was replaced. First, we considered the gender of ethnic minority adolescents and their parents' occupations. The full sample was categorized into boys and girls based on gender. The ethnic minority adolescents were divided into two subgroups according to their parents' occupations: nonagricultural and other. Table 6 provides the different subgroups regression results using the fixed effects OLS models. When the control variables and fixed effects were controlled for, as indicated in Table 6, the toilet accessibility was significantly and positively related to the physical health of ethnic minority adolescents in Fig. 1 Moderating effect of family living arrangement. a The results are estimated by the fixed effects (individual, year, and county) ordinary least squares model added the interaction term of toilet accessibility and family living arrangement. b "1" means the adolescents living with parents, and "0" indicates the adolescents living with grandparents, relatives, and other. c CI means confidence interval. d The 95% confidential interval of adolescents who live with parents is [0.029 -0.443], and the average marginal effect is 0.236. The 95% confidential interval of the adolescents who live with grandparents, relatives, and others is [0.334 -1.008], and the average marginal effect is 0.673 Table 4 Results of difference-in-differences combined with propensity score matching by employing three matching algorithms a Standard errors are in the parentheses b *** Significance level at 1% c The coefficient of Toilet accessibility <unk> After is <unk> (i.e., DID value), the coefficient of Toilet denotes <unk> 2, and the coefficient of After is <unk> 3 in Eq. (2) d The element number of the nearest-neighbor matching with a caliper was 1, the radius was set to 0.01 in radius matching, and kernel matching used default kernels and bandwidth the subgroups of boys (<unk> = 0.333, p <unk> 0.05) and girls (<unk> = 0.366, p <unk> 0.01), and subgroups of parents with non-agricultural occupations (<unk> = 0.480, p <unk> 0.01) and other (<unk> = 0.261, p <unk> 0.05). Thus, these results are con- sistent with those estimated by using the entire sample. Second, because the dependent variable of ethnic minority adolescents' physical health on a five-point Likert scale can also be viewed as ordered and discrete data, we used an alternative estimation method of the fixed effects ordered probit model to confirm the effect of household toilet accessibility on ethnic minority adolescents' physical health. Table 7 displays the results of the fixed effects ordered probit model. These results demonstrated that household toilet accessibility continued to have a significant and positive effect on adolescents' physical health (<unk> = 0.911, p <unk> 0.01). Third, the dependent variable of adolescents' physical health was replaced with their self-rated physical health and body height. Table 7 shows the results of the robustness check for substituting the dependent variable using the fixed effects OLS model. The results indicated that household toilet could have a significant and positive effect on adolescents' self-rated physical health (<unk> = 0.258, p <unk> 0.01) and body height (<unk> = 1.896, p <unk> 0.01), suggesting that the positive effect of household toilet accessibility on ethnic minority adolescents' physical health was valid across different methods and models. --- Discussion The toilet is one of the most significant factors in expanding the human lifespan [40]. A previous study has argued that the contribution of modern sanitary facilities to public health is more important than that of antibiotics, vaccines, or anesthesia [4]. Adolescents are at a crucial stage of growth and development, and the surrounding sanitary environment has a profound effect on their health. The effect of household toilet accessibility on the physical health of adolescents in China's ethnic minority areas has been underfocused and underexamined in the existing literature. Using two waves of data from the CEPS, we examined the effect of household toilet accessibility on the physical health of adolescents from Chinese ethnic minority groups. We found that household toilet accessibility had a significantly positive effect on the physical health of ethnic minority adolescents. Our results are similar to those of previous studies that focused on the influences of family or school sanitary facilities on the health of adolescents in South Asia, South America, and Europe [5][6][7][8][9][10]. Although these conclusions are consistent, the toilet accessibilities in China's ethnic minority areas cannot be improved instantaneously. There is a blind spot in technological rationality; namely, the civilization and history, which is often the internal logic behind the failure of modern technology projects that aim to improve people's quality of life [16]. When focusing on toilet issues in China's ethnic minority areas, the implementation of the "Toilet Revolution" policy and related projects should be founded on a complete understanding and respect for the civilization and history of the particular ethnic group. For instance, local government can use successful toilet improvement cases to publicize the health benefits of using household toilets and gradually guide people to adopt more hygienic public health practices. Moreover, the government can try to rely on the local elites in ethnic minority areas as role models to motivate people's enthusiasm for toilet improvements. To promote ethnic minority adolescents' physical health more comprehensively, it is not sufficient to simply provide household toilets; the quality and dignity of household toilets are also crucial. Compared with the soil cesspit, the toilet types of cement cesspit, squat toilet, and flush toilet could more effectively improve ethnic minority adolescents' physical health, especially the flush toilet. However, the harsh climatic and environmental conditions of ethnic minority areas are not conducive to the use of flush toilets, and climatic and environmental barriers are difficult to control. This may explain why there are fewer flush toilets in these ethnic minority areas. To more efficiently improve ethnic minority adolescents' physical health, alternatives for flush toilets should be widely developed. For example, central and local governments could proactively promote environment-friendly toilets (such as vacuum toilets and water-free toilets) for households in ethnic minority areas, rather than limiting the usage of these environment-friendly toilets to scenic spots [41]. Such toilets can work just as well as flushing toilets and simultaneously have the advantages of saving water, reducing noxious odors, and effective disposal [42]. Meanwhile, local governments need to provide households in ethnic minority areas with a method of installing environment-friendly toilets and establish a long-term maintenance system for these toilets by utilizing market mechanisms. Moreover, in the "Toilet Revolution", ethnic minority adolescents living alone and with grandparents, relatives, and others should receive more attention than those living with their parents. In the analysis of the moderating effect of the family living arrangement, we discovered that the physical health of adolescents living without parents was more likely to be significantly affected by household toilet accessibility than the health of those living with their parents. China has over six million left-behind children under the age of 16 whose parents have left rural areas for urban employment in cities [43,44]. Of these, 96% lived with their grandparents [44]. To a certain extent, the health and lifestyles of most adolescents living without parents may be influenced by their grandparents, whom themselves have a poor understanding of health and hygiene. Therefore, if the household sanitary facilities of ethnic minority adolescents who live without parents can be improved and their school health education is strengthened, their physical health may be significantly improved. We also found that among the regression coefficients of the control variables, the regression coefficient of parents' occupations was large and significant. This implies that parents' occupations have a significant influence on children's physical health. Compared to parents who are farmers or unemployed, parents who are engaged in non-agricultural work may directly improve the family economic and provide better opportunities for children to obtain healthcare services. Therefore, it is so economically beneficial for parents to have non-agricultural jobs to promote the physical health of their children. Furthermore, in the robustness regression that used a gendered split sample, the variable of parents' occupations was very significant for girls' physical health, but not for boys. China has a historical tradition of prioritizing boys over girls; however, this situation has gradually disappeared with the improvement in parents' educational levels and economic conditions [45]. Thus, if adolescents' parents are engaged in non-agricultural jobs, they may have higher educational levels and better economic conditions than parents who are farmers or unemployed, and their children, especially girls, can be given better access to health services and have good physical health. This study had some limitations that deserve mention. First, each ethnic group has its own toilet civilization; therefore, we recommend improvements in future research. Focusing on a specific ethnic minority group can reveal the history and development of household sanitary facilities in greater depth and provide more practical policy implications. Thus, future studies should investigate the effect of household toilet accessibility on adolescents' physical health from the perspective of a particular ethnic group. Second, restricted by the data availability, we utilized adolescents' close friends' lifestyle behavior as an alternative variable for adolescents' lifestyle behavior based on peer influence. If the data are available, future studies should directly control for variables related to adolescents' lifestyle behaviors. --- Conclusions Employing two waves of nationally representative data from the CEPS, this study explored the effect of household toilet accessibility on Chinese ethnic minority adolescents' physical health. By using the fixed effects OLS model and DID combined with the PSM method, we found that there was a statistically significant and positive impact of household toilet accessibility on Chinese ethnic minority adolescents' physical health. Compared with soil cesspits, the toilet types of cement cesspit, squat toilet, and flush toilet could improve adolescents' physical health substantially, and the flush toilets could improve adolescents' physical health the most. Moreover, the family living arrangement played a moderating role in the effect of household toilet accessibility on adolescents' physical health. The physical health of ethnic minority adolescents who lived without their parents was more evidently and significantly affected by household toilet accessibility than that of those who lived with their parents. Therefore, improving household toilet accessibility in China's ethnic minority areas is essential for promoting adolescents' health. --- Availability of data and materials The study utilizes secondary sources of data that are freely available in the public domain through http:// www. cnsda. org/ index. php?r= proje cts/ view& id= 61662 993. Those who wish to access the data may register at the above link and thereafter can download the required data free of cost. --- Abbreviations --- CEPS --- Declarations Ethics approval and consent to participate All data utilized in this study are secondary sources of data and freely available in the public domain. The ethics approval and consent to participate of this study was not applicable. All the methods were carried out in accordance with relevant guidelines and regulations. --- Consent for publication Not applicable. --- Competing interests The authors declare no conflict of interests. • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background Accessing household toilets is vital for adolescent health during crucial stages of growth and development; however, some ethnic minority families in China lack toilets. The relationship between household toilet accessibility and the physical health of adolescents in China's ethnic minority areas has been underexamined.Using two waves of data from the China Education Panel Survey (2013 and 2014), this study explored the effect of household toilet accessibility on Chinese ethnic minority adolescents' physical health. The sample consisted of 576 ethnic minority adolescents with an average age of approximately 13 years. To test the health effect of household toilet accessibility on ethnic minority adolescents, a fixed effects ordinary least squares model and difference-indifferences combined with propensity score matching method were used.The results of the fixed effects model and difference-in-differences combined with propensity score matching showed a significant and positive effect of household toilet accessibility on adolescents' physical health. The heterogeneity analysis indicated that among the toilet types of soil cesspits, cement cesspits, squat toilets, and flush toilets, the flush toilets were proven to be the most efficient in improving adolescents' physical health. Moreover, the family living arrangement was proven to play a moderating role in the effect of household toilet accessibility on the physical health of ethnic minority adolescents. The average marginal effects of household toilet accessibility on the physical health of ethnic minority adolescents who lived without parents were more evident and significant than those who lived with their parents.Promoting household toilet accessibility in China's ethnic minority areas is essential for improving adolescents' health. Moreover, having a household toilet is not sufficient, the quality and dignity of household toilets are also important.

Introduction Sexually transmitted infections (STIs) are a growing issue in the United States, with over 26 million new cases in 2018 alone [1]. Emerging adults (EAs; age 18-25) comprise 25 percent of the sexually active adult population and yet they make up 50 percent of all new STI cases [2]. Emerging adulthood is a transitional time, often marked by personal and sexual exploration [3] and the development of personal health behaviors [4] as adolescents navigate their way into adulthood. During this period, EAs are more likely to partake in "risky" sexual behavior, such as having multiple and simultaneous sexual partners and using protection less frequently [5] perpetuating the spread of STIs [6,7]. Additionally, the Centers for Disease Control and Prevention (CDC) has noted that EAs' lack of STI screening increases their risk of contraction [1]. Several demographic characteristics may lead to this behavior [8], including a sense of invulnerability and perceived barriers to testing. The low testing rates among emerging adults, and low levels of disclosure have been attributed in part to shame and stigma [8]. However, some research has found a lack of shame or stigma toward STIs in general, talking about STI testing with health care providers, and getting tested [9,10]. Perceived shame (the expectation of society's negative reaction) and stigma (the perception of failing to meet social standards) involve the culture an individual is part of. The lack of these reactions may be due to a variety of factors, including the "hook-up culture" among this demographic, specifically on college campuses [11], increased sex education, and/or the perception that STIs are treatable, if not curable [1]. There has also been increased awareness surrounding HPV and its prevalence in recent years [2]. Although this is concentrated on one specific STI, such awareness may decrease the shame and stigma surrounding the medical and communicative aspect (specifically talking to health care providers) of STIs as a group. Thus, unwillingness to get tested may not come from felt shame and stigma of the STI. Despite many EAs engaging in, or being surrounded by, a "hook-up culture" [11], this demographic remains uneasy when talking about sexual behaviors, sexuality, and STIs to sexual partners and/or parents when seeking support. Noar and colleagues [12] argue that open communication can promote safe sex practices and healthier decision making. However, sex communication, condom negotiation, and STI disclosure are considered by many to be daunting and uncomfortable tasks. Therefore, the aim of this study is to delve into sex communication between EAs and their parents, monogamous sexual partners, and casual sexual partners, as the shame and stigma may potentially be more closely related to the process of engaging in conversation about STIs and disclosing an STI status than the STI itself. --- Literature Review 2.1. Emerging Adults Arnett [13,14] argues that continued reliance on parental figures, in conjunction with falling within the age range of 18-25, defines an individual as an emerging adult. Oftentimes this dependency is connected to the notion that EAs are prolonging their time until full adulthood and are attending college longer, and thus in need of financial support [14]. Relevant to this study, EAs also continue their dependence on parents' health insurance plans. The onset of the Affordable Care Act (ACA) allows EAs to remain under their parents' health insurance through 26 years of age. Immediately following the approval of the ACA, more than three million EAs who did not have coverage before 2010 were able to get medical coverage on their parents' plan [15]. EAs have the highest rate of uninsured than any age group at 30% and have the lowest rate of access to employer-based insurance as they transition into the work force in entry-level and part-time jobs, or even take positions with small businesses unable to provide employer-sponsored insurance [16]. Thus, many EAs remain on their parents' health insurance plans as long as they can due to this lack of access, as well as the lower or no cost to them, since many parents pay the premium [17]. By even subjective definitions [13], these 18-26 year olds remain steadfast within emerging adulthood. As these EAs increase the amount of financial dependence on parents, they may also be welcoming more parental involvement, specifically in times of illness or injury. That is, continued reliance on parents during emerging adulthood blurs boundaries of what information may be deemed private, and when that information is then financially supported by parents (e.g., getting parent-paid health insurance to cover STI tests or treatments), the boundaries around private information, even those around sensitive and highly personal facts, go completely out of focus. Thus, although EAs desire more restrictive boundaries around their personal and private information, like sexual behaviors, continued reliance financially and through insurance makes privacy boundaries difficult to navigate [18]. Moreover, parents may realize the importance of refraining from interfering, but providing support can be perceived by parents as permission to co-own personal details, including problematic behaviors and sexual activity [19]. On the other hand, EAs continue to feel tension [18] surrounding their relationship with their parents as it transforms during the emerging adulthood years. EAs may feel obligated to disclose sensitive information like sexual activity to receive support and/or create intimacy with a parent, and yet also want to withhold personally private information to gain, or maintain, autonomy and independence [18]. --- STI-Related Shame and Stigma Shame and stigma have the potential to discourage STI testing and treatment. Healthrelated stigma refers to the "expectations of society's negative reaction to the health condition" [20], potentially resulting in discrimination because of a particular trait. Shame is a more internally motivated reaction, defined as "a negative emotion elicited when a person experiences failure in relation to a personal or social standard, feels responsible for this failure, and believes that the failure reflects self-inadequacy" [8,21] Research has suggested that stigma and shame related to STIs are common in the U.S. [22]. Specifically, felt STI shame and stigma have also been tied to failure to test for STIs [21,23], delaying care [8,21], and discussing sexually related topics with a romantic partner or health care provider [24,25]. Morris et al. [22] continued this line of research, finding similar results among male African American low-income youth, where stigma was a greater indicator of avoiding treatment and testing than shame [22]. However, much of the research remains within specific demographics [22,26,27]. Past research using traditional measures of assessing shame and stigma towards STIs (i.e., assessing treatment, health support seeking, and testing) have shown sufficient variance of stigma and shame related to STIs [28]. However, in more recent studies, the measure for shame and stigma towards STIs remained skewed to the left from study to study [9,10]. That is, most of the participants of these studies experienced less shame or stigma towards STIs on average. This may be due to a variety of factors, including increased sex education and understanding of the infections. Many STIs are now treatable, allowing the infection to subside [1,29]. There is also increased knowledge that infections such as HPV-despite being dangerous, causing various forms of cancer, if the immune system is unable to protect itself against the virus-often go unnoticed by the individual and can disappear over time [2]. The CDC [2] indicates that over fifty percent of sexually active individuals in the United States are infected with HPV at some point in their life. Such prevalence, lack of symptoms, and ease with which to treat may decrease the shame and stigma surrounding STIs. The culture surrounding sex during emerging adulthood may have also decreased the role of shame and stigma. EAs may be more exploratory than established adults when it comes to risky behaviors, including sex [3]. Additionally, EAs are increasingly partaking in the hook-up culture, and engaging in sexual behaviors outside of committed relationships [30]. This behavior is widely accepted and typical on many college campuses [30,31], which could indicate that STIs are less of a concern than more pressing social issues (e.g., to be perceived as casual about sex). However, the hook-up culture does compound the risk of infection. Those who hook up are more likely to have multiple and simultaneous sexual partners [5] and are less likely to use, or ask for, protection [7]. Breaking these "hook-up norms" is stigmatized, potentially more than the STIs that come of it, despite the widespread dissatisfaction with them [31]. The traditional measure of STI shame and stigma [8] is to assess the stigma of getting tested for STIs and the perception of individuals who test positive. Due to the hook-up culture and increased STI education among EAs, such a measure may no longer be sufficient. That is, if EAs are better able to manage their impression [32] in a health care setting when seeking testing or treatment, the shame and stigma of an STI may be more readily observed in other communicative processes. For example, this demographic has been shown to experience uneasiness when talking about sexual behaviors, sexuality, and STIs with those close to them. In fact, despite widespread understanding that open communication about sexual behaviors can promote safer sexual decision-making [12,33], sex communication (e.g., asking questions, disclosure of STIs, asking for help) with parents and sexual partners remains limited in nature due to being condemned as taboo, uncomfortable, and daunting. --- Shame Resilience Theory Brown argues that shame, and thus similarly, stigma, is a painful experience or feeling coming from the idea that we are flawed or unworthy for some reason [34]. STIs have long been categorized as health issues that not only encourage people to discredit others, but also to feeling discredited themselves following an STI diagnosis. Goffman explains that stigma is deeply discrediting [35]; thus, as shame and stigma are social and communicative by nature [36], people feel discredited by others judging or discriminating them due to their stigmatizing flaw. Charmaz elaborates on these ideas of shame and stigma since illness makes people vulnerable to negative social identifications, but also their own selfdefinitions [37]. They feel discredit-able, and that is enough for them to perceive felt shame and stigma. Brown proposes that people can become resilient to shame and stigma by decreasing their own feelings of being trapped, powerless, and isolated [34]. She continues that resilience comes with an increase in experiencing empathy and connection with others [35], an inherently communicative process. This connection allows for people of similar identities to have "mutual support, shared experiences, and the freedom to explore and create options" [34]. Although not all EAs are diagnosed with an STI, it is possible that EAs are already connected to one another by not only being of a similar demographic (i.e., similar age range, reliance on parents, in school, etc.), but also being a part of the hook-up culture. Shame and stigma literature usually examines how people cope with and communicate about their perceived experiences when it comes to a diagnosis, disability, skin color, or religion. However, potentially being young and seemingly promiscuous is stigmatizing in and of itself, and others stigmatize them based on this supposed blemish of character (e.g., promiscuous sex). Based on shame resilience theory [34], EAs may recognize this vulnerability to the "outside world," form connections, are somewhat isolated on college campuses, and are empathetic of the outcomes of their behaviors, including those that may cause more stigmatizing identities, like an STI diagnosis. In other words, EAs may preemptively be resilient to STI shame and stigma due to the resilience they developed as a coping mechanism to deal with older adults' stigmatization of their college lifestyle. Thus, this resilience allows EAs to face other shame and stigma traditionally associated with additional stressful events, like disclosing an STI diagnosis or getting tested. --- Research Questions Research surrounding sex communication has focused on the notion and the extent to which it is uncomfortable [38], as well as its effect on behavior [39]. Conversely, research concerning STIs focuses on the outcomes of not disclosing: avoiding disclosure altogether and/or avoiding getting tested [21,23]. Thus, there is a gap in the literature regarding the shame and stigma associated with the idea of sex communication, conversations surrounding safe sex with sexual partners, and the disclosure of STIs. Disclosure is both critical and complicated. A meta-analysis emphasized the positive outcomes of disclosing, including the positive relationship to social support, self-competence, reduced anxiety, and decreased problem behavior [40]. In terms of deciding to disclose, this study found that timing, opportunity, cognitive ability of the receiver, preparation of the message, and advantages and disadvantages of disclosure all affect the decision. Researchers are aware that there is shame and stigma associated with STIs but have not yet examined whether there is shame and stigma associated with the act of disclosing a status within interpersonal relationships, or the variables surrounding the communicative process. Moreover, given the lack of research in shame resilience theory in the context of STIs, it is unknown whether EAs have in fact developed a level of resilience to shame and stigma surrounding their lifestyle, which could include STIs, nor is it understood whether the receiver of the disclosure impacts the resilience or felt shame and stigma. Research [40] also does not address how shame and stigma confound other processes like disclosure (e.g., revealing sexual activity to a parent). or other related communicative interactions around sex (e.g., safe sex talks). Therefore, although research has been conducted on the role of stigma on delaying STI treatment and care, little research has delved into the stigma and shame surrounding disclosure of an STI, and the following research questions are posed: RQ1: How do EAs perceive they will be treated by others if specific others find out they have been (a) tested for STIs; and (b) the results of those STI tests? RQ2: Are there differences in the perceived shame and stigma associated with disclosing a STI to a (a) parent; (b) monogamous sexual partner; or (c) casual sexual partner? RQ3: How do the shame and stigma related to STIs impact communicative processes, like communication about STIs, disclosing an STI diagnosis, and sex communication? RQ4: What is the relationship between shame and stigma related to STIs and EAs' perception of their knowledge surrounding STIs? Additionally, how does that impact communication with others? --- Materials and Methods --- Participants Participants included 462 undergraduate students, of ages 18-25 (M = 20.09; SD = 1.29), from a large American Southwestern university. Undergraduate students were chosen to represent the EA population because approximately half of adults between 18 to 24 have enrolled in or have completed college studies [41], and the college hook-up culture was also a contributing factor within this study. Another motivation was the ability to employ convenience sampling, which in this case allowed for lowered cost and research time and a high participant turnout. Students were recruited through announcements in their courses for extra credit. In order to participate, they must have been between the ages of 18 and 25 years, which is classified as the emerging adulthood stage [13,14]. The majority of the participants were female (n = 359; 80.7%). Over half of the sample identified as non-Hispanic and White (n = 276; 62%) followed by Hispanic or Latino (n = 57; 12.8%). Demographic characteristics and subsequent analyses are based on 444 of the participants, as 18 did not complete several measures. The majority of the participants were sexually active, with over half (n = 350; 78.7%%) having engaged in vaginal sex; 86.7% (n =386) indicated having received or given oral sex. There was an even distribution of single (n = 178; 40.0%) participants and those who identified as being in a monogamous, long-term relationship (n = 197; 44.3%). Some claimed to be in casual relationships or dating multiple people at the same time (n = 64; 14.4%), while one participant indicated s/he was married and another that s/he was divorced. Lastly, well over half of the participants indicated they belonged to some religious organization (n = 276; 62.0%), identifying primarily as Catholic (n = 71; 16.0%), Baptist or Methodist (n = 66; 14.8%), or non-denominational Christian (n = 72; 16.2%). Relatedly, approximately half of these individuals are also engaging in religiously affiliated activities at least once a week (n = 143; 32.2%). This is consistent with the religious demographics for the geographical area. --- Procedures Recruitment emails and website posts included a link to the survey, hosted online by Qualtrics. After completing a consent form, the first portion of the questionnaire included questions pertaining to participants' past sexual behaviors and safe sex practices. Participants then completed a short health literacy measure, as well as an STI knowledge quiz. Using a variety of scale measures, participants were asked to report on their ability to communicate about STIs, as well as their perception of how others would treat and/or support them if others knew they were getting tested for, or had contracted, an STI. Participants' shame and stigma was assessed to examine their feelings towards STI diagnoses, getting tested for STIs, and disclosing a hypothetical status to parents, romantic monogamous partner, or casual partner. Lastly, demographic information was collected, including age, sex, religious affiliation, relationship status, health insurance coverage, and current living situation (i.e., whether they live in a dorm, apartment, or home, and whether they have roommates). Before exiting the site, participants were debriefed of this study's goals. As this study is part of a larger data set, the full survey took approximately 20 min, depending on the pace of the participant. --- Measures --- Perception of Treatment from Others This measure was adapted from an assessment tool developed by Pandya and colleagues [42] to assess perceived impact of disclosing a mental illness. Participants were asked to indicate whether they thought they would be treated worse, better, or not differently by specific groups of people following certain disclosures. First, they indicated treatment when the other party knew they had gotten tested, and then if the other party had found out positive results of the test. As per Pandya et al. [42], groups included police, employer, co-workers, extended family, roommates, health care providers, friends, partners, parents, and places of worship. --- Perception of Available Social and Medical Support As with disclosure experiences in the previous section, those who received the firstperson questionnaire were only asked to respond to this measure if they indicated they had ever been diagnosed with an STI. Additionally, adapted from Pandya et al. [42] to address STI status rather than mental illness, this measure had participants use a 7-point Likert scale (1 = strongly disagree; 7 = strongly agree) to indicate how much they agreed with eight statements pertaining to their ability to rely on health care providers and the health care system, family, friends, and partners. For example, "I know I can rely on my parents, siblings, or extended family to help me when I need it," or "People I relied on became distant after they learned I'd been diagnosed with an STI." Reliability was adequate (<unk> = 0.71; M = 3.77; SD = 0.65). --- Willingness to Disclose of STI Status Participants were first asked "if they had been, or were to be, diagnosed with an STI, would they disclose this information" to their parent, monogamous partner, or casual sexual partner. If they indicated no, they were then asked to use a 7-point Likert scale (1 = strongly disagree; 7 = strongly agree) to indicate how much they agreed with three items asking why they did or would not disclose an STI status: "I do not want to burden my parent(s) with the knowledge that I had contracted an STI," "I did not think it was important to disclose to my parent(s)," and "I consider my STI status a private matter." For the scale asking about disclosure to parents, participants also responded to the item, "I do not keep in touch with my parent(s)." This phrasing and method were used when asking about disclosing to a parent (<unk> = 0.61; M = 3.78; SD = 0.80), a monogamous romantic partner (<unk> = 0.81; M = 2.86; SD = 1.07), and a casual sexual partner (<unk> = 0.76; M = 3.17; SD = 0.92). This measure was adapted from a tool developed by Aktan-Collan et al. [43] to assess disclosing genetic risk with family members. The subject matter was the primary change, but the receiver of the disclosure was also adjusted to include all three potential confidants indicated above. --- Communication about Sex and STIs Twenty items were used to assess communication about sex, tapping into the general attitudes and experiences participants had had talking about sex with their parents and their romantic and/or sexual partners [44]. Eight were specific to talking with parents about sex (e.g., "My parent and I have discussed abstinence as a way of preventing STIs"), while the remaining twelve addressed communication with dating or sexual partners (e.g., "I have discussed the use of condoms with my partner as a STIs prevention method"). Participants were asked to indicate how much they agreed or disagreed with each of the statements using a 7-point Likert-type scale (1 = strongly disagree; 7 = strongly agree). Consistent with past research, the parent version was internally consistent (<unk> = 0.94; M = 2.80; SD = 1.12). After running a principal axis factor analysis with varimax rotation, the dating or sexual partner version yielded three factors or themes: general discussions about STIs with partners, speaking openly and honestly with partners, and abstinence as a form of protection. Item 12, "using a condom would be embarrassing" was removed due to low item-total correlation (0.04). Despite three factors with eigenvalues over one, the remaining eleven items had relatively high inter-item reliability (<unk> = 0.90; M = 3.09; SD = 0.84). --- Shame and Stigma towards STIs This scale assesses reactions to STIs and STI-related testing [8]. Using a 7-point Likerttype scale (1 = strongly disagree; 7 = strongly agree), participants responded to five items pertaining to stigma towards STIs (e.g., "Getting a sexually transmitted infection would make me feel lonely," and "Getting a sexually transmitted infection means I have poor morals"). The other six items indicated a sense of contamination and shame towards an STI diagnosis (e.g., "people with sexually transmitted infections should be ashamed of themselves"). A principal axis factor analysis with varimax rotation indicated both factors were present: stigma and shame. There was high reliability amongst the items (<unk> = 0.90; M = 2.57; SD = 0.76). --- Shame and Stigma of Disclosing an STI Similar to the measure evaluating stigma towards STIs, a measure was developed based on the aforementioned [8] scale for the purpose of this study to examine EAs' felt stigma towards disclosing an STI to parents, monogamous romantic partner, or casual sexual partner. Items were adapted from the above-mentioned measure to address each confidant. For example, "I would feel dirty to tell my parent I had been diagnosed with an STI," or "If I told a romantic partner I had been diagnosed with an STI, they would assume I had poor morals." Each of the measures were relatively high in reliability (parent: <unk> = 0.92; M = 3.40; SD = 1.04; a monogamous romantic partner: <unk> = 0.91; M = 3.38; SD = 0.90; a casual sexual partner: <unk> = 0.91; M = 3.43; SD = 0.90). --- Demographic Information Several additional data points were collected as possible covariates within the analysis, such as gender, age, and ethnicity. Participants were asked to indicate whether they were currently sexually active as well as their sexual health practices (e.g., condom usage and consistency). Participants were also asked about their current relationship status, insurance coverage, religious affiliation, religiosity (frequency of participating in religious events), current residence (e.g., dorm versus apartment), and whether they had roommates. --- Results --- Preliminary Analyses Descriptive analyses were conducted to examine the mean, standard deviation, and distribution of all primary variables. Unless otherwise noted, scales were checked for normality, unidimensionality through the use of exploratory factor analysis with varimax rotation. Although some of the variables showed more than one factor, tests for internal reliability displayed consistency and relatively good reliability. Therefore, an aggregate measure was created by summing the items and taking the mean to create a scale for each variable. Bivariate correlations between all aggregates and summed variables were also examined before testing the research questions (see Table 1). To utilize all available data, expectation maximization (EM) was used. EM is a maximum likelihood procedure in which the parameters are estimated, then missing values are estimated [45,46]. Additionally, EM infers values based on the likelihood under the normal distribution and is advantageous as it produces nearly unbiased estimates of means, variances, and co-variances [46]. --- Perception of How Others May Treat EAs (RQ1) The first research question asks how EAs perceives others in their life and community may treat them if they find out they had been tested for an STI and if they found out the results of said tests. In most cases, EAs thought they would be treated no differently if loved ones or community members found out they had been tested for STIs (see Figure 1). Over half of the participants did perceive that those from their religious community may treat them worse (59.7%). Over 40% also perceived casual sexual partners and extended families may also treat them worse. The main difference between these two groups, however, is that 27.6% thought casual sexual partners may treat them better, compared to only 5% of extended family members (i.e., 51.6% of EAs perceived their family would treat them no differently). On the opposite end, EAs perceived doctors and other health care providers would likely treat them better (43.3% compared to 11.8% who thought their HCP would treat them worse). The outcomes were relatively similar if their loved ones and community members found out about STI results, with the majority of EAs perceiving most people would treat them no differently (see Figure 2). --- Associations between Disclosing to Parents, Monogamous, or Casual Partner (RQ2) The second research question asked the differences between perceived shame and stigma associated with disclosing STIs to parents, monogamous sexual partners, and casual sexual partners. To explore this question, one-way repeated-measures ANOVA was used in order to test for significant differences between traditional forms of STI-related shame and stigma and the shame and stigma related to disclosing to a parent, monogamous romantic partner, or casual sexual partner. Mauchly's Test indicated that the assumption of sphericity had been violated, W = 0.96, <unk> 2 (2) = 20.17, p <unk> 0.001, therefore degrees of freedom were corrected using Huynh-Feldt estimates of sphericity (<unk> = 0.96). The --- Associations between Disclosing to Parents, Monogamous, or Casual Partner (RQ2) The second research question asked the differences between perceived shame and stigma associated with disclosing STIs to parents, monogamous sexual partners, and casual sexual partners. To explore this question, one-way repeated-measures ANOVA was used in order to test for significant differences between traditional forms of STI-related shame and stigma and the shame and stigma related to disclosing to a parent, monogamous romantic partner, or casual sexual partner. Mauchly's Test indicated that the assumption of sphericity had been violated, W = 0.96, <unk> 2 (2) = 20.17, p <unk> 0.001, therefore degrees of freedom were corrected using Huynh-Feldt estimates of sphericity (<unk> = 0.96). The results show there was a significant difference of perceived shame and stigma depending on who EAs were to disclose an STI status to, controlling for the traditional measure of STI shame and stigma, F(1.91, 840.20) = 4.74), p = 0.01. Specifically, at p <unk> 0.001, there was a significant difference between perceived stigma and shame when disclosing an STI diagnosis to a monogamous partner in comparison to disclosing an STI diagnosis to parents, controlling for the traditional measure of general STI shame and stigma (Fortenberry et al., 2002). There were also significant differences between perceived shame and stigma when disclosing to a monogamous partner compared to a casual sexual partner. However, there were no significant differences between perceived shame and stigma of disclosing an STI status to parents compared to perceived shame and stigma of disclosing to a casual sexual partner (see Table 2). --- Relationships between Shame and Stigma and Sex Communication (RQ3) The third research question asked how STI-related shame and stigma impact other communicative processes, like communicating about STIs, actually disclosing an STI diagnosis, and safe sex communication. To explore RQ2, several regression analyses were conducted with shame and stigma pertaining to STIs being the independent variables, controlling for sexual experience (oral and vaginal), whether participants had been tested for an STI before, whether participants had tested positive for an STI before, and religiosity by placing the variables into Block 1. The first text placed the aggregate of shame and stigma into Block 2 and parent communication as the dependent variable. The overall model was significant (F(5,184) = 2.43, p = 0.04, adjusted R 2 = 0.04, R 2 change = 0.03), as was the standardized coefficient specific to STI-related shame and stigma on EA's communication with parents about STIs (ß = -0.16, p = 0.03). Thus, the shame and stigma, along with the control variables, accounted for 6.2% of the variance; shame and stigma accounting for 2.5% of the variance on its own. The second test placed partner communication about STIs as the dependent variable. The overall model was once again significant (F(5,184) = 3.24, p = 0.008, adjusted R 2 = 0.06, R 2 change = 0.03), as was the standardized coefficient (ß = -0.19, p = 0.01). The independent variables accounted for 8.1% of the variance; specifically shame and stigma accounting for 3.4% on its own. Both of these relationships were negative, indicating the more shame and stigma surrounding STIs an EA perceives, the less likely they are to engage in conversation with their parents or any sexual partner about STIs. This research question also digs into whether shame and stigma surrounding STIs impacts disclosing STI diagnoses to parents, monogamous romantic partners, and casual sexual partners, as well as to why EAs would or would not disclose that information. Running whether EAs disclosed (or would disclose given the necessity) an STI status to their parents on shame and stigma surrounding STIs yielded a non-significant overall model (F(5,85) = 0.75 p = 0.59, adjusted R 2 = -0.01, R 2 change = 0.004). The test running why they would or would not disclose an STI diagnosis was also non-significant (F(5,32) = 0.42, p = 0.83, adjusted R 2 = -0.09, R 2 change = 0.004). Similarly, the relationship between perceived STI shame and stigma and disclosing to a monogamous romantic partner was non-significant (F(5,86) = 1.50, p = 0.20, adjusted R 2 = 0.03, R 2 change = 0.002). Lastly, the relationship between perceived STI shame and stigma and whether EAs would or would not disclose and STI to a casual sexual partner was significant (F(5,86) = 2.48, p = 0.03, adjusted R 2 = 0.08, R 2 change = 0.01), but the standardized coefficient was not (ß = 0.07, p = 0.54), indicating that the specific independent variable of shame and stigma was not significantly associated, as it only accounted for 1% of the variance. --- Partial Mediations (RQ4) The last research question digs into EAs' perceived knowledge of STIs and how that impacts not only their perceived shame and stigma surrounding STIs, but whether they get tested for STIs, as well as how these issues in turn impact communication with others. A regression analysis controlling for sexual experience (oral and vaginal), whether participants had tested positive for an STI before, and religiosity by placing the variables into Block 1. Knowledge perception of STIs was placed into Block 2, and the shame and stigma aggregate was entered as the dependent variable. The overall model was significant (F(5,86) = 2.70, p = 0.20, adjusted R 2 = 0.09, R 2 change = 0.05), as was the standardized coefficient (ß = -0.22, p = 0.04). The overall model regressing getting tested for STIs on perceived knowledge was significant (F(5,86) = 25.30, p <unk> 0.001, adjusted R 2 = 0.19, R 2 change = 0.02), but the standardized coefficient was not (ß = -0.14, p = 0.16). Lastly, a regression was conducted examining whether those who perceive shame and stigma around STIs get tested for STIs. The overall model was significant (F(5,184) = 11.33, p <unk> 0.001, adjusted R 2 = 0.22, R 2 change = 0.02), as was the standardized coefficients, (ß = 0.16, p = 0.02). To explore whether these relationships were a partially mediated model, I ran a Sobel test [47], which barely reached significance (z = -1.59, p = 0.10). To explore the second half of RQ4, a similar set of tests was conducted using the same first regression (STI knowledge and shame and stigma) with shame and stigma as the independent variable in Block 2, placing communication about STIs with parent as the dependent variable, and then also communication with partners. The Sobel tests were significant, both with parental communication about STIs (z = 1.89, p = 0.05), and partner communication (z = 1.95, p = 0.05). The results to the individual regression analyses can be found in Table 3, with change in R 2 indicating the amount of variance each variable accounted for. --- Discussion The purpose of this study was to examine EAs' perception of shame and stigma surrounding STI disclosure and communication. Results suggest that although the EAs in this study may not feel immense shame and stigma about STIs (e.g., getting tested or treated for) specifically-the scale was relatively normally distributed-they would feel shame and stigma in some situations. Just thinking that some members of their community knew they got tested for STIs, or found out the results of said test, left EAs feeling as though they would likely be treated differently. However, this study did not indicate EAs' concern about such treatment (e.g., better or worse). That is, we did not measure if EAs would care if they were treated worse, or if it would impact them in any way. This study also found that if EAs had to actually engage in a conversation about

Emerging adulthood is identified as a time of identity exploration, during which emerging adults (EAs) may engage in sexual exploration and risky behaviors, potentially resulting in the contraction of a sexually transmitted infection (STI). Many EAs, do not disclose their status to partners or those who can provide social support, like parents. Nor do they often get tested. This may be due to the changing status of stigma surrounding STIs. This study examines traditional measures of the stigma/shame of STI diagnoses, treatment, and testing, and their relevance alongside both increased opportunities for casual sex and not only heightened education surrounding STIs, but also heightened prevalence of STIs in the U.S. Results show EAs perceived that if their community found out they got tested, they would likely be treated differently. They also felt they would be uncomfortable disclosing an STI to parents as well as to sexual partners. However, disclosing to a monogamous partner yielded less felt shame and stigma by EAs. Lastly, stigma/shame was associated with STI communication, as well as with overall perception of STI knowledge, and getting tested. Further explanation of the results and possible implications of this study are discussed.

STIs (z = 1.89, p = 0.05), and partner communication (z = 1.95, p = 0.05). The results to the individual regression analyses can be found in Table 3, with change in R 2 indicating the amount of variance each variable accounted for. --- Discussion The purpose of this study was to examine EAs' perception of shame and stigma surrounding STI disclosure and communication. Results suggest that although the EAs in this study may not feel immense shame and stigma about STIs (e.g., getting tested or treated for) specifically-the scale was relatively normally distributed-they would feel shame and stigma in some situations. Just thinking that some members of their community knew they got tested for STIs, or found out the results of said test, left EAs feeling as though they would likely be treated differently. However, this study did not indicate EAs' concern about such treatment (e.g., better or worse). That is, we did not measure if EAs would care if they were treated worse, or if it would impact them in any way. This study also found that if EAs had to actually engage in a conversation about STIs and/or disclose a hypothetical STI status they would likely experience some discomfort. This is the case when EAs disclose to parents (for social/financial/informational/health care support), monogamous sexual partners, and casual sexual partners. Interestingly, disclosing an STI status to a monogamous sexual partner yields less felt shame and stigma as compared to disclosing to either parents or a casual sexual partner. Although there is little research on disclosure of STI statuses to monogamous versus casual sexual partners, plenty of health websites (e.g., WebMD, HealthLine) provide guidance on when and how to reveal your STI status to a sexual partner. For example, WebMD tells people to be in a private space, straightforward, calm, and sincere. The site also states to stray away from disclosing "in the midst of a passionate embrace," and to plan ahead [48]. Planning and potentially having a script ready makes theoretical sense, however, for the life of an EA amidst the hook-up culture, such preemptive thought may be more difficult. This may be one reason why disclosing to a monogamous romantic partner carries less perceived stigma with it. That is, EAs can think and plan and develop a script to sit down with their partner and disclose their status before engaging in sexual activity, as compared to casual sexual partners in the hook-up culture that may involve alcohol, drugs, and more impromptu sexual encounters. Monogamous partners also have more trust, security, and stability, allowing people to feel supported [49]. This becomes even more important for those with a chronic STIs to disclose [49]. Thus, it makes sense that if EAs feel supported and secure in their relationships, they perceive less stigma at disclosing an STI. It is also likely that EAs would be expected to disclose STIs early on in a monogamous relationship, questioning just how much trust is present. However, if EAs see a future monogamous romantic relationship with this person, although the trust would not be at the same level as a well-established partnership, it is likely they see that trust grow, and disclosing an STI status could help build the trust and intimacy within the relationship, just as disclosing any private information does [50]. We should note, much discussion of monogamous relationships does so in the traditional, binary and ideological sense. Thus, it is imperative to replicate this study allowing participants to identify what monogamy meant to them, whether it was long-term and binary, or some agreed-upon relational formation (e.g., polyamorous) that may still allow for trust and security but may not be "traditional". It may also move the field forward to explore these ideas of shame and stigma surrounding STIs and sex communication beyond casual and monogamous sex, examining outcomes and perception of felt stigma in consensually non-monogamous partners. With other sexual relationships, people tend to manage their self-relevant information in order to make positive impressions on others [51]. Thus, although sharing and exchanging personal information is imperative to develop relationships, at an early stage, people may be highly cognizant of the impression they give off [52], especially within emerging adulthood and college, when coolness is an imperative factor of everyday life [53]. Therefore, disclosing an STI status to casual sexual partners, although an important part of sex communication and sexual interactions, makes impression management more difficult. It also makes sense that disclosing an STI status to someone who is basically a stranger, with a lack of trust and security within the relationship, comes with felt stigma and shame for EAs. Future studies should be more specific with types of causal sexual partners, as this study did allow participants to think of casual sexual partners as whatever they perceive them to be. For some, casual sexual partners can be friends with benefits or first-time hook ups with good friends, in addition to almost strangers. Thus, our explanation of having less trust within casual sexual interactions may not always be the case. This study also shed light on how shame and stigma impacted the overall communicative process of talking about sex and STIs with parents and sexual partners. Shame and stigma were significantly and negatively associated with parental communication. That is, the more shame and stigma EAs experienced or perceived around STIs, the less likely they were to talk about STIs and sex with their parents, even when in need of social support of varying nature (e.g., financial, informational, emotional). There was a similar relationship (i.e., negative and significant) between STI-related shame and stigma and partner communication. As this is a regression analysis, it could be argued that the relationship could go either way, since a decrease in parental communication about sex and STIs yields higher perceived shame and stigma surrounding STIs. There were no relationships, however, between shame and stigma and whether they have, or would, disclose an STI to parents or sexual partners. Lastly, the relationship between STI knowledge and getting tested for STIs seemed to be mediated by shame and stigma. Similarly, the relationships between shame and stigma and getting tested were partially mediated by communication with parents, and in another model, mediated by communication with partners about STIs and sex. That is, an increase in knowledge about STIs decreased STI-related shame and stigma, which increased the likelihood to get tested for STIs. Additionally, as STI-related shame and stigma increased, communication with parents or partners about STIs or safe sex decreased, which decreased the likelihood of getting tested for STIs. This finding is novel to the field as past research has primarily only argued that STI-related shame and stigma, alone, were enough to prevent getting, or find the results of, STI tests [23,54]. This study also revealed this to be true, (F(5,184) = 11.33, p <unk> 0.001, adjusted R 2 = 0.22, R 2 change = 0.02; ß = 0.16, p = 0.02), however, the relationships were partially, but significantly, mediated by parental and partner communication, with highly significant pathways and a significant overall Sobel test statistic. In other words, having conversations about STIs, STI risks, and safe sex practices with parents can increase the likelihood of getting STI tests when sexually active. Having partner communication about sex, safe sex, STIs, and STI risks can also increase the likelihood EAs get tested for STIs, which is imperative with the nature of the hook-up culture. The present research is a useful first step in identifying theoretically valuable understandings about how shame and stigma surrounding STIs have shifted within the hook-up culture, which impacts EAs' disclosure and communicative processes. --- Limitations and Future Directions Although this study aims to make unique contributions to communication and health scholarship, there are limitations. First, this study relied on an undergraduate student research pool. Although this pool includes the student body of a large Southwestern university that is relatively diverse in nature, this sample is rather homogenous. Most participants identified as White (Non-Hispanic; 62%) and female (80.7%), and they averaged 20.1 years of age (SD = 1.29), with few participants under 19 years old (n = 39) or over 22 (n = 11). Increasing the variance of the sample would allow for a more inclusive understanding of the communicative processes surrounding STI shame, stigma, and disclosure. Furthermore, having a more diverse population may allow for effects that differ due to differences in culture surrounding sex. Future research may benefit from adding some participation requirements, as well as measures that were not considered prior to examining participants' responses during the analyses. First, it may be of interest to conduct a replication of this study with the limitation of participants who have experienced STI testing, or an STI diagnosis. Including those who have been diagnosed and actually experienced the process of disclosing may allow for more thorough within-group comparisons and a different perspective on the communicative processes. Comparing a replication of this study to the results provided here would allow us to see the differences in disclosure processes, as well as how diagnosed individuals perceive their peers' perceptions of STIs, communication about sex, and the process of disclosing. The timing or context of the contraction of the STI may also be an important variable to consider. That is, many participants remarked that if they had cheated and contracted the STI, they would not expect their partner to support them, whereas if they had contracted an STI prior to their current partner, they would expect more support. Similarly, research may want to consider separately disclosure of the STI status to the partner that potentially gave the participant an STI, as well as the current partner, if they are in fact different individuals. Lastly, future research should dig into the relationships between perceived shame and stigma surrounding STIs and the ability to have safe sex conversations with both monogamous romantic and casual sexual partners. That is, exploring how EAs weigh risking getting an STI-one that may just go away on its own and show no symptoms-from new sexual partners and the stigma and complications (e.g., showing lack of trust, love, or even subtly telling a partner you have an STI) of condom negotiation. Lastly, it is important to note that the survey distributed did not specify STI type in the verbiage of the questions. Communication (e.g., uncertainty, information management) and disclosure theory has long argued that how someone assesses the problem or issue impacts how they then proceed. The health disclosure decision-making model [55] argues that we assess five factors surrounding our health-related issue: stigma surrounding it, preparation required to deal with the health diagnosis, actual prognosis and outcomes, symptoms related to the diagnosis, and its relevance to others. In the case of STIs, it could be argued each of these is important to consider when exploring whether EAs would disclose their STI status to their parent or a sexual partner. Potentially, most importantly, the actual prognosis and symptoms. That is, if an STI is chronic, EAs may feel a need to disclose their STI to others, especially if it impacts the other person by way of sexual contact or that they will need to provide support down the road. With a chronic STI, infected individuals may weigh the benefit of disclosing information to their parents more than the risks associated with having that type of conversation. A chronic STI is also likely to impact everyday life when it comes to self-esteem as it relates to sexual identity, desirability, and ability to form strong intimate relationships in the future. Chronic STIs may also require more social support from parents, like emotional support to cope with any psychological issues that may arise, or financial support for consistent and life-long treatments. However, when the majority of STIs are treatable and/or show no symptoms [2], even when EAs require treatment, these often dissipate as the infection ceases to exist. Thus, EAs may be able to manage their infection independently without the need to disclose to a parent or any sexual partners. Therefore, future research should indicate type of STI, chronic or acute, when exploring perceived stigma and shame. With this research setting the groundwork to better understand EAs' perception of stigma and shame surrounding STIs, future research would benefit from exploring the other four aspects of information people assess when deciding to disclose personal health information [55]. Doing so could also shed light on EAs' perception on STI prevalence and treatments available, and how they weigh the costs and benefits of disclosing any potential STIs with condom negotiation conversations, as well as STI disclosure. --- Conclusions This study provides insight into the way EAs are experiencing STIs, including those who have not been diagnosed with an infection. Past research provides evidence for the idea that shame and stigma surrounding STIs can delay treatment for people of all ages [23,54,56,57], potentially decreasing overall physical health, thus increasing the likelihood of spreading to others. However, if we continue to conceptualize and operationalize shame and stigma surrounding STIs using the traditional definitions, we may be missing out on a major component that impacts individuals' engagement in healthy sexual behaviors, including condom negotiation conversations with potential sexual partners and support-seeking behaviors after a diagnosis. Furthermore, the operationalization of the traditional measure is quite strongly worded (e.g., "I have poor morals," or "getting an STI would be embarrassing"). Thus, many of these EAs may have some shame related to STIs, but not to that extent, considering the low amount of variance of this measure. This study therefore helps to extend our understanding and conceptualization of what the stigma surrounding STIs is and how it may differ for EAs and therefore impact the likelihood of their engagement in healthy sexual behaviors. The results of this study can also help to thoroughly develop the operationalization used for shame and stigma about STIs (i.e., using items worded less strongly and creating encompassing processes surrounding STIs beyond treatment and health care providers), allowing researchers to more fully understand the experience, differences that varying ages may face, and motivations behind sexual health conversations and seeking the support necessary to get tested and adhere to treatment regiments. The results of this study provide an opportunity to develop stronger, theory-based health interventions targeting EAs and their sexual behaviors, safe sex practices, and STI testing. Specifically, knowing that EAs' perceived shame and stigma surrounding STIs may be how health interventions encourage this demographic to get tested and treated. Rather than arguing issues currently used as costs (e.g., general prevalence, long term effects if gone untreated), health interventions can flip the rhetoric using these now non-issues as benefits of getting tested. That is, due to prevalence, getting tested and treated are easier and more private. This study does show that EAs still think many people in their lives may treat them differently if an STI status was disclosed. Therefore, knowing the importance of impression management during emerging adulthood may be another way STI testing and STI communication campaigns can encourage EAs to get tested in the first place, allowing EAs to avoid the uncomfortable conversation of disclosing an STI to their romantic partner. Moreover, this study provided insight into the differences in shame and stigma EAs feel concerning STIs. As sexual exploration is commonplace [30] for EAs and within the hook-up culture [11] it is important to consider the communicative processes surrounding EAs' perception of STIs. Seemingly they are willing to get tested and find out the results, however, taking the step to engage in conversation about STIs is more difficult. Thus, although it is important for EAs to get tested and get treated, talking about STIs can help to prevent spread and contraction [23]. The results of this study in conjunction with past research argue for more targeted health-promotion tactics that also encourage talk, the starting point towards increasing condom usage, consent, and healthier sexual relationships within the EA population [12]. Understanding how EAs perceive communication about STIs with their parents is another component to consider when developing interventions and programs. Conceptualizing this type of talk is also important when we consider the role health insurance, and the Affordable Care Act, has on the relationship between parents and EAs. That is, many individuals may remain on their parents' health insurance until they are 26 years old. Any services they receive, including STI testing, will be noted on their parents' health insurance itemized bills, disclosing EAs' private information to them. Health-promotion initiatives may therefore want to focus on assisting EAs in becoming more assertive about engaging in communication about sex with their parents, especially as it pertains to safe sex practices and health care surrounding sex (e.g., STI testing). Such conversations with parents may guide EAs to practice safe sex behaviors [12]; furthermore, having positive, open discussions about sex can assist the EA in building relationships and increasing sexual confidence [58]. Thus, the EA may become more assertive about engaging in communication about sex with romantic partners, thus reducing risky behavior enactment and the consequences associated with those behaviors. Clearly, this project only begins to scratch the surface of our understanding of how shame and stigma towards STIs has changed, especially among EAs who are part of the "hook-up culture." Though people used to experience shame and stigma when getting tested and finding out the results [23], hence it became the traditional measure primarily used in research [8], this study suggests we may need to focus more on the disclosure process. Disclosing an STI status not only allows EAs to receive social support in its various forms from parents and romantic partners, but also allows EAs to initiate conversations with partners about safe sex practices. --- 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 IRB protocol at time of approval. --- Conflicts of Interest: The author declares no conflict of interest.

Emerging adulthood is identified as a time of identity exploration, during which emerging adults (EAs) may engage in sexual exploration and risky behaviors, potentially resulting in the contraction of a sexually transmitted infection (STI). Many EAs, do not disclose their status to partners or those who can provide social support, like parents. Nor do they often get tested. This may be due to the changing status of stigma surrounding STIs. This study examines traditional measures of the stigma/shame of STI diagnoses, treatment, and testing, and their relevance alongside both increased opportunities for casual sex and not only heightened education surrounding STIs, but also heightened prevalence of STIs in the U.S. Results show EAs perceived that if their community found out they got tested, they would likely be treated differently. They also felt they would be uncomfortable disclosing an STI to parents as well as to sexual partners. However, disclosing to a monogamous partner yielded less felt shame and stigma by EAs. Lastly, stigma/shame was associated with STI communication, as well as with overall perception of STI knowledge, and getting tested. Further explanation of the results and possible implications of this study are discussed.

Background Children under 15 years of age account for approximately 3.3 million of the people living with human immune virus (HIV)/acquired immune deficiency syndrome (AIDS) worldwide, and almost 90% of all HIV infected children live in sub-Saharan Africa including Ghana [1]. In Ghana the prevalence of HIV in young people from 15-24 years is 1.2% [2]. Currently, children infected with HIV through mother-to-child transmission once diagnosed and in care, live longer and reach adolescence because of greater access to antiretroviral therapy (ART) in the past decade in West Africa [3,4]. Thus, with the start of national programme for providing care and free highly active antiretroviral therapy (HAART) to children, there has been a significant reduction in morbidity and mortality of HIV infected children and more of them are surviving through childhood into adolescence with improved quality of life [3]. Along with the increased survival of children infected with HIV, disclosure of HIV status to children remains a complex and a critical clinical issue in the care of HIV infected children [4]. World Health organization (WHO) and American Academy of Pediatrics (AAP) strongly encourage disclosing HIV infection status to school aged children and younger children. They should be informed incrementally to accommodate their cognitive skills and emotional maturity [5,6]. Telling children and adolescents about their HIV infection is a challenging dilemma because children are often asymptomatic in the early stages of HIV infection, yet require daily medications and close monitoring [7]. Caregivers and healthcare workers are presented with group of challenges around disclosure, including deciding on what is in the child's best interest and when, why and how information about his/her HIV positive status should be shared with him/her. Caregivers are reluctant to disclose the HIV positive status to their children for fear of social rejection and isolation, parental sense of guilt, and fear that the child would not keep diagnosis to themselves [8,9]. Disclosure of HIV status helps HIV infected adolescents to understand their status and the need for treatment, thereby promoting their participation and responsibility for their care [10]. The rate of disclosure remains low especially in resource limited countries despite the growing evidence of the benefits of disclosure [11]. The inability of most caregivers to handle disclosure has defined the three main patterns of disclosure: complete, partial, and non-disclosure. Complete disclosure of HIV status has been associated with improved adherence to ART [12]. Whereas partial and non-disclosure can strain the relationship between the caregiver and the child. In such situations, force and persuasion are often used to get the child to adhere to treatment; these may results in purposeful rebellion and non-adherence by the child [12]. There is a lot of controversy about what and when to tell children living with HIV about their diagnosis. Many families do not think their children can handle the information and they do not want to upset them because they have already been through so much [8]. Several studies have documented the benefits of disclosure of HIV positive status to HIV infected children and adolescents, including psychological benefits as well as positive effects on the clinical course of the disease [6][7][8][9]. The American Academy of Pediatrics recommends that children and adolescents with HIV should be told their diagnosis [6]. However, published rates of disclosure in children from resource rich countries vary widely, from 18 to 77% [6][7][8][9]12], partly due to the lack of conclusive guidelines on when and how to disclose the diagnosis of HIV to children. Though limited, studies available on rates and effects of HIV disclosure on children are mainly from resource rich countries. There is paucity of data on disclosure from resource limited settings, where the majority of HIV infected children live. Moreover, studies from resource limited settings have mainly been qualitative in design with small samples sizes making it difficult to generalize the findings. This study found out the knowledge and disclosure of HIV status among adolescents and young adults attending adolescent HIV clinic. --- Methods A cross-sectional study was carried out in July 2013 among HIV infected adolescent and young adults attending an adolescent club meeting in Accra. The Adolescent HIV Care programme at Fevers unit of Korle-Bu Teaching Hospital was established in June 2012 and provides comprehensive HIV/AIDS care and management of opportunistic infections. The Unit serves as the national referral centre for HIV-infected patients. Ever since the first case of HIV was diagnosed in Ghana in 1986, the Unit has provided care and support to persons living with HIV/AIDS. Provision of comprehensive HIV/AIDS care and treatment started at the Unit in December 2003 in line with the scale-up of access to treatment by the National AIDS Control Programme (NACP) and her partners. Patients are referred from the paediatric HIV clinic to the adult clinic at age twelve. Thus the adult clinic serves both adolescents and adults. An adolescent clinic was started as best practice when it was recognized that their needs and expectations were not being met at the adult clinic. The clinic was supported by adult physicians since it proved difficult to get the paediatricians to come on board. The adolescents and young adults referred to the clinic present with the following modes of diagnosis: have mothers known to be HIV seropositive during pregnancy through the programme for the prevention of mother-tochild transmission (PMTCT), others are discovered to be infected with HIV after presenting with an AIDS-defining illness, or are diagnosed after either a symptomatic sibling or parent was found to be HIV positive. Majority of them were infected perinatally. At the time of the study, the adolescent clinic was 13 months old and had about 70 HIV infected adolescents and young adults. Korle-Bu Teaching Hospital gave the approval for the formation of the adolescent group as well as the study. All the young people recruited into the study received medical care for their HIV infection through the clinic and were all invited to voluntarily participate in a general survey on HIV infection at one of the club meetings (Self-selection). Following the self-selection, written informed consent was obtained from those aged 16 years and above and assent and permission from care givers for those less than 16 years after briefing on the purpose, use, and significance of the study. Those who did not consent were excluded from the survey but continued to participate in the fun games, music and dance. A structured questionnaire was then used to collect data. The questionnaire was anonymous and solicited information on respondents' background as well as their general knowledge on the causes, mode of transmission of HIV, HIV treatment and disclosure. The questionnaire had three parts. The first part was closed ended and contained questions on the background of the respondents. The second part contained questions on general knowledge and awareness of HIV/AIDS found in HIV/AIDS awareness leaflets used by the Ghana AIDS Commission for information, education and communication. The final part of the questionnaire sought respondents' views on HIV status, disclosure and treatment. Here, information was collected on disclosure status, age at disclosure, treatment compliance, and socio-demographic data, family type (biologic parents, adoptive/foster, extended), and care-givers HIV status. Service providers and care givers were independently interviewed in a focused group discussion on HIV disclosure. The data collected was entered and analyzed using Statistical Package for the Social Sciences (SPSS®; IBM Corporation, Armonk, NY, USA) version 16 with the key issues being presented in summary frequency tables as descriptive statistics. --- Results --- Demography of respondents A total of thirty-four (34) young person's living with HIV from age 13 to 22 were self-recruited to participate in the study out of 36 at the adolescent meeting. Thirty four adolescents and young adults did not attend the adolescent meeting. The mean age of respondents was 16.91 <unk> 2.5, of which 61.8% (21/34) were females. All participants were still in school and the distribution is showed in Figure 1. In relation to religion, 94% (32/34) were Christians and 6% (2/34) were Muslims. The demographic characteristics of the absentees from the meeting were similar to those who attended. Of the 8 care givers in the FGD 7 were females and 1 male whereas 75% (6/8) of the service providers in the FGD were females. --- HIV knowledge Eighty-five percent (29/34) of respondents were aware that young people of their age could fall sick whilst 5.8% (2/34) thought otherwise and 8.8% (3/34) did not give any response. Over 91% (31/34) of adolescents had heard of HIV as compared to 8.8% (3/34) who had no clue. Moreover, 70.6% (24/34) knew someone with HIV whilst the remaining 29.4% (10/34) did not. Forty-five percent thought that adolescents were not at risk of getting HIV. Fifteen percent (5/34) were not sure whether infected patients could continue with their education. Sub-analysis did not show any statistical difference in HIV knowledge in terms age, sex and educational status. --- Modes of transmission Majority of the respondents 90.4% (multiple responses) were able to correctly identify one or more mode of HIV of transmission with the remaining 9.6% were not able to provide any correct answers. Out of those who correctly identified the modes of transmission, 66.7% knew that HIV could be transmitted through unprotected sexual intercourse with infected persons and 63.6% knew about mother to child transmission. Twenty three percent (19/ 83 multiple responses) said HIV can be transmitted by sharing sharp needles/syringes with infected persons. Overall, 29% (10/34) participants think HIV can be transmitted through physical contact. --- The prevalence of HIV disclosure and antiretroviral therapy HIV disclosure status rate was 52.9% (18/34), 20.6% (7/ 34) said they did not have HIV, 20.6% (7/34) did not know whether they had it or not and 5.9% (2/34) did not give any specific answer. Ninety three percent (13/14) of those who did not know their HIV status were between 13 to 15 years. Fifty percent (17/34) of the respondents were on antiretroviral whilst 5.9% (2/34) were not, however 26.5% (9/34) did not know whether they were on ARVs or not. One person who said he was HIV negative but was on ARVs and 17.6% (6/34) did not give any response. In general, 35% (6/17) of the respondents who knew they were on ARVs admitted missing some of their doses Figure 2. From the FGDs, most of the disclosures happened after ART initiation and were planned. The persons involved in this process varied across individual families. It was sometimes disclosed by family members, particularly care givers or parents after an interview between the latter and a psychologist. None of the respondents did find out their status by chance or through unplanned events. --- Limitation The study was carried out in a single adolescent clinic with a small sample size which may affect generalizability of our findings. --- Discussion --- Knowledge of HIV The study revealed that 90.4% were able to correctly identify one or more mode of HIV transmission with the remaining 9.6% not able to provide any correct answer. The fact that 10% of respondents were ignorant on mode of HIV transmission even when they were HIV positive poses serious threats to HIV/AIDS prevention and control in this age group. It was however not clear whether this was just ignorance or denial. This study is in line with findings from other studies done in developing countries such as South Africa and Ethiopia as only about half knew much about their HIV including their status even when on anti-retroviral therapy which was quiet poor [13]. The key to HIV awareness hinges on knowledge of what HIV and AIDS is, and what exposures and risky behaviors can potentially lead to contracting the virus. Knowledge of HIV among infected adolescents and young adults' is therefore crucial not only in preventing the spread of the virus, but also in addressing the threats posed by HIV/ AIDS. There is the need to strengthen education to young persons on HIV as well as support care givers to disclose HIV status and to support young people to adhere to ARVs for better outcomes. --- Status of disclosure The prevalence of disclosure among our study population was 52.9% (18/34). This is however within the middle bounds of the range (18-77%) reported in studies from resource rich countries. Our finding is consistent but slightly higher than the reported prevalence of disclosure from Thailand, Zambia, and Uganda of 30.1%, 31.8%, and 29%, respectively [6,7,12,14]. Of the children who did not know their HIV positive status, more than two third reportedly were being told of other related diagnosis like TB, malaria, cholera and the rest did not have any clue at all about their illness. Reasons cited by the caregivers for not disclosing HIV status to their wards were consistent with that of studies in resourcelimited countries; namely stigmatization and isolation, parental sense of guilt and fear that the child would not keep diagnosis to themselves [10,15]. As part of standard of care, disclosure is made to adolescents before transfer to the adult clinic. We are unable to tell if these were missed, disclosure was made but not accepted or that adolescents and care givers were still in denial. Interestingly, some caregivers for fear of stigma reverse what healthcare providers say as soon as they leave the presence of healthcare providers. However, several studies from both resource rich and resource limited countries suggest that the disclosure of HIV status to HIV infected adolescents and young adults has immense psychological benefits and positive effects on the clinical course of the disease [5,12,16,17]. Our findings and that of other studies in the subregion of low rate of disclosure underscore the need for a systematic review of the hurdles to disclosure and implementation of locally and culturally sensitive interventions programs to promote disclosure. Nearly a third of the caregivers expected the healthcare providers alone to disclose the HIV status of their children. This is contrary to European study where majority of caregivers believed that as parents it was their responsibility to disclose the diagnosis to the child [14]. This may be partly due to the cultural differences in relation to communication between parents and children on issues and the lack of disclosure skills. The healthcare providers agreed that disclosure is the prerogative of the caregivers and children should be told their HIV status. These contrasting views are consistent with that of studies from resource rich countries [14]. The expectation of some caregivers that healthcare providers should disclose the HIV status to the HIV infected adolescents may lead to unnecessary delays in disclosure. As reiterated by the healthcare providers, training in disclosure counseling and a disclosure program adopted by the clinic would allow healthcare providers to better advice caregivers on how to disclose. With caregivers who lack disclosure skills, a counselor assisted or supported disclosure session may suffice [10]. --- Transmission of HIV/AIDS Ninety percent of respondents were able to clearly point out that HIV/AIDS may be transmitted through having unprotected sex with an infected partner, being born to a mother who is infected with HIV, or through the infected mother's breast milk. Which is similar to the 2008 Ghana Demographic and Health survey (GDHS) report of 97% adult knowledge on HIV transmission [18]. Almost all the adolescents and young adults were also aware that using unsterilized instruments or sharing personal items like toothbrushes could lead to acquiring the virus. Their awareness has a positive impact on prevention of HIV/ AIDS considering that sharing items like blades, toothbrushes, pins and needles, cooking and pocket knives, hair and nail clippers, and other sharp objects is a major cause of transmission in less developed countries [16]. The small proportion (10/34) of participants believed that physical contact with infected persons or sharing public space with infected persons could be a source of transmission of HIV/AIDS. This might explain the relatively high stigmatization and discrimination against PLWHA by their colleagues within some communities in Sub Saharan Africa [4]. This is in sharp contrast to the content of educational materials and information provided by the Ghana AIDS Commission, National HIV/AIDS Control Program, and the mass media on HIV/AIDS transmission. These materials clearly explain that it is a scientifically proven fact that physical contact, without the exchange of bodily fluids with an infected person, cannot transmit the virus. In fact, a lot of HIV/AIDS awareness messages in Ghana focus on this point because of the realization that poor awareness of this fact leads to high stigmatization, seclusion, and discrimination of PLWHA. Such perceptions also explain the case made by World Bank that PLWHA in rural areas and less developed countries are sometimes denied access to public transport, swimming in public pools, fetching water from public standpipes, engaging in contact sports, and other activities that involve physical proximity [19]. Believing that physical contact and sharing public space can lead to HIV infection may result in the adolescent and young person self-isolation and self-stigmatization. He or she may not want to socialize and interact for fear of transmitting the virus or becoming infected. These leads to poor social skills, lack of support and subsequently affect adherence and also school work. If such perceptions are still being held by some people, then prevention and even management of new and existing cases of HIV/AIDS may be seriously hindered. --- Conclusions In conclusion, 47% of HIV infected adolescents in the study were not aware of their HIV status, though strong beneficial effect of HIV disclosure on retention in care after ART initiation beyond the age of ten is known. Disclosure of HIV status to adolescents and young people is dependent on a complex mix of factors and most practitioners recommend an age and developmentally appropriate disclosure thus it is highly individualized. The knowledge and awareness of HIV was 91% compared to 97% of adults in the most recent Ghana demographic and health Survey. There is the need to support care givers to disclose HIV status and to support young people to adhere to ARVs for better outcomes. --- Consent Written informed consent was obtained from the patient's guardian/parent/next of kin for the publication of this report and any accompanying images. --- Competing interests The authors declare that they have no competing interests. --- Authors' contributions Conceived and designed the study: EK, ML, GBN, AOA, MS and AB Performed the experiments: EK, ML, GBN, AOA, AB and MS. Analyzed the data: ML, EK, GBN and AOA. Wrote the paper: GBN, EK and ML. All authors read and approved the final manuscript.

Background: In Ghana it is estimated that 1.2% of HIV infections occur in young people aged 15-24 but the representation in our clinics is small. Adherence to treatment, appointment keeping and knowledge of HIV status remains a challenge. Disclosure has been shown to result in better adherence to therapy, good clinical outcomes, psychological adjustment and reduction in the risk of HIV transmission when the young person becomes sexually active. A baseline study was conducted to ascertain if adolescents and young adults knew their HIV status and their knowledge on HIV. Informed consent and assent were obtained from willing participants. Self-administered questionnaires on general knowledge of HIV, HIV treatment and disclosure were collected and analyzed. Results: Thirty-four young persons participated in the study. The mean age was 16.9 ± SD 2.5 and 62% (21/32) were female. All of them were still in school. Eighty-five percent were aware that young people their age could fall sick, 91% had heard of HIV, 70% knew someone with HIV and 45% thought that adolescents were not at risk of HIV. On modes of HIV transmission, 66.7% knew HIV was transmitted through sex and 63.6% knew about mother to child transmission. Fifty three percent (18/34) knew their HIV status, 50% (17/34) were on antiretroviral and 35% (6/17) of them admitted to missing ARV doses. One person who said he was HIV negative and another who did not know his status were both on ARVs. Conclusion: Disclosure of HIV status to adolescents and young people is dependent on a complex mix of factors and most practitioners recommend an age and developmentally appropriate disclosure. Thus it is highly individualized. The knowledge and awareness of HIV was 91% compared to 97% of adults in the most recent Ghana Demographic and Health Survey however only about two thirds had acceptable in depth knowledge on HIV. Only half knew their HIV status which was not the best considering their ages. There is the need to strengthen education to young persons with HIV, support adhere to ARVs for better outcomes and assist care givers to disclose HIV status to them.

Materials and Methods In a retrospective national cohort study, based on compulsory registers with data collected prospectively, we included everybody born in Norway between 1967 and 1992 (1490279 individuals, 2584 with clefts). This cohort was followed until the year 2010, when the youngest individuals were 18 years old. In order to ensure that the individuals were not affected by unknown syndromes or diseases, we excluded all individuals with any chronic medical condition, or who had other birth defects than clefts, hydroceles and dislocated hips. Individuals with oral clefts who were included in the study are said to have isolated clefts. --- Results Isolated cleft patients are similar to the general population regarding education, income and social class. Isolated cleft patients have lower fertility than the background population, but considering only married couples this difference in fertility disappeared. --- Conclusions An oral cleft did not appear to affect future socioeconomic status or chances of becoming a parent for children born in Norway. An exception was males with cleft lip and palate, but differences were small. --- Introduction Oral cleft is a common category of congenital anomalies. Worldwide the prevalence is about 1.7 per 1000 live births [1]. There are three major sub-groups with different etiology, severity and treatment: Cleft lip only (CLO), the more severe cleft lip with cleft palate (CLP), and cleft palate only (CPO) [1]. Early and multidisciplinary treatment is important [2]. Although clefts can be repaired surgically, children born with clefts may experience difficulties. They may have problems with breast feeding [3], speech and language development may be delayed [4], and hearing may be impaired [5], which may affect education and social integration during childhood. Treatment is extensive and time consuming, and teasing or lack of satisfaction with facial appearance may have a negative effect on self-esteem [6]. More generally, the presence of a cleft at birth is suspected to adversely affect quality of life in a number of areas, like health and mental health [7,8], well-being and social life [9]. However, there is limited information on how an oral cleft affects the ability of otherwise healthy adolescents to cope with adult life, and previous studies are inconsistent [10]. Clefts are associated with other disorders, like genetic syndromes and developmental delays [11]. Clinical experience suggests that such disorders may be hard to observe directly, but may be seen indirectly through the presence of other medical problems. Congenital anomalies in addition to the cleft are commonly used as a crude indicator of underlying disorders. Clefts without additional congenital anomalies are referred to as isolated. Some individuals with apparently isolated clefts experience cognitive difficulties later in life [12]. Even if a medical diagnosis is not received until adolescence, this still suggests that the cleft was not really isolated [13]. A study that attempts to predict the consequences in adult life of an isolated oral cleft and its treatment should not include individuals with such diagnoses. In order to study socio-economic status and reproduction among adults who were born with an isolated oral cleft, clinical files of all cleft cases treated in Norway from 1967 to 1992 were linked with compulsory national registries. All individuals with other congenital anomalies than oral clefts or any chronic medical diagnoses by the age of 18 were excluded. A cohort consisting of more than two thousand individuals with isolated cleft and 1.3 million unaffected individuals was followed from the age of 18 through the year 2010. --- Methods --- Study design All children born in Norway from 1967 to 1992 were considered eligible for the study. This comprised a total of 1 490 401 individuals, including 2 860 in the cleft group (Fig 1). Information was obtained by linking clinical registries of all cleft patients treated in Norway with the following compulsory national registries: The Medical Birth Registry [14], the National Education Database [15], the National Registry [16], the National Insurance Scheme [17] the Norwegian Labor and Welfare Organization [18] and the Norwegian Tax Administration [19]. The unique personal identification number given to all Norwegian citizens simplified the merging procedures. A total of 166 in the cleft group could not be reached (e.g., unknown address, change of name), and 167 declined participation in the study (Fig 1). Individuals who were not present in all registries or did not reside in Norway in 2010 were excluded. In order to minimize the presence of unobserved disorders we excluded individuals who were registered with any chronic medical condition before the age of 18, or any congenital anomaly other than oral cleft, hydrocele, or hip-dysplasia (Fig 1). The study cohort included a reference group of 1 331 094 unaffected individuals and a group of 2077 individuals with apparently isolated clefts, and was followed from the age of 18 through the year 2010. The treatment of oral clefts in Norway has since the early 1960's been coordinated by two regional centers (Haukeland University Hospital, Bergen and Oslo University Hospital, Rikshospitalet, Oslo). The centers have cooperated in the documentation, treatment and followup of cleft patients during the whole study period. Experienced clinicians have carried out a thorough and systematic registration of clefts, including comorbidity and classification of cleft morphology. At any given time, treatment at the highest level available has been provided to the patients, free of charge. The Medical Birth Registry has collected information on all births in Norway since 1967. The registry provided data on birth year, sex, survival status, congenital anomalies, parity, maternal age, and maternal marital status. Congenital anomalies are coded according to the standards of International Classification of Diseases (ICD) [20]. The Medical Birth Registry also provided data on whether an individual became a parent during the study period. The National Education Database provided information on the education of all individuals in the study and their parents. For parents the following four education levels were used: Master's degree or PhD (at least five years at university or college); Bachelor's degree (at least three years at university or college); Completed High School (Intermediate level); Not completed High School (Basic level). Parent's educational level was recorded when the offspring was 16 years old, and was defined as the level of the parent with the highest education. The Central Population Registry contained information on parental immigration status (two, one, or no immigrant parents) and marital status of the study participants (married/separated/divorced/widowed vs. single). All residents of Norway are covered by the National Insurance Scheme, which may grant financial support for medical conditions demanding special needs for care or attendance. We collected any such diagnoses that were given before the age of 18. The Norwegian Labor and Welfare Organization record job related income at an annual basis. Type of occupation was recorded by the Norwegian Tax Administration. Individuals who were employed in 2010 were categorized according to Erikson et al [21], adapted to Norwegian data by Flemmen & Andersen [22]. The categories were professional, characterized by responsibility, autonomy or authority (e.g., nurses, lawyers, CEOs, teachers, researchers, and engineers); manual, characterized by less responsibility, autonomy and authority, including skilled and unskilled labor (e.g., hairdressers, carpenters, nurse's aides, cooks, shop assistants, and maids); and others, for individuals employed in all other types of occupations (e.g., farmers, lower level store and restaurant managers, nurse's aides with leadership responsibilities, and the self-employed) [23]. --- Statistical analysis We followed the cohort from the age of 18, and compared each of the groups of isolated CLO, CLP, and CPO with the reference group of unaffected individuals. For most of the analyses we also stratified by sex. The outcome variables were educational achievement at ages 17, 21, and 25, median job related income, job related income in the upper 2006 quartile (yes/no), job related income in the lower 2006 quartile (yes/no), type of occupation, parenting at least one child (yes/no), and marital status (married/not married). Information on all outcome variables except income was available through 2010. The most recent year of recorded income was 2006. Some individuals had no income in 2006 (mostly young people), and were therefore excluded from the income analysis (n = 229 137). Regarding educational achievement, three educational outcomes were examined: completed basic education by age 17, completed intermediate level education by age 21, and completed higher education by age 25. Basic education consisted of 9-10 years of compulsory school, usually ending the year a pupil turned 16, making this analysis not completely prospective. Intermediate level education required basic education, was optional, and lasted 2-5 years. This was either preparation for skilled labor (carpenter, plumber, hairdresser, cook etc.), preparation for university studies, or both. Higher education consisted of at least three years in college or university (i.e., at least a bachelor's degree). When considering intermediate and higher education, individuals who had not yet reached the ages of 21 (n = 158 892) and 25 (n = 354 494), respectively, were excluded. Thus, only individuals old enough to have had the possibility of attaining a certain level of education were included in the analyses. Individuals without employment in 2010 (e.g., unemployed, student) were excluded from analyses regarding job type (n = 278 183). A Poisson regression with robust standard errors [24] was applied when the outcomes were academic achievement, marital status, reproduction, type of occupation, and high/low job related income. This is a numerically robust way to obtain relative risks (RRs) with 95% confidence intervals. Job related income was analyzed using median regression [25], and percent change in median income was reported. Analyses were adjusted for sex, parity, parental immigration status, marital status of mother at birth (single vs. married/cohabitating), parental level of education when participant was 16 years old, maternal age, and birth year. To account for non-linear effects of age at the end of follow-up, birth year was treated as a categorical variable, with each year being its own category. Analyses were conducted using R version 3.2.3 [26] and Stata 13 [27]. letters were sent to each of them, where they were given the opportunity to withdraw from the study. Data were stored on a secure server without information on names or personal identification numbers. --- Results There were no significant differences in the attainment of basic or high education between the isolated cleft groups and the unaffected reference group (Fig 2). However, marginally reduced proportions in the CLP and CPO groups had completed intermediate education by age 21. In analyses stratified on sex, the completion of intermediate education was reduced among males in the CLP group (RR = 0.92, 95% CI 0.86 to 0.99) and among females in the CPO group (RR = 0.92, 95% CI 0.85 to 0.99). About 78.7% of the study cohort was registered with an occupation in 2010 (Table 1). The remaining 21.3% consisted of students, unemployed, retired persons, and people who were too sick to work. Fig 3 shows that the proportion in the manual occupation category was increased for females in the CPO group (RR = 1.39, 95% CI 1.22-1.59), and that males in the CLP group had a decreased probability of employment in the professional category (RR = 0.87, 95% CI 0.78-0.98). There were no significant differences between the cleft groups and the reference group regarding their chances of a low or a high income (Fig 4). However, for males there were minor differences in median income. Males in the CLO group had a median income about 2.3 percent higher than the males in the reference group (95% CI 0.8 to 3.8), whereas the CLP group (-5.7 percent, 95% CI -9.6 to -1.9) and CPO group (-3.6 percent, 95% CI -6.2 to -0.9) had lower median income. The estimated reproduction rate was slightly lower for the CLP and CPO groups. In the sex stratified analyses, however, a reduction was significant only for males in the CLP group (RR = 0.82, 95% CI 0.75-0.90). A similar pattern was found when considering the chance of getting married (Fig 5). However, among individuals who were married, there was no evidence of a reduced chance of becoming parents in any of the cleft groups. --- Discussion This national population based cohort study suggests that individuals who were born with an apparently isolated oral cleft have prospects for future socio-economic status, including chances of becoming a parent, that are similar to those in the reference group. For the CLO group, the similarity was remarkable. The CPO group was also similar to the reference group, although males had a slightly lower median income, and women had a somewhat lower chance of completing intermediate education by age 21. Women in the CPO group were also overrepresented in the manual occupation category. The CLP group had some signs of problems, but this seemed to be confined to the males, who were underrepresented in the professional occupation category, had a significantly reduced proportion with intermediate education by age 21, and a lower median income. The chance of marrying or having own children was also lower than in the reference group. These results emerge after extensive attempts to remove cases with other underlying medical problems that sometimes come together with clefts. The good prospects of the cleft groups may be attributed to clinical management and follow-up. For the CLP group, which includes morphologically more severe clefts [28], there were still some apparent challenges, particularly for males. Studies related to the academic achievement in the cleft groups have often focused on children and adolescents, and have reported ambiguous results. Two large studies with more than 500 cleft cases suggested that school results of children born with clefts were slightly worse than those of unaffected individuals [29,30]. The first study found that the CPO and CLP groups did worse than the CLO and reference group [29], whereas all groups lagged behind the reference in the second [30]. Both studies accounted for social confounders and to some degree for potential syndromes. A smaller study (n = 256 cleft cases), comparing children born with cleft and their unaffected siblings, reported no differences in academic achievement [31]. Roberts et al., 2012 [10] conducted a relevant meta-analysis of several measures of cognitive functioning and development of individuals born with clefts. When accounting for publication bias and heterogeneity between samples, the only measure where the cleft groups significantly lagged behind unaffected individuals was language development. These studies accounted for the impact of disorders associated with congenital anomalies that were detected at birth, but did not account for disorders that were detected later in childhood. Proper adjustment for social background factors was often lacking and may contribute to differences in results between studies. Two previous Norwegian studies have shown that males born with clefts are less likely than unaffected individuals to become a parent [32,33]. Similar results have been found in a study of Danish women [34]. However, neither of these adjusted for social or medical factors. A small questionnaire based study, published by Ramstad et al. in 1995, indicated that Norwegians born with CLP were less likely to marry than unaffected individuals, and when they did marry, they did so at higher age [35]. We confirm that such differences probably exist, but mainly among males. When we included only those who got married, the proportion who became parents was not reduced for any cleft group. This suggests that the reduced fertility was not biological but related to the likelihood of having a partner. Ramstad et al. [35] also found that married men and single women with CLP had lower income than random controls. Another small questionnaire based study from 1975 (196 cleft cases, 190 unaffected siblings, 209 random controls) found that there were no differences between the cleft group and the controls regarding income and occupational status [36]. Unaffected siblings, however, had a higher income. Both these studies used data that is at least thirty years old today, so it is not obvious that their findings are still relevant. Educational attainment, reproduction, marital status, and income may affect quality of life, which in turn has been found by several studies to be negatively associated with oral clefting [7,9]. However, these studies did not account for other chronic medical conditions, outcomes were based on questionnaires as opposed to registry data, and they generally had a low number of participants (n<unk>500) [8]. Because of the differences in methodology and outcomes, the negative association between quality of life and oral clefting cannot be confirmed nor rejected in the current study. The strengths of this study were the population based design and the large sample size. Information about additional congenital anomalies and chronic medical conditions was available, in addition to several markers of social background. Data was prospectively collected, and loss to follow-up was minimized because the study was based on several compulsory national registries. The clinical registries ensured access to data on all patients treated for clefts in Norway during the study period, except for small portion who declined participation in the study or could not be reached. Proper surgical treatment is important to ensure good results regarding appearance and functioning of individuals born with an oral cleft. The structure of the Norwegian health care system ensures that clefts are detected early, and that high-quality treatment is provided. One possible limitation of our study was that the proportion of unregistered couples has increased with time, so marital status was probably a better indicator of relationship status for people born early in the study period. We did not have information on parental income or parental type of occupation, and we lacked information on relevant maternal lifestyle factors (e.g., tobacco smoking, alcohol intake, or use of folate supplements). Although we used available information to remove individuals with associated medical disorders, we may still have included some in our cohort. The registration of congenital anomalies in the Medical Birth Registry is not complete (see e.g., Kubon et al. 2007 andMorris et al. 2014 [37, 38]), and we had access to information on chronic medical conditions only for individuals whose parents had requested financial support from the National Insurance Scheme. Failure to exclude such cases should inflate the difference between the cleft groups and the reference group, and could explain some of the marginal differences we find. Still, an oral cleft did not appear to affect future socio-economic status in Norway. An exception was males with cleft lip and palate, but the differences were small. --- We are not allowed by Norwegian law to make the data available, as the datasets used in the study contain sensitive patient information. The regional ethical committee for medical and health research ethics does not allow for public deposition of the data. Readers can apply for access and permission to analyze data from each of the registries involved (The Medical Birth Registry: http://www.fhi.no/artikler/?id=94819; FD-Trygd http:// www.ssb.no/omssb/tjenester-og-verktoy/data-tilforskning/fd-trygd). --- Ethics The study was approved by the Regional Committee for Medical and Health Research Ethics. None of the above mentioned registries objected to this approval (the Medical Birth Registry, the National Education Database, the National Registry, the National Insurance Scheme, the Norwegian Labor and Welfare Organization, and the Norwegian Tax Administration). Written consent was provided. Cleft cases born between 1967 and 1992 had all turned 18 in 2010 and --- Author Contributions Conceptualization: EB <unk>S DM RTL <unk>AH. --- Data curation: EB <unk>AH. Formal analysis: EB AMSA <unk>AH.

It has been reported that people born with orofacial clefts do worse in life than their peers regarding a range of social markers, such as academic achievement and reproduction. We have compared otherwise healthy individuals with and without clefts, to investigate if these differences are due to the cleft or other background factors.

Introduction Scientific research has a key role to play in developing sustainability solutions. However, integrating science into decision-making processes about sustainability (or any complex or "wicked" issue) alongside the many actors, institutions, types of knowledge, jurisdictions, political processes, and other social issues remains a significant challenge (Cook et al. 2010;McCright and Dunlap 2011;Cvitanovic et al. 2014;Addison et al. 2015;Hering 2015;Cairney 2016;Clark et al. 2016a). Yet, effective solutions have to account for this tangle of overlapping and shifting issues. Moreover, sustainability challenges cannot wait for a slow diffusion of solutions from the scientific community that may or may not be useful (Kates et al. 2001). Indeed, the United Nations Foresight report ranked "Re-connecting Science to Policy" as the fourth out of 21 top challenges for sustainability in the twenty-first century (UNEP 2012). This is one voice in a chorus of calls over the last few decades to update and re-shape what constitutes useful science for highly complex social problems such as sustainability (Funtowicz and Ravetz 1993;Lubchenco 1998;Gibbons 1999;Guston 2004;Fazey et al. 2018). This goes beyond the ability of scientists to communicate their research findings more eloquently. Instead, these calls emphasize finding ways for society to "speak back to science" (Gibbons 1999). In response, there has been considerable academic interest in identifying principles and processes that might build a more dynamic relationship between science, policy, and society (e.g., Kates et al. 2001;Cash et al. 2003;Clark et al. 2016a;Dilling and Lemos 2011;Lemos et al. 2012;Miller et al. 2014;Cvitanovic et al. 2016;Marshall et al. 2017;Fazey et al. 2018). More recently, practitioners of many kinds have been trying to build on those principles (e.g., Guston 2001;McGreavy et al. 2013;Reed et al. 2014;Bednarek et al. 2015Bednarek et al., 2016;;Wyborn 2015;Clark et al. 2016b). However, the emergence of these "boundary spanners"-organizations and individuals that work specifically in the science-policy interface-may be outpacing our understanding of how best to enable effective relationships between science and policy in practice. Some recent efforts have sought to address the challenge of creating feasible and effective practices for boundary spanning (e.g., Bednarek et al. 2016). To build on these, in May 2017, The Pew Charitable Trusts convened a meeting of boundary-spanning practitioners (the authors) focused on sustainability and environmental issues to (1) reflect on progress to date, (2) develop a value proposition for boundary-spanning activities, and (3) identify the challenges and opportunities to mainstreaming boundary-spanning activities more broadly. The participants represented more than 130 years of cumulative practical experience operating at the interface of science and decision-making in 11 countries. This perspective represents the outcomes of this workshop. In sharing our experiences, we hope to contribute a practitioners' perspective to the discourse regarding the relationship between science, policy, and society. We present our perspective in three parts. First, we articulate our shared definition of the practice of boundary spanning and the types of activities it encompasses. As part of this discussion, we identify the core features of boundary spanning that distinguish it from other approaches to improving the use of research in policy, including communicating more effectively about research results (science communication), addressing socially-relevant research questions based on a researcher's conception of usefulness (applied science), or advocating for policy changes that reflect research results (advocacy). Second, we propose specific value propositions for boundary spanning based on our observations practicing it. Finally, we reflect on opportunities to more effectively bring these activities to the fore of mainstream research, training, and funding efforts. Although the ideas presented in this paper are focused on the role of scientific research in policy (because that is where the workshop participants have the most experience), we recognize the importance of considering science in conjunction with other kinds of knowledge relevant to a decision-making process (e.g., traditional knowledge). Indeed, as boundary spanners, we often account for multiple interests and sources of knowledge, recognizing that decision makers rarely use research evidence in isolation. In addition, although the workshop participants have worked throughout the world, the examples we discuss in this perspective are mainly from western contexts. We recognize, however, that approaches to integrate science and policy are highly context specific (across both space and time) and have unique opportunities and challenges in different geographic settings. We encourage other practitioners, working with different kinds of knowledge in different cultural settings and sectors, to build on our efforts and share their experiences. --- What is boundary spanning? Boundary spanning as a concept first emerged in the 1970s in the business and organizational management literature which sought to identify organizational characteristics (e.g., specific functions or roles) that facilitate knowledge exchange between two or more organizations (e.g., Aldrich and Herker 1977;Leifer and Delbecq 1978). More recently, the importance of constructive knowledge exchange has been taken up by those trying to understand how to address "wicked problems" or complex social challenges such as sustainability (Guston 2001;Brown et al. 2010). The idea is that solutions for wicked problems have to account for many dimensions of "knowing and learning" (Kates et al. 2001). This includes the ways different actors engaged in, or affected by, an issue view the cause of a problem, their institutional and political incentives, how they feel about each other, how they view the relevance and credibility of available evidence, how they access and understand evidence, and how they view potential solutions and their viability. Indeed, solutions generated without accounting for all these moving parts are not likely to align with the information needs within a decision process. Drawing on these features, the academic literature, and our collective experiences, we define the practice of boundary spanning as 'work to enable exchange between the production and use of knowledge to support evidence-informed decision-making in a specific context' and boundary spanners 'as individuals or organizations that specifically and actively facilitate this process'. Essentially, boundary spanners dedicate their time to creating and enabling effective knowledge exchange. We recognize that knowledge production and use are not immutable categories; individuals and organizations can play multiple or shifting roles in producing or using knowledge within the same process (e.g., a decision maker who uses research in their decision-making could also provide knowledge about an issue). We also note that boundary spanning is more than just a one-to-one matching process between production and use (e.g., it involves more than just a decision maker articulating a specific question or need). As we have described earlier, accounting for the broader context of actors, perspectives, values, contested evidence, decision-making history, and power dynamics is critical in shaping a productive knowledge exchange process. We contend that with sufficient time, resources, and expertise dedicated to it, boundary spanning can sustain productive interactions between science, policy, and society, lead to increasingly useful science, and ultimately build capacity for science to inform decision-making about sustainability. While some may prefer terms other than "boundary spanning," we use it as a starting point for a discussion of the practice of connecting science and policy. Our definition of boundary spanning encompasses a spectrum of roles and organizational configurations. In some cases, an individual researcher can act as a boundary spanner and work to understand and reflect user needs in their research program, as well as to create opportunities for themselves to engage in a decision-making process. Given the intensity and scope of the work required, however, we have found that boundary spanners are more likely to act in a full-time capacity as an expert intermediary, rather than being engaged directly in research, or to work within a team of researchers and boundary spanners to create integrated, solution-based research programs. In Table 1, we provide examples of a variety of boundary spanners and organizations, each with different boundary functions. For example, an individual could work with a research institute to help researchers to reflect user needs in their research programs and facilitate effective policy engagement (Cvitanovic et al. 2017). At the level of organizations, teams of boundary spanners may work together as a collective to divide the work into more manageable parts and fulfill different needs within the knowledge exchange process (e.g., California Ocean Science Trust). In other cases, a team of boundary spanners may focus on building capacity among scientists and decision makers to engage in boundary spanning (e.g., COMPASS; Smith et al. 2013). In yet other configurations, university-based centers focus on solution-driven collaborations of teams of researchers and boundary spanners who can engage with users and develop relevant research (e.g., Mitchell Center; Hart et al. 2015). Some funding agencies, through their grant-making, actively match the production of science with specific decision-making needs and context using boundary spanners (e.g., the Lenfest Ocean Program; Bednarek et al. 2015). While components of boundary spanning are similar to other roles at the interface between science and decisionmaking (e.g., science communication, applied science, and advocacy), we believe that several features help distinguish it as a distinct practice. First, boundary spanners recognize that effective communication is necessary but not sufficient in connecting science and policy. Instead, they tend to focus on interactive and regular exchanges aimed at understanding what research would be most useful and why, and how other actors and sources of knowledge factor into the decisionmaking process, rather than packaging research for transmission to potential "users" (Cvitanovic et al. 2015a). These iterative exchanges help refine the mutual understanding about research questions and the type of research that is most needed. They also help build the relationships and broader social formations that are necessary to facilitate the uptake of that research (Jasanoff 2004). These exchanges also differentiate boundary spanning from applied science. While research aimed at solving specific problems is a critical part of generating sustainability solutions, applied researchers do not typically have the resources (e.g., time, money, etc.) to actively engage users in developing or implementing research in an iterative way (e.g., Cvitanovic et al. 2016). Third, rather than acting as advocates for specific research results or policy changes, boundary spanners aim to foster trust that they, and in many cases, the scientists and others with whom they work are not pushing an agenda or choosing research findings to fit a particular position (Lacey et al. 2018). In this way, they strive to act in accordance with Pielke Jr's (2007) description of an honest broker, whereby they do not advocate for a single cause or predetermined outcome. Instead, they aim to consider and offer multiple available options and perspectives, and to cultivate a process of knowledge creation and exchange that can be viewed as rigorous, credible, and legitimate (Cash et al. 2003). We are not implying that boundary spanners are value-free and neutral. However, they aim to be reflective and comprehensive about identifying perspectives and values within a process, including their own and those of the scientists involved, so that those values are explicitly recognized and accounted for whenever possible. --- A value proposition for boundary spanning We view boundary spanning as a distinct and emerging practice. Thus, we believe it is useful to understand the mechanisms by which it contributes to more productive relationships between science and policy, both to improve its practice and understanding of its role in knowledge exchange. In this section, we outline four potential benefits of dedicating time and expertise to boundary spanning and illustrate these by drawing on examples from our collective body of work. We developed these from reflections during the workshop on our experiences as boundary spanners. First, our experiences suggest that regular and sustained boundary spanning can help increase the efficiency by which scientific research is tailored for consideration within decision-making. Our observations within the sustainability sector, as well as within other sectors (e.g., education), suggest that research "designed for action" targeted for specific contexts is more likely to be considered in decision-making 2007). For example, in research projects associated with the Mitchell Center for Sustainability Solutions in Maine, US, team members often found that it was essential to work with intended decision makers throughout a research process to ensure that the research questions were relevant and those who might use the results were involved in ways that would support their ability to eventually to do so. This group found that having a system for assessing research needs and preferences for partnerships supported effective tailoring of research question and design (Bieluch et al. 2017). We propose that these efforts can also reduce the risks of science not meeting decision maker needs and mismatches between the timing of research and decision making, or decisions moving forward without sufficient evidence to inform them. An example of this can be seen in the Australian context of the Cooperative Research Centres (CRCs), specifically the eWater CRC. Through consistent boundary spanning, such as brokering between science and policy and communication management, the project was able to adapt research focus areas and align decision support tool development with both national and state level policy needs to support the development of formal policy instruments for the health of Great Barrier Reef (Carroll et al. 2012). Second, we suggest that boundary spanning can increase the potential for durable decision processes and policy. This is not to say that the goal is to support decisions that are static or unchangeable. Rather, we mean decision-making processes that can integrate new evidence and perspectives, including through periods of change, such as changes in governance or unexpected conflicts that may arise (recognizing that political processes can change decision-making dynamics at any point). We contend that boundary spanners can support process durability in two ways. First, boundary spanners' focus on facilitating knowledge exchange means that they assess how different actors understand and process information, and aim to cultivate meaningful, trusted and sometimes sustained relationships among those involved. Based on our experience, we suggest that creating and nurturing this knowledge exchange infrastructure can help actors in the process (including scientists) absorb new information and account for conflicting evidence without derailing an entire process. Similarly, boundary spanners aim to identify and account for contradictory evidence and divergent perspectives as early in the process as possible. We suggest that this comprehensive scanning function may help manage the risk that either a single ideology will shape policy, leaving it likely to be dismantled to meet a different agenda, or that contradictory scientific or other knowledge will reverse a decision. Third, we suggest that dedicated boundary spanning can help increase the legitimacy (Cash et al. 2003) and social robustness (Gibbons 1999) of science, or the degree to which science is accepted among a diverse set of actors and is relevant for societal challenges. Boundary spanners specifically aim to increase permeability between science and policy to promote "testing and retesting" of the usefulness of scientific knowledge (see Gibbons 1999). Our experiences suggest that boundary spanning can result in those involved better understanding the role and value of multiple sources of knowledge, including science, and feeling that their perspectives have been considered. We contend that this could decrease the potential for science to be seen as a vehicle for pushing a particular viewpoint at the expense of other perspectives (and in turn, decreasing the potential for contesting it). We propose that this focus on legitimacy may also help to mitigate the politicization of science, which can be aggravated by scientists advocating for their "rightful" role or favorite research findings (Sarewitz 2016). COMPASS, a boundary organization in the U.S., for example, focuses on empowering scientists to directly share what they know, without making direct policy recommendations. COMPASS boundary spanners coach scientists to be ready for the "what should we do?" question that inevitably comes from policymakers, by arming them to explain how science can help make the implications of a range of policy options more transparent. Scientists see close up how their science can be used in decisions and they become more open to sharing the Federal funding from the National Oceanic and Atmospheric Administration (NOAA) supports regionally focused research centers that coproduce relevant and useful climate science products, working directly with stakeholder groups in an end-to-end process that meets their needs, and helps to build resilience and adaptive capacity Parris et al. (2016) substance of what they know without an agenda. As a result, scientists become trusted resources and connectors to others with relevant knowledge in their academic communities. This is particularly pertinent at the moment given a growing tension between the public questioning of the value of science in public discourse and the urgency within the scientific community to ensure its status and role in decision-making (e.g., McCright and Dunlap 2011;Gillard 2016). Fourth, we contend that, by comprehensively scanning the policy context, boundary spanners can identify current and emerging opportunities for science to inform policy, i.e., policy windows. This, in turn, may increase the opportunities for decision makers, scientists, and others to identify if, when, and how research may be able to meet a decision-making need or fit within a local context (e.g., Gibbons 1999;Cvitanovic et al. 2017;Kettle et al. 2017). Indeed, as described by Rose et al. (2017), the ability to create and capitalize quickly on new policy windows significantly increases the likelihood that a decision will be evidence informed. Finally, if the relationship-building function of boundary spanning is successful, new policy windows may open. This may be especially true if the relationships can be sustained (Honig et al. 2014). In our work, we have found sustained engagement processes leading to new policy windows and research and policy agenda alignment, even after "final" decisions are made. For example, the Lenfest Ocean Program has found that in some cases, grantees and resource managers are interested in continuing to collaborate even after a project has run its course. Most commonly, they express interest in identifying next steps for the research to be used within decision-making as well as follow-up research that could help decision makers. --- Mainstreaming boundary spanning Our experiences suggest that boundary spanning can contribute to sustainability solutions. However, our experiences have also revealed that there are multiple challenges to realizing this potential, which we canvass here. This is not meant to be a comprehensive review or analysis, but rather, a list of some key challenges and opportunities workshop participants identified as pressing. Moreover, we recognize that some of these challenges and potential opportunities could vary depending on cultural context (see Kates et al. 2001). We aim for this discussion to complement other emerging efforts to raise critical issues for boundary spanning and to spur meaningful conversations about what it might take to build capacity at the science-policy interface across the globe (see Schwartz et al. 2018;Fazey et al. 2018). First, recognize boundary spanning as a distinct practice and reconfigure organizational structures accordingly. This recommendation is based on our observation that boundary spanning is confused with, and thus implemented as part of, complementary activities such as science communication, assumed to be an activity that a scientist takes on in addition to their full-time research program, or believed to be an activity that an individual can manage for an entire organization. In our experience, this tends to hinder the essential integrative function and potential of boundary spanning by not allocating sufficient time, resources, or expertise to the effort. It can also constrain professional development and job opportunities, for example, by limiting the skills profiles sought through hiring processes. This is not to say that scientists cannot be boundary spanners in addition to being researchers. However, based on our experiences, we believe that we need a much clearer conception of who is best suited to which roles, and the time and expertise necessary for it. Not all sustainability scientists desire to fill this function, excel at it, or have the time for it. We recognize that re-configuring jobs and organizations to allow for distinct boundary roles is a significant undertaking (see Keeler et al. 2017). However, we also have a critical opportunity to re-shape our institutions to more effectively address sustainability challenges. Indeed, some research institutions are already transforming themselves by organizing around solving specific problems rather than disciplinary lines (e.g., Hart et al. 2015). We urge that the role of expert boundary spanners be a critical part of the conversations about how institutions might more effectively address sustainability challenges. Second, develop new approaches to training and professional development that emphasize the skills needed to work at the science and policy interface. It is commonly assumed that boundary spanners will primarily emerge from the scientific community, and in many cases, that these boundary spanners will be researchers who engage at the science-policy interface in addition to their existing research efforts. As we describe above, however, working at the science-policy interface can be a full time and long-term enterprise and often requires a different skill set. This includes "practical knowledge", or a keen ability to read social cues, facilitate diverse viewpoints, and navigate complex politics (Cairney 2016), and systems or meta-thinking rather than a singular focus on an issue (see Addison et al. 2013;Bernstein et al. 2017, Schwartz et al. 2017). However, training programs for scientists (of all career stages) to engage in the science-policy interface still tend to focus on first generating research considered high quality within academia and then improving scientists' ability to communicate that research, and training programs aimed specifically at boundary-spanners are still emerging. Thus, at the least, training programs for scientists interested in policy engagement need to be reconfigured to reflect the realities of working at the science-policy interface. This could include, for example, understanding how to meaningfully reflect user needs in a research program. Training should also focus on scientists' ability to think as critically about the science-policy interface as they do about their own research programs. This includes reflecting on how their values and perspectives might influence their views on the kind of research that might be useful within a policy process (Bernstein et al. 2017). Given that the time commitments and skill sets needed, however, we also suggest broadening our conception of who might act as a boundary spanners and provide appropriate training (e.g., systems thinking) for expert, and potentially full time, boundary spanners. Third, develop and implement measures of success that appropriately account for boundary-spanning activities. At present, career progression among many boundary spanners remains tied to traditional metrics (e.g., numbers of peer-reviewed publications and citations) or expectations of significant policy change. These overlook the work undertaken by boundary spanners conflate it with advocacy, and have been shown to undermine the extent to which boundary spanning can occur (Shanley and López 2009;Cvitanovic et al. 2015b). We need to describe rigorous boundary spanning and its outcomes in more detail, even it if is challenging, for example, to assess how boundary spanning changes relationships between science and policy and helps facilitate science-informed policy deliberations (see Nutley et al. 2007). --- Conclusions Scientific knowledge, alongside other forms of knowledge, has an important contribution to make in addressing contemporary sustainability challenges. Based on our collective experiences, we contend that boundary spanning as a distinct practice can play a critical role in facilitating this contribution, by reconciling the production and use of scientific knowledge to support sustainability policy and solutions. We believe that boundary spanning has the potential to increase the efficiency by which scientific evidence informs policy, foster the capacity to absorb new evidence and perspectives, enhance research relevance for societal challenges, and open new policy windows. By offering these propositions for the value of boundary spanning, we hope to encourage a more robust and critical conversation about how best to achieve evidence-informed decision making in practice. We encourage colleagues to test and refine our value proposition, as well as to offer new and different ones. We have also identified a number of changes which would enhance our ability to realize the potential of boundary spanning. We do not mean to imply that everyone interested in connecting science and policy more effectively needs to be a boundary spanner. There is a wide spectrum of roles across the science-policy interface. Rather, we feel that we need to better address what functions are necessary at the science-policy interface, how these roles can best be filled, and how to provide support for them. We recognize that institutional norms and professional development conventions are difficult to shift, but without these changes, we believe that opportunities to support evidence-informed decision making and sustainability solutions will be constrained.

Cultivating a more dynamic relationship between science and policy is essential for responding to complex social challenges such as sustainability. One approach to doing so is to "span the boundaries" between science and decision making and create a more comprehensive and inclusive knowledge exchange process. The exact definition and role of boundary spanning, however, can be nebulous. Indeed, boundary spanning often gets conflated and confused with other approaches to connecting science and policy, such as science communication, applied science, and advocacy, which can hinder progress in the field of boundary spanning. To help overcome this, in this perspective, we present the outcomes from a recent workshop of boundary-spanning practitioners gathered to (1) articulate a definition of what it means to work at this interface ("boundary spanning") and the types of activities it encompasses; (2) present a value proposition of these efforts to build better relationships between science and policy; and (3) identify opportunities to more effectively mainstream boundary-spanning activities. Drawing on our collective experiences, we suggest that boundary spanning has the potential to increase the efficiency by which useful research is produced, foster the capacity to absorb new evidence and perspectives into sustainability decisionmaking, enhance research relevance for societal challenges, and open new policy windows. We provide examples from our work that illustrate this potential. By offering these propositions for the value of boundary spanning, we hope to encourage a more robust discussion of how to achieve evidence-informed decision-making for sustainability.

factor hindering their participation in public outreach [10]. The same barriers still apply to social media outreach, even though it initially promised a fast and easy way to reach the public. A major problem is that both faculty and administrators consider outreach a volunteer activity without academic reward or incentive for participation. Other priorities such as teaching, research, grant writing, are given precedence. In a more recent survey of 97 scientists [11], respondents felt the pressure of the academy on research productivity left little time for public engagement. Despite the relative ease of using social media platforms, scientists cited the same concern about time restraints as a reason they did not use social media [12]. It seems clear that any strategy to incorporate more scientists into outreach, including social media, must address the problem of the clock. --- The potential and reach of Facebook The role of Facebook as a source of news and information has become increasingly important as the rise of "fake news" has made clear. Over 66% of Facebook users receive news shared through the site-higher than any other platform except for Reddit [2]. Despite the expanding influence of Facebook, it has received far less attention as a tool for science outreach than Twitter and blogging [8,9,[13][14][15][16][17][18][19][20][21][22][23]. Indeed, few surveyed scientists believe Facebook is an effective form of online science communication [24]. A sentiment echoed in a study on an institutional Facebook page that concluded "Facebook pages do not offer appropriate social context for learning." [25] The peril of ignoring this platform becomes evident when you consider how many people use Facebook-1,790,000,000 monthly users as of the third quarter of 2016, with 81.7% of daily active users residing outside of the US and Canada [26]. Scientists would do well to consider not just how many people are likely to encounter the torrent of baseless scientific information that's being spread on the site but also what they can do to counter it. And that's where the payoff for using Facebook as a tool for science outreach comes in. The key to Facebook is the networks that individuals form on the platform: adult users, on average, connect with 338 friends through Facebook. Daily engagement with Facebook by users appears common, though users primarily consume rather than actively participate in discourse. The average user spends 21 minutes and 6% of their digital time on Facebook [27]. In the US, the daily time spent on Facebook increases to 40 minutes [27] with women visiting the site more often and younger users spending more time on the site [28]. Scientists should note that while Facebook usage is high in both total numbers and frequency of usage, many users may only passively consume rather than actively participate in discourse. Only 44% of users per day liked content posted by their friends, only 31% commented on these posts, and only 10% post status updates to Facebook on a daily basis [29]. However, most Facebook users are actively engaging with their networks on a daily basis: 65% of Facebook users frequently or sometimes share, post, or comment on Facebook. This active engagement is greater than other social media platforms such as Instagram and Twitter [29]. Still, the numbers of likes, shares, and comments may not be the most effective metrics to gauge impact of science-related posts because the passive consumption and exposure to new topics can shift behaviors and perspectives [30]. Key to success is understanding that the reasons people use Facebook vary particularly by gender. A factor analysis identified seven unique uses for Facebook: social connection, shared identities, content, social investigation, social network surfing, and status updating [28]. While the number of male and female users of Facebook are relatively equal [27], usage differs with gender. As a whole, men and women interact with Facebook to view photos and videos from friends (47%), share information with many people at once (46%), read updates from others (39%), and see humorous content (39%) [29]. However, women are more likely to use Facebook to view photos or videos, see entertaining or funny posts, and share information with a large audience [29]. --- The Facebook network of scientists Academics appear to interact with Facebook often-nearly 40% in science and engineering and over 50% in social sciences, arts, and the humanities visit daily [31]. An additional 40-50% are aware of the sites but do not visit regularly [31]. A more recent study finds that 82% of respondents used Facebook [24]. Facebook, in terms of awareness and usage, only falls behind research profiling sites such as Google Scholar, ResearchGate, and Linkedin for scientists and engineers [31]. More recently, however, scientists appear to be more heavily favoring the use of Twitter [24]. In November of 2016, I surveyed scientists via several social media sites examining their usage and behavior on Facebook, including network size and the sharing of science (S1 Text, Supporting Methods). Given the dissemination on social media of the survey through the author's connections, the respondents are biased toward the fields of biological sciences (21%), marine and aquatic science (22%), environmental and conservation science (11%), and ecology and evolution (31%). Another bias may occur because respondents self-reported posting frequencies and may under-or over-report actual usage. Of the 203 scientists who responded, response rate declined with career stage and gender (and response was greater among females). The average number of reported friends was 519.33, but the median was lower at 428. Differences in network size were not tied to scientific field, gender, or career stage (S2 Text. Supporting Results). Of these friends, on average 27.5% were reported to be scientists. However, three distinctive clusters of Facebook scientist users were identified: those who connect with nonscientists (most common), mainly other scientists (rare), or a mixture of the two (common). Interestingly, earlier career scientists were much more likely to have Facebook networks that contained nonscientists. Senior scientists were more likely to include scientists as Facebook friends, potentially reflecting a shift in the view of Facebook as a professional networking tool. However, as scientists rise in their scientific careers, their connections with other scientists increase and deepen because of either exposure to new scientific networks or potential isolation within the "ivory tower." Interestingly, the networks of scientists who self-identified as not having a "traditionally-defined" career within academia (i.e., not labeled as a level of a professor or equivalent) also included more nonscientific members. Scientists' responses to the survey on Twitter suggest a professional verses personal division on Facebook is fundamental in terms in connections as well as posting habits. Although prior work has shown that many scientists (88%) indicated that they "regularly use Facebook for personal communication where science is shared with interested friends and family," the survey findings here suggest scientists posting frequencies may be low. The mean number of posts to Facebook reported by survey participants was 16 per month, though most researchers reported well below 6 posts per month. The mean percentage of science posts to Facebook reported by survey participants was 23.6% of posts per month. Many survey respondents posted about science much less than this, with 75% of respondents posting about science less than 33% of total their total monthly Facebook posts. Comments by scientists on social media supported the notion that many scientists have turned to Twitter instead of Facebook for science outreach. "Posting science to Facebook had become obsolete because of Twitter" "#Twitter is my primary social media for #science" "I used to post more science related posts on Facebook before joining Twitter" The scientists surveyed here infrequently post about science (mean = 23.6%). This finding is similar to the finding of another survey reporting that only 25% of scientists posted frequently (52% posted occasionally) about science [24]. Many of those posts were reported not to address the scientist's own research programs (mean = 17.7% of all science related Facebook posts, Fig 1F). Survey participants had varying practices on sharing science related to their expertise and field. On average, 45.6% of all science-related posts on Facebook were related to the discipline of the researcher. Survey participants also varied in their propensity for sharing culturally controversial science topics (Fig 1H). On average, 40.1% of all science-related posts on Facebook by participants were on controversial topics (e.g., climate change, vaccines, evolution, genetically modified organisms). Survey participants displayed departing practices on addressing controversial science topics with Facebook networks, i.e., with participant groups posting infrequently, moderately frequently, and frequently. Another recent survey found that many scientists are hesitant to engage other Facebook users to correct misrepresentations of science, with only 18% frequently and 40% occasionally posting corrections [24]. --- Becoming a Nerd of trust This study adds to prior work suggesting that scientists appear to be heavily represented on Facebook and many use it regularly to connect with people outside of science [24,32]. Prior research suggests many scientists already believe in the utility of Facebook to share science with personally connected and interested colleagues, family, and friends [24]. But despite this belief and overall usage of Facebook, the survey results here suggest scientists are still missing a rich opportunity to discuss science with the nonscientists in their networks by actually posting and engaging with their networks. The sample sizes here are limited, but if these results here are representative of the larger population of scientists on Facebook, many scientists may not be taking advantage of the platform as an outreach tool. Here, I argue that Facebook represents an unparalleled and overlooked opportunity. Models of science outreach on Facebook do exist. For example, several Facebook groups focus on specific patient populations. In particular, these Facebook groups have made it easier for researchers and medical doctors to find patients with rare diseases [33]. However, one study noted in the case of diabetes support groups that while clinically inaccurate recommendations were rare, 27% of posts featured advertising for non-FDA approved, "natural" products, further highlighting the importance and need of scientific experts on Facebook [34]. Established organizations also attract a substantial Facebook following. For example, Scientific American currently has 2,851,129 followers and the National Institute of Health has 312,875. Even independent groups such as my own Deep-Sea News and Dr. Andrew Thaler's Southern Fried Science receive 15,549 and 8,254 respectively. Indeed, 33% of scientist responding to a survey indicated they administered a Facebook group [24]. The exposure and coordination of social movements including the recent March for Science also testify to the power of Facebook. However, I advocate here that these efforts while successful, at least as measured by followers, are not the same as individual scientists engaging with their personal networks on Facebook, and great opportunity exists in the latter. --- Your personal Facebook audience is large and listening A large audience already exists on Facebook and the personal networks of individual scientists can be quite substantial. In the survey administered for this study, individual networks ranged for most scientists from 223.5 to 706.0. This is an audience that will likely far exceed any fledgling blog or Twitter account in the first years. Not only is this audience large but is personally connected to you. The personal relationships that people have on Facebook often transcend the online world, i.e., the depth of connection is greater than other social media outlets. The promise of Facebook is that many scientists are already a "Nerd of Trust" within their network of family and friends [35]. These connections can often traverse ideological affiliations with nearly 20% of liberals maintaining Facebook friendships with conservatives and vice versa [36]. The composition of these social networks is the most important factor that determines the content encountered by social media users [36]. On average the people in an individual's personal Facebook network, because of familiarity, trust and value their judgment, especially in their specific field [37,38]. Moreover, Facebook is the new public forum where individuals ask each other for input, e.g., asking a mechanic friend for automotive help, a workout partner for fitness tips, an audiophile for best songs in a genre. This back and forth engagement is an ideal vehicle for online science outreach because it occurs in a community with respect for individual expertise. As this relationship grows, I have personally experienced friends commenting on or asking specific scientific questions on current topics as well as posting current scientific research on my page. It is the latter part that represents one of the most exciting aspects-nonscientists engaging and sharing science. While posting frequency is a delicate balance between service and annoyance, this reflects the general nature of Facebook; individuals posting about their individual interests. Facebook streams are filled with mentions of politics, fitness, recipes, internet memes, quotes, and humorous videos. People primarily interact with Facebook to connect with the lives of others and for content [28,29,39]. A scientist's life includes science, and posting about the process of that as both a passion and vocation is reasonable. Facebook users predominantly claim their identities implicitly rather than explicitly, i.e., "show rather than tell" [40], a medium thus well suited for science outreach. --- Sharing information is easy and important Ultimately, scientists need to engage with the social media venues they are already using and enthusiastic about. Facebook is the low-hanging fruit-many scientists already have accounts and are active on a daily to weekly basis. Likewise, it's easy to post updates, links, photos and videos, especially compared with blogging, and to save time by automating cross-posts from a personal blog or Twitter account, e.g., a service like dlvr.it. In social media, the role of scientists to make others aware of information and filter this information could potentially be as valuable as generating new content, i.e., a blog post. In the era where fake news and alternate facts are now common, scientists have the expertise and skill set-and some might say, the responsibility [41]-to efficiently and effectively vet online content for scientific accuracy. Any scientist can quickly post a comment or share a link to correct misinformation in the news or on a conversation thread with minimal effort. Scientists should be cautious and respectful in how they respond to misinformation as some strategies may actually reinforce preconceived ideas [42]. However, providing alternative narratives and repeated messages can reduce, but not eliminate, the impact of misinformation [42]. --- Facebook for science outreach: The way forward Realizing the promise of using Facebook for science outreach may require overcoming cultural and technical barriers. Funders may not consider sharing science on a personal Facebook account as a legitimate form of science outreach. Public outreach is part of the broader impacts statement required for a National Science Foundation grant. But reviewers are likely to prefer that research-related content be shared via a blog rather than a personal Facebook account, even though the audience on Facebook is likely to be far greater than the traffic of most fledgling blogs. Another potential issue-and one that could help resolve funders' hesitation to value social media outreach-is the difficulty of evaluating engagement. Metrics for Facebook are needed that quantify the quality and quantity of engagement with scientific content posted to Facebook. Currently accessing Facebook data is difficult and can often incur a fee. To convince scientists and their funders that it's worth the effort to counter the proliferation of pseudoscience where it's most widely disseminated, we need both serious conversations about the legitimacy of personal Facebook accounts for science outreach and the metrics to gauge their success. --- Supporting information S1

Arguably, the dissemination of science communication has recently entered a new age in which science must compete for public attention with fake news, alternate facts, and pseudoscience. This clash is particularly evident on social media. Facebook has taken a prime role in disseminating fake news, alternate facts, and pseudoscience, but is often ignored in the context of science outreach, especially among individual scientists. Based on new survey data, scientists appear in large Facebook networks but seldom post information about general science, their own scientific research, or culturally controversial topics in science. The typical individual scientist's audience is large and personally connected, potentially leading to both a broad and deep engagement in science. Moreover, this media values individual expertise, allowing scientists to serve as a "Nerd of Trust" for their online friend and family networks. Science outreach via social media demands a renewed interest, and Facebook may be an overlooked high-return, low-risk science outreach tool in which scientists can play a valuable role to combat disinformation.Over the last year, a new political and cultural climate arose in which the prevalence and dissemination of "fake news," "alternate facts," and "pseudoscience" rose considerably. The proliferation of "fake news"-the fabrication of sensationalized stories that imitate the style and appearance of real news articles and are published on sites that mimic legitimate outlets [1]-is likely a reflection of the fact that an increasing proportion of the public get their news through social media. Nearly 62% of adults in the United States in 2016 received news from social media, up from 49% in 2012 [2]. The necessity for scientists to engage with the public online is perhaps greater than ever, and over the last decade, scientists have seen increasing calls, from both within and outside the field, to engage with the public [3-5], especially through social media [6][7][8][9]. The ease, time, and financial costs of starting a social media account, combined with the potential for very large audiences, makes social media outreach seem very promising. But it's a big leap from creating a social media account to building a high-profile social media presence buoyed by original content and an engaged audience, which often requires a long-term commitment [9]. Scientists name the considerable time investment as the number one obstacle to participating in public outreach. In a 2008 survey of 325 scientists, "lack of time" was the unanimous

Introduction Oral health is considered an important part of populations' health and wellbeing [1]. However, while prevalence of oral disease is increasing in many low-and middle income countries [1], in many countries less attention has been given to oral health [2]. A recent literature review showed that dental caries as a major oral health problem is markedly increasing worldwide [3]. In addition, the global trend in lifestyles towards increased consumption of sugar and alcohol as well as smoking suggest that burden of oral disease will persist in the future in many countries [4]. Beside lifestyle factors, the social rank (SR) of individuals and regions are significantly associated with oral health status [5][6][7][8][9]. Additionally, there are in general social inequalities in access and utilization of dental services [10][11][12][13]. In other words, people with a lower SR not only have poorer oral health status, but they may also have lower access to oral health care resources. These issues brought the WHO to encourage countries to incorporate oral health as an integral part of policies for prevention of non-communicable diseases and to promote the accessibility and availability of oral health services especially for poor and disadvantaged populations [1]. In Iran, dental care is mainly provided by the private sector in cities and by the public sector in rural areas. Since 1997, oral health has been integrated into the primary health care (PHC) network within Iran. Following this, dental care is delivered in four levels in the country, from primary prevention in rural areas to specialists' care in the cities [14]. Moreover, during past decades, the number of dentists has increased substantially in the country from 3,500 in 1990 to 11,000 in 2,000 [14]. Based on latest available data from the Iran Medical Council (IMC) [15], more than 20,000 dentists are currently practicing in the public and private sectors in Iran (only 20% of dentists work in public sector and remaining 80% have private practices [13]). In 2012 approximately 1,400 students were admitted to dental schools across the country, a figure which rose steadily through the preceding decade, as demonstrated by the increase in dental schools, which has risen from 18 in 2000 to 44 in 2012 [14,16]. It is well-established that increasing the number of health inputs, including dentists, does not necessarily result in improved health, but how they are distributed is also a determinant factor [17]. Actually, access to health care and distribution of resources within health sector is considered as one of the social determinants of health [18,19]. Although studies in Iran have examined the association between demographic and SR with population's oral health status [20][21][22] and utilization of dental services [23], little attention has been paid to the distribution of dentists across the provinces in Iran. To fill this gap of knowledge, this study examined availability of dentists in Iran using various disparity measures which evaluate different aspects of disparity. We focused on following two research questions: How were the dentists distributed across the provinces in 2009? And was this distribution associated with the provinces' SR? These questions are among policy interests as equity in access to health care is a common goal of policy-makers in all countries. Moreover, response to these questions is relevant for health resource (here dentists) allocation policies across the provinces in Iran. --- Method and Materials --- Study Setting Iran, a lower-middle-income country, is located in the Eastern Mediterranean Region with an area of 1,648,000 km sq. Based on the census data in 2011, a population of about 75 million people are living in 31 provinces in Iran [24]. --- Data Sources and Variables The data on the distribution of population at the province level were obtained from the National Organization for Civil Registration [25]. The data on the provinces' SR were obtained from the Statistical Centre of Iran and the President Deputy of Strategic Planning and Control [24,26]. The data on the number of dentists practicing in public and private sectors in each province were gathered from the Iran Medical Council [15]. In the current study, the number of dentists per 100,000 people (DPR) was used as the indicator for availability of dental care resources for the population in each province. --- Disparity Measures We evaluated two types of disparities in the current study: pure and social. In pure disparity, we examined how dentists were distributed across the provinces in Iran regardless the provinces' socioeconomic characteristics. Then, we examined whether social disparity among the provinces can explain the distribution of dentists across the provinces. Three different measures were used to examine the pure disparity: Gini coefficient, index of dissimilarity (ID) and Gaswirth index of disparity (GID). These measures respond to different research questions. The relative index of inequality (RII) was used to measure the social disparity. These measures are defined as follows: These are commonly used in assessing the pure disparity in distribution of health care resources [17,[27][28][29]. Lorenz curve is used to compare distribution of specific health variable with perfect equality (diagonal line). This curve plots the cumulative share of population ranked by health variable, in an increasing order, against the cumulative share of health variable. The further the distance from diagonal line implies the greater degree of disparity. The Gini coefficient is equal to twice the area between the Lorenz curve and diagonal. Its value ranges from 0 (perfect equality) to one (maximum possible inequality). We used the formula proposed by Brown [30] to calculate Gini coefficient: 1 (1) where G is Gini coefficient; Y i is cumulative share of dentists in ith province; X i is cumulative share of population (ranked by DPR) in ith province; and k is the total number of provinces. This measure takes into account the distribution of health variable in the entire population. --- Index of Dissimilarity (ID) This index estimates the proportion of total health variable, which would need to be redistributed across provinces to achieve a situation of perfect equality [31] and is calculated through following formula: 1 2 | |(2) where p ip is ith province's share of population; p ih is ith province's share of health variable; and n is the total number of provinces. --- Gaswirth Index of Disparity (GID) This index measures relative increase in health variable to bring the entire population to the level received by reference group [32]. It is recommended that the rate of the best group should be used as reference rate. It is calculated in two steps: first, the fraction of entire population that is under-served relative to the best group is calculated as follows: (3) where <unk> i is ith province's share of population; p i is a measure of health variable in ith province; and k is total number of provinces. Then the GID is calculated as GID=U/p. The p is the mean of health variable (dentists, here) in the entire population. While ID estimates that how available dentists should be redistributed (among geographical units or provinces) to achieve an equal distribution (equal DPR across provinces), the GID estimates that how many new dentists should be added to current number of dentists in each province to reach the DPR equal to level of province with the highest DPR. --- Social Disparities --- Relative Index of Inequality (RII) It is a regression-based method to measure the social disparities in health. RII takes into account the population distribution across social groups [28]. After determining the relative position of the population in the provinces ranked by socioeconomic position, the number of dentists in the provinces was regressed on these relative ranks using negative binomial regression with a robust variance. In this case, an RII value greater (lesser) than 1 show that dentists are more available in the provinces with higher (lower) SR. To account for effect of demographic confounders (age and gender), the adjusted RII was also calculated by including the proportion of female in the population, proportion of people younger than 15 years old and proportion of people older than 65 years old as covariates in our regression. The reason for including these covariates in the analysis was that the previous studies showed that age and gender are significantly associated with demand for dental services in the country [23,33]. --- Data Analysis In the current study, the geographic unit of analysis was 30 provinces in Iran (as there were 30 provinces in the country in 2009 and Tehran province was split in two provinces later). To rank the provinces, we used the Human Development Index (HDI), average total expenditures per head (TXH) and average non-food expenditure per head (NFXH). Moreover, the unemployment rate for people of 10 years and older and urbanization rate were used as proxies of SR in correlation analysis. To account for economies of scale in household expenditures in calculating the TXH and NFXH, the household's total and non-food expenditures were divided on the equivalent scaled household size by raising household size to the power 0.56 [34]. To calculate the social disparities, the HDI was used as main variable and TXH and NFXH used in the sensitivity analysis. Furthermore, in another sensitivity analysis, Tehran province was excluded from the analysis to examine the pure and social disparities across the remaining provinces. The reason for this was that Tehran has special situation as the capital of the country and being the centre of economic, social and political activities. The pairwise correlations between DPR and each of the social ranking variables were calculated to examine if there is any linear relationship between the distribution of dentists and the provinces' SR. Data were analyzed using Excel 2010 and STATA version 11. --- Results Figure 1 shows how dentists were distributed across the provinces in Iran in 2009. On average, there were 28 dentists per 100,000 population in the country (range 7-71). There were substantial differences across the country and value of DPR was 11 times higher for Tehran (with the highest DPR) compared to Northern Khorasan (with the lowest DPR). Only three out of 30 provinces had a DPR equal or greater than the country's average (i.e., Yazd, Isfahan and Tehran). As it can be seen from Figure 1, most provinces (60%) had a DPR value of 10 to 20. Table 1 shows the pure and between-area social disparity measures in the distribution of dentists across the country. The Gini coefficient was equal to 0.39 which implies substantial disparity across the country. The Lorenz curve corresponded to this Gini coefficient is been shown in Figure 2. The value of the ID implies that in order to have an equal distribution, 6,193 of current dentists should be redistributed in the country. On the other hand, results of GID show that about 43% of the populations in the country are under-served when compared with Tehran's population and a number of 31,583 new dentists (GID <unk> current number of dentists) are needed in the country to bring the DPR to the level available for Tehran province. Both unadjusted and adjusted RII (ranked by HDI) showed that there were social disparities in the distribution of dentists in favor of better-off provinces. The same results were obtained when THX was used to rank the provinces in the sensitivity analysis. However, when we used NFXH for ranking, the adjusted RII was not statistically significant. Although excluding Tehran province from the sample decreased the magnitude of pure and social disparities, significant disparities remained in the distribution of dentists across the provinces. Table 2 shows that how dentists should be distributed in each province based on the results of ID and GID. Based on the results from ID, the main redistribution should happen from Tehran province where approximately 61% of the current dentists within this province should be redistributed to other provinces. On the other hand, if government wants to increase the availability of dentists in all provinces to the level of Tehran province (i.e., 71 dentists per 100,000 population), then the lowest and the highest number of new dentists are needed in Ilam and Khorasan Razavi provinces, respectively. Table 3 presents the pairwise correlations between DPR and SR across the provinces in Iran. It can be seen that there were strong positive correlations between the density of dentists and the provinces' SR. Moreover, the provinces with higher proportion of population living in urban areas had higher density of dentists. The association between unemployment rate and DPR was small and statistically non-significant. --- Discussion The current study is the first national study that assessed the availability of dentists across the provinces in Iran. The results showed that while availability of dentists in Iran is higher than global average (22 per 100,000 population [35]), there were substantial pure and social disparities in the distribution of dentists across the country, and generally dentists were located in the provinces with better SR. To achieve an equal pure distribution of dentists across the country, about three out of 10 dentists should be redistributed from the over-served provinces to the under-served ones. The previous national and international studies reported that people with lower SR had poorer oral health than people with higher SR [5][6][7][8][9][20][21][22]36]. The results of our study imply that the lower availability of dentists for these people may partly explain the disparity in oral health. There are some potential explanations for the pure and social disparities in the distribution of dentists in Iran. Firstly, about 80% of dentists are working in private sector [14] and dental services are generally provided with high copayments by users. Since people with higher SR have higher capacity to pay and/or higher knowledge about the importance of oral health, then they have potentially higher demand for dental services than their counterpart with poor SR [37][38][39]. Secondly, as it was shown by a recent study in Iran [40], better employment and social opportunities are determinant factors in decision-making for their career among dentists. It seems that the provinces with better SR offer better opportunities for dentists and hence dentists are more motivated to practice at these provinces. Thirdly, several studies have demonstrated that new physicians are more likely to practice in the region (or states) where they have finished their medical school or residency training [41,42]. Considering the fact that 20% of dental schools are located in two provinces with the highest DPR and better social position (i.e., Tehran and Isfahan), this can partly explain the pure and social disparities in the distribution of dentists in the country. Of course, it should be noted that graduated dentists cannot practice in metropolitan areas like Tehran and Isfahan immediately after graduation and they have to work for a few years (3-7 years depending on deprivation and remoteness of the city) in smaller cities to qualify to practice in larger cities. The unequal distribution of dentists has also been reported in the other countries. Kruger et al. [43] reported significant geographic disparities in private dental practice in Western Australia with a higher density in the regions with better SR. In another study in Japan, a Gini coefficient equal to 0.255 was reported for the distribution of dentists in year 2000 [44]. Moreover, the WHO reported a range of less than 0.5 to 40 dentists per 100,000 population around world implying substantial disparity in the distribution of dentist at a global level [35]. Compared to other health resources, distribution of dentists are more unequal across the provinces in Iran. For example, previous studies have shown that the Gini coefficient for specialist physicians, nurses, active hospital beds, rural health houses and pre-hospital trauma care facilities were equal or less than 0.20 in Iran [17,45,46]. The results of the current study should be interpreted in light of some limitations. Firstly, potential incompleteness and measurement errors in the registry data utilized in this study may bias the results. If number of dentists were underestimated (overestimated) for the provinces with lower SR, then there is overestimation (underestimation) in our social disparity. Secondly, the current study is an ecological study at province level. It means that the observed pure and social disparities are between-provinces and it is not valid for within-province variations. Hence, generalizability of the results to smaller geographic units or individuals is limited. Thirdly, prevalence of dental caries and periodontal diseases are among the main determinants of need for the dental services which in turn is a determinant factor in the distribution of dentists. As it was shown in the previous studies [47,48], there are geographic disparities in the distribution of these disorders, implying the need for controlling these factors when examining the distribution of dentists. However, the lack of data on these disorders confined us to control for them. It should be noted that the previous studies in Iran [20][21][22] have shown that people with lower SR have higher need of dental services. This implies that the disparity of dentists may be more profound than what was found in our study. Fourthly, this is a descriptive cross-sectional study which implies that any causal inference from the results should be avoided. Despite these limitations, the findings of this study provide valuable information for health care policy makers. It is suggested that the current policies should be reviewed and some new policies developed to narrow the pure and social disparities in the distribution of dentists in the country. Some potential policies include: allocating a number of training positions at universities for students from the provinces with lower SR to practice in their home provinces after graduation, offering economics incentives for dentists if they practice in the provinces with lower SR and remote areas, promoting coverage of dental services by health insurance system (in terms of population, services and costs) in the country. Promoting insurance coverage may also decrease income uncertainty for the dentists in these provinces. --- Conclusions This study demonstrated that there are substantial pure and social disparities in the distribution of dentists across the provinces in Iran. Generally dentists are located in the provinces with better SR. It is suggested that the results of this study be considered in making decisions on the dental service system in Iran including the dentistry education and health insurance system. In addition, further analyses are needed to explain these pure and social disparities in the distribution of dentists in Iran. --- Conflict of interest The authors declare that they have no conflict of interest.

During past decades, the number of dentists has continuously increased in Iran. Beside the quantity, the distribution of dentists affects the oral health status of population. The current study aimed to assess the pure and social disparities in distribution of dentists across the provinces in Iran in 2009. Data on provinces' characteristics, including population and social situation, were obtained from multiple sources. The disparity measures (including Gini coefficient, index of dissimilarity, Gaswirth index of disparity and relative index of inequality (RII)) and pairwise correlations were used to evaluate the pure and social disparities in the number of dentists in Iran. On average, there were 28 dentists per 100,000 population in the country. There were substantial pure disparities in the distribution of dentists across the provinces in Iran. The unadjusted and adjusted RII values were 3.82 and 2.13, respectively; indicating area social disparity in favor of people

INTRODUCTION Childhood adversity is a common challenge faced by children worldwide (McLaughlin, 2016). The prevalence of exposure to childhood adversity has been found to be 38-39% across high-, middle, and low-income countries (Kessler et al., 2010). Much research also demonstrates that adversity in childhood could have lifelong costs that impair learning, behavior, and health (Green et al., 2010;Shonkoff et al., 2012;Appleton et al., 2017;Edalati et al., 2017;Jakubowski et al., 2018;You et al., 2019). Despite the extensive amount of evidence linking childhood adversity to negative health consequences, debates and inconsistencies still remain regarding the definitions, measurement, and time-varying consequences of childhood adversity (Oh et al., 2018a), particularly when the contexts in which childhood adversity is defined vary. Most of the existing studies of childhood adversity focus on western countries (Björkenstam et al., 2019), perhaps because of the richness of data in these countries. However, the social contexts of developing countries could be sharply different from those in western countries. Even within developing countries, there is a high degree of heterogeneity in social contexts. Therefore, whether the findings obtained from particular countries in the existing literature still hold in other countries remains unclear. This uncertainty calls for more empirical studies based on developing countries, particularly the longitudinal studies that reveal the complex relationships between cumulative childhood adversity and child development outcomes over a long time span. In China, there have been some longitudinal studies about early exposure to adversity and subsequent developmental outcomes. For example, Chen et al. (2012) examine the reciprocal direct and indirect effects among aggression, peer relationships, and depression based on a sample of 1,162 children in Beijing from the third grade to the sixth grade (ages 9-12). They find that children's initial aggression and peer isolation positively contribute to later depression, which suggests that early adverse social-behavioral conditions exacerbate children's developmental problems. Similarly, relying on the life history framework, Lu and Chang (2019) analyze a longitudinal sample of 198 rural adolescents in China. They find that adverse environmental factors, such as parental separation, are positively associated with subsequent aggression and risktaking behaviors. Based on a 6-year longitudinal sample of 1,245 adolescents from 9 counties including China, Chang et al. (2019) investigate how environmental harshness and unpredictability (measured by unsafe neighborhood, negative life events, family chaos, and family income change) affects adolescents' developmental outcomes. Results show that environmental harshness and unpredictability longitudinally corresponds to more externalizing problems and lower academic performance, and such results are invariant to countries. In another study of 206 Chinese adolescents in rural areas, Chang and Lu (2018) examine the associations of family risk factors (such as stress, parental absence, and exposure to mortality and morbidity) and subsequent psychosocial outcomes after 18 months. Results demonstrate that family risk factors are significantly linked to more risky behaviors as well as academic underperformance. In the Chinese context, while such studies provide valuable insights on the associations between early adversity and later developmental outcomes, due to data constraints, such longitudinal studies are mainly focused on childhood and adolescence. It remains unclear how the influences of early adversity will vary or persist over a long time span from childhood to adulthood. Thus, more research is needed to reveal the long-term associations between childhood adversity and developmental outcomes not only in childhood and adolescence but also in adulthood. Drawing on a 15-year longitudinal dataset of children in rural China, this study tries to fill these gaps. I will examine the cumulative adversity, its different domains, and its associations with mental health problems in childhood, adolescence, and adulthood. The following sections will first lay the theoretical background for research questions, and then describe the data and methods. After presenting the empirical results, this paper concludes with a discussion. --- THEORETICAL BACKGROUND --- Definition and Measurement of (Cumulative) Childhood Adversity Childhood adversity is often defined as the adverse childhood experiences or even trauma that could impair children's health and wellbeing over a long time period (e.g., Patterson et al., 2014;Reid et al., 2017;Racine et al., 2020). Since adversity might take place in different aspects of life, researchers have developed many comprehensive measures, such as Adverse Childhood Experiences (ACEs), to capture the complexity of childhood adversity. While some have used the weighting methods to differentiate the importance of various adverse events, "weighting of events, either through regression-based techniques or by independent judges, does not typically improve correlations with outcomes" (Turner and Butler, 2003, p. 95). Also, "previous findings have suggested that the number of adverse events within a specified time period is more important than the novelty or types of events, and that adolescents are at greatest risk when simultaneously experiencing multiple adaptive challenges" (Davidson and Adams, 2013, p. 534). Therefore, cumulative childhood adversity, which is a composite score of a series of adverse experiences, has been extensively used in the existing research (e.g., Clark et al., 2010;Ford et al., 2011;Tan et al., 2017;Danielson and Sanders, 2018). Cumulative childhood adversity covers a wide range of childhood life, which should include at least four domains. First, cumulative childhood adversity is most frequently measured by family disruption or dysfunction such as parental separation, divorce, abuse, or neglect (Schilling et al., 2008;Clark et al., 2010;Danielson and Sanders, 2018;Edalati et al., 2020). Second, it is also often measured by household low socioeconomic status such as economic hardship (Surtees and Wainwright, 2007;Benjet et al., 2009;Tan et al., 2017) and parents' low levels of education and occupation (Wheaton et al., 1997;Yazgan et al., 2021). In addition to these two domains, another domain that could be integrated into cumulative childhood adversity is physical issue since physical health problems could cause huge or even lifelong stress on children. For example, in the existing literature, cumulative childhood adversity has incorporated such measures of children's physical conditions as hospitalization, chronic disease, or poor health (Turner and Lloyd, 1995;Wheaton et al., 1997;Surtees and Wainwright, 2007;Davidson and Adams, 2013;Shen et al., 2017). Finally, educational adversity also contributes to cumulative childhood adversity since schooling and education comprises a critical part of childhood life. Children "spend more time in school than any other setting except their bed" (Eccles and Roeser, 2011, p. 225). Also, poor academic performance could result in tremendous mental distress among students (Chen et al., 2000;Roeser and Eccles, 2000;Davidson and Adams, 2013;Huang, 2015). For instance, research has showed that children with poor academic performance feel more pressure from parents, receive more criticism from teachers, and get less friendliness from peers (Shen, 2020). For these reasons, the domain of education deserves consideration when constructing cumulative childhood adversity. While cumulative childhood adversity provides a convenient tool to investigate the consequences of overall adverse experiences in childhood, a single score might mask some critical, differential information between different types of adversity. Some research has shown that "categorization based on adversity type did appear to result in varying strength of association between each index and mental health outcomes" (Schilling et al., 2008(Schilling et al.,, p. 1141)). Therefore, in addition to cumulative childhood adversity, its different domains also need to be examined, which will reveal whether these different domains have equal or differential influences on child development. --- Health Consequences of Cumulative Childhood Adversity Research to date has demonstrated a series of health and developmental problems associated with cumulative childhood adversity. For example, a meta-analysis of 35 studies shows that cumulative childhood adversity corresponds to delays in cognitive development, infection, and sleep disruption at age 20 (Oh et al., 2018b). Cumulative childhood adversity is also linked to mental health problems in later life periods. Such mental health problems might include anxiety, depression, eating disorders, self-harm behaviors, internalizing problems, externalizing problems, antisocial behavior, and personality disorder, among others (Turner and Butler, 2003;Schilling et al., 2008;Putnam et al., 2013;McLaughlin, 2016;Björkenstam et al., 2017Björkenstam et al.,, 2021;;Steine et al., 2017;Hébert et al., 2018). Undoubtedly, there has been an extensive amount of evidence showing the detrimental effects of cumulative childhood adversity on adolescents' and adults' mental health. In contrast, much less evidence is presented in the existing literature regarding whether such detrimental effects of childhood adversity change across different life stages and whether different domains of childhood adversity contribute equally to mental health in different life periods. More research is needed to address these issues. --- The Mediating Role of Educational Attainment and Moderating Role of Gender In addition to mental health problems, research to date has found that cumulative childhood adversity impairs academic functioning like doing homework, staying calm, and curiosity in learning (Tan et al., 2017). Childhood adversity might also lower status attainment such as educational attainment (Haas, 2006;Shen et al., 2017). Some research has pointed out that "childhood adversity is associated with worse outcomes through lower adult socioeconomic status " (Jakubowski et al., 2018, p. 702). Since educational attainment is a critical component of adult socioeconomic status (e.g., King and Bearman, 2011) and also closely related to mental health (e.g., Esch et al., 2014;Mojtabai et al., 2015), it is necessary to investigate whether educational attainment mediates the relationship between childhood adversity and mental health problems in adulthood. Furthermore, due to historical and cultural reasons, girls are often a particularly vulnerable group compared with boys. For instance, girls are more likely than boys to have abuse experiences (Mathews et al., 2017) and suffer depression (Kessler, 2003). Research on childhood adversity also finds that girls are more likely than boys to experience childhood adversity (Baglivio et al., 2014). Despite such gender differences in childhood adversity and mental health, whether girls and boys have the same relationship between childhood adversity and mental health over life periods remain unclear. --- RESEARCH QUESTIONS As many previous studies (e.g., Schilling et al., 2008;Montez and Hayward, 2014), the present study adopts a life course perspective to investigate how childhood adversity relates to mental health over a long life course from childhood to adulthood. The theoretical framework is illustrated in Figure 1. Capitalizing on a 15-year longitudinal dataset that traced the development of children from childhood to early adulthood in rural China, I will investigate the following research questions: 1. What are the associations between cumulative childhood adversity and children's mental health over a long time span including childhood, adolescence, and adulthood? 2. Do different domains of childhood adversity have similar or differential weights (or effects) on children's short-term and long-term mental health? 3. Could educational attainment, as a critical marker of socioeconomic status, mediate the potential association between cumulative childhood adversity and mental health in adulthood? 4. Are there gender differences in the associations between cumulative childhood adversity and children's short-term and long-term mental health? --- MATERIALS AND METHODS --- Participants The data came from the Gansu Survey of Children and Families (GSCF, 2000(GSCF,, 2004(GSCF,, 2015)). GSCF is a longitudinal study of 2,000 children in 100 rural villages in Gansu. Gansu was a rural province in Northwest China where over 60% of the population resided in rural areas in 2010 (Shen et al., 2021). The percentage of rural population was even higher when the data was first collected in the year 2000. Thus, although the sample was not a representative sample of all children in China, they came from a less developed and more impoverished area and were more likely to experience adversity than the average Chinese children. Moreover, to my knowledge, this is the only available longitudinal data in China that has focused on child development an education and traced their life trajectories from childhood to adulthood in 15 years. For this reason, the data was suitable for this study. The sampled children were first interviewed between ages 9 and 12 in the year 2000, and last interviewed in early adulthood in the year 2015. In the 2000 and 2004 waves, questionnaires were administered at schools and in homes to children, teachers, school principals, mothers, and household heads. In 2015, to reduce non-response and attrition due to outmigration, interviews were conducted in family homes during Chinese New Year, when out-migrants were likely to return home for family reunions during the festival. For respondents who did not return home during the festival, basic demographic and education information from the household head and selfreported information through phone interviews were collected. This resulted in 1,613 out of the initial 2,000 respondents successfully followed in 2015. In the 2000 and 2004 waves, both children's internalizing and externalizing problems were measured, while in 2015, the now-adults' depression and selfesteem were measured as the indicators of their mental health. --- Measures In the present study, childhood adversity consisted of four domains that covered a wide range of childhood life experience: socioeconomic hardship, family disruption, physical issue, and academic setback. Each domain included three binary indicators with 1 denoting a particular adversity and 0 denoting the lack of that particular adversity. All adversity indicators were measured in the year 2000. Socioeconomic hardship measured the socioeconomic status of a household. In extensive research, socioeconomic status comprises three indicators-education, income, and occupation (e.g., Kessler, 1982;Duncan et al., 2002;Baker, 2014;Ayoub et al., 2018). As the sample were drawn from rural areas where all children's parents were farmers, their occupation had no variation and thus only education and income were utilized for socioeconomic hardship. Specifically, socioeconomic hardship was measured by maternal education, paternal education, and income insufficiency. Back to 2000 in the study site, the educational levels of most parents were really low. For instance, in the sample, 51% of mothers and 24% of fathers received no education at all. Thus, no education instead of a low level of education (such as primary education) was selected as the indicator of adversity. A parent who received no education was coded as 1 which indicated adversity for children. Back to the year 2000 in rural Gansu, many households still suffered from poverty and some even relied on borrowing money to afford children's education (Shen and Hannum, 2020). Therefore, income insufficiency was another measure of socioeconomic hardship. It was obtained from the question "whether your household income in the past year was sufficient" in mother's questionnaire. An answer of "no" was coded as 1 for income insufficiency. Family disruption has been found to have long-term negative influences on children's status attainment and mental health (e.g., Biblarz and Raftery, 1993;Gilman et al., 2003;Steele et al., 2009). In the existing research, family disruption is defined as the "discontinuation of cohabitation between the child's biological parents" (Eriksen et al., 2017(Eriksen et al.,, p. 1080)). It often includes the separation and divorce of parents (e.g., Somers et al., 2011). In GSCF, family disruption measured both the formal and informal separations of parents. Formal separation denoted the dissolution of marriage such as divorce, while informal separation denoted the separation due to migration. Therefore, family disruption included three indicators: maternal migration, paternal migration, and parental divorce. A parent was defined as being in migration if he or she was absent from home for at least 6 months in the past year (Graham and Jordan, 2011;Sun and Wang, 2016;Viet Nguyen, 2016). Physical issue denoted the physiological problems that may directly or potentially harm children's physical, emotional, and academic functioning. In GSCF, physical issue was measured by whether children had any chronic disease, insufficient breakfast, or myopia. Children who cannot have enough breakfast, either because of poverty or because of eating habit, tend to have low protein intake (Chitra and Reddy, 2007). "Protein malnutrition is prevalent in the developing parts of the world and children are the most affected" (Navam et al., 2014, p. 1). In China, research has shown that children who had no or little breakfast would have higher risks for malnutrition (Li et al., 2018). In GSCF, 15% of sampled children reported that they could not have breakfast enough to feel full. In this context, insufficient breakfast was taken as a marker of adversity since it was strongly related to malnutrition of children. Myopia among children and youth has become a global problem (Leo, 2017). Yet it is rarely considered as adversity because glasses to correct myopia are often easily accessible and affordable, particularly in urban areas. However, in the context of impoverished rural areas in this study, people either lacked the awareness or had insufficient financial resources to correct for children's myopia. For instance, in this study, 20% of the sampled children reported that they had difficulties in reading blackboard or doing homework due to short-sightedness. For this reason, myopia was included as an indicator of adversity. Academic setback denoted the problems and difficulties children encountered in their schooling and education, such as poor school performance (Davidson and Adams, 2013;Shen et al., 2017). This was measured by two questions in the data: whether children perceived their language (i.e., Chinese) performance was very poor and whether they perceived their math performance was very poor. An answer of "yes" was coded as 1 to denote a particular educational adversity. Furthermore, some children might experience grade retention. Grade retention, often due to falling behind peers, could result in a series of mental health problems such as low self-concept, misconduct in school, and depressive symptoms (Demanet and Van Houtte, 2013;Klapproth et al., 2016;Hu and Hannum, 2020). It was also adopted as an indicator of cumulative adversity in past research (e.g., Wheaton et al., 1997). Thus, grade retention was also included as a third marker of educational adversity. These measures of (cumulative) childhood adversity, although together describing the total adverse events children encountered in particular domains of childhood life, were not necessarily correlated with each other. For instance, a child with poor math performance might not have any chronic disease, and poor math performance was also not indicative of whether parents were divorced. Therefore, Cronbach's reliability score was not calculated for adversity measures. This is consistent with the practice in many existing studies (e.g., Schilling et al., 2008;Montez and Hayward, 2014;Björkenstam et al., 2018). In fact, the reliability of adversity measures concerns more about whether the reported adversity was truly experienced by children. Many past studies adopted retrospective data on childhood adversity and showed a good reliability (Surtees and Wainwright, 2007). Childhood adversity in GSCF were all measured in the year 2000 when children were still in childhood and when such events just took place. Thus, those adversity measures should be more reliable than the retrospective recalls. In 2000 and 2004, children's mental health was measured by their internalizing and externalizing behavioral problems. Internalizing and externalizing problems were measured by a set of questions adapted from the Youth-Self Report (Achenbach, 1991), edited through local piloting for cultural relevance (Liu, 2003). Each type of behavioral problems contained 18 questions, and each question was rated on a 4point Likert scale (1-4: totally disagree, disagree somewhat, agree somewhat, and fully agree). A summative score was calculated for each type of behavioral problems, with a higher score implying more behavioral problems. For internalizing problems, the Cronbach's alpha scores in 2000 and 2004 were 0.834 and 0.802, respectively. For externalizing problems, the Cronbach's alpha scores in 2000 and 2004 were both 0.870. In 2000, the sampled children were in their late childhood. 4 years later in 2004, they entered their adolescence. Thus, the 2000 and 2004 measures of behavioral problems depicted their mental health from childhood to adolescence. In 2015, the respondents' mental health was measured by their depression and self-esteem. Depression in 2015 was measured by a shortened (10-item) form of the CES-D scale, validated and used in many studies (Andresen et al., 1994;Baron et al., 2017;González et al., 2017). For each question, a Likert scale of 1-4 denoted never, occasionally, sometimes, and often for the frequency of each depressive symptom. Thus, higher values indicated a higher likelihood or level of depression. The Cronbach's reliability score was 0.737. Self-esteem in 2015 was measured by the widely used and validated 10-item Rosenberg Self-esteem Scale (e.g., Schmitt and Allik, 2005;Zhang and Kong, 2021). Each item was rated on a 4-point Likert scale denoting totally disagree, disagree somewhat, agree somewhat, and fully agree. A summative score was calculated and a higher score indicated a higher level of self-esteem. The Cronbach's reliability score was 0.722. Educational attainment in 2015 was measured by the total years of education that children had attained, transformed from a categorical variable of children's highest level of education. According to children's age, all sampled children should have completed formal education by 2015. Thus, the 2015 measure represented their stable or fixed educational attainment. In addition, children's age in 2000 and gender (1 for female and 0 for male) were used as control variables. Children's age was calculated by taking the difference between the survey year 2000 and children's birth year reported in the household questionnaire. 1 The descriptive statistics for all variables are listed in Table 1. --- Statistical Analysis This study utilized structural equation models (SEM) for statistical analysis. SEM has the following advantages. First, it can estimate how childhood adversity corresponds to different mental health problems simultaneously. Second, it has a convenient and powerful technique of handling missing data, i.e., the full information maximum likelihood (FIML) method. Different from the conventional multiple imputation method, the results estimated by FIML are unaffected by the imputation model and also asymptotically efficient (Allison, 2015). Third, the multiple group analysis in SEM can estimate the same model for different subgroups simultaneously, which facilitates the comparison between girls and boys for the relationships between childhood adversity and mental health outcomes. Given these advantages, SEM is a preferred method for this study. For each life stage (e.g., childhood, adolescence, and adulthood), a model with mental health problems as endogenous variables was estimated. In each model, the error terms of the two types of mental health problems in the same life stage were also 1 Due to this age calculation formula (i.e., 2000-birth year), some children aged 12 (e.g., those born in November of 1987 and interviewed in July of 2000) could be rounded up to 13. Also, due to some unknown errors in response (e.g., respondents might confuse the lunar year and solar year when reporting birth year) or data input, there were 6 observations whose calculated ages fell out of children's true age range. The ages of these 6 observations were treated as missing values in this study. I also excluded these 6 observations from the sample for a robustness check. Whether the ages of 6 observations were treated as missing values or excluded from analytical sample, results are always consistent without any substantial change. variable for the gender difference model). For each life stage, another model that used the four domains of childhood adversity was also estimated in order to show the differential contributions of these four domains. Next, educational attainment was added to the adulthood model to examine whether it mediates the relationship between childhood adversity and adults' mental health. As for gender differences, the same model was estimated for both girls and boys, and then their parameters were compared. The childhood and adolescence models used all the 2,000 observations. In 2015, only 1,613 respondents were followed, and thus the adulthood model only analyzed these 1,613 observations. FIML was used for parameter estimation and the handling of missing data. All models were saturated models and thus no goodness-offit test statistics were reported. To adjust for non-normality in dependent variables, robust standard errors were calculated for all models (Chou et al., 1991;Hu et al., 1992). All the SEM models were estimated by STATA 17. --- RESULTS Table 1 depicts the descriptive statistics. The average of cumulative childhood is 2.022, which indicates that the overall frequency of childhood adversity is not high. The most common childhood adversity is socioeconomic hardship. This is not surprising given that the sample was collected in rural areas. Academic setback is the second most frequent childhood adversity. This result illustrates the importance of incorporating educational adversity into the construct of childhood adversity when children spend so much time in school. Table 2 and Figure 2 show that after adjusting for children's age and gender, cumulative childhood adversity is significantly associated with both internalizing and externalizing problems in childhood. However, when childhood adversity is decomposed to four domains, not all domains matter for children's behavioral problems. Academic setback has the strongest association with children's behavioral problems. The second strongest association exists between physical issue and children's behavioral problems. In contrast, socioeconomic hardship and family disruption have no significant relationship with children's behavioral problems in childhood. When children become adolescents, the results are listed in Table 3 andFigure 3. Surprisingly, neither cumulative childhood adversity nor any of its four domains measured in childhood has a significant relationship with behavioral problems in adolescence. One particular reason might be that when children enter adolescence and middle school, their living environments change dramatically and thus their behavioral problems also change. Such sharp changes render the previous significant relationship between childhood adversity and behavioral problems no longer significant. Table 4 and Figure 4 show that more cumulative childhood adversity is significantly associated with higher depression and lower self-esteem in adulthood. Yet, this significant association mainly results from socioeconomic hardship in childhood. Moreover, more academic setback in childhood significantly reduces adult self-esteem but not adult depression. The significant association between cumulative childhood adversity and adult mental health problems in Table 4 seems to contradict the results in Table 3. If the effect of childhood adversity on mental health has diminished in adolescence, how could the effect become significant in adulthood? Of course, one reason could be that the measures of mental health problems in adolescence and adulthood used in this study are not exactly the same. But another reason could be the mediating role of educational attainment. As shown in Table 5 and Figure 5, after educational attainment is included in the model, cumulative childhood adversity no longer has a direct, significant relationship with either depression or self-esteem in adulthood. Rather, cumulative childhood adversity has significant, indirect effects on adult depression and self-esteem via educational attainment. Such indirect effect sizes are comparable to the effect sizes identified in Table 4 (i.e., the adulthood model without educational attainment). Finally, Table 6 describes the gender differences in the relationships between cumulative childhood adversity and mental health problems across different life stages. Consistent with previous models in each life stage, for both girls and boys, cumulative childhood adversity is significantly linked to more behavioral problems in childhood, as well as high depression and low self-esteem in adulthood, but is not significantly related to behavioral problems in adolescence. More importantly, there are no significant gender differences in the relationships between cumulative childhood adversity and mental health problems across different life stages. (The significant gender difference in the association between cumulative childhood adversity and externalizing problems in 2004 is more like a statistical artifact since the coefficients for both girls and boys are non-significant.) A further analysis shows that there is no gender difference in the occurrence of cumulative childhood adversity either (t-test p-value = 0.157). --- DISCUSSION This study focuses on cumulative childhood adversity and examines its associations with mental health problems in childhood, adolescence, and adulthood, capitalizing on a unique, 15-year longitudinal dataset in rural China. Consistent with previous literature (e.g., Turner and Butler, 2003;McLaughlin, 2016;Steine et al., 2017), cumulative childhood adversity is associated with various mental health problems in both short and long terms. However, such associations vary across different life stages and different domains of childhood adversity. Different from a 45-year study of birth cohorts in Britain that finds unattenuated effects of childhood adversity on psychological problems over time (Clark et al., 2010), this study in China shows that cumulative childhood adversity has no significant relationship with behavioral problems in adolescence, although that relationship is significant at first in childhood. Indeed, the samples of these two studies are not comparable. But the different results regarding the timevarying effects of childhood adversity again reminded us of the importance of social contexts in which childhood adversity is studied. Furthermore, this study shows the different values of childhood adversity as a summative score and as separate domains. As a cumulative score, it facilitates the discovery of the overall effect of childhood adversity. As separate domains, childhood adversity could reveal the unequal effects of different types of childhood adversity at different times. For example, this study finds that in childhood, physical issue and academic setback have the strongest associations with behavioral problems while in adulthood only socioeconomic hardship matters. Similarly, another study finds that family dysfunction and maltreat have different effects, with the former being more significant in early childhood and the latter being more significant in adolescence (Oh et al., 2018b). These pieces of evidence demonstrate that we cannot simply understand childhood adversity as a homogenous set of adverse experiences even if we continuously use cumulative childhood adversity. This study also demonstrates that the effects of childhood adversity might be complex and implicit. In the long term, even if childhood adversity has no direct effect on adult mental health, it could indirectly affect adult mental health through educational attainment. This result also illuminates the importance of including education in the examination of childhood adversity. Previous research has found that among different types of childhood adversity, dropping out or failing out of school has one of the strongest effect on adult's health (Kuhlman et al., 2018) and low school grades further heighten the risk for mental disorders among those who have experienced childhood adversity (Björkenstam et al., 2017). Similar findings are also observed in this present study. For example, academic setback has a strong association with not only children's behavioral problems but also adults' selfesteem 15 years later. Therefore, education-whether taken as a component of childhood adversity or a mediating or moderating role of childhood adversity's effect-needs to be considered for the study of childhood adversity. Some previous studies have found gender differences in childhood adversity experiences and mental health problems (e.g., Kessler, 2003;Mathews et al., 2017). However, this is not the case in the present study. In the present study, no gender difference is found in either the occurrence of childhood adversity or the association between childhood adversity and mental health problems. This further illustrates the complex relations between childhood adversity and mental health in varying contexts, and calls for more studies on the gender differences regarding both childhood adversity and its health consequences. Of course, the differences between findings in the present study and those in other research might result from different samples. Admittedly, the sample analyzed in this study was drawn from rural China and might not be representative of the Chinese children. Thus, the results in the present study cannot be simply generalized to a different social setting. As pointed out in the introduction, more studies based on empirical evidence from developing countries are needed in order to better reveal the influences of childhood adversity in different contexts. Another limitation in this study is that it has identified the associations between childhood adversity and mental health problems in different life stages, yet such associations are not causations. Thus, whether childhood adversity causes such mental health problems or their associations are just a result of some unobserved confounders remains unclear. Future research could develop better research design to identify more causal interpretations of the complex relationships between childhood adversity and mental health 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/s. --- ETHICS STATEMENT The studies involving human participants were reviewed and approved by Harvard University and the University of Pennsylvania. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin. --- AUTHOR CONTRIBUTIONS WS contributes to data analysis and manuscript writing. --- FUNDING The Gansu Survey of Children and Families (GSCF) was supported by the Hong Kong Research Grants Council (HSSPFS 36600014). Earlier waves of data were supported by the United Kingdom Economic and Social Research Council and Department for International Development (ESRC RES-167-25-0250), the Spencer Foundation Small and Major Grants Programs, the World Bank, and NIH Grants 1R01TW005930-01

Capitalizing on a 15-year longitudinal dataset of 9-12 years old children in rural China, this study adopts a life course perspective and analyzes cumulative childhood adversity and its associations with mental health problems from childhood to adulthood. Four domains of childhood life are selected to construct cumulative childhood adversity: socioeconomic hardship, family disruption, physical issue, and academic setback. Overall, cumulative childhood adversity significantly associates with children's internalizing and externalizing problems as well as adults' depression and self-esteem. However, cumulative childhood adversity has no significant relationship with internalizing and externalizing problems in adolescence. Furthermore, different domains of childhood adversity matter differently for mental health problems in different life stages. Physical issue and academic setback have the strongest association with internalizing and externalizing problems in childhood, while only socioeconomic hardship has a significant relationship with depression and self-esteem in adulthood. The relationship between cumulative childhood adversity and adult mental health problems is fully mediated by educational attainment. Finally, there is no gender difference in either the occurrence of cumulative childhood adversity or the association between cumulative childhood adversity and mental health problems.

a result of some unobserved confounders remains unclear. Future research could develop better research design to identify more causal interpretations of the complex relationships between childhood adversity and mental health 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/s. --- ETHICS STATEMENT The studies involving human participants were reviewed and approved by Harvard University and the University of Pennsylvania. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin. --- AUTHOR CONTRIBUTIONS WS contributes to data analysis and manuscript writing. --- FUNDING The Gansu Survey of Children and Families (GSCF) was supported by the Hong Kong Research Grants Council (HSSPFS 36600014). Earlier waves of data were supported by the United Kingdom Economic and Social Research Council and Department for International Development (ESRC RES-167-25-0250), the Spencer Foundation Small and Major Grants Programs, the World Bank, and NIH Grants 1R01TW005930-01 and 5R01TW005930-02. The writing and submission of the manuscript was supported by the Improvement on Competitiveness in Hiring New Faculties Funding Scheme of the Chinese University of Hong Kong. --- Conflict of Interest: The author declares 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.

Capitalizing on a 15-year longitudinal dataset of 9-12 years old children in rural China, this study adopts a life course perspective and analyzes cumulative childhood adversity and its associations with mental health problems from childhood to adulthood. Four domains of childhood life are selected to construct cumulative childhood adversity: socioeconomic hardship, family disruption, physical issue, and academic setback. Overall, cumulative childhood adversity significantly associates with children's internalizing and externalizing problems as well as adults' depression and self-esteem. However, cumulative childhood adversity has no significant relationship with internalizing and externalizing problems in adolescence. Furthermore, different domains of childhood adversity matter differently for mental health problems in different life stages. Physical issue and academic setback have the strongest association with internalizing and externalizing problems in childhood, while only socioeconomic hardship has a significant relationship with depression and self-esteem in adulthood. The relationship between cumulative childhood adversity and adult mental health problems is fully mediated by educational attainment. Finally, there is no gender difference in either the occurrence of cumulative childhood adversity or the association between cumulative childhood adversity and mental health problems.

The idea that work encroaches on the lives of its employees is hardly new. Four decades ago, Lewis Coser (1974) characterized work as a "greedy institution" that has little compunction about extracting time and attention from its workers. As early theorizing about inter-role conflict and role strains foreshadowed (Goode, 1960;Pearlin, 1983), some of this 'greed' is channeled through what has been called work-family role blurring-that is, the integration of behaviors and thoughts associated with work and family roles (Desrochers, Hilton, & Larwood, 2005;Glavin & Schieman, 2012). Rapid changes in communication technologies and their increasing use in the workplace have modified employees' accessibility (Bittman, Brown, & Wajcman, 2009;Valcour & Hunter, 2005). This has led to greater permeability in the boundaries between work and nonwork life and the integration of these once more separate spheres (Olson-Buchanan and Boswell 2006). Work-family scholars have shown how these changes are due in part to the changing composition of the labor market and, more importantly, the transformation of work itselfespecially the greater prominence of flexible arrangements in the temporal parameters of work (Towers, Duxbury, Higgins, & Thomas, 2006;MacEachen et al., 2008). Despite this, attempts to systematically document those workplace conditions that are associated with permeability and role blurring are either rare or incomplete (see Glavin & Schieman, 2012 for an exception). In the present study, we advance on that research by systematically unpacking the multifarious ways in which conditions in high-pressure work contexts are associated with role blurring. The global economy fuels the demand for flexible schedules and remote work optionsand enhanced communication technologies facilitate the completion of work "anytime, anywhere" (Chesley, 2005). While these technologies often improve personal and organizational functioning, they also generate disruptions outside of work that require border management (Batt & Valcour, 2003;Kossek & Lambert, 2005;Hyman, Scholarios & Baldry, 2005). As the pace, intensity, and complexity of work expands over time (Green, 2005;Kelliher & Anderson, 2010;Maume & Purcell, 2007), these processes engender a normative intrusion of work contact, via proliferated communication technologies, which leads to an apt contemporary characterization of work as the needy institution. Chesley's (2014) depiction of this trend captures the downsides: "It is possible that the instantaneous nature of technologically facilitated communication and information tasks may be changing social expectations about acceptable response times at work and elsewhere in ways that outstrip the new efficiencies of these innovations" (p. 17). These processes and outcomes bring new challenges to what Moen (2015) calls "work-life quality." Role blurring involves a complex set of interrelated activities that include behavioral indicators (e.g., multitasking on work and family activities; being contacted for work purposes at home) and psychological indicators (e.g., experiencing work-related thoughts while at home) (Voydanoff, 2007). Collectively, these indicators denote high role blurring, since individuals who experience all of them are likely to make less of a distinction between their work and family roles. We expand the scope of prior research by refining a previously developed role blurring measure (see Glavin & Schieman 2012) in an effort to include a more comprehensive account of the extent that workers send and receive work-related communications outside of regular working hours. Here, we argue that these technologically assisted behaviors represent an important manifestation of role blurring in the contemporary workforce that deserves further assessment (Batt & Valcour, 2003). It is important to underscore that each of the individual components of role blurring are at once distinctive and yet collectively tap into one underlying theme (described in detail below). A worker might experience frequent work contact and engage in work-family multitasking while at home-but these activities on their own might not fully capture the concept of role blurring. According to Desrochers and Sargent (2004), behaviors and thoughts about work-related matters occur while at home in ways that challenge the separation of work and family domains. Ultimately, its potentially harmful implications for work-family conflict encourage analyses of the contexts that contribute to role blurring (Allen, Cho, Meier, 2014;Chesley, 2014;Dex & Bond, 2005;Voydanoff, 2007). In the present study, we evaluate the possible influences of one of the most pernicious workplace stressors-job pressure. Glavin and Schieman (2012) established that exposure to job pressures represents a key predictor of role blurring. Our study explicitly extends those findings by evaluating the potential status contingencies in that association. We ask: Does higher statusas measured by education, occupation, job authority, earnings, and job control-modify the association between job pressures and role blurring? We test two competing hypotheses. The status advantage hypothesis predicts that higher status should weaken the positive association between job pressure and role blurring. By contrast, the stress of higher status hypothesis predicts that the positive association between job pressure and role blurring is stronger among those with higher status. In supplemental analyses, we also evaluate if gender functions as an additional contingency in these associations. To test our hypotheses, we analyze data from a national sample of workers-the 2011 Canadian Work, Stress, and Health Study (CANWSH)which includes individuals across diverse statuses, occupations, sectors, and work conditions. --- THEORETICAL FRAMEWORK AND HYPOTHESES The Pressure-Status Nexus and Blurred Work-Family Boundaries Sociologists have long been interested in the interrelationships among different and sometimes-conflicting social roles (Pearlin, 1983). As the nature of work changes, so too does the particular configuration of those social roles and the competing priorities within them (Jacobs & Gerson, 2004;Ruppanner & Huffman, 2014). Moreover, the changing dynamics between roles has spawned a vast body of conceptual language (e.g., "boundary dynamics," "blending," "balance," "fit," and so on) that simultaneously informs theory and analyses (Allen et al., 2014;Moen 2015). In Border Theory, which formally articulates the temporal, physical, and psychological characteristics that define the borders between work and non-work roles (Clark, 2000), the concept of permeability is central; it involves the degree to which aspects of one domain are able to enter others (Ashforth et al., 2000). The rapid rise and use of communication technologies generates new challenges for permeability and the integration of work and nonwork domains (Bittman et al., 2009;Chesley et al., 2003). Expanded access to e-mail, laptops, and "smart phone" tools-what Duxbury and colleagues (2006) refer to as "work extending technology"-can enhance flexibility and enable remote work, but these tools also foster permeability that intensifies pressure to accomplish tasks outside traditional workplace parameters (Kelliher & Anderson, 2010;Matusik & Mickel, 2011;Valcour & Hunter, 2005). As Border Theory predicts, these boundary-spanning demands blur the boundaries that separate work and non-work life and require adjustments from workers (Boswell & Olson-Buchanan, 2007;Chesley, 2014;Kossek & Lambert, 2005;Voydanoff, 2005). This gives rise to the potentially problematic enactment of role-related activities beyond the usual spatial, temporal, or psychological parameters of work. We argue that job pressure represents one of the most salient determinants of permeability and, by extension, increases the need for frequent role blurring. Individuals who report job pressure feel overwhelmed by the amount of work they have to do; they have to work on too many tasks at the same time; the demands of their job exceed the time they have to do the work (Diestel & Schmidt, 2009;Kristensen et al., 2004;Schieman, 2013). Collectively, these pressures and their links to workload, intensification, and overwork increase workers' sense that they must devote additional time to work (Cha & Weeden, 2014;Chesley, 2014;Duxbury, Lyons, & Higgins, 2008;Duxbury et al., 2004;Green, 2006;Matusik & Mickel, 2011;Moen, Lam, Ammons, & Kelly, 2013). Ultimately, this kind of exploitation-excessively demanding work with insufficient time to complete it-requires increased efforts and attention that extend beyond the conventional parameters of the workplace (especially in contexts where "ideal worker" norms are potent). We propose then that, by its very nature, job pressure creates the requirement for permeable work-family borders and an amplification of 'deviant' forms of role enactment. These are the means through which the needy institution ensnares its workers. Job pressure is highly relevant in work-stress research partly because of its welldocumented pervasiveness in the population. For example, Galinsky and her colleagues (2005) observe that approximately 9 in 10 American workers agree somewhat or strongly that they experience one or more of the following: "my job requires that I work very fast," "my job requires that I work very hard," or "I never have enough time to get everything done on my job." Similarly, approximately one-third of Canadian workers report that they frequently feel overwhelmed by work and that the demands of their job exceed the time to do it; another 40 percent of Canadian workers feel that they are frequently required to work on too many tasks simultaneously (Schieman, 2013). Moreover, the study of job pressure is important given that workers who experience high levels of job pressure tend to encounter more unfavorable personal and social outcomes, including risks to role functioning, health, and well-being (Ducharme & Martin, 2000;Galinsky, et al., 2005;Glavin et al., 2011;Koltai and Schieman 2015). The prediction of a positive association between job pressure and role blurring is important in its own right. In fact, there is surprisingly limited research that has documented the association using comprehensive role blurring measures and nationally representative survey data; one exception involves our previous research (Glavin & Schieman 2012), which found that job pressures were associated with increased role blurring in a national sample of US workers. To our knowledge, however, no studies have subsequently investigated potential contingencies in the relationship between job pressure and role blurring. We propose that association may be more complex and multifaceted, and outline theoretical reasons for status as a moderator of this relationship. The question then becomes: What form does such moderation take and is it uniform across various dimensions of status? Below, we articulate our rationale in a process that we call the pressure-status nexus, and the competing hypotheses that emerge from it: the status advantage versus stress of higher status hypotheses. Figure 1 illustrates the main components of this conceptual model. --- [INSERT FIGURE 1 ABOUT HERE] --- Status Advantage or Stress of Higher Status? While previous research finds that higher status conditions are associated with greater role blurring (Glavin & Schieman, 2012), we ask: Does higher status modify the relationship between job pressure and role blurring? On the one hand, the status advantage hypothesis predicts that workers with higher status should be better able to withstand the impact of job pressure in ways that minimize role-blurring activities. By contrast, the stress of higher status hypothesis (Schieman, Whitestone, & Van Gundy, 2006;Schieman, Milkie, & Glavin, 2009) predicts that job pressure is more strongly coupled with role blurring among workers with higher status. Moen and her colleagues'(2013) show that "work-family stress and time strains driven by conditions at work, especially heavy temporal demands that are no longer bound by space or time" exemplify the experience of professionals (p. 102). Applied here, these ideas imply that professional workers and those with more education, job authority, personal earnings, and job control might experience a weaker relationship between job pressure and role blurring. Of the two hypotheses, the status advantage view more closely parallels what has been referred to as the 'buffering hypothesis' in the job demands-control (JD-C) and the job demandresources (JD-R) models (Bakker & Demerouti, 2007;Karasek, 1979;Koltai & Schieman 2015). In the sociological study of stress, a number of scholars have referenced similar dynamics in discussions of coping (Thoits, 1995;Wheaton, 1985), especially with respect to social supportbased resources (Sloan 2012). While those ideas refer to the ways that resources buffer against the distress associated with stressors, we follow Wheaton's (1996) encouragement to evaluate the significance of buffering for the relationships between stressors. Among the forms of status that we assess here, education is the most clearly linked with human capital (Becker, 1964;DiTomaso & Parks-Yancy, 2014), as well as other forms of job-related resources like selfdirected work (Kohn, 1976;Kohn & Schooler, 1982;Mirowsky & Ross, 2003). Ross and Mirowsky (2010) describe education as "a special resource, because it indicates resourcefulness, or the general ability to meet situations effectively" (3). These resource advantages involve the development and deployment of skills, knowledge, cognitive flexibility, and social connections (Mirowsky & Ross, 2005). From the status advantage perspective, these facets of education should provide tools that help workers minimize the role-blurring activities associated with job pressure. The status advantage hypothesis therefore predicts that education should have the strongest buffering effect in the positive association between job pressure and role blurring. An alternative perspective-the stress of higher status hypothesis-articulates the possibility that some statuses might exacerbate the positive association between job pressure and role blurring. This view evolves from Blair-Loy's (2009) concept of the "work devotion schema" and the ways that workers who feel greater commitment are likely to expend extra effort and feel more ambition to succeed (Benson & Brown, 2007). In her qualitative study of "unusually successful women," Blair-Loy (2009) observed that the work devotion ideology might become "purer"-and therefore more consequential-among those with higher status (see pp. 291/307). As Boswell and Olson-Buchanan (2007: 595) observe: "Staying connected after hours may be seen as a means to get ahead in the organization and profession more generally." And Moen and her colleagues (2013) assert that higher status workers might be more receptive to and accepting of role blurring. Given ideal worker norms and their implications for expectations about full commitment to work (Acker, 1990;Blair-Loy, 2009;Moen & Roehling, 2005;Williams, 2000), we propose that the stress of higher status predictions may be particularly pertinent for workers with professional occupational status, greater job authority, and higher earnings. While forms of higher status like job authority are thought to be "highly coveted resources" and aspects of workplace rewards (Smith, 2002:511), greater responsibilities for the direction and content of the work of others may correspond with downsides like role blurring (Schieman & Reid, 2009). Regarding the dimensions of job control, Border Theory and the JD-R model both emphasize resources that involve flexibility in the timing and location of work (Bakker & Geurts, 2004;Voydanoff, 2005). In our analyses, schedule control and job autonomy align with this resource perspective. Schedule control is the degree of control that workers have over the scheduling of their work hours or the capacity to select the times that they start and/or finish work (Golden, 2001), while job autonomy entails the extent that workers have the freedom and responsibility to decide how their work gets done (Kalleberg, 2011). The conventional characterization of schedule control and job autonomy underscores the ways that they should help workers navigate the work-family border (Jacobs & Gerson, 2004)-as Perlow and Kelly (2014) suggest, these are features of an "accommodation model" in which flexibility is provided by employers to foster employees' management of work demands (see Munsch, Ridgeway, & Williams, 2014). While there are advantages to these job-related resources, one possible drawback entails greater border permeability and role blurring (Glavin & Schieman, 2012;Chesley, 2014;Moen et al, 2013;Schieman & Glavin, 2008). The resource perspective predicts that schedule control and job autonomy should facilitate more efficacious forms of role blending and integration, thereby potentially mitigating the 'interruption' dimensions that make it problematic (Allen et al. 2014). Voydanoff (2005) underscores this point in her account of "boundary-spanning resources," asserting that the "flexibility provided by resources such as...work schedule flexibility...reduces work-family conflict and perceived stress by increasing one's ability to perform work activities while also meeting family responsibilities" (p. 492-493). While schedule control and autonomy align more clearly with the typical description of'resources,' and should therefore attenuate a positive relationship between job pressure and role blurring, another job characteristic-challenging work-presents a more complicated scenario. Challenging work involves activities that require workers to do different things on their job, learn new things, and use their skills and abilities. This concept blends interrelated themes that researchers have labeled with numerous terms like "job enrichment," "task variety," "skill discretion," "self-directed work," "complexity," "non-routine work," and "creative work" (Bakker & Geurts, 2004;Ford & Wooldridge, 2012;Hackman & Oldham, 1976;Karasek, 1979;Mirowsky & Ross, 2007). While challenging work is a key dimension of job control, it also often reflects higher status in the workplace (DiTomaso & Parks-Yancy, 2014; Schieman, 2013). From this view, it might function more like other forms of status described above (e.g., job authority) in its potential to exacerbate the positive association between job pressure and role blurring. In sum, while conventional thinking tends to favor the status advantage hypothesis prediction that higher status should weaken the association between job pressures and role blurring, the stress of higher status hypothesis provides an alternative: Higher status might strengthen the positive association between job pressure and role blurring-or, fail to attenuate it. Moreover, different forms of job control also yield competing predictions about attenuation versus exacerbation of the expected positive association between job pressure and role blurring. --- Gender and the Pressure-Status Nexus Adding to the complex moderating scenarios described above, it is also possible that the interrelationships articulated in the pressure-status nexus and our two competing hypotheses differ for women and men. Previous efforts to investigate potential gender differences in role blurring have been hampered by a dearth of studies that utilize nationally representative samples of workers. The limited available evidence, however, does suggest that men report more role blurring than women, and that job resources that entail control over the nature and timing of work are more predictive of greater role blurring among men (Glavin & Schieman, 2012;Schieman & Glavin, 2008;Voydanoff, 2005). However, research that identifies gender differences in the interrelationships between job pressures and role blurring is generally lacking. The theoretical basis for anticipating gender differences in our focal hypotheses revolves around the long-standing articulation of the gendered nature of work and the work-family border (Damaske, Ecklund, Lincoln, & White, 2014;Yaish & Stier 2009). In particular, this provides a gendered view of border theory and its description of permeability in ways that help elaborate on the hypotheses described above. The broad claim is that women and men tend to experience the work-family interface differently (Winslow, 2005); in this regard, different socio-cultural orientations shape the realities of permeability and, by extension, the extent and forms of roleblurring activities. From a traditional view of gender roles, the meanings, values, and obligations might continue to differ for women and men. Decades ago, Pleck (1977) proposed that because women have traditionally prioritized family over work, they might be less likely to experience interference from work into the home. By contrast, the traditional notion that men should be "good providers" for their family implies that they might be more likely to prioritize work and, when required, allow work matters into the household (Duxbury & Higgins, 1991;Pleck, 1985). This gendered model of the ideal worker involves an unequal weighting of family versus work priorities such that the needs of work are favored (Acker, 1990;Davies & Frink, (2014;Kanter, 1977;Williams, 2000). In this regard, men who engage in work-family role blurringespecially in the context of high job pressures-reinforce the masculine stereotype. This is not to suggest that the family is less important for men, but rather that the work role is highly salient and therefore, when work calls, men should answer. These claims have direct implications for the pressure-status nexus: Some indicators of higher status position-especially professional occupational status, greater job authority, and higher earnings-might more strongly exacerbate the relationship between job pressure and role blurring among men. In high-pressure contexts, ideal worker norms might intensify. It is therefore plausible that, among men, higher status work increasingly requires them to blur boundaries-with amplified expectations to take work home in ways that both reflect and the subsequently increase the permeability of the work-family border. A parallel but somewhat different argument describes the ways that men are increasingly contributing time and energy to the non-economic aspects of fatherhood, and in this regard have perceived themselves as 'central participants' in activities in the home sphere (Barnett & Rivers, 1996;Coltrane, 1996;Gerson, 1993;Grzywacz & Marks, 2000;Townsend, 2002). If men have become less likely to define home-related role activities as supplemental to their work role, then they might also experience the greatest need for role blurring-especially when they are situated in the pressure-status nexus. By contrast, when higher status women experience similarly high levels of job pressure, the manner in which this relates to permeability-and ultimately, role blurring-might be more constricted. In the context of cultural norms and values associated with being a 'good mother,' women might attempt to keep work and family roles more segmented and therefore limit role blurring. Interpreting these ideas in the framework of our hypotheses implies that the stress of higher status hypothesis might more accurately characterize the experience of men. Analytically, this requires a three-way interaction test to evaluate if the job pressure <unk> status interactions for women and men-that is, whether or not the pressure-status nexus is a gendered experience that yields divergent consequences for work-family role blurring. (2) at least 18 years of age; (3) currently in a paid job or operating an incomeproducing business; (4) employed in the civilian labour force; and 5) living in a non-institutional residence. In households with more than one eligible person, the "next birthday" method was used to randomly select a participant. Calls were made to a regionally stratified unclustered random probability sample generated by random-digit-dial methods. Interviews were conducted in English or French and averaged 35 minutes. A $20 gift card was offered as incentive. The full sample was 6,004, with a response rate of approximately 40 percent. 1 We obtained a final analytical sample of 4,527 cases after excluding the self-employed, those working fewer than 20 hours per week, those with missing values on the dependent variable, and after weighting the data in accordance with the gender, age, education, and marital status composition of the Canadian workforce. --- METHODS --- Sample --- MEASURES --- Work-Family Role Blurring We developed a new five-item index that assesses frequency of role blurring activities. This index builds on our previous three-item measure that examined the frequency that workers are contacted about work-related matters outside of normal work hours, engage in work-family multitasking, and have work-related thoughts outside of the workplace (Glavin and Schieman 2012). As part of the current index, participants were asked: (1) "How often were you called about work-related matters when you were not at work?" (2) "How often did you read jobrelated email or text messages when you were not at work?" (3) "How often did you contact people about work-related matters when you were not at work?" (4) "How often do you try to work on job tasks and home tasks at the same time while you are at home?" and (5) "How often did you think about work-related things when you were not working?" This index builds upon the previous index by assessing the reciprocal nature of both receiving and sending work-related communications. The two new items (3 and 4) extend the conceptual scope of the index by including the respondents' initiation of role blurring-rather than simply the passive receipt of after-hours contact-and therefore taps the exchange element of these dynamics. Using the past three months as the orienting time frame, we provided these response choices: "never" (coded 1), "rarely" (2), "sometimes" (3), "often" (4), and "very often" (5). Factor analysis (not shown) confirms the items tap one underlying construct and the scale has good reliability (<unk> =.80). We averaged the responses to the five items to create the work-family role-blurring index. --- Job Pressure We use three items to measure job pressure that are similar to those used in other previously published research on similar themes like "workload" or "quantitative demands" (Kristensen et al., 2004;Van den Broek, Vansteenkiste, De Witte, & Lens, 2008). The items ask about the following experiences in the past 3 months: "How often did you feel overwhelmed by how much you had to do at work?" "How often did you have to work on too many tasks at the same time?" "How often did the demands of your job exceeded the time you have to do the work?" Response choices are: "never" (1), "rarely" (2), "sometimes" (3), "often" (4), and "very often" (5). Responses are averaged, with higher scores indicating more job pressure (<unk> =.85). --- Statuses Education. We coded educational attainment as follows: (1) "less than high school," (2) "high school or GED," (3) "associate/2-year degree," (4) "some college, no degree earned," (5) "4-year university degree," and (6) "graduate or professional degree." Professional occupation. To assess occupation, we asked: "What kind of work do you do? That is, what is your occupation?" Using the open-ended information provided, occupations are coded into thirty-three categories using the 2006 Canadian National Occupation Classification. We compare individuals in professional or executive occupations with a category that combines all others (technical, service, sales, administrative and production). Job authority. Responses to three items are used to assess job authority: "Do you supervise or manage anyone as part of your job?" "Do you influence or set the rate of pay received by others?"; "Do you have the authority to hire or fire others?" Workers with no authority are compared to workers with supervisory or managerial authority (Elliott & Smith, 2004;Schieman, Schafer, & McIvor, 2013;Smith, 2012). Individuals who responded "no" to all three items are classified as having no authority (the reference group). Participants who responded "yes" only to the "supervise or manage" item are classified as having supervisory authority. Those who supervise others and have some degree of sanctioning authority-that is, they set pay or can hire/fire others-are classified as having managerial authority. Personal income. One item is used to measure personal income. It assesses total personal earnings in the previous year from all sources. We coded income into quartiles. Job autonomy. Three items are used to assess job autonomy. Participants are asked the extent that they agree or disagree with the following statements: "I have the freedom to decide what I do on my job," "It is basically my own responsibility to decide how my job gets done," and "I have a lot of say about what happens on my job." Response choices are coded "strongly disagree" (1), "somewhat disagree" (2), "somewhat agree" (3), and "strongly agree" (4). Responses are averaged; higher scores reflect more job autonomy (<unk> =.76). Schedule control. One item asks: "Who usually decides when you start and finish work each day?" The "someone else" response choice is coded 1, "you are able to decide within limits" is coded 2, and "you are entirely free to decide" is coded 3. Challenging work. Five items measure challenging work: "My job requires that I keep learning new things," "My job requires that I be creative," "My job lets me use my skills and abilities," "The work I do on my job is meaningful to me," and "I get to do a lot of different things on my job." Response choices are: "strongly disagree" (1), "somewhat disagree" (2), "somewhat agree" (3), and "strongly agree" (4). Responses are averaged, with higher scores reflecting more challenging work (<unk> =.79). 2 role blurring; importantly, each of those effects holds net of all the others. These patterns are consistent with previous research examining the work-related antecedents of role blurring (Glavin & Schieman, 2012). For income, the patterns are non-linear: Only those in the highest earning quartile report significantly more role blurring compared to the lowest earners (the reference group). For job autonomy, the quadratic term indicates that the association between autonomy and role blurring is flat up to a point beyond which, at higher levels of autonomy, the relationship between autonomy and role blurring becomes increasingly stronger. Among the control measures we observe that men, younger workers, and those with children at home report more frequent role blurring. Collectively, these patterns indicate that work-family role blurring is unequally distributed in the population of Canadian workers-and clearly demonstrates that workers with higher levels of status tend to experience significantly more of it. --- [INSERT TABLE 1 ABOUT HERE] Turning to the interaction coefficients, models 2 and 3, respectively, demonstrate that the association between job pressure and role blurring is significantly more positive among those with higher levels of education and professional status (p <unk>.001). While job pressure is strongly related to role blurring, these interactions suggest that higher education and professional status do not buffer the impact of job pressure on role blurring-patterns that are contrary to the status advantage hypothesis predictions. Instead, these findings are more consistent with the stress of higher status hypothesis. Figure 2 illustrates the results for the education-based contingency. --- [INSERT FIGURE 2 ABOUT HERE] In model 4, the statistically significant pressure <unk> manager interaction coefficient indicates that the positive relationship between job pressure and role blurring is stronger among managers compared to workers with no job authority (p <unk>.001). Importantly, however, having only supervisory status does not moderate the pressure-role blurring association-this represents an important distinction between supervisory versus managerial authority, with the latter reflecting higher levels of control over organizational resources and, by extension, greater status. Likewise, as model 5 demonstrates, job pressure is also more strongly associated with elevated role blurring among workers in the highest income quartile (p <unk>.001). Taken together, these two core dimensions of socioeconomic status and positional location in the workplace amplify rather than attenuate the positive association between job pressure and role blurring-and, like education and professional status, are consistent with the stress of higher status hypothesis. The significant job pressure <unk> job autonomy 2 coefficient (model 6) suggests that the moderating effect of autonomy is non-linear. The positive association between job pressures and role blurring is strongest among those with low levels of autonomy. Put differently, the gap in role blurring is widest among individuals with lower levels of job pressure, and that gap diminishes at an accelerated rate as levels of pressure increase-with a steeper slope observed for those with low job autonomy. Similarly, the job pressure <unk> some schedule control interaction coefficient (model 7) indicates that pressure is more strongly associated with role blurring among workers who have some schedule control (compared to those with no schedule control). By comparison, however, having full schedule control does not moderate the association. --- Gender and the Pressure-Status Nexus In supplemental analyses, we tested whether any of the statistically significant two-way interactions that we reported in Table 1 are further contingent upon gender-that is, do the observed stress of higher status effects differ for women and men? As Table 2 indicates, only the three-way terms involving professional status, managerial status, and income are statistically significant. That is, job pressure is more strongly associated with work-family role blurring among professionals, managers, and high earners-but in each case, those patterns are more representative of men's experience. Figures 3,4, and 5 show each of those findings. Collectively, these results demonstrate that the stress of higher status hypothesis-as represented in the pressure-status nexus-more accurately reflects the experience of men compared to women. --- [INSERT TABLE 2 AND FIGURES 3 -5 ABOUT HERE] --- DISCUSSION Feeling overwhelmed by the amount of work one has to complete and the squeeze of excessive tasks and insufficient time represent quintessential forms of exploitation experienced by the modern-day worker. These kinds of job pressures are important to study because of their negative consequences for workers' performance and well-being; however, in this paper we consider an additional possibility in the study of job pressure-the proclivity for workplace demands to facilitate increased permeability of the work-family border and role blurring. This element of the work-life boundary dynamic generates the potential for 'deviant' forms of role enactment and provides a basis for unexpected patterns of emotional inequality. Confirming previous observations among Americans (Glavin & Schieman, 2012), and expanding upon an earlier role blurring measure, here we demonstrate a strong positive association between job pressure and the frequency of work-family role-blurring activities in a large national sample of Canadian workers. In addition, two other discoveries make novel contributions beyond that previous study. First, we document that each form of status-education, professional status, job authority, and income-are related positively and independently to more frequent role blurring. These patterns lend additional empirical support for the stress of higher status thesis (Glavin & Schieman, 2012;Schieman et al. 2009). Second, we examine the ways that job pressures relate to role blurring across lower, moderate, and higher statuses; the relationship between job pressure and role blurring does indeed vary considerably across these different levels of status. The set of particular combinations of job pressure and status-what we call the pressurestatus nexus-are among the most compelling findings in our analyses. Specifically, we demonstrate that

Always cite the published version, so the author(s) will receive recognition through services that track citation counts, e.g. Scopus. If you need to cite the page number of the TSpace version (original manuscript or accepted manuscript) because you cannot access the published version, then cite the TSpace version in addition to the published version using the permanent URI (handle) found on the record page.

for 'deviant' forms of role enactment and provides a basis for unexpected patterns of emotional inequality. Confirming previous observations among Americans (Glavin & Schieman, 2012), and expanding upon an earlier role blurring measure, here we demonstrate a strong positive association between job pressure and the frequency of work-family role-blurring activities in a large national sample of Canadian workers. In addition, two other discoveries make novel contributions beyond that previous study. First, we document that each form of status-education, professional status, job authority, and income-are related positively and independently to more frequent role blurring. These patterns lend additional empirical support for the stress of higher status thesis (Glavin & Schieman, 2012;Schieman et al. 2009). Second, we examine the ways that job pressures relate to role blurring across lower, moderate, and higher statuses; the relationship between job pressure and role blurring does indeed vary considerably across these different levels of status. The set of particular combinations of job pressure and status-what we call the pressurestatus nexus-are among the most compelling findings in our analyses. Specifically, we demonstrate that job pressure is more strongly associated with frequent role blurring among individuals with higher education, professional status, managerial authority, and higher earnings. Moreover, these patterns hold steady even after we account for three forms of job control or jobrelated resources: schedule control, job autonomy, and challenging work. And, importantly, the results also remain net of adjustments for the extent of overwork-that is, the commonlydocumented experience of working longer hours among those with higher status (Cha & Weeden, 2014). Taken together, we interpret these findings as being consistent with the stress of higher status hypothesis. The fact that individuals embedded in higher status conditions experience a significantly stronger coupling of job pressure and role blurring directly contradicts the status advantage hypothesis and its prediction that higher statuses should attenuate the positive relationship between job pressure and work-family role blurring. The relevance of these findings extend beyond the narrow parameters of the work-family border by intersecting with broader debates about ideal worker norms and their consequences for worker functioning, particularly across levels of socioeconomic standing (Correll, Kelly, O'Connor, and Williams, 2014). Higher status workers might experience a greater devotion or commitment to their jobs and therefore are more willing-and possible often expected-to allow work-related tasks to spillover outside conventional parameters. Role blurring may simply be an implicit (or even explicit) part of the job description for today's higher status workers; the greater dedication implies an acceptance of the price of ambition-one that involves blurred work-family boundaries (Benson & Brown, 2007;Blair-Loy, 2003). However, it is worth underscoring here that we are not simply describing the conditions of higher status workers, but rather elaborating on their potential role enactments and dynamics in high-pressured contexts. Another way to view our results is to imagine that if we randomly selected any two individuals in our sample who experience higher levels of job pressure, the worker with the higher status is more likely to engage in frequent role blurring. This kind of role blurring might be the 'new normal' because of greater ambition, or because higher status workers-when under more intense time pressure and role overload-might feel more compelled to blur the boundaries of work and non-work life to "get ahead" or maintain their higher status (Moen et al., 2013). But, to be clear, these patterns do not imply that the job pressures experienced by lower status workers are somehow less painful or detrimental. Rather, lower status workers might be less likely to have the kind of content or demands in their work activities that foster role-blurring activities, or their experience of pressure might be more constrained to the particular spatial and temporal parameters of the workplace. While high job pressures might manifest in other negative outcomes for lower status workers (e.g., conflict, poorer health, dissatisfaction), it might manifest less directly in the stress associated with role blurring. Certainly, though, much more research-both qualitative and quantitative-is needed to understand the complex link between job pressure and role blurring across the entire socioeconomic and occupational spectrum. In addition to the basic patterns of the pressure-status nexus, we expand upon prior findings by demonstrating three gender-contingent patterns. Consistent with our previous observations (Glavin and Schieman 2012), we first confirm that men report more frequent role blurring than women; yet, it is the gendered nature of the relationship between job pressures and role blurring that is especially novel. Job pressure is more strongly associated with frequent role blurring among professionals, managers, and higher earners-and those relationships are more apparent among men compared to women. These findings reinforce the notion of a gendered model of the ideal worker norm in which women and men experience a different weighting of work-versus family-related expectations and responsibilities. At least in this form of pressure-status nexus, there is evidence that men tend to be more likely to blur the borders between work and family. Our findings in no way imply that higher status women in high-pressured job contexts are not also engaging in frequent work-family role-blurring activities. In fact, one finding stands out: Job pressure is the strongest predictor of role blurring among both women and men-especially those with higher status. However, it is the relative comparison between men and women that reveals the additional gender gap that reinforces the masculine stereotype. It may be that the nature of the work or the type of job differs for women and men in such a way that when work contact occurs, for example, men are more likely to feel that they should respond. There is some suggestion in prior research that men with job authority report a wider span of responsibility at work. According to Wright and colleagues (1995), "because of the real power associated with positions in authority hierarchies, gender inequalities in authority may constitute one of the key mechanisms that sustain gender inequalities in workplace outcomes" (p. 407). Marsden and colleagues (1993) suggest that women have also tended to hold jobs with fewer "commitmentenhancing features" (e.g., supervisory positions), but even when women hold similar positions of job authority as men, there is some indication that the scope of their authority is more narrow. Prior research also suggests that women tend to cluster in positions of more circumscribed authority, in which they have a narrower band of responsibilities for overseeing other workers with fewer of the other benefits typically associated with authority (Reskin & Ross, 1992). Collectively, these ideas imply that women in managerial positions might therefore be better able to maintain less permeable borders-even in high-pressured contexts. While ideal worker norms are influential, higher status men might also feel norms regarding the ideal father, and therefore desire to contribute more time and energy to the non-economic aspects of fatherhood (Barnett & Rivers, 1996;Coltrane, 1996;Gerson, 1993;Grzywacz & Marks, 2000;Townsend, 2002). Future research could benefit from a deeper exploration of ideal worker norms and their influence on the gendered experiences of the pressure-status nexus. As for the three job-related resources-schedule control, autonomy, and challenging work-each one is associated with more frequent role blurring. This point is worth underscoring: These job resources are each independently related to greater exposure to a potential stressor. Moreover, only schedule control and autonomy modify the effect of job pressure-and those patterns are not linear. For schedule control, the positive relationship between job pressure and role blurring is stronger among those with some schedule control (compared to none). Taken together, these patterns have implications for notions of 'flexibility' and'resources' in both Border Theory and the JD-R model, and their predictions about role blurring (Bakker & Geurts, 2004;Voydanoff, 2005). In particular, the conventional characterization of schedule control and job autonomy underscores the ways that they should enable workers to navigate the work-family border (Jacobs & Gerson, 2004). If that were true, it could be argued that these would also help attenuate the positive link between job pressure and role blurring. For workers with some schedule control, however, that relationship is actually stronger (not weaker, as the status advantage hypothesis predicted). Why does having only some schedule control seem to amplify the blurred boundaries that come with pressure? It may be that individuals with some schedule control have more opportunities than those with no schedule control to engage in role blurringthat is, they can choose their own start and end times (within limits). In this context, 'work creep' might be more problematic compared to those with a more rigid schedule. Flexibility might come with expectations that work can be done elsewhere, at other times. This is another salient feature of the stress of higher status that connects to ideal worker norms and work devotion. --- CONCLUSION While it is possible that role blurring may be functional or even desirable-a means to reducing commuting costs, for example-it is worth noting that previous research has shown a positive association between role blurring at home and work-family conflict (Glavin & Schieman 2012;Voydanoff 2005). Indeed, for some workers, role blurring may initially begin as part of an attempt to deal with excessive job pressures and reduce inter-role conflict; paradoxically, however, these behaviors might expand the reach of work demands into non-work domains in ways that normalize rather than ameliorate these pressures. Future research should test these possibilities by exploring the temporal pathways that connect job pressures, role blurring and work-family conflict. It is also worth investigating the temporal dimensions among these relationships: Is the link between job pressures and role blurring unidirectional? Or is it possible that role blurring contributes to increased perceptions of pressures that produce an upward spiral of overwork? Does role blurring facilitate the ways that work intensification generates inter-role conflict-and do higher status work conditions engender such processes? Much remains unknown about the ways that these dynamics unfold over time for workers across statuses. Our study also contributes to the ongoing debate and discussion about the concept of role blurring itself. The earlier work of Desrochers and colleagues (2005), in which they identified the "Work-Family Integration-Blurring Scale" (WFIBS), provides a foundational framework. They conceptualized work-family role blurring as "as the experience of difficulty in distinguishing one's work from one's family roles" (p. 460). The WFIBS is a three item index consisting of the following: (1) "It is often difficult to tell where my work life ends and my family life begins"; (2) "I tend to integrate my work and family duties when I work at home"; (3) and "In my life, there is a clear boundary between my career and my role as a parent." By comparison, our conceptualization and operationalization of role blurring is more oriented towards specific activities, especially related to after-hours work contact and multitasking. We view our measure as a possible determinant of the WFIBS-that is, for example, people who frequently send and receive work-related communications after hours should be more likely to (subsequently) have difficulties figuring out where work life ends and family life begins or experience a lack of clarity in the boundary between one's career versus parenting. Beyond questions about overlap with the WFIBS, a recent piece by Allen and colleagues (2014) further contends that some of the items in our earlier measure (see Glavin and Schieman 2012) "are similar to items used to capture interruptions as well as a lack of psychological detachment from work" (p. 109). Allen and colleagues raise important questions about the distinctions between and/or interrelationships among different constructs. We agree that these lines can seem somewhat blurry. In our view, however, "distractions," "interruptions," or the "inability to detach from work" are potential consequences of role blurring-but they are not necessarily the same as role blurring activities. For example, Desrochers and colleagues (2005) measure disruptions with a single item: "When I work at home, distractions often make it difficult to attend to my work" (p. 453). Here, there is a clear conceptual distinction between role blurring and its consequences (e.g., distraction/interruption), but once again there is the suggestion of an empirical relationship between them. For example, workers who frequently send emails after work hours might subsequently experience interruptions or remain "cognitively preoccupied" with their work well beyond the normal temporal parameters of work. We submit that these are empirical questions that require tests of the relationships among different constructs: When do role-blurring activities, such as the sending and receiving of work-related communications or work-family multitasking, create disruptions or interruptions at home? Do role-blurring activities automatically result in workers having difficulties psychologically disengaging with work? This is fertile terrain for future research. We also think it might be unwise to assume that role blurring necessarily equates to work detachment problems. Some might characterize detachment from work as a 'natural' (and desirable) state, but there are a number of vocational callings where there is little clear distinction between work and non-work roles (e.g. the clergy, on-duty soldiers, and perhaps even some politicians). For these individuals, the blurring of roles should not automatically represent a failure to detach from work, but it might simply be the way they experience the work-family interface, integration versus segmentation, and their own various role identities. Thus, individuals might experience role blurring that is not necessarily indicative of work detachment problems. Certainly, in many cases, role blurring could be a strong predictor of detachment problems-but this is yet another question that deserves greater scrutiny in future research. Finally, it is essential to acknowledge that the pressure-status nexus occurs within what Moen (2015) has recently referred to as the "institutional/organizational turn," that is, a consideration of the structural constraints placed around micro-/individual-level processes. While our account is admittedly individual-focused, we recognize that the pressure-status nexus occurs within broader institutional/organizational contexts that shape work-life boundary dynamics. The workplace culture might shape the degree to which the pressure-status nexus facilitates role blurring. For instance, some organizational cultures might simply expect frequent work contact outside of regular working hours, while other organizations endeavor to restrain the greedy impulse of after hours contact. The former encourages more frequent deviant role enactments, while the latter discourages it. Unfortunately, we do not have measures that might help account for any influence that organizational culture, but future inquiry might consider it. This could illuminate the point at which the 'potential stressor' of role blurring becomes what Wheaton (1996) calls a'stressful stressor'-one that actually has an impact on health and well-being. --- Control Variables Gender is coded 0 for men and 1 for women. Participants' age is coded in years. Married is coded 1 for married and 0 for all others. Those with children at home are coded 1 and 0 for all others. Private sector jobs are coded 1 and all others are coded 0. Work hours compare part-time (fewer than 35 hours) with those who report full-time (35 to 49 hours) or overwork (50-plus). All analyses adjust for region (e.g., Ontario versus British Columbia, Quebec, Alberta, and so on). --- Plan of Analyses Ordinary least squares (OLS) regression techniques are used to test models. In Model 1 of Table 1, role blurring is regressed on job pressure, statuses, and control variables. Subsequent models test for interaction terms for job pressure <unk> education (model 2), professional status (model 3), job authority (model 4), income (model 5), job autonomy (model 6), and schedule control (model 7). Only the job pressure <unk> challenging work coefficient was statistically insignificant and is therefore not reported in the table. In supplemental analyses to evaluate gender contingencies among the two-way interactions, we tested three-way interaction terms; this involved creating job pressure <unk> status <unk> gender terms. For the sake of presentation, however, in Table 2 we only report the three statistically significant three-way interactions. The Appendix table provides basic descriptive statistics of study variables for women and men. --- RESULTS --- The Pressure-Status Nexus The results in model 1 of Table 1 indicate that job pressure is associated strongly and positively with role blurring. In addition, those with more education, professional status, supervisory or managerial status, schedule control, and challenging work report more frequent

Always cite the published version, so the author(s) will receive recognition through services that track citation counts, e.g. Scopus. If you need to cite the page number of the TSpace version (original manuscript or accepted manuscript) because you cannot access the published version, then cite the TSpace version in addition to the published version using the permanent URI (handle) found on the record page.

Introduction The past decade has evidenced effective approaches to reducing maternal and neonatal mortality in settings with resource strained healthcare infrastructure by increasing access to facilities, services, and life-saving interventions [1,2]. Consequently, global maternal mortality has remarkably declined [1,3], though inequities persist particularly for disadvantaged populations [4]. As maternal and child healthcare access has improved in low-and middle-income countries (LMIC), attention has turned to improving services with the provision of high quality, respectful care [3,[5][6][7]. This is particularly important given that increasing access and demand for quality facility-based maternal health services is a core determinant of positive maternal outcomes. Recent multi-level, systems-integrated interventions have demonstrated some success with increasing service coverage, but still have limited impacts on perceived positive experiences with care [8][9][10][11][12][13]. Specifically, one multi-country investigation found that health service users in 12 LMIC countries had low expectations of care quality, likely resulting in lower demand for higher quality services and inflating satisfaction ratings [14]. More work is needed to further understand how pregnancy and childbirth services are experienced, how satisfaction is articulated and evaluated, and, most importantly, how these findings can improve care provision to optimize quality and subsequent satisfaction and drive future service use for mothers and children [8]. The World Health Organization, building from Donabedian and Hulton's frameworks [15,16], proposed a multi-dimensional framework for maternal and newborn quality of care [17]. Recent systematic reviews of qualitative studies of maternal healthcare satisfaction [6,18] highlight important determinants of satisfaction and demonstrated the utility of qualitative work to uncover the important nuances and descriptions of poor-quality care, illuminating the voices of women to propel change. The authors highlight needs for further research including a need to assess the relative strength of various determinants and further investigations into contextual histories of women's care navigation to help prioritize recommendations and improvements for maternal health services [7,[19][20][21][22][23]. Importantly, descriptions of experiences of (in) equitable and people-centered care remain thin, despite these domains remaining important drivers of care satisfaction and future healthcare seeking decisions among women. Within these, effective communication, respect and dignity, and emotional support have been less articulated in the literature. Instead, most work has focused on mistreatment and abuses (physical, verbal, and emotional) and health systems limitations in the provision of care (e.g., available resources, staffing, adequate referral, facility quality, etc) [23,24]. Ethiopia made substantial progress towards several Millennium Development Goals, including reducing under-five mortality and improving maternal health (Goals 4 and 5) [25][26][27][28][29], and recent multi-level healthcare quality improvement initiatives have demonstrated early promise in their efficacy [8,30,31]. These initiatives respond to the complex difficulty of improving maternal health service utilization indicators including antenatal care visits, facility-based births, and important follow up postpartum care. In the 2019 Demographic and Health Survey, 52% of women did not deliver at a health facility and less than half (43%) received the recommended number of antenatal care visits [26]. In a qualitative study in four Ethiopian regions (Amhara, Tigray, Oromia and SNNPR), Kaba et al. found that the home environment was prioritized as a safe and comfortable space, affording women important experiences of positive birth-related rituals [32]. This was preferred vis-à-vis health facilities, where births were seen as medicalized, problematized, and access to important home-based rituals were denied. Other qualitative studies demonstrate women's preferences for home deliveries in the context of structural barriers to accessing facilities and their poor-quality services with limited supplies [33,34]. Shiferaw et al. found that women preferred the care provided by traditional birth attendants, felt facility-based delivery was unnecessary, and had previous negative experiences in health facilities [35]. While these studies highlight important barriers to care, there is limited research illuminating what satisfactory care means, under what contexts, and how existing expectations and relationships shape maternal perceptions of quality. We seek to fill this gap by presenting birth narratives from women who delivered in government health facilities in Ethiopia. The purpose of this study is to describe women's experiences, perceived satisfaction, and the contextual events that shape expectations with her maternal health care services. --- Methods --- Study setting Ethiopia is a geographically and culturally diverse country split into administrative units including Regions (generally with a referral hospital), Zones (with at least one general hospital), Woredas (with a primary hospital and on average five health centers), and Kebeles (with at least one health post). The government health system in Ethiopia is a decentralized model. Health posts (HPs) are the smallest service unit staffed by a government salaried Health Extension Workers (HEWs) and provide basic health services. HEWs are responsible for providing healthcare and education related to family health, vaccination, infectious disease prevention, and select curative services. HEWs can be a pregnant woman's first and most common interface with the health system. HPs have an affiliated health center (HC) that provide basic emergency obstetric and newborn care [36]. HCs have an affiliated primary hospital that provides comprehensive emergency obstetric and newborn care [37]. Affiliated HCs, HPs and a primary hospital form a Primary Health Care Unit that collaborate to deliver services and, aggregate and share health information. This study was conducted in the regions of Oromia and Southern Nations, Nationalities, and Peoples' (SNNP). While Amharic is the national language, over 70 languages are spoken throughout Ethiopia. In predominantly rural Oromia and SNNP, estimated literacy rates among women are 48 and 53% [38], respectively. The median age of marriage is 17 and 18 years, and the age of first birth is 19 and 20 years for women in Oromia and SNNP, respectively. Ethnographic work illuminates the complex aspects of the cultural context, agency, aspirations, and expectations of motherhood across Ethiopia. Marriage and motherhood are valued sources of feminine identity and often, a requisite to attain social respect and acceptance. However, vis-à-vis rapid social-economic shifts, a woman's transition into motherhood is contested with new emerging modern ideals [39]. --- Study design and sampling This study was part of a mixed-methods program evaluation of a district-wide health systems quality improvement intervention implemented by the Institute for Healthcare Improvement (IHI) Ethiopia in partnership with the Ethiopian Ministry of Health. This qualitative study sought to understand how mothers perceived and described their experiences and satisfaction with MCH services received at government facilities. The intervention targeted health staff and used a woreda-wide systems-embedded quality improvement (QI) approach to provide woreda leaders and health facilities additional clinical training in MHC best practices, QI training and coaching of QI collaboratives to identify quality gaps and implement locally derived and cost-effective solutions and provide data quality improvement assistance for the health information systems. Detailed information related to the nationally integrated intervention is published elsewhere [8,30]. Qualitive data in this analysis was generated from four baseline study Woredas collected between March and April 2018 prior to the invention's full implementation. In consultation with local health leaders and study staff at each site, we purposively identified and selected one hospital and HC in each woreda that best reflected the catchment area's patient burden, service quality, staffing, and challenges (e.g., supply chain) to achieve variation in birth experiences. Childbirth care registries were used to purposively sample five mothers per facility. Our purposive sample frame was theoretically derived to maximize participant variation by parity and age, as we expected that these factors influenced a woman's care experience. Given the hilly terrain and limited formal roads, facility staff, HEW, and/or community guides assisted the research team to find and approach mothers for consent. Inclusion criteria were being a woman aged 18 to 45 years who lived in the selected study woredas and delivered a live newborn at government health center or hospital participating in the IHI intervention two to six-months before their interview. Four female research assistants (RAs) with graduate level training in public health, conducted the interviews. The RAs were based in Addis Ababa and unaffiliated with the health facilities in the study sties. RAs received training from Ethiopian (ASE, DWK) and US-based (AH, HGR, CB) investigators with expertise in qualitative research. For example, RAs explored their own positionality and the power-dynamics that may exist with participants. For example, two RAs that were not mothers opened their interviews by explaining that they had never experienced birth or motherhood, positioning the participant with the power and expertise for them to learn from. During piloting, RAs provided critical input on engaged data collection and communication, focusing on specific strategies to explain the study and begin the conversation to encourage participant comfort and positioning them as an expert of whom we are learning from. Other rapport building efforts included the female RAs practicing active listening and asking questions to help engage and facilitate dialogue with participants and be responsive to their local context and styles of communication. Interviews were also conducted in and around their home in privacy, where they felt more comfortable to discuss their healthcare experiences without the near presence of clinicians. Alongside IHI project coordinators and local investigators, RAs (two in each region) engaged with research sites to receive formal health system and community approvals. Thereafter, RAs met with health facility staff and reviewed facility birth registries to create a sampling list of eligible participants. Participants selected the language for the interview. Interviewers in Oromia conducted interviews in Amharic and Afan Oromo, languages that were spoken by the RAs. Interviewers in SNNPR spoke Amharic and used interpreters to interviews participants in Wolaitaa, Dorze or Gamo as they did not speak those languages fluently. Semi-structured interviews (see Supplement for interview guide) elicited mothers' experiences of their most recent pregnancy and delivery, focusing on various domains of healthcare satisfaction including staff interactions (communication, skill, trust), services (type, quality, privacy), and facility (cleanliness, equipment, distance) informed by the Donabedian Framework [15]. To contextualize expectations for healthcare services and subsequently, healthcare satisfaction, we explored women's past delivery experiences, her recommendations for service quality improvement, and influences in her social and community network that shaped her use of government health facilities. We assessed perceived satisfaction by asking participants if they would use the government facility in the future and/or recommend MHC services to women they knew. Participants provided verbal informed consent to participate in an audio-recorded interview. After each interview, RAs wrote structured field notes on interview characteristics, key themes and reflected on their immediate reactions and recommended foci of subsequent interviews based on information saturation. RAs and the study team regularly debriefed after each interview. Interviewer audios were first transcribed, verbatim, in their original language. Subsequently, transcripts not conducted in Amharic were translated from their original language to Amharic, and then translated to English for analysis. This research was ethically approved by the University of North Carolina at Chapel Hill's Institutional Review Board and obtained a program evaluation waiver from the Ethics Committee of Ethiopian Public Health Association. All methods were carried out in the protocol outlined in the ethical approval application. --- Data analysis We used an inductive analytic approach to iteratively analyze field notes and transcripts [40]. Data analysis began during fieldwork, with fieldnotes and team debriefs via conference calls. Informed by field notes, team debriefs facilitated discussions and reflections of immediate observations and impressions as well as domains and emerging themes to assess saturation. Following data collection, we conducted a rapid preliminary analysis of field notes and key observations from team debriefs. In-depth analysis began after all interview transcripts were translated to English. We developed a codebook informed by our preliminary analysis and an initial review of transcripts. Next, we systematically coded transcripts using ATLAS.ti (v8), first double-coding single transcripts and meeting to address coding disagreements and subsequently update code definitions. We did this iteratively until no new disagreements were identified. Half the transcripts were double coded by [BLINDED] and [BLINDED], and the remaining were single coded. Inductive and deductive codes applied and followed by axial coding to relate themes to satisfaction domains. We then wrote delivery experience narrative summaries for each transcript using a matrix. Lastly, we broadly categorized, compared, and contrasted maternal healthcare processes, experiences, and narratives around satisfaction with matrices. --- Results In total, we interviewed 41 participants: 20 in Oromia and 21 in SNPPR. In Oromia, one interview was conducted in Amharic and 19 in Afan Oromo. In SNNPR, five interviews were conducted in Wolaitaa, 12 in Amharic, three in Dorze, and one in Gamo. Participant details can be found in Table 1. The average age was 28 years and nearly one quarter of women were first time mothers. Among multiparous participants (n = 31), the average number of live births was about three and nine Live births, mean (range) 2.9 (1-9) --- Parity Primiparous (%) 10 (24.4) --- Prior home birth (%) 9 (29) Health Facility Visits, mean (range) 3.9 (1-7) had a prior home birth. Nearly 47% of participants could not read and write. We first present women's descriptions of MHC experiences and then describe the complex array of experiences that built women's expectations for and reactions to maternal health services. We found that a woman's appraisal of her birth experience is largely contingent on these expectations and may be perpetuated by tolerance for poor quality care when women are not adequately engaged by their providers. Our results are visualized in Fig. 1, emphasizing the contextual importance of not only direct birth-related determinants of childcare delivery services (e.g., maternal/newborn survival, health staff communication, and facility conditions), but other determinants more indirectly affecting her experience, including her prenatal care and prior births, social learning related to childbirth, and the authoritative knowledge and culture of the healthcare environment. We conceptualize our findings in a circle, rather than a linear diagram, to emphasize the inextricable impacts these experiences and expectations bring to a mother's appraisal of her care. --- Overall appraisal of MHC experience Women who described relatively uncomplicated births and delivered healthy newborns generally said their MHC experience met their expectations (e.g., maternal and newborn survival) but expressed ambivalence with the services they received. Care experiences and quality were generally framed as the technical provision of healthcare services. While women reported gratitude and relief following mother-infant survival, secondary aspects that influenced their experience included provider communication and facility amenities. Common themes in descriptions of poor experiences with care included perceived neglect and long delays for care, poor rapport with health providers, and health system constraints. Women's experiences were enhanced by supportive communication and behavioral demonstrations of commitment from health staff. We first describe the primary drivers of women's expectations, and then describe various nuanced factors (e.g., individual-level, facility-level and external factors such as social support and economic resources) that shape her experience. --- Survival as a seminal expectation In their delivery narratives, women expressed survival, their own and of their newborn, during delivery to frame her birth experience. Nearly half of mothers spontaneously said surviving birth delivery influenced their general satisfaction with their delivery care. Mothers expressed gratitude to the facility and its providers following a safe delivery, even when describing negative experiences, such as neglect and verbal abuse, including ridicule and threats. Despite being insulted or threatened, some of these women still felt gracious for services because they prioritized a safe delivery, "Even though they insult us, the delivery service they gave is the great thing. They treat us [at the hospital]. Below God, they save us from death. But we became angry when they insult us. " (106, 18 years old, firsttime mother, Oromia) The woman quoted above uses a 'we' statement, extrapolating her own experience as a common one. Another participant stated: "I say my child [is] saved and [I] keep silent". Some women may have tolerated poor provider treatment because they depended on staff to deliver safely and may have felt vulnerable, scared and without a choice. Mothers highlighted that providers were tasked with facilitating a safe delivery, framing the care they expected from them as technical and unidimensional (e.g., delivery resulting in healthy baby and mother). Provider encounters were described as'services' and 'procedures', suggesting transactions rather than collaboration and partnership. Therefore, perceived provider abuse did not seem to affect overall perceptions of satisfaction with delivery services when there was a positive outcome as many prioritized delivering safely over how they were treated. Relatedly, a focus on survival created ambivalence around poor facility conditions such as lack of drinking water and food. Few facilities had piped water, and women frequently said drinking water was sparse. One woman did not mind the challenge of basic supplies, "I didn't mind because I want to be alive and I was begging them to save me and they told me not worry...then they told me I had a son. I was happy... it was a nice room. But there is water problem. " (404, 21 years old, Mother of 1, SNNP) Ultimately, surviving delivery and having a healthy baby were the most salient factors that informed women's reports of their experience. --- Factors facilitating positive experiences of care Provider treatment and communication Provider care and communication were secondary drivers of women's satisfaction during delivery. Women wanted to feel cared for and emphasized the importance of human connection within their MHC services, --- "...a person has to think, care about another person. If a person does not care about another per-son, then what makes us human? A person has to think about another person, they must be thinking us, caring for us, " (206, 25 years old, first-time mother, SNPP) Supportive communication was particularly important to women who felt embarrassed displaying physical vulnerability in front of providers that might otherwise be considered socially improper outside the delivery context, as reflected by one participant, "The doctors and nurses welcomed me in a good way. Even when I was sick, vomiting, and embarrassed, they told me not to [feel embarrassed]... The nurses help a lot...they encourage me to get it out [screaming] and after I gave birth, I felt relief and happiness, " (410, 28 years old, a mother of three, SNNP). Some women described their appreciation for providers' patience and understanding during delivery. One mother recalled shouting at providers when her delivery took longer than she expected and worried a prolonged delivery signaled "problems" for her child, "I was shouting at them. What they were saying and what I was thinking was not the same...I was hating them, I was asking, "Why you are saying, 'You have to wait for some time?' I wanted it [delivery] done immediately, but that's not how it works. " (206, 25 years old, first-time mother, SNNP) She appreciated her providers' calm reaction to her stress and that they explained things to ease her worry. Mothers appreciated when providers routinely monitored or accompanied them throughout their labor and delivery, as reflected by 104, a 40-year-old, mother of four, Oromia, "She [health staff ] never forgot me and observed me now and then". Women felt safe and reassured when health staff frequently physically checked in on them. In turn, many women disliked feeling "lonely", or alone, for long periods while laboring. In addition to physical presence, women noted that feeling a connection to the health staff, or having some familiarity with them, made them feel safer, happier, and supported. "Well, the midwives were girls and they would take care of me, they would come and give me support, they would tell me that it would be okay, when you forget also they would call you using pet names and stuff, it was very good. They would say "It's ok, you are almost done" and they would really make me feel better... Even when I call them in the middle of the night when I have some pain they would come and treat me. They treated me very nicely. " (202, 30 years old, first-time mother, SNNP) Such authentic human connections, through supportive emotional care, were valued and created trust. Participants also felt more comfortable using MHC services when providers took measures to safeguard privacy and confidentiality during delivery. For example, while some women may have felt uncomfortable delivering alongside other women in facilities with limited delivery space, they generally expected and trusted providers to maintain details of their delivery private from others not allowed in the delivery room. Women described walls, covered windows, and privacy curtains as maintaining privacy during labor. Trust in the health staff, developed through previous encounters, professional integrity expectations, and practices of 'taking care', enabled confidence that the details of their delivery would remain private, "Well, the nurses can keep a secret; they can keep anything a secret. Other people cannot keep a secret. But they [the nurses] have professional responsibility, and they know everything, they (Other people) will say, 'This person did this and that', they might say something because it will be new for them. But for the professionals, it is not new. So they can keep a lot of secrets. " (206, 25 years old, first-time mother, SNNP) Women felt treated poorly when providers neglected or dismissed their concerns during labor. One mother described initial provider judgement and the need to advocate for admission when she arrived laboring, "Immediately when I reached [the hospital], they started to check me and the doctor asked me, 'Why did you come? Your time is not reached. You would be around eight to eight and half months'. I told him, 'No, I know my months' and the amniotic fluid had broken. Then they saw me again and it was labor time. I gave birth there. " (110, 21 years old, first-time mother, Oromia) This mother described feeling "shocked" and "disappointed" after being treated this way by a male staff, where she had to defend her presence and need for care. She credited delivering safely to an attentive female provider who took her concerns seriously. Similarly, another woman described the cold transactional care provided by a midwife, "She only let me in and checked me, that was the boundary of our relationship...She didn't greet me... she told us to lay down, and move something on our body, and then they told us to leave the room...I don't believe in her [knowledge], if she gave me some advice, I would believe in her. " (306, 21 years old, first-time mother, Oromia). As noted earlier, women often differentiated poor interpersonal care and communication from quality appraisal of their services if they had a healthy delivery, as the woman above had. Mothers valued warm interaction and wanted staff to demonstrate their expertise by providing information and advisement. Women recommended improvements in staff interpersonal interactions and privacy (particularly in the hospitals). When asked if women would recommend to friends or other women to use the health facility, they almost exclusively said yes, even if they had a negative experience. This first-time mother reflects this stating, "If they [other women] ask me, I will advise them to go to health center...Despite their behavior, the doctors have knowledge, I don't have any doubt in their knowledge. " (306, 21 years old, first-time mother, Oromia). Another mother (408, age unknown, mother of eight, SNNP) said she didn't tell other mothers about the negative aspects of her care because, "if I told them, they may not go [to the health facility]". These women supported their recommendation by elaborating that the safety and emergency services provided in the hospital were worth enduring disrespect. --- Facility structure and conditions Physically clean facility conditions supported positive birth experiences. Some women highlighted the access to janitorial services as well as help cleaning themselves after delivery as a reason to deliver at a facility rather than at home. The attention to cleanliness gave mothers comfort, as one described, "I think what they are doing is an excellent job, especially during labor the cleaning lady cleans everything right away...it makes us so comfortable. " (410, 28 years old, mother of three, SNNP). Alternatively, women cited poor sanitation, such as soiled beds, floors, and walls, and structures providing inadequate privacy and confidentiality structures as areas that they felt could be improved (thought notably, did not drive overall dissatisfaction strongly). Pungent smells made some women feel sick, made it difficult to eat, and propelled women to quickly leave the facility following childbirth. "There was bad smell in the room; you couldn't stay a minute there. Doctors even talked to the janitors about cleaning. They told us it is not good for you and your baby, better to go to your home... I was happy for my safe delivery but I was disappointed by the room quality. " (108, 25 years old, mother of two, Oromia) Some women felt that poor facility sanitation reduced overall service quality during delivery, "The quality of the hospital was not good because they were not cleaning the bed on which women stayed during delivery" (306, 21 years old, first-time mother, Oromia). Poor sanitation and lack of cleaning staff also extended to the post-natal waiting rooms, where women described blood on the beds and floors. Mothers suggested more cleaning practices, testing, and supplies, such as more mattresses for the birthing beds and ultrasound machines. This too, directly related to transactional care. The tangible and often mechanical receipt of services played an integral role in the assessment of maternal and neonatal health services. Services tied to comfort and 'care' were deeply appreciated, though not expected. For example, when available, women appreciated receiving bedpans, blankets, food, drinks and functional assistance (bathing, assistance using the restroom) throughout delivery. For example, 104, appreciated the privacy and assistance with waste removal, an unexpected service, "To receive urine, they put something beside you. They leave you alone assuring you to feel free. What could be more comfortable than this! During delivery I was on a high bed and there is a receiver for waste under the bed where it drains. What more than this can she do?" (104, 40 years old, mother of four, Oromia) Women with higher risk pregnancies also appreciated having access to a maternal waiting home (structures near the health facility typically built in a traditional way, emulating an Ethiopian home to increase comfort for longer stays for high-risk women) as they waited to deliver. Other women appreciated receiving food, drinks, and having their clothes washed. These additional unexpected services, and often the staff that provided them, benchmarked moments of comfort and feeling cared for. Some women did not seem to expect services to help them throughout the delivery process. Women appreciated this help, which consequently increased comfort and satisfaction with their delivery experience. Interestingly, the lack of these services, however, did not typically diminish a participant's positive experience with delivery services as many did not expect to receive some of these services to begin with. --- Social support and satisfaction Material and social support supplemented women's experience and subsequent'satisfaction' with MHC, suggesting satisfaction with delivery experience is extended beyond facility factors and providers. For example, friends, families, and neighbors provided mothers with food, clothing, and blankets that were in short supply at facilities during delivery, as described below, "There was leather on the bed which attracts cold. You feel cold when you sleep on it. There was nothing except that leather. [There was no blanket], I took it [blanket] from my home. Additionally, my husband sent me [another] blanket from home. " (-307, 31 years old, mother of two, Oromia). Relatives also provided women with food, medicine and clean clothes, supplies they did not anticipate receiving at a health facility, reflecting the central role of external support during the delivery experience. Women expected staff to provide medicine and medical services but not necessarily any additional provision of comfort. Women expected support such as encouragement, accompaniment, or assistance bathing from family and friends throughout delivery, though not necessarily from staff. When birth companions and family engagement was experienced during labor and delivery, women noted their gratitude, suggesting that it was unexpected that it would be allowed. For example, 104, appreciated having her agabaatte (husband's sister-in-law) support her in the delivery room, --- "I felt great pleasure for her [agabaate] coming in. The midwife allowed my "agabaatte" to come in after seeing that I was becoming tired. While I was in misery, I grasped her and pushed the baby out with effort. I felt as if she inserted her hands into my body and took out the baby" (104, 40 years old, mother of four, Oromia) However, women varied on their ability or desire to have family, friends, or neighbors present during delivery. For example, many women believed staff did not permit others in the delivery room, "Nobody was allowed to enter there, " said, 106, 18 years old, a first-time mother, Oromia, who did not bother asking, believing it was against the rules. 106 is from a distant community and had no family support throughout pregnancy and delivery. She relied on a neighbor to support her throughout labor, "She was staying with me when there were no doctors with me but when doctors came, she stayed outside. She helped me greatly after I gave birth too, preparing coffee, hot drink like 'atmit', preparing food for me and things like that. " Other women did not want relatives or neighbors present, even if permitted, viewing it as impractical or unnecessary. "Why would they go in when there are nurses? There are three or four nurses, so they will not allow them, and I don't want them there either", said 201, 29 years old, mother of three, SNNP, again echoing the perception that staff did not permit others in the delivery room. Some women did not want others in the delivery room because they felt embarrassed, ashamed or uncomfortable delivering in front of others. Other women distrusted relatives and neighbors with their privacy. Therefore, variations in preferences for social support alongside different perceptions on what staff would allow created differences in women's perceptions of the quality of care they received. Some wanted more privacy and the belief that others were not allowed in the delivery room helped maintain their privacy, while others assumed their companions would have remain outside, and their preference was not considered when assessing their perceptions of the quality of care. Material and social buffers then, play important and complex roles in perceived quality of care, contingent on what women expected and understood a priori. --- Health extension workers created opportunity for increased quality care and satisfaction HEWs were often a mother's s first facilitator for using government health services for MHC, particularly in rural areas, providing initial education about pregnancy and childbirth and advising on what services she can access. Thus, HEWs set mothers' initial expectations for anticipated services, particularly that a facility delivery would afford them a safer and more hygienic childbirth. One mother explained, --- "She [HEW] teaches us how our health will get improved if we visit the health center. That my baby will be healthier and I will be more relaxed with my body...I will know how my baby is doing inside. " (406, 25, mother of five, SNNP) HEWs also provided education around warning signs and complications to propel women with more autonomy in their care. The relationship HEWs developed with mothers contributed to women's satisfaction with delivery by providing multiple forms of support and advocacy that complemented the services provided at the facility. For example, 402, 30 years old, mother of three, SNNP, described the HEW as someone who, "could find a solution for any problem that existed. " HEW support in labor and delivery was described as both instrumental and emotional. HEWs were frequently the first point of contact and played a key role in facilitating access to and from HCs and hospitals and helping women quickly obtain or navigate care upon arriving at facilities, as again illustrated by 406, "She came to the health center and asked me how I was doing. Then she asked why I didn't call her and, when I left the health center, she took me to my home paying for the motorcycle. She told me to take the things given by the government, like oil and bread flour. " This narrative highlights a disconnect between what HEWs expect to do for mothers (and the resources they know are available to them) and what resources and support mothers understand is available to them, suggesting important equity gaps. HEWs also emotionally supported women throughout delivery. 210, 18, a first-time mother, SNNP, liked that HEWs accompanied women to the clinic so they did not have to go alone, "She is a good person. She goes with us. We follow her teaching. " Other women appreciated when HEWs accompanied women throughout delivery, fulfilling a support role not often given by the health staff. HEW's encouraged women to relax and aided with pushing in the delivery room. One mother felt good knowing that the HEW did not leave her site during delivery. Others described how the HEWs brought them blankets during delivery. Disparities in HEW relationships and support existed, as some women did not engage with HEWs beyond a couple limited interactions, thus not affording them a buffer if there was insufficient emotional support from health staff. --- Past experiences shaped delivery care expectations In addition to the individual, facility and social support factors that influenced satisfaction, women's narratives of satisfaction with their latest birth were inextricably linked to their past births and antenatal care. Five women that had delivered previously in a government health facility reported positive past experiences using healthcare services. 204, 26 years old, a mother of two, delivered her second child at the same health center as her first because she felt staff went beyond their scope of work to care for her, --- "[This health center] is really good. When I was giving birth to my first baby, I went in the morning and when it got to lunchtime, I asked him [the staff ] not to end his shift. He was a night duty so he should leave at 9:00 in the morning, but he stayed for me without even eating his breakfast, when he said there is another person on duty and he had to leave, I asked him to stay. So, he stayed [with me] without eating lunch. He stayed with me until I gave birth. " (206, 26 years old, mother of two, SNNPR) Feeling 'cared for' and experiencing health staff committing themselves to their care encouraged women to return for future MHC services. The relationships these mothers established with the healthcare providers (e.g., midwife, HEWs, etc.) helped build trust and familiarity with the care processes and procedures of childbirth. Their previous safe delivery made them more confident of a subsequent safe childbirth, reinforcing expectations that delivery at the facility would result in a healthy infant. Multiparous mothers with a prior homebirth and a facility delivery naturally contrasted their experiences, underlining what mattered most: maternal and infant survival. A mother of five described, "I had near death experience because the umbilical cord was not cut. The baby was crying laying on the floor and the person who was supposed to cut the umbilical cord lost the blade and I was scared. I didn't want that experience again, so I decided to go [to the health center]. " (406, 25 years old, mother of five, SNNP) She then recounted after her delivery at the facility, --- "They [health staff ] took good care of him [her son]. They cleaned me and transferred me to another bed, and I was happy. If God gave me another child, the health center is my choice...they [the health staff ] are doing a great job. " Facilities brought feelings of safety and security for this mother, and she expected that her baby's health would be ensured. These women said delivering at a health facility afforded access to

Background: Globally, amidst increased utilization of facility-based maternal care services, there is continued need to better understand women's experience of care in places of birth. Quantitative surveys may not sufficiently characterize satisfaction with maternal healthcare (MHC) in local context, limiting their interpretation and applicability. The purpose of this study is to untangle how contextual and cultural expectations shape women's care experience and what women mean by satisfaction in two Ethiopian regions. Methods: Health center and hospital childbirth care registries were used to identify and interview 41 women who had delivered a live newborn within a six-month period. We used a semi-structured interview guide informed by the Donabedian framework to elicit women's experiences with MHC and delivery, any prior delivery experiences, and recommendations to improve MHC. We used an inductive analytical approach to compare and contrast MHC processes, experiences, and satisfaction. Results: Maternal and newborn survival and safety were central to women's descriptions of their MHC experiences. Women nearly exclusively described healthy and safe deliveries with healthy outcomes as 'satisfactory' . The texture behind this 'satisfaction' , however, was shaped by what mothers bring to their delivery experiences, creating expectations from events including past births, experiences with antenatal care, and social and community influences. Secondary to the absence of adverse outcomes, health provider's interpersonal behaviors (e.g., supportive communication and behavioral demonstrations of commitment to their births) and the facility's amenities (e.g., bathing, cleaning, water, coffee, etc) enhanced women's experiences. Finally, at the social and community levels, we found that family support and material resources may significantly buffer against negative experiences and facilitate women's overall satisfaction, even in the context of poor-quality facilities and limited resources.

, reinforcing expectations that delivery at the facility would result in a healthy infant. Multiparous mothers with a prior homebirth and a facility delivery naturally contrasted their experiences, underlining what mattered most: maternal and infant survival. A mother of five described, "I had near death experience because the umbilical cord was not cut. The baby was crying laying on the floor and the person who was supposed to cut the umbilical cord lost the blade and I was scared. I didn't want that experience again, so I decided to go [to the health center]. " (406, 25 years old, mother of five, SNNP) She then recounted after her delivery at the facility, --- "They [health staff ] took good care of him [her son]. They cleaned me and transferred me to another bed, and I was happy. If God gave me another child, the health center is my choice...they [the health staff ] are doing a great job. " Facilities brought feelings of safety and security for this mother, and she expected that her baby's health would be ensured. These women said delivering at a health facility afforded access to trained medical staff that could provide life-saving procedures if they experienced a medical emergency and provide medications to alleviate pain, as described by this mother: --- "They gave me an injection [medication to alleviate pain]. When I gave birth at my home, I was feeling sick. I was feeling stomach sickness for one week at that time. But when I gave birth at hospital, they gave me syringe, medicine, and glucose. So, it was [a] better [birth experience] here in hospital" (303, 39 years old, mother of three, Oromia) While most women highlighted the safety and reduced fear of facility deliveries, they also described comfort and reduced stress with home delivery, which included receiving family support, trusting those present, and the allowance for preferred practices (e.g., standing during labor). Women with prior homebirths liked being surrounded by family, receiving food and material comforts (e.g., clean blankets), and, importantly, privacy. Facility deliveries carried the expectation that privacy, traditional preferences (e.g., standing), and warm home comforts would not be afforded to them. This, however, did not appear to reduce their satisfaction with facility deliver because the expectations were adjusted. Antenatal care (ANC) visits at health facilities and community-based support received from the government health system through HEWs, also generated expectations for the current delivery. The interpersonal relationships defined and developed throughout prenatal care played a complex role, both creating trust and frustration. ANC visits allowed a woman to learn how to efficiently engage with the health system to optimize their care. For example, a mother of one initially received ANC care at the closest health center, but switched to the hospital, "Because doctors helped you in good manner. The doctors in the hospital treated you in good manner. But, in the health center, health workers even insult you. " (302, 25 years old, mother of one, Oromia) Experiencing disrespectful care propelled this mother to find providers that treated her better and, after feeling treated well in her ANC care, she expected kind treatment during delivery. During ANC, women also learned to bring appropriate supplies and documents (blankets, food, and health card) to improve their experience. Several women said physically carrying their health card was a necessity for both efficient care and proper treatment. Failure to have a health card may mean not receiving care at all, or, certainly, receiving exasperation, rudeness, and a longer wait. ANC, thus, allowed women to accumulate patient capital and agency -women with few ANC visits on her health card wielded less power in the delivery room, stating that they may be blamed or mistreated because they did not come for care. Women expected health staff to "do something" during ANC to feel reassured about their baby's health, referring to tangible testing procedures including ultrasounds, blood, urine, and blood pressure tests, and checking the position of the baby. A first-time mother from SNNP explained, --- "knowing that my baby is in good health, also that my health [is good made me happy]. And I believe that nothing is more important [to know] than this (404)". Women not provided testing expressed frustration and providers, "did nothing" for them. The phrase "did nothing" was common across many interviews. Women seemed to prefer and expect a tangible test above and beyond receiving education and information. Many mothers indicated that they never wanted or needed to ask the staff any questions, perhaps indicating their trust in the "tests" and staff's to detect any issues. When women were treated with kindness and respect, they were overwhelmingly appreciative, suggesting that kindness and respect were not necessarily expected, as illustrated by one woman 410, who described and compared her experiences with a male doctor and a female clinician (specific clinical title unknown), "At first it was a woman who checked me and second round it was a man, all of them knew their job very well but the man was more knowledgeable.... the female didn't ask a lot of questions but the man was a doctor, and asked me questions a lot.... he asked if things are favorable at home, like the food and if I drink a lot of water, which will be very good for myself and the baby...He also gave advice to keep stress away because the stress will hurt the baby so much.... I was comfortable even though he didn't speak my language (woletiega) and I was worried I might not understand what he is saying, but I speak and listen to Amharic, so it was nice of him to think of his patients because he wants us to understand. " (410, 28 years old, mother of three, SNNP) As with delivery, staff kindness, patience, and dignified communication could forgive infrastructural barriers and poor-quality care, like longer wait times. For example, 403, 30 years old, mother of two, SNNP, said, "there is a long line, but they examine us properly". Other women stressed that when staff were able to give them 'human dignity' (kibir) throughout ANC, they were quite happy. One first time mother, after stating her initial fear from rumors that the staff were rude, relayed, "she was treating me very nicely, she was very playful and making me laugh. She was comforting me saying there is no problem, it was very nice. (202, 30 years old, mother of 1, SNNP)" Another primiparous mother said that staff explaining treatment options for possible concerns was in itself empowering, "If for instance they say your blood is low, and I need to eat and drink to increase it. It's nice when they tell me before it can cause a problem, if your weight is low then you are told it has to be like this then you eat to increase it. When they say, this can lead to be a problem to the baby, its blood pressure might go high or low and that dangerous for the baby so you to check your self and take whatever is needed, it's really nice. It's really nice when you are told you need to do this because you need it. (110, 21 years old, mother of 1, Oromia)" ANC therefore set expectations for what is 'normal' or routine MHC care, such as infrastructural expectations (e.g., wait times), service expectations (e.g., ultrasounds and testing), and type of staff treatment (in some cases straightforward and impersonal, while in others, empowering, respectful, and educational). ANC also served as a critical time when mothers developed relationships with the health system itself prior to delivery. These 'benchmarks' of expectation and developed relationships follow mothers into the delivery, shaping their expectations for birth and, ultimately, their subsequent satisfaction. --- Discussion This study elicited birth narratives and perceptions of care from 41 women that recently delivered in government health centers or hospitals in two Ethiopian districts. For all women in the study, the seminal driver of a positive birth experience was maternal and infant survival and safety. Deliveries with healthy outcomes, especially their infants, were nearly exclusively described as'satisfactory,'when asked directly, but these women went on to describe instances of provider disrespect, neglect, and poor facility quality (e.g., no running water, overwhelming odor, etc). Although women express'satisfaction', they highlight simultaneous negative experiences (e.g., anger, mistrust, neglect) as well as empowering ones (patient communication, behavioral expressions of caring, and social support). Women described maternal and infant survival as a priority over poor experiential quality care and abuse and mistreatment. Material and social support from friends, family, neighbors, and HEWs supplemented women's experience and subsequent appraisal of their MHC, particularly when they did not receive, or expected to receive, this type of support from health facilities and its staff. We highlight that women described their experiences of their latest birth contextually with other prior events, such as ANC healthcare encounters and past home or facility-based births. This expectation benchmarking shapes their expectations for birth and, ultimately, their subsequent appraisal of their experience. Previous studies have highlighted the importance of maternal and infant survival in healthcare satisfaction, particularly in low-and middle-income settings where under-five mortality remains relatively higher [41]. Cham et al. (2009) found that maternal survival following an obstetric emergency was associated with satisfactory care, even, in some cases, when the newborn did not survive [42]. In line with this, Roder-DeWan et al. highlight that low expectations drive high care quality assessments [14]. Alternatively, a 2015 systematic review focused on LMIC found that infant survival and birth outcomes were a'minor determinant' (e.g., reported in fewer than five studies) of maternal satisfaction with healthcare. In our study, women highlighted this as a deeply important condition for satisfaction, dismissing, excusing, or ignoring disrespectful or abusive care if healthy birth outcomes were met. This underlines that when expectations of care are low, the cultural capital given to health providers and birthing facilities silences women from communicating and engaging in their own care beyond expecting the bare minimum [43]. O'Donnell et al. highlighted that women may not expect or advocate for respectful care, fearing that confronting or disagreeing with staff so may incur poor outcomes [44]. Similar findings were highlighted in Rwanda where despite improvements in quality of care and neonatal mortality, patient satisfaction was not significantly improved [11]. Beyond the explanation that O'Donnell proposed, the authors also highlight that the health systems intervention may have not had adequate focus on improving dignity and respect during patient contact. These explanations may be the case in these study sites in Ethiopia, though components of the larger intervention (fully implemented after this data collection) did focus on some specific strategies to increase provider delivery of dignified and respectful maternity care [45]. In concert with this extant literature, our results also highlight that more opportunities may be indicated for more patient education and time for information sharing during healthcare encounters to increase the relative expectations and subsequent power of maternal patients. While women in our study did not describe fear, they did describe'remaining silent' in instances of receiving poor care. Particularly in rural areas (where our data was collected), women may not be afforded opportunities to exhibit agency, and the existing culture of the health system may reinforce disempowerment and complacency with care. Beyond maternal and newborn survival, mothers noted that interpersonal communication and demonstrations of 'caring' by health and facility staff impacted their experience, though not necessarily overall satisfaction. Patience, emotional support, and communication certainly increase positive birth experiences [46], but midwives and other health staff encounter various barriers, both structural and cultural, to implement these elements into their care [47]. The positive aspects of care that mothers highlight are important contributions to the maternal care literature given the focus to date on disrespect and abuse. Other interpersonal aspects of care, such as decision making and active participation in childbirth, were not explicitly highlighted by the mothers in our study. Shakibazadeh conducted a review and found that women in LMIC were less likely to demand and prioritize these aspects of care compared to women in high-income countries [48]. Women may not be empowered enough to be actively involved in their facility-birth. One explanation may be that many women are unable to accumulate the cultural health capital valued by health staff so that they are left with sub-optimal health care relationships [43]. For example, women in our study may not have deployed the verbal competencies, interactional styles, or futureoriented decision-making approaches that providers are trained to respond to. Thus, they experienced less provider communication and fewer opportunities to attain agency in their own care. There is, however, healthcare provider consensus that provider communication and perceived experience of care are important dimensions of high-quality care delivery, and more work is needed by care workers themselves to better implement holistically respectful care [49]. Indeed, Ethiopia has targeted compassionate and respectful care as a core pillar of their Health Sector Transformation Plan, and respectful maternal care trainings are being tested and implemented nationwide with some success [45], as have other interventions in Tanzania [50] and Kenya [51], thought important challenges remain in practice [52,53]. Mothers in our study noted special quality of care when birth attendants provided them continuous accompaniment, attention, privacy, and support. Supporting the introduction of a birth companion into the labor process was a focus of the larger quality improvement intervention and national maternal quality care initiative [30,45]. These factors and improvement approaches have been highlighted in other contexts as well [54][55][56] and while respectful maternity care trainings address the patient communication domains, provider absenteeism and overwhelmed and overburdened systems may continue to make it difficult to pragmatically provide such care [57]. To date, conceptual frameworks of maternal satisfaction focus mostly on point-of-care aspects. While this is undoubtably important, our findings highlight that what women bring to their birth experience, including their previous experience with their antenatal care, previous births, and social exposure to birthing norms and experiences of their social networks all shape expectations. In line with theories related to cultural health capital, the expectations and responsibilities women place on health providers, combined with the expectations placed on them as 'compliant patients' by the medical system, privilege the actions and treatment deployed by the more powerful providers [43]. Bradley points to health providers need to maintain power and control, particularly in resource strained contexts with immense pressure to perform well on paper, further contributing to existing inequities in quality care [47]. Considerations of these domains must be incorporated in MCH quality of care frameworks in order to create environments where healthcare systems attend to histories of inequity, discrimination, and other deeply important experiences that women bring with them into the delivery room. This also extends to growing efforts to create patient centered care and derive quality metrics [18,[58][59][60]. Our results support the need to further develop these valid metrics for more routine use to continuously improve personcentered care. Our study has limitations. Our findings may not be representative given our purposive sampling of women in two regions of Ethiopia with distinct cultural and geographical contexts. We excluded women that had stillbirths or early neonatal mortality from our study and this limits our ability to reflect on satisfaction among this population. However, our findings may provide transferable insights and ideas that could inform research in other settings. Interview audio, depending on the interview language, was double and sometimes triple transcribed as we conducted data analysis in English. As such, it is possible that cultural and linguistical nuances in communication, such as idioms, tone, etc. were omitted or changed during the translation and interpretation process. Next, the personal characteristics such as class, education, maternity status, age, and cultural background between the participants and interview team may have influenced power dynamics and therefore, how some participants responded to questions. For example, it is possible some participants felt less or more comfortable expressing and sharing their experiences with the research team, particularly if health staff, HEWs, interpreters or others aided the team to find or interview women. Additionally, communities and a mother's positionality deeply shape what they may feel comfortable sharing in a formal research context. Although our team took communityinformed steps to build rapport, create comfort, and ensure confidentiality, some mothers may feel uncomfortable sharing negative experiences or expressing dissatisfaction with MHC services or specific individuals that live and/or work in their community. Future research could elicit maternal experiences with antenatal and postpartum care are warranted, as these perinatal health encounters also contribute to a woman's perception of her maternal healthcare, as well as her future decision making and influences on her broader social networks. --- Conclusion Our findings extend the literature on maternal satisfaction in healthcare by highlighting the importance of understanding past experiences and expectations, as well as the social and instrumental resources available to her during the delivery process. Our finding that newborn and maternal survival drove positive reflections on birth experiences, suggests that assessments conducted shortly following delivery may be overly influenced by these outcomes and not fully assess the overall experience. Novel findings, such as the power and agency some women accumulated through antenatal care, examples of empowering patient-provider communication, and the expectations women brought into the delivery room highlight the importance of engaging and assessing quality in maternal health care in multi-dimensional ways in order to understand the complexities of patient experience. Improved assessment of experiential quality as a key healthcare quality outcome may be more relevant that measuring overall satisfaction, particularly in contexts with limited health resources. --- Abbreviations LMIC: Low and middle-income countries; SNNPR: Southern Nations, Nationalities, and Peoples' Region; HEW: Health Extension Worker; HP: Health Post; HC: Health Center; QI: Quality improvement; RA: Research Assistant; IHI: Institute for Healthcare Improvement; ANC: Antenatal care visits; MHC: Maternal Healthcare. --- Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12884-022-04381-z. --- Additional file 1. Interview guide Authors' contributions KS, HM, CB, and ASE developed the research design. CB, HGR, AKH, and ASE developed the data collection procedures, analytic approach, and conducted the analysis. AKH and HGR drafted the initial version of the manuscript. CB, KS and HM conceptualized the larger study and contributed to manuscript writing. ASE and DWK managed the data. All authors contributed to the interpretation of the results and reviewed and provided substantial revisions to the drafts of the manuscript. The author(s) read and approved the final manuscript. --- Funding The study was supported by a grant from the Bill & Melinda Gates Foundation. AH is supported by the National Institute of Mental Health (K01MH125142). AH and KS were supported by institutional grants awarded to the Carolina Population Center at the University of North Carolina at Chapel Hill from the National Institute of Child Health and Development (T32HD007168 and P2CHD050924). --- Availability of data and materials Data is available upon reasonable request directed to the author Professor Clare Barrington (cbarr ing@ email. unc. edu). --- Declarations Ethics approval and consent to participate This research was ethically approved by the University of North Carolina at Chapel Hill's Institutional Review Board and obtained a program evaluation waiver from the Ethics Committee of Ethiopian Public Health Association. Given that written informed consent may not be understood by illiterate or partially literate participants, verbal informed consent was obtained from all study participants. A written informed consent form was available and were used as a guide to obtaining verbal informed consent. This procedure was approved by the Ethics Committee of Ethiopian Public Health Association. All methods were carried out in accordance with relevant guidelines and regulations. --- Consent for publication Not applicable. --- Competing interests We declare no competing interests. • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Globally, amidst increased utilization of facility-based maternal care services, there is continued need to better understand women's experience of care in places of birth. Quantitative surveys may not sufficiently characterize satisfaction with maternal healthcare (MHC) in local context, limiting their interpretation and applicability. The purpose of this study is to untangle how contextual and cultural expectations shape women's care experience and what women mean by satisfaction in two Ethiopian regions. Methods: Health center and hospital childbirth care registries were used to identify and interview 41 women who had delivered a live newborn within a six-month period. We used a semi-structured interview guide informed by the Donabedian framework to elicit women's experiences with MHC and delivery, any prior delivery experiences, and recommendations to improve MHC. We used an inductive analytical approach to compare and contrast MHC processes, experiences, and satisfaction. Results: Maternal and newborn survival and safety were central to women's descriptions of their MHC experiences. Women nearly exclusively described healthy and safe deliveries with healthy outcomes as 'satisfactory' . The texture behind this 'satisfaction' , however, was shaped by what mothers bring to their delivery experiences, creating expectations from events including past births, experiences with antenatal care, and social and community influences. Secondary to the absence of adverse outcomes, health provider's interpersonal behaviors (e.g., supportive communication and behavioral demonstrations of commitment to their births) and the facility's amenities (e.g., bathing, cleaning, water, coffee, etc) enhanced women's experiences. Finally, at the social and community levels, we found that family support and material resources may significantly buffer against negative experiences and facilitate women's overall satisfaction, even in the context of poor-quality facilities and limited resources.

Introduction Mass vaccination represents a crucial strategy in preventing and mitigating the spread of infectious diseases through the establishment of herd immunity (John and Samuel 2000), protecting even unvaccinated individuals. Despite its demonstrated benefits, vaccine hesitancy remains a persistent issue, compounded by challenges posed by the ongoing COVID-19 pandemic (WHO 2019;Hudson and Montelpare 2021;Cascini et al. 2021;Coustasse et al. 2021;de Miguel-Arribas et al. 2022;Jentsch et al. 2021). Heightened concerns regarding vaccine safety and efficacy have exacerbated the long-standing challenge of voluntary vaccination (Larson 2013;Determann et al. 2014;Kennedy 2020). Of particular interest, the study of how social networks influence public health behavior and vaccine choices is a significant area of research (Christakis and Fowler 2013). The dynamic spread of changing opinions on vaccination and its impact on disease outbreaks through social networks constitutes a "dueling contagion" process (Fu et al. 2017). It is imperative to recognize that social contagions can also perpetuate non-socially optimal health behaviors. For instance, the diffusion of vaccine scares among parents through social networks has led to a decline in infant vaccination rates, resulting in a surge in childhood diseases (Jansen et al. 2003). Both pro-and anti-vaccine opinions are not only transmitted through personal interactions (Shaham et al. 2020) but also through digital social media platforms (Salathé and Khandelwal 2011;Nayar et al. 2019). A comprehensive understanding of these spreading mechanisms (see, for example, a comprehensive review in Ref. Pastor-Satorras et al. (2015)) is vital to leverage the positive impacts of social contagion and mitigate its adverse effects in public health efforts (Campbell and Salathé 2013;Salathé and Bonhoeffer 2008). In recent years, the role of social factors in epidemiology has attracted growing attention (Bauch and Galvani 2013;Nayar et al. 2019). Researchers have used behavior-disease interaction models to investigate the impact of various factors, such as vaccine scares or heightened disease awareness, on vaccine compliance (Bauch 2005;Fu et al. 2011;Cardillo et al. 2013;Zhang et al. 2012;Bhattacharyya et al. 2019). We refer to Refs. Bhattacharyya and Bauch (2012), Wang et al. (2016), Bedson et al. (2021) for systematic reviews in this regard. Going beyond interacting diffusion of the same sort (Gomez et al. 2013;Nguyen et al. 2018;Chang and Fu 2019), prior studies have considered the spread of disease, health behavior, and/or information can interact in a close feedback manner, either through the same network (single-layered) (Funk et al. 2009) or through multilayer networks (Granell et al. 2013;Wang et al. 2014;Liu et al. 2016;Pan et al. 2019;Kahana and Yamin 2021). In the latter, spread of infectious disease on one layer interacts with the diffusion of health behavior on the other (Mao and Yang 2012), or coupled with a third layer of information diffusion (Mao 2014). Motivated by the empirical work using social network targeting for improving public health interventions (Kim et al. 2015), here we propose and compare a number of network-based targeting algorithms that aim to maximize the influence of initial seeds who are early supporters of health intervention (e.g., vaccination). This problem is also known as influence/diffusion maximization: how to select the seed users so that the total number of triggered adopters can be maximized. It is common practice to identify and target influential individuals based on various centrality measures (Masuda 2009;Zhao et al. 2014;Gupta et al. 2016) in both real-life networks and social media (Morone and Makse 2015). Despite that such targeting concept is similar to identify superspreaders and targeted vaccination (Pastor-Satorras and Vespignani 2002;Cohen et al. 2003) in epidemiology, much is unknown about diffusion maximization on multiplex networks. In this work, we fill this theoretical gap and study a variety of targeting algorithms in a multiplex setting and assess their effectiveness of mitigating an ongoing epidemic. Specifically, we consider a two-layer multiplex, one with a social contagion network of opinion formation and the other with a spatial network of disease transmission. The novelty of our approach is twofold. First, we consider heuristic context-dependent targeting methods explicitly using multiplex networks by coupling the dynamics of social influence and behavior changes with an ongoing epidemic. Second, aside from such interplay of concurrent disease and opinion propagation, we also consider dueling contagions of opposing views on the influence network (not just the pro-vaccine opinion but also the anti-vaccine together with neutral ones) and study how they impact targeting effectiveness. In so doing, our work represents a step forward in the field of network-based interventions, by improving our understanding of how to optimize targeting interventions with potential applications across a wide range of domains. Our present work extends beyond our previous preliminary study (Fügenschuh and Fu 2023) with two key contributions. First, when measuring targeting effectiveness, we not only consider the final size of the epidemic (how the disease spreading is suppressed overall), but also the peak of the epidemic (how the targeting intervention flattens the epidemic curve). Second, we test the robustness of our targeting algorithms against variations of modeling parameters including initial conditions, network size, and density. Our results help improve our understanding of how to optimize targeting interventions by making them more practical and scalable in the real world. --- Model and methods --- Model overview The multiplex we study consists of two network layers, each containing the same set of individuals but with different connections between them. The first layer represents an individual's typical environment, including in-person interactions with family, coworkers, friends, and others during daily activities (including leisure time or hobbies). We assume a degree-regular network of these contacts among all individuals, and represent these relationships using a square lattice graph. On this layer, we simulate the spread of a disease. The second layer represents a virtual social network and provides a platform for opinion exchange (Mastroeni et al. 2019). The connections between individuals on this layer are degree-heterogeneous, generated by the Barabási-Albert network model (Barabási and Albert 1999). Such network structure allows the position of an individual to differently influence the dynamics of opinion dissemination. Past research has shown that single-layered scale-free networks promote both vaccination behavior and effective immunization (Cardillo et al. 2013). In this work we explicitly consider a dueling contagion process (Fu et al. 2017) on two-layered networks as follows. The simulation alternates probabilistically between disease spread on one layer and opinion formation (and vaccine uptake) on the other. In each iteration, an individual is chosen at random to update their current states. The next step is to determine which layer the process should continue on. As the dynamics of the processes on the two layers are different, the probability of choosing a layer need not be equal. To this end, we introduce the parameter <unk> to control the relative time scales of biological and social contagions: with probability <unk>, a state update of that individual takes place on the disease transmission layer, and otherwise with probability 1 -<unk>, an opinion state update occurs on the other network layer (see Algorithm 1). A full overview of all simulation parameters can be found in Table 1. Once a network layer is selected for update, the corresponding process is continued based on the individual's state in that layer (see Fig. 1), as described in further detail below. --- The opinion contagion layer and network targeting algorithms Our objective is to study networks targeting methods that harness the social contagion of pro-vaccine opinions, ultimately increasing the willingness of the population to prevent an epidemic through vaccinations. Therefore, we simulate an exchange of views on the subject of vaccines. Based on the way the individuals represent their opinions, we consider a voter model with a single discrete variable with more-in our case three-states, according to the classification given in Ref. Sîrbu et al. (2017), Jedrzejewski and Sznajd-Weron (2019). In our setting, an individual can have one of the three discrete opinions: anti-, neutral or pro-vaccine, indicated by the values -1, 0 and 1 respectively. The number of supporters and opponents at the start is given by the parameters <unk> + and <unk> -respectively. To assess the impact of the supporters' exact positions in the network on the opinion formation process and consequentially on the epidemic, we consider the following network-based targeting methods for the initial placement of pro-vaccine opinions. 1. Hubs: The supporters are chosen from the vertices with the highest degree, referred to as hubs. 2. Betweenness: The probability of a vertex being chosen as a supporter is based on its betweenness centrality. The higher the betweenness value, the greater the likelihood of being selected. 3. High-degree: Pro-vaccine opinions are assigned according to the degree distribution of the vertices. Again, the higher the degree, the higher the probability of being selected. 4. Random: Vertices are randomly selected from the entire population with a uniform probability. 5. Hub-community: A heuristic approach is used, starting by filling the supporter set with a vertex of the highest degree and its neighbors. Then, new vertices are repeatedly added to the supporter set by choosing neighbors of supporters with the highest degree in the opinion layer as long as they have not been chosen yet and there are places available to do so. 6. Low-degree: The least connected vertices have a higher probability of being chosen as vaccination supporters. Pro-vaccine opinions are assigned based on the degree distribution of the vertices, with the lower the degree, the higher the probability of being selected. 7. Hub-neighbors: The individuals selected are those adjacent to hubs, but not themselves hubs. 8. Mod-community: Supporters are members of communities determined using the Clauset-Newman-Moore greedy modularity maximization algorithm (Clauset et al. 2004), designed to find communities in scale-free networks, among others. Communities are selected in the order of their size until the desired number of supporters is reached. 9. Coworkers: Pro-vaccine opinions are assigned to vertices that form a connected rectangular-shaped sub-grid on the disease layer. After the initial pro-vaccine opinions are assigned, the individuals with anti-vaccine views are randomly selected from the remaining nodes. The rest of the individuals are set to neutral. We use the parameter O s to designate the targeting method used in the simulation. Figure 2 demonstrates how the placement of vaccine supporters varies with the different targeting methods. The network, which remains constant for all methods, is comprised of vertices arranged on a 25 <unk> 25 lattice for the disease layer, while the opinion layer is generated using the Barabási-Albert preferential attachment model (Barabási and Albert 1999) (in which each new node is connected to existing nodes with one edge, m = 1 ). Both pro-and anti-vaccine opinions are present at a proportion of 10% of the total population. The sizes of vertices are scaled based on their degree and the edges connecting the vaccine supporters are displayed to give an indication of the strength of connections within the assigned pro-vaccine group. Note that all other edges have been omitted for clarity. Opinion updating rule After the initial views are assigned, the opinion formation process unfolds (Castellano et al. 2009). The opinion of a selected individual is updated using one of the following five updating rules. The focal individual adopts the opinion 1. Of a randomly chosen neighbor (random) Sood and Redner (2005)) or 2. Of the majority of the neighbors (max) Fu and Wang (2008) or 3. Indicated by the sign of the sum of the opinions of the neighbors (sum) or 4. The closest to the average of the neighboring opinions (mean) Kozitsin (2022). We also implement a hybrid method of opinion adoption that incorporates all above updating rules by randomly choosing one at each iteration step. We use the parameter O a to specify the opinion updating rule used in the simulation. In addition, it is often difficult for individuals to accept opposing opinions (Deffuant et al. 2002), which is why we use the parameter <unk> to represent the probability of ulti- mately accepting the calculated opposing opinion. This assumption of an inertia effect when revisiting contrary opinions aligns with prior experimental observations (Pitz 1969;Traulsen et al. 2010) and modeling studies (Stark et al. 2008;Zhang et al. 2011;Wang et al. 2020). --- The disease transmission layer The epidemic spreading process we consider on the disease layer is based on the Susceptible-Infected-Recovered (SIR) model, with an extension to include the state of immunization (Hens et al. 2012;Newman 2002). Individuals who opt to receive the vaccine, in particular the initial supporters in the opinion layer, once vaccinated, will remain in immunized state throughout the epidemic simulation. Infection seeds are prescribed by a predefined number of individuals in the population-given by the parameter <unk>-who, of course, are not vaccinated. Once the simulation is started and if it is the turn of the disease layer to update, the state of the chosen individual is updated according to one of the following rules. (At each iteration, an individual can only record at most a single state change at a given time.) 1. A susceptible pro-vaccine supporter is vaccinated without hesitancy, whereas a susceptible individual with neutral attitude will get vaccinated with a predefined very low probability <unk>. In contrast, a susceptible anti-vaccine individual will never choose to be vaccinated. 2. If not vaccinated, a susceptible can get infected, provided that there are active infections in the immediate neighborhood. As a matter of fact, the more infected neighbors the higher risk of infection (Pastor-Satorras et al. 2015). Thus, the parameter <unk> that represents the base probability of getting infected, increases by the percentage of the infections in the neighborhood: <unk> 1 + I u N u, where N u is the number of all neigh- bors of the focal susceptible u and I u is the number of infectious neighbors. 3. An infected recovers with a predefined probability <unk>. 4. Once vaccinated or recovered, individuals remain in their assumed state until the end of the simulation. In our simulations, we focus on exploring how the initial pro-vaccine opinion individuals should be placed on the opinion layer in order to achieve the most efficient diffusion of vaccine support views in terms of suppressing the epidemic spreading on the disease layer. Thus we mainly concentrate on varying the parameters that govern social contagion on the opinion layer, including the initial number of individuals holding proand anti-vaccine opinions as well as targeting methods for the pro-vaccine supporter placement and specific assumptions of individuals' view adoption. Simulation results, averaged over 100 realizations, are shown as boxplots. Unless noted otherwise, we use default model parameter values given in Table 1. --- Results To begin with, we show the results of a typical simulation run initialized with the default values as given Table 1, which gives us an intuitive understanding of how the contagion process on the opinion layer can be harnessed to control disease spread (Fig. 3). On the left panels of Fig. 3, we show the simultaneous progression of both social and biological contagions-on the opinion layer (upper left) and on the disease layer (lower left). For comparison, the panels on the right display the epidemic spread with the same initial model parameters as in the left panels but without the influence of the opinion layer: The individuals are either vaccinated only in the initial phase (upper right) or not at all (lower right). To assess the impact of the epidemic and compare various scenarios, we keep track of the number of individuals who contract the illness during each simulation run. Based on the three scenarios of disease transmission provided, we see that when the spread of the disease is suppressed by increasing vaccinations based on opinion Fig. 3 Impact of dueling contagions. Shown is an example of an opinion formation process (upper left) which is concurrent with a disease spread course (lower left). Such multiplex disease-behavior interactions can lead to the outgrowth of vaccination supporters and thus the suppression of the disease transmission, as compared to scenarios having a constant rate of vaccination (upper right) or absent of any immunized individuals (lower right). All other parameters are set to be default values as given in Table 1 formation, 54% of the population becomes infected. On the other hand, 87% of the population becomes infected when the number of immunized individuals remains constant throughout (no social contagion of vaccination), and 99% if there is no vaccination at all. In the latter scenario, the epidemic lasts significantly longer, with a duration of 60,000 iterations compared to 45,000 iterations when vaccinations are administered. In evaluating the effectiveness of targeted interventions, we not only consider the final size of the epidemic (Fügenschuh and Fu 2023), but also the peak as a measure of disease mitigation effort. In the following, we present a comprehensive analysis of the effectiveness of targeting algorithms, and we validate the robustness of these findings through simulations that take into account varying model parameters such as initial conditions, network size, and density. --- Comparison of different network targeting algorithms along with opinion updating rules Here we investigate and compare the extent to which the spread of the disease can be impacted jointly by the initial placement of pro-vaccine supporters ( O s ) and the spe- cific opinion adoption method ( O a ) applied to the opinion layer. To this end, we vary the parameter combinations governing O s and O a and set all others to defaults, as given in Table 1. The simulation results are presented in Fig. 4. The upper plot displays boxplots of the peak of the epidemic, indicating the maximum percentage of infected individuals during all iterations. In the lower plot, the boxplots show the final epidemic size, i.e. the total percentage of individuals ever infected across all iterations. Corresponding to each initial placement method of vaccine supporters (as ordered on the x-axis), each of the five opinion adoption methods that are grouped together is distinguished by different colors. The evaluation of the impact of different opinion adoption methods on the spread of pro-vaccine support demonstrates that both the initial placement of vaccine supporters on the opinion layer and the chosen opinion adoption method have a crucial impact on the effectiveness of mitigating the epidemic. It is clear that the vertices with the highest network centrality values are the most influential in this social contagion process, and that the potential number of nodes that a pro-vaccine supporter can influence is a critical factor. However, when the best-connected nodes are part of a like-minded community, as seen in the hub-community and mod-community cases, their performance is similar to or even worse than the completely random approach. The density of connections between the initially placed vaccine supporters does not appear to significantly affect the spread of pro-vaccine opinions. In the case of coworkers, it is apparent that seeding a pro-vaccine opinion within a physically connected group, such as coworkers, does not result in an effective spread of pro-vaccine views. Our simulations also revealed that the scenario where low-degree vertices acting as hub neighbors could spread provaccine opinions to their hub neighbors and trigger a ripple effect through the entire network is unlikely. The assessment of the opinion adoption methods O a reveals a consistent ordering of corresponding results (the colored boxplots in Fig. 4) per initial placement method for pro-vaccine supports, indicated by the x-axis. Among all five different opinion updating rules considered, the one yielding the most impact on disease mitigation appears to be taking the sign of the sum of neighboring opinions (indicated by the purple color Fig. 4). This method is followed by the random approach, which exhibits a noticeable gap in performance. The methods max and mean trail closely behind and perform almost equally with a slight disadvantage compared to the other majority approach. The exceptional case of mod-community-based targeting highlights the limitation in exchanging or spreading opinions within a closed like-minded community, regardless of the specific opinion adoption method used. This is due to the restricted flow of information, which results in saturation when this method of targeting is applied. --- Impact of varying initial conditions of opinion formation The outcome of an epidemic is shaped by the combination of the initial number of individuals holding pro-vaccine and anti-vaccine opinions at the outbreak. Specifically, we quantify the effect of these initial conditions on the infection curve by altering the parameters <unk> + and <unk> -, which denote the percentage of individuals holding pro-and anti- vaccine opinions, respectively, at the beginning of each simulation. The top and bottom Fig. 4 Comparison of effectiveness of different targeting algorithms. Shown are the peak of infections (the maximum percentage of infections over all iterations of the epidemic spreading process) (top) and the final epidemic size (the total percentage of individuals ever infected (bottom), grouped for each targeting method of initial placement of pro-vaccine supporters (as indicated on the x-axis ) with respect to different opinion updating rules (corresponding to the color map in the legend). Each boxplot is based on 100 independent simulations panels of Fig. 5 depict the results of simulations with different targeting methods for the pro-vaccine opinion messengers, with O s set to high-degree in the top panel and random in the bottom panel. All parameters other than <unk> + and <unk> -are fixed to their default values as specified in Table 1, and each simulation run starts with the generation of a fresh pair of networks for both layers. In Fig. 5, each boxplot corresponds to a combination of initial conditions <unk> + and <unk> - with values taking from the set <unk>5, 20, 15, 20, 25<unk>%, as indicated on the x-axis. On the y-axis, we have the total percentage of infected individuals (final epidemic size). As a Fig. 5 Robustness of network-based targeting. Shown is the final epidemic size (the percentage of individuals ever infected) for varying initial proportions of individuals holding pro-vaccine <unk> + and anti-vaccine <unk> -, with the targeting method O s set to high-degree in the top panel and random in the bottom. The insets show the case of <unk> + = <unk> -. Degree-based targeting provides more effective interventions than random placement. Each boxplot is based on 100 independent simulations comparison, we also show the base case where every individual has neutral attitude towards vaccine, that is, <unk> + = <unk> -= 0. The insets of Fig. 5 show how the final epidemic size decreases with an increasing equal presence of pro-and anti-vaccine individuals, including values of <unk> + = <unk> -beyond 25%. As expected, for a fixed proportion, <unk> +, of individuals seeded as pro-vaccine support- ers, increases in anti-vaccine opinion individuals <unk> -can cause greater final epidemic sizes, suggesting an approximately linear relationship with positive slopes. However, a larger initial presence of pro-vaccine individuals <unk> + leads to a greater extent of disease mitigation that strongly counteracts the negative impact due to the increase in <unk> -. For example, for <unk> + = 20%, the increase in total infections for bigger <unk> -is limited, exhibit- ing a much smaller slope compared to the scenario <unk> + = 5%. Regarding the symmetric initial conditions <unk> + = <unk> -, the final epidemic size decreases with increasing <unk> + = <unk> -; in particular, as <unk> + = <unk> -increases beyond 30%, the spread of disease will be significantly suppressed. At <unk> + = <unk> -= 50% we see that essentially little or no disease outbreaks can unfold. These results hold for both targeting methods: high-degree-based and random. The former is much more effective than the latter, which is inline with Fig. 4. Overall, we confirm similar impact of varying initial conditions for other targeting methods O s. --- Impact of varying couplings of disease transmission and opinion contagion The time scale parameter <unk> controls the relative time scales governing the dynam- ics of disease transmission and opinion contagion on the two layers. The strength of interlayer coupling determined by <unk> may impact the effectiveness of targeting meth- ods. Intuitively, greater <unk> values render more frequent updating on the disease layer, thereby representing less favorable conditions for network targeting. We thus evaluate the robustness of our three representative targeting methods O s, including the most effective "high-degree", the moderate one "random" and the least effective "coworkers", by adjusting <unk> values beyond its default value, which is <unk> = 0.75, as used in our simulations. Figure 6 shows that increases in <unk> compromise the effectiveness of targeting, resulting in larger sizes of infections. These results confirm that a larger <unk> corresponds to a more challenging condition for interventions. We also see that "high-degree"-based targeting is robust to increasing <unk> as the increase in disease inci- dences is slow, exhibiting a concave upward curvature. In contrast, the other two Fig. 6 Comparing the effectiveness of selected O s -methods by using the total percentage of infections while varying the time scale parameter <unk>. Each boxplot is based on 100 simulations. Other model parameters as in Table 1 targeting methods are more sensitive to the increase in <unk>, showing a concave down- ward curvature. For instance, the least effective "coworkers"-based targeting shows a much sharper rise than others, while the most effective "high-degree"-based targeting starts slowly and steepens at higher <unk> values, but still is lower than the other two even at <unk> = 90%. In addition, we vary the disease transmission rate <unk> and assess how targeting methods may depend on <unk>. For a larger <unk>, the disease is more contagious, making it more diffi- cult to contain. Overall, the effectiveness of the three aforementioned targeting methods shows qualitatively the same pattern in their changes while having their order of effectiveness attained: for each O s, the total number of infections increases with <unk> and is satu- rated at higher <unk> values (Fig. 7). Regarding the opinion layer, we also vary the parameter <unk>-namely, the probability of adopting a contrary opinion. Accepting the opposite view works both ways; that is, switching from pro-vaccine to anti-vaccine and vice versa is equally likely, both with probability <unk>. Therefore, as demonstrated in Fig. 8 specifically for the "hubs"-based targeting, the parameter <unk> does not significantly influence the effec- tiveness of targeting methods. Neither does the effect of opinion adoption methods O a on containing the disease depend much on <unk>. For example, we see that "random"-based opinion adoption remains in between the other two opinion updating rules concerning Fig. 7 Comparing the effectiveness of selected O s -methods by using the total percentage of infections while varying the disease transmission rate <unk>. Each boxplot is based on 100 simulations. Other model parameters as in Table 1 Fig. 8 Comparing the effectiveness of the "hubs"-based targeting method for selected opinion updating rules O a (as indicated in the legend), using the total percentage of infections while varying the probability <unk> of adopting a contrary opinion. Each boxplot is based on 100 simulations. Other model parameters as in Table 1 their ultimate impact on disease mitigation. For each O a, we observe a slight decrease in the total number of infections as <unk> increases. --- Impact of network size The previous simulations have been conducted with a fixed network size of 2500 individuals. Figure 9 provides insight into the sensitivity of our simulation to larger network sizes. Beginning with the default 50 <unk> 50 for the bottom layer of square lattice and top layer of Barabàsi-Albert network, the remaining ticks on the x-axis represent network sizes of 10000, 22500, 40000, and 62500 nodes, respectively. To compare the impact of initial supporter placement methods (denoted as O s ) in these test scenarios, we focus on three targeting algorithms O s, each of which corresponds to high-degree, random, and coworkers, as indicated by the legend in Fig. 9. Each boxplot summarizes the results of 10 independent simulations. We observe an intriguing finite size effect: as the number of nodes in the layers increases, the final epidemic size decreases with smaller fluctuations, as shown in the left plot of Fig. 9. The relative position between the boxplots of different colors, representing different targeting methods, remains unchanged: high-degree-based targeting remains most effective in mitigating the disease impact, followed by random and then coworkers. This result further underscores the robustness of the simulation with regard to network size in comparing effectiveness of targeting methods. Overall, we confirm that the relative positions of the boxplots obtained for different targeting methods O s as well as initial conditions of (<unk> -, <unk> + )-pairs remain qualitatively the same as in Fig. 4 and 5, respectively. It is important to consider computing times when working with larger network sizes. The right plot in Fig. 9 shows that the simulation times, which were conducted on a 3.2 GHz 16-Core Intel Xeon W with Turbo Boost up to 4.4 GHz and 768 GB RAM, increase steadily, approximately quadratically with the number of nodes. Starting with approximately one minute for a 50 <unk> 50 square lattice/Barabàsi-Albert network, the compu- tation time can reach over three hours for a 250 <unk> 250 network. We also find that the computation time is less sensitive to the specific targeting method O s. Fig. 9 Quantifying network size effect. Shown are the epidemic size (the percentage of individuals ever infected) (left panel) and the running times of the simulations (right panel) as a function of the total number of nodes in the network layers. Each boxplot is based on 100 independent simulations Furthermore, we perform a comprehensive comparison of simulation times with respect to specific targeting methods O s and opinion update methods O a using the same default network size (Fig. 10). Except for betweenness-based targeting, the computation times roughly track the trend of the final epidemic size (total infections) presented in Fig. 4. Generally and intuitively, the fewer infections there will be, the sooner the simulation will end. --- Impact of average degree of the opinion layer Lastly, we investigate the robustness of our simulation results with respect to varying the average degree of the opinion network layer. Keeping all other parameters fixed, we vary the number of edges added from a new node to existing ones (as indicated by the parameter m in Table 1) from m = 1 (average degree k = 2, density d = 0.0008 ) to m = 4 ( k = 8, density d = 0.0032 ) when generating the Barabási-Albert network as the opinion formation layer in our model. As shown in Fig. 11, the effectiveness of disease mitigation by targeting high-degree, random, and coworkers nodes as initial pro-vaccine supporters remains largely unchanged. We also see that increasing the density of the opinion layer (via the parameter m) appears to yield noticeably yet insignificant lesser final epidemic size. Taken together, the density of the opinion layer has little, if any, impact on the social contagion process, whereas it is primarily the very structure of peer influence network and the network-based targeting methods that jointly play an important role in shaping the opinion propagation process. --- Discussions and conclusion In this work we investigate how various centrality-based targeting algorithms seeding the opinion influence network can impact the social contagion of pro-vaccine support as a means to control epidemic spreading on the other disease layer. For simplicity, the disease layer uses regular square lattices to capture the spatial transmission dynamics of local community neighborhoods. However, the use of degree-regular lattices leads to a lack of network heterogeneity, which can impact the threshold for disease outbreaks. On the other hand, by varying the transmission rate <unk> in our extended simulations on spatial lattices, the resultant effective basic reproductive ratio can approach values expected in degree-heterogenous networks (cf. the bottom right panel in Fig. 3 which shows a total 99% infection). In this context, our findings offer preliminary insights for studying more nuanced scenarios in multilayer networks (Kivelä et al. 2014). While our model does not explicitly account for interlayer dependence during network generation, our simulations and targeting algorithms, such as those focusing on hub-community and co-workers ones, do incorporate certain aspects of interlayer correlations to some extent (Kivelä et al. 2014). This is qualitatively evidenced by the varying levels of connectivity of selected targeting seeds and their corresponding spatial distributions for each targeting algorithm, as shown in Fig. 2. Nonetheless, future work is needed to quantify how interlayer network characteristics, for example, selection of initial seeds based on a combined centrality measure from both layers, can enhance the efficiency of multiplex targeting strategies. While we consider a range of widely used opinion updating rules (Sobkowicz 2009), it is helpful for future work to incorporate more realistic contagion models (Dodds and Watts 2004;Campbell and Salathé 2013) and high-order interactions (Iacopini et al. 2019;Barrat et al. 2022), such as hypergraphs (Barrat et al. 2022). The targeting algorithms also need to account for other factors, such as homophily ('birds of a feather to flock together') (Centola 2011), and the presence of central top-down campaign influence (Wang et al. 2020) apart from peer influence. By incorporating these extensions, we can enhance public confidence in vaccines at a time when coverage from child vaccinations to COVID-19 vaccinations may be declining or stagnant. Our model primarily focuses on an ongoing epidemic where healthy and susceptible individuals exchange opinions on vaccine uptake when they are not infected. It is thus important to note that vaccine uptake is only considered for individuals who have not contracted the disease. Once individuals are infected, vaccination is no longer a consideration. However, a meaningful extension of our current model is to incorporate the reciprocal relationship between infection status and vaccine opinion. Specifically, it is not only peer influence that matters, but also an individual's own experience with the disease, and particularly how their health outcomes, impacted by contracting the disease, shape their opinion about the necessity of vaccinations. Previous research suggests that a higher perceived cost of infection leads to

Incorporating social factors into disease prevention and control efforts is an important undertaking of behavioral epidemiology. The interplay between disease transmission and human health behaviors, such as vaccine uptake, results in complex dynamics of biological and social contagions. Maximizing intervention adoptions via networkbased targeting algorithms by harnessing the power of social contagion for behavior and attitude changes largely remains a challenge. Here we address this issue by considering a multiplex network setting. Individuals are situated on two layers of networks: the disease transmission network layer and the peer influence network layer. The disease spreads through direct close contacts while vaccine views and uptake behaviors spread interpersonally within a potentially virtual network. The results of our comprehensive simulations show that network-based targeting with pro-vaccine supporters as initial seeds significantly influences vaccine adoption rates and reduces the extent of an epidemic outbreak. Network targeting interventions are much more effective by selecting individuals with a central position in the opinion network as compared to those grouped in a community or connected professionally. Our findings provide insight into network-based interventions to increase vaccine confidence and demand during an ongoing epidemic.

al. 2022). The targeting algorithms also need to account for other factors, such as homophily ('birds of a feather to flock together') (Centola 2011), and the presence of central top-down campaign influence (Wang et al. 2020) apart from peer influence. By incorporating these extensions, we can enhance public confidence in vaccines at a time when coverage from child vaccinations to COVID-19 vaccinations may be declining or stagnant. Our model primarily focuses on an ongoing epidemic where healthy and susceptible individuals exchange opinions on vaccine uptake when they are not infected. It is thus important to note that vaccine uptake is only considered for individuals who have not contracted the disease. Once individuals are infected, vaccination is no longer a consideration. However, a meaningful extension of our current model is to incorporate the reciprocal relationship between infection status and vaccine opinion. Specifically, it is not only peer influence that matters, but also an individual's own experience with the disease, and particularly how their health outcomes, impacted by contracting the disease, shape their opinion about the necessity of vaccinations. Previous research suggests that a higher perceived cost of infection leads to a greater tendency to vaccinate (Fu et al. 2011). Thus, incorporating such experiential learning feedback, beyond just word-of-mouth, into the disease and behavior interactions could provide a more comprehensive understanding of the dueling contagion dynamics. The present work assumes a challenging scenario where the exchange and acceptance of views is much slower than disease spreading (using the time scale parameter <unk> = 0.75 ). To verify the robustness of our findings, we conducted a comprehensive sensitivity analysis on key model parameters. These include the disease transmission rates <unk>, the probability <unk> of adopting a contradictory opinion, and the impact of varying the time scale parameter <unk>. The parameter <unk> essentially controls the extent of interdependence (or coupling) of disease transmission and opinion contagion, thereby impacting the effectiveness network-based targeting algorithms (see Fig. 6). It is worth noting that a prompt response from the population (smaller <unk> valuers) would create favorable conditions for mitigating the spread of disease (Fu et al. 2017). Moreover, it is promising for future work to consider the impact of external shocks in the form of vaccine scares or skepticism, in order to overcome the hysteresis effect previously discovered in Ref. Chen and Fu (2019). The multiplex network targeting algorithms explored in this study can be further refined to identify individuals that are not only susceptible, but also responsive to, interventions. Keeping these in mind when optimizing network-based targeting methods, we will be able to harness the social contagion of vaccine knowledge and positive attitudes towards vaccination, with the goal of overcoming the hysteresis effect and increasing vaccination rates in areas of need. In conclusion, our results demonstrate that network-based targeting algorithms seeding the opinion layer with pro-vaccine supporters can greatly enhance attitude and behavior changes that are needed to control the spread of disease. Among those considered, targeting hubs-individuals with the highest degrees in the influence network-yields the most effective intervention that is not only able to flatten the curve with the smallest peak of infections but also able to curb the outbreak of the disease with the least number of total infections. The betweenness-based approach is similarly effective, but slightly lags behind the hub-based method. On the contrary, targeting groups of closely connected individuals either in the influence network (mod-community) or in the physical network (co-workers) leads to the least effective intervention due to the saturation effect, which is similarly observed in the field experiment (Kim et al. 2015). Therefore, our study offers simulation-based insights to enhance the targeting effectiveness for future experiments. --- Author contributions MF and FF conceived the project, performed research, and wrote the manuscript. MF performed simulations and prepared figures. All authors read and approved the final manuscript. --- Availability of data and materials All simulation data has been included in the main text. Source code to reproduce the simulation results is available from the first author on reasonable request. --- Declarations Ethics approval and consent to participate Not applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Incorporating social factors into disease prevention and control efforts is an important undertaking of behavioral epidemiology. The interplay between disease transmission and human health behaviors, such as vaccine uptake, results in complex dynamics of biological and social contagions. Maximizing intervention adoptions via networkbased targeting algorithms by harnessing the power of social contagion for behavior and attitude changes largely remains a challenge. Here we address this issue by considering a multiplex network setting. Individuals are situated on two layers of networks: the disease transmission network layer and the peer influence network layer. The disease spreads through direct close contacts while vaccine views and uptake behaviors spread interpersonally within a potentially virtual network. The results of our comprehensive simulations show that network-based targeting with pro-vaccine supporters as initial seeds significantly influences vaccine adoption rates and reduces the extent of an epidemic outbreak. Network targeting interventions are much more effective by selecting individuals with a central position in the opinion network as compared to those grouped in a community or connected professionally. Our findings provide insight into network-based interventions to increase vaccine confidence and demand during an ongoing epidemic.

Background The coronavirus disease 2019 (COVID-19) first broke out in Wuhan, China, in December 2019, which has since affected more than 200 countries. According to the data of the World Health Organization (WHO), 75 million cases and 1.6 million deaths due to COVID-19 have been confirmed worldwide as of December 2020 [1]. Social isolation has been the cornerstone of effective public response to this health crisis [2] because this method can prevent transmission to close contacts and other people. By late January, several cities in China have begun to implement stringent social isolation policies. Then, by the middle of February 2020, such policies have restricted the movement of more than 500 million persons across 80 Chinese cities [3]. Although evidence that mass quarantine is an effective preventive measure has been provided, scholars have called for caution in implementing social isolation due to its adverse effects on mental health [4]. Such mental health problems include, but not limited to, anxiety disorder and depression. These problems occur mainly because of fear of infection, distress and boredom, reduced social and physical contact, and disruption of normal life during social isolation [5]. Previous research showed that the negative effects of social isolation may continue up to 3 years after the isolation period, which is similar to the influence of post-traumatic stress disorder [6]. Among the various psychological problems, depression is the most common mental health issue related to COVID-19 [7]. Several factors could influence depression among individuals subjected to mandatory isolation. First, perceived threat and risk of public health emergencies, including perceived susceptibility and severity, may influence depression levels among such individuals [8]. Perceived susceptibility refers to the likelihood that people feel that they will be infected, whereas perceived severity indicates that people consider the consequences of infection as fatal [9]. High levels of perceived risk may be associated with increased depressive symptoms [10]. Second, the media can provide many channels for spreading news and information about the COVID-19 pandemic and forming one's understanding of such information. However, differences of stances and opinions across these media channels may exist, and the tone of media reports may range from positive and encouraging (i.e., medical staff exerting heroic efforts on the front line in the fight against the COVID-19 pandemic) to negative and desperate (i.e., severe shortage of personal protective equipment for healthcare workers). People subjected to mandatory social isolation who are affected by negative media reports are more likely to suffer from mental health problems [11]. The existing research rarely mentions the impact of public health services on the mental health of people subjected to mandatory isolation. During the COVID-19 pandemic, China was the first country to implement social distancing policies, which were supported by multiple public health services. Community-based public health service providers are responsible for providing daily support to individuals under mandatory isolation. The main public health services include delivery of daily meals, provision of essentials, monitoring of body temperature, and provision of primary guidelines for seeking medical help as necessary [12]. Frontline public health workers providing these services were deeply involved in the lives of people under mandatory social isolation and thus may play a vital role in their mental health. The provision of high-quality people-oriented public health services may alleviate the depressive symptoms of people under mandatory social isolation, whereas low-quality community-based public health services may exacerbate their depressive symptoms. In addition, the quality of public health services procured may moderate the associations between perceived risk and perceived tone of media coverage and depressive symptoms among individuals under mandatory social isolation. In line with the theory of people-centered public health services, public health workers are encouraged to show empathy, respect, engagement, individualized focus, and coordinated care to enable people to live a meaningful life [13]. Calling for humanism in medicine and holistic healing has been a longstanding principle [14]. This principle aims to place human beings at the center, express sympathy to those suffering, and promote better an understanding and experience of medical services. In this manner, people-oriented public health services could play salient roles in improving mental health among people subjected to mandatory social isolation. For individuals with high-risk perception, highquality people-oriented public health services may provide a sense of comfort and security, which are associated with less depressive symptoms. By contrast, lowquality public health services may aggravate negative feelings and are associated with increased depressive symptoms. Similarly, high-quality people-oriented public health services could alleviate negative feelings among individuals who perceive a negative tone from media reports, which is related to less depressive symptoms. Moreover, high-quality people-oriented public health services can enhance the positive perception of media coverage. Thus, such services would be associated with less depressive symptoms among individuals subjected to mandatory social isolation. The objective of this study is threefold: (1) to examine depression among the isolated population, (2) to investigate the association among perceived risk, perceived tone of media reports, and depression during COVID-19, and (3) to explore the moderating role of peoplecentered public health services on the association between perceived risk and tone of media reports on depressive symptoms during the COVID-19 pandemic. --- Methods --- Research context Considering the rapid spread of COVID-19, many provincial and municipal Chinese governments have taken the highest level of response to major public health emergencies since the end of January 2020 [15]. These responses included strict measures in conducting a comprehensive screening of people arriving from cities with a high disease burden. Shenzhen City in Guangdong Province is one of the cities implementing such strict response measures. The city is home to 4 million permanent residents and a floating population of 8.5 million. All residents who visited or stayed in key pandemic areas (e.g., Hubei Province) in the past 14 days before returning to Shenzhen were required to undergo 14 days of mandatory social isolation. Individuals who have been in close contact with patients diagnosed positive for COVID-19 were also required to undergo stringent quarantine. As of February 28, 2020, 418 positive cases of COVID-19 were confirmed in Shenzhen. Among them, 141 were from Hubei Province, which was the hardest hit area, and 261 cases were local [16]. Unlike many countries that encouraged citizens to self-isolate, China implemented mandatory social isolation for these groups who visited or stayed in key pandemic areas from late January to late April of 2020. As of February 21, 2020, the number of residents in Shenzhen who were under home or centralized social isolation reached approximately 25,000. According to the regulations in Shenzhen, the difference between homebased and centralized isolation is whether the person required for mandatory social isolation has the conditions for home isolation, such as whether each home for isolation has a single room and independent bathroom and whether community management is available for the place. If the home does not meet the conditions for home-based mandatory social isolation, the individual needs to be isolated in a hotel. Regardless of homebased or centralized isolation, public health worker's services were not different. However, people may feel more familiar and comfortable with the home environment. --- Sample collection The study conducted an online cross-sectional survey among individuals subjected to mandatory home or centralized social isolation in Shenzhen from February 28 to March 6, 2020. To be eligible for the study, participants must be above 16 years old and currently under or have experienced and completed mandatory home or centralized social isolation during the survey period. At the time of the survey, conducting a field investigation was not possible or allowed. Thus, this study conducted online surveys as this method provides a unique opportunity for research in the COVID-19 era and is the tool of choice among researchers [17]. First, out of 10 districts in Shenzhen, two districts, namely, Luohu and Longgang, were randomly selected for investigation. Luohu District is located in the south part of Shenzhen, which is very close to Hong Kong with a well-developed economy. Longgang District lies in the north part of Shenzhen, which is far from Shenzhen's Bay Area with a relatively underdeveloped economy. The geographical area of Longgang District is 388.22 m 2, which is much larger than that of Luohu District (78.75 m 2 ). Both districts are composed of a large number of permanent and floating populations. As of the end of 2018, Luohu District had a total population of 1.03 million and a permanent population of 0.6 million. The permanent population of Longgang District exceeded 2.3 million, and of which, the permanent population was only 0.72 million [18]. Second, to reach the individuals subjected to mandatory social isolation who were considered a part of the "hidden population" [19,20], the researchers cooperated with public health workers in local communities to distribute the online questionnaire. The survey questionnaire was initially published in the working groups of the two districts for pandemic management through WeChat (China's most popular social media platform). A total of 67 public health service workers in the local communities have seen the survey link and forwarded the questionnaire to individuals subjected to mandatory social isolation. Specifically, as public health service workers were responsible for providing daily necessities and monitoring health status (i.e., body temperature), they had access to the contact information of individuals under mandatory isolation. Moreover, WeChat is easy to use and can transmit pictures to facilitate monitoring. Thus, public health workers and people under mandatory social isolation tend to be friends on WeChat. The local public health workers sent the survey links or quick response codes containing the survey link to the participants through WeChat. The participants read the informed consent before answering the survey and then voluntarily and anonymously completed the survey. The majority of participants spent 3-5 min to fill the questionnaire (Additional file 1). Because the online survey tool reminded respondents to answer all the required questions, the study contained no missing data. Overtly incorrect and unreasonable entries, however, were removed and considered as missing data. According to backstage data of the survey link, 65 people did not complete the questionnaire after entering the survey's webpage. The response rate was 84%. The sample size was calculated based on the assumption that people under home or centralized social isolation would present 10% more depressive symptoms than regular individuals during the COVID-19 pandemic [21]. Therefore, a sample size of 335 would ensure that statistical analysis with <unk> =.05 has an 80% ability to examine the differences in depressive symptoms of people under mandatory social isolation. The study obtained approval from the institutional review board of the Shenzhen Center for Disease Control and Prevention. --- Measurements Depression Depression refers to various negative psychological symptoms, such as depressive mood, loss of interest, fatigue, difficulty in paying attention, and suicidal ideation. Depression was measured using the Patient Health Questionnaire Depression scale (PHQ-9) [22], which was previously validated for use among Chinese adults. The participants rated the extent to which they experienced nine psychological symptoms using a four-point scale, ranging from 0 = "not at all" to 3 = "nearly every day." The reliability of the scale as used in the study was acceptable with a Cronbach's alpha coefficient of.89. The scores of the nine items were summed for analysis. Referring to other studies using PHQ-9, this study also considered 0-4 points as no depression, 5-9 points as mild depression, 10-14 points as moderate depression, 15-19 points as moderately severe depression, and 20-27 points as a severe depression [22]. --- Perceived risk Perceived risk indicates individuals' subjective perception of certain risks, which are specifically represented by perceived severity and perceived susceptibility. In the present study, perceived severity was assessed by the perception of how long the pandemic would continue to influence people's life. The participants rated their responses using a five-point scale, ranging from 1 = "Less than 1 week" to 5 = "More than 6 months." In addition, perceived susceptibility was measured by the perception of the possibility that one could contract the virus during social isolation. The participants rated the probabilities from 1 = "highly unlikely" to 5 = "highly likely." --- Perceived tone of media coverage The perceived tone of media coverage showcases the tone or emotions perceived from the media coverage on COVID-19. The perceived tone of media coverage was evaluated using seven items on a bipolar semantic scale. The seven pairs of opposite adjectives included negative versus positive, critical versus encouraging, complaining versus forgiving, nonreflective versus reflective, worried versus composed, indifferent versus touching, and timid versus brave. The participants rated their responses using a seven-point scale, ranging from -3 to 3. The Cronbach's alpha coefficient reached.91, which indicated excellent reliability. The scores of the seven items were averaged for analysis. --- People-oriented public health services People-oriented public health services measured whether public health officials and workers that were designated to support the daily life routine of people under mandatory home or centralized social isolation were understanding, caring, and trustworthy. In total, three statements were presented, namely, "Public health service workers responded to my question in ways that I can understand," "Public health service workers cared about my feelings and emotions," and "I perceived the public health service workers as trustworthy." The participants rated their responses using a five-point scale, ranging from 1 = "Strongly disagree" to 5 = "Strongly agree." Reliability was considered excellent with a Cronbach's alpha coefficient of.95. The scores for each item were averaged and further categorized people-oriented public health services into high (average scores higher than 4), medium (average scores between 3 and 4) and low (average scores below 3) quality. --- Control variables The study controlled for demographic variables, such as age (continuous variable), sex (female = 0; male = 1), level of education (categorical variable from primary education to master's degree or above), and monthly income (categorical variable from 0 to more than 30,000 RMB). Previous studies found that media exposure and online social support could influence depressive symptoms [23,24]. Therefore, other factors such as participants' time spent on COVID-related news and social support received online were controlled for in the analysis [25]. Time spent on COVID-related news was measured as the time that the participants spent paying attention to COVID-related news, which ranges from a few (less than 1 h) to many (more than 7 h) times per day. Online social support was measured through participant reports on information, emotional, instrumental, and esteem support received from others online. Using a five-point scale, the participants rated whether statements, such as "When I feel scared, I turn online to my relatives and/or friends to talk about my feelings," were similar to their experiences. In addition, the duration of social isolation, venue of social isolation, infection status of family members, isolation status of family members, and separation from family members/friends during isolation were also used as control variables. The duration of social isolation pertained to the start and end dates of social isolation. If the respondent was isolated during the survey, then the duration was measured by subtracting the start date from the date of participation in the survey. The venue of social isolation was measured by whether the participants were isolated at home or a hotel. The infection status of family members was measured by whether the participant has family members confirmed to be infected with COVID-19 virus. Moreover, the isolation status of family members was measured by whether the participant had family members who were under mandatory social isolation. Finally, the participants reported whether they lived with family members/friends or they were with family members during the mandatory social isolation process to measure separation state from family/friends during isolation. --- Statistical analysis The study first described the demographic characteristics, risk perceptions, and depressive symptoms of the participants. Multiple regression was then conducted to examine the main effects of perceptions of susceptibility, severity, tone of media coverage, and people-oriented public health services on depression. These variables were mean-centered to avoid multicollinearity before conducting moderation analyses. Finally, three rounds of stepwise multiple regression were performed to examine the moderating effects, which can clearly determine the effects of various factors and those of the interaction terms [26]. In the stepwise regressions, the first, second, and third layers mainly included control variables, main research factors, and interactive items, respectively. Three sets of interaction terms, namely, interactions between perceived susceptibility and people-centered public health services; perceived severity and peoplecentered public health services; perceived tone of media and people-centered public health services, were independently added to the model. Age, sex, education, monthly income, time spent on COVID-related news, duration of social isolation, venue of social isolation, infection status of family members, isolation status of family members, status of separation from family/friends during isolation, and online social support were controlled for in all regression models. In case of missing data, since the missing data were random and caused by input errors, we retained the cases if the missing values were not the primary exposure or outcome variables. We transformed the missing data and applied the mean replacement method for further analysis and modeling [27]. Data analyses were performed using SPSS Statistics (IBM SPSS version 26.0). Statistical significance was set at p-values less than or equal to.05. --- Results --- Characteristics of the participants Among the 340 participants under mandatory social isolation, 196 (57.65%) were male. The average age was 35.51 years (SD = 8.37), which ranged from 17 to 68 years, and more than half (55.58%) held a bachelor's degree or above. Moreover, the monthly income for the majority of the participants (57.35%) was more than 1100 USD (8000 RMB). Table 1 provides a list of the demographic characteristics. The participants scored an average of 2.19 (SD = 3.57) in the depression scale, which indicates generally low levels of depression. In this study, a total of 284 (83.53%) participants reported no depression, whereas 41 (12.06%), 11 (3.24%), 2 (.58%), and 2 (.58%) pointed to mild, moderate, moderately severe, and severe levels of depression, respectively. Furthermore, 12 (3.53%), 103 (30.29%), and 225 (66.18%) of the participants indicated the quality of people-oriented public health services as low, medium, and high, respectively. Regarding risk perception, the average score for perceived severity was 2.55 (SD =.87), and the average score for perceived susceptibility was 1.36 (SD =.54). Perceived tone of media coverage was mainly positive (M = 1.97, SD = 1.05). Finally, the average score for online social support was 2.83 (SD =.74). --- Main effects on depressive symptoms Perceived susceptibility was significantly associated with depressive symptoms (b = 1.04, p <unk>.01), whereas the opposite is true for perceived severity (b =.16, ns). The results indicate that individuals under mandatory social isolation who perceived a high possibility of becoming infected were more likely to report depressive symptoms. In other words, a 1-unit change in perceived susceptibility was associated with a 1.04-unit increase in depressive symptoms. In addition, the perceived tone of media coverage was negatively associated with depressive symptoms (b = -.46, p <unk>.05), which indicated that a positively perceived media tone was related to less depressive symptoms. That is, a 1-unit increase in perceived positive media tone was associated with a.46-unit decrease in depressive symptoms. Moreover, the results for regression demonstrated that people-oriented public health services were not directly correlated with depression (b = -.60, ns). The adjusted R-square of the main model was.065, which indicated that these factors could explain 6.5% of depressive symptoms. --- Interaction effects on depression Model 2 in Table 2 indicates that the interaction term between perceived susceptibility and people-oriented public health services was significant, with the model's explanatory power increased to 8.1% in adjusted Rsquare. The interaction between perceived susceptibility and people-oriented public health services exerted a significant effect on depressive symptoms (b = -1.33, p <unk>.05). In other words, a 1-unit increase in the quality of people-oriented public health services corresponded to a 1.33-unit decrease in the positive effect of perceived susceptibility on depressive symptoms. For socially isolated individuals suspected of having COVID-19, the high quality of people-oriented public health services attenuated depressive symptoms (Fig. 1). The study found that the interaction term between perceived severity and Online social support M = 2.83, SD = 0.74 a Note: Six (1.76%) participants inserted unreasonable dates for the start date of mandatory social isolation, so their duration of mandatory social isolation could not be calculated and was considered as missing data. The mean was used to replace these missing data for further analysis Note: Model 1 examined the main effects of perceived susceptibility, perceived severity, perceived tones of media coverage, and people-oriented public health services on depression. Model 2 examined the effects of the interaction between perceived susceptibility and people-oriented public health services on depressive symptoms. Model 3 examined the effects of the interaction between perceived severity and people-oriented public health services on depression. Model 4 assessed the effects of the interaction between perceived tone of media coverage and people-oriented public health services on depressive symptoms Fig. 1 The effects of interaction between perceived susceptibility and people-oriented public health services on depressive symptoms quality of public health services was also significant, with the model's adjusted R-square increased to 7.4%. The interaction between perceived severity and peopleoriented public health services exerted a significant effect on depressive symptoms (b = -.79, p <unk>.05). That is, a 1-unit increase in the quality of people-oriented public health services corresponded to a.79-unit decrease in the positive effect of perceived severity on depressive symptoms (Fig. 2). Furthermore, for socially isolated individuals who perceived the COVID-19 pandemic as severe and fatal, the high quality of people-oriented public health services alleviated depressive symptoms. The interaction term between perceived tone of media coverage and people-oriented public health services was significant, with an 8.4% increase in adjusted R-square. In addition, the interaction between perceived tone of media coverage and people-oriented public health services yielded a significant effect on depressive symptoms (b =.87, p <unk>.01). Figure 3 illustrates that the low quality of people-oriented public health services was associated with increased depressive symptoms among individuals subjected to social isolation who perceived a negative tone in media coverage. That is, a 1-unit increase in the quality of people-oriented public health services was related to a.87-unit decrease in the positive effect of a perceived negative tone of media on depressive symptoms. However, for participants who perceived a positive tone in media coverage, the high quality of people-oriented public health services was associated with less depressive symptoms. An additional unit of quality of people-oriented public health services was related to a.87-unit increase in the negative effect of perceived positive tone of media coverage on depressive symptoms. --- Discussion Public health measures, such as social isolation, physical distancing, and quarantine, are widely considered an essential part of controlling the spread of COVID-19. This study addressed the possible mental health problems among people subjected to mandatory home or centralized social isolation. The study results indicated that perceived susceptibility to infection and perceived tone of media reports were associated with depressive symptoms. Additionally, the quality of people-oriented public health services negatively moderated the associations among perceived susceptibility, perceived severity, and depressive symptoms. However, such a quality positively moderated the association between perceived tone of media coverage and depressive symptoms. The study emphasized the importance of people-oriented public health services during mandatory social isolation and extended scholarly understanding of the interaction among risk perception, the role of media reports, and the quality of public health services. People subjected to mandatory isolation generally reported mild symptoms of depression. In the study, the levels of depression of the participants were relatively low compared with those of other research findings during the Ebola and SARS outbreaks [28]. Specifically, Tang et al. [20] conducted a study on quarantined populations during the COVID-19 pandemic in China. They found that people subjected to mandatory isolation in unaffected areas (i.e., all areas except Wuhan) were more depressed compared with those in affected areas (i.e., Wuhan) and those not quarantined. They further explained that the stringent screening conducted in unaffected areas among residents from Wuhan resulted in extreme depression and anxiety. However, our study Fig. 2 The effects of interaction between perceived severity and people-oriented public health services on depressive symptoms revealed an entirely different scenario because the majority of our participants were not depressed. In fact, their levels of depression were lower than those during non-emergency periods [20,29]. This result can be due to many reasons. First, although people were subjected to mandatory social isolation, their mentality was likely relatively relaxed because they had moved into an unaffected area. Second, most participants subjected to mandatory social isolation stated in the survey that they believed that the pandemic would be brought under control in less than half a year. Moreover, according to the WHO, COVID-19 is more deadly for the elderly population and those with chronic illnesses and/or poor health conditions compared with other diseases [30]. This news and information may have alleviated depression among people subjected to mandatory social isolation. Third, Shenzhen's well-known service quality and efficient government management capabilities may have provided these individuals with a sense of security and safety despite the mandatory social isolation [31]. Similar to this study, Chen et al. [19] conducted a survey of people subjected to mandatory social isolation in Guangzhou, China, during the pandemic and found that the majority of the participants did not exhibit symptoms of depression. That said, although the findings of the present study seem inconsistent with those of the previous studies [32], all results essentially verify that adequate and effective public health services are associated with low levels of depression. This study revealed the risk perception factors associated with depression symptoms among people under home or centralized mandatory isolation. Perceived susceptibility, which pertains to the fear and worry of being infected, could be described as a haunted feeling during social isolation, particularly for those with physical symptoms that may be related to COVID-19 [2]. Meanwhile, under mandatory social isolation, information disseminated through media coverage largely influenced people's perceptions. Thus, the perceived positive or negative tone of media coverage could establish an overall picture of the fight against the pandemic, thereby enhancing or undermining confidence in the successful response to COVID-19. People-oriented public health service was identified as an important moderating factor for mental health. Previous studies have emphasized public health services during epidemics and their roles in providing instrumental and informational support for people under mandatory social isolation [5,33]. Although public health services in most countries may become overwhelmed during the epidemic, the need to maintain high-quality peopleoriented services continues to reduce public panic and increase public trust. When people feel understood and cared for by others, they feel less frustrated particularly because public health service providers are representative of local governments and public health management departments [34]. Importantly, people-oriented services enable people under mandatory social isolation to actively cope with the situation and weaken the perceived stigma and xenophobia associated with becoming a potential source of COVID-19 infection [2]. This study has certain theoretical and practical implications. People-oriented services can enhance personal coping strategies and reduce negative emotions during the fight against the pandemic [35]. In this study, the concept of people-oriented public health services during pandemics extends the concept of people-centered medical services to the emergency [36]. People-oriented Fig. 3 The effects of interaction between perceived tone of media coverage and people oriented public health services on depressive symptoms public health services can be used as a complementary measure for improving the effectiveness of treatment and reducing potential mental health burdens. In addition, these services are typically provided by workers at the community level. Therefore, public health emergency preparedness should include relevant training to address the mental health of the public and prevent diseases simultaneously and to further practice peopleoriented public health services in the future [37]. The study has several advantages. First, in the early stages of the pandemic, where people are mainly concerned with the response, prevention, and control efforts exerted by the government, the study focused on the mental health issues of people under mandatory social isolation and investigated their depressive symptoms. Second, the study emphasized the correlation between people-centered public health services and positive perceptions, which was also associated with low levels of depression. Third, the study focused on the interaction between the elements of interpersonal communication and factors of mass communication. Furthermore, the study controlled for many potential influencing factors when exploring mental health during the health crisis. These considerations have built a relatively comprehensive model that can be used to evaluate mental health during public health emergencies. --- Limitations Despite its strengths, this study has its limitations. First, the random sampling method cannot be used because of the shortage of manpower and resources during the peak time of the pandemic [20]. Therefore, this study conducted a convenience sampling method, constraining its representativeness and generalizability. Individuals with severe mental health problems may also have refused to answer the survey. Thus, the mental health problems of people under mandatory social isolation may have been underestimated. Future studies with sufficient resources can consider using random sampling and door-to-door methods to capture the status of individual mental health problems during epidemics. With the advancement of technology and the Internet of things, other unobtrusive methods (e.g., wearable devices and machine learning techniques) may also assist in identifying individuals with severe mental health problems [38]. Second, the study used a cross-sectional survey; thus, the findings should be interpreted with caution. The temporal sequence between people-centered public health services and depression cannot be determined. Some of the associations may be reversed, and the causal relationships cannot be guaranteed using this type of study design. Hence, future studies using longitudinal surveys or experimental methods are necessary to further explore the causal relationships between mental health and health crises. Third, the researchers were unable to assess the mental health conditions of the participants prior to the survey. Therefore, whether or how their mental health status before the quarantine influenced their current depressive symptoms cannot be determined. In addition, endogenous problems may occur because the measurements were self-reported, and the influence of other factors cannot be excluded. Finally, the study examined the symptoms of depression during the period of social isolation. Future research can further explore the correlates of mental health status post-quarantine or postpandemic. --- Conclusions This study examined the depressive symptoms among people under mandatory home or centralized social isolation in Shenzhen, China. The results emphasized the moderating role of people-oriented public health services in the associations among perceived susceptibility, perceived severity, perceived tone of media coverage, and depressive symptoms during the COVID-19 pandemic. Despite the pressure to fight the pandemic, peopleoriented public health services for people under mandatory social isolation should be promoted. --- Availability of data and materials The datasets used and analyzed during the current study are available from the corresponding author on reasonable request. --- Supplementary Information The online version contains supplementary material available at https://doi. org/10.1186/s12889-021-11457-6. Additional files 1. Survey questionnaire. --- Authors' contributions BC and XL contributed to the study design and conception. NW, MW, and QM participated in the coordination of the study. BC and YW analyzed the data. BC and DW drafted the manuscript, with feedback from HB, JT, and XL. All authors read and approved the final manuscript. --- Declarations --- Ethics approval and consent to participate This study has received approval from the institutional review board of Shenzhen Center for Disease Control and Prevention. Written informed consent has been obtained from all participants. --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Public health measures, such as social isolation, are vital to control the spread of the coronavirus disease 2019 ), but such measures may increase the risk of depression. Thus, this study examines the influencing and moderating factors of depressive symptoms among individuals subjected to mandatory social isolation. Methods: An online cross-sectional survey was conducted to collect data from people under mandatory home or centralized social isolation in Shenzhen, China, from February 28 to March 6, 2020. The perceived risk of infection with COVID-19, perceived tone of media coverage, perceived quality of people-oriented public health services, and their depressive symptoms were assessed. Three rounds of stepwise multiple regression were performed to examine the moderating effects after controlling various variables, such as demographics, duration and venue of mandatory social isolation, infection and isolation status of family, time spent on COVID-related news, and online social support. Results: Among the 340 participants, 57.6% were men, the average age was 35.5 years old (SD = 8.37), and 55.6% held a bachelor's degree or above. Individuals subjected to mandatory social isolation generally reported low levels of depressive symptoms. Perceived susceptibility to infection was relatively low, whereas perceived tone of media coverage was mainly positive. In terms of perceived quality of public health services, 12 (3.5%), 103 (30.3%), and 225 (66.2%) participants reported low, medium, and high quality of people-oriented services, respectively. Perceived susceptibility was positively associated with depression, whereas perceived tone of media coverage was negatively associated. The quality of people-centered public health services moderated the association between perceived risk and depressive symptoms and between perceived tone of media coverage and depressive symptoms.

Introduction Humanitarian health has been suggested as a key research priority and an essential part of global health initiatives in emergencies [1]. In 2020, nearly 168 million people worldwide are estimated to be in need of assistance or protection due to humanitarian emergencies, such as conflicts or natural disasters [2]. Needs assessment is one of the fundamental cores of the humanitarian health response, both in long-lasting humanitarian settings and after sudden-onset disasters. A proper and well-designed needs assessment lays the foundation for a coherent, efficient and trustworthy humanitarian response to any emergency [3]. Many of the humanitarian emergencies are long-lasting crises, and the majority of all people considered as affected by humanitarian emergencies live in designated areas, such as camps [3]. Health and well-being among people living in such camps include a wide range of potential health problems: infectious diseases, chronic conditions, injuries, malnutrition, gender-based violence, mental health problems and disruption of cultural and social conventions [4]. The daily life of affected people living in camps is fraught with unmet basic needs [5]. Having a higher level of perceived needs has been found to predict a greater level of psychological distress [6]. Therefore, a reliable assessment of perceived needs can be said to be the fundament in order to understand mental health and other health problems among people in vulnerable situations [7]. The development and availability of scientifically and contextual feasible instruments to assess health and needs in humanitarian emergencies is strongly needed [1]. The Humanitarian Emergency Settings Perceived Needs (HESPER) scale was developed to provide a quick and reliable way to assess the perceived needs of affected people in humanitarian emergencies, including complex emergencies, conflicts and natural disasters [8]. The HES-PER scale was developed by the World Health Organization and Institute of Psychiatry at King's College London, based on literature studies, experts on humanitarian assessments and several pilot-and field tests including different samples of populations affected by different kinds of humanitarian emergencies. A detailed description of the development and testing of the scale has been reported elsewhere [8]. The original HESPER was designed to gather data through individual face-to-face interviews and paper surveys. Today, an increasing part of the world population has access to Internet connections. About 93% of all displaced people in the world have access to a mobile network, and many have access to the Internet, through a mobile connection, broadband in schools, community Internet cafés or other sources. Additionally, in rural areas, the coverage and quality are progressively improving [9]. Internet-based data gathering offers quicker data collection and analysis and fewer internal dropouts and processing errors, and is often a more economical alternative to other types of surveys [10]. To combine the strengths of Internet-based data collections and a scale measuring perceived needs among humanitarian populations, we developed HESPER Web, a self-administrated web-based version of the original HESPER [11]. The HESPER Web can be administrated through a web link and answered on a computer, tablet or mobile phone and the first psychometric evaluation of HESPER Web showed very good reliability and feasibility among a study sample of asylum seekers in Sweden [11]. In order to further evaluate the HESPER Web, a field test in a large scale humanitarian context was necessary. This study had three aims; (1) to evaluate the reliability and feasibility of HESPER Web in a large-scale humanitarian context, (2) to compare the demographics and means of the perceived needs of a random walk method study sample and a convenient, self-selected study sample recruited though social media and (3) to describe the perceived needs within the study sample. --- Materials and Methods The study consisted of two parts: an alternate forms reliability evaluation and a feasibility evaluation. When analyzing the reported needs, data from both the alternate forms evaluation and the feasibility evaluation were used. --- Study Setting The Dadaab refugee camp in eastern Kenya has been operative for about 26 years and consists of three smaller camps, Dagahaley, Ifo and Hagadera. In February 2019, the camps hosted about 205,000 refugees [6]. The absolute majority of the Dadaab population are refugees from Somalia, and about 50% of all people living in Dadaab are male. The United Nations High Commissioner for Refugees (UNHCR) is the operational manager of the camps, and all services, such as housing, food, water supply, basic health care and schools, are free for registered refugees [12]. The Hagadera refugee camp houses about 83,940 people, where 50% are male. Hagadera has 10 schools. In a household survey conducted in 2017, 43% of all households reported English as their first language [13]. The Internet is available through a 3G connection (via mobile networks) or by broadband in the 10 schools, one adult literacy centre and one ICT training centre [13]. The study context for this study was the Hagadera refugee camp and the data collection was conducted in February 2019. --- Instrumentation The HESPER scale consists of 26 fixed questions covering physical, psychological and social needs [8]. The ratings are made by interviewers in a face-to-face interview with affected persons by asking whether a certain need is perceived as a "serious problem" or not. In addition, the affected person can add other needs if not covered by the original 26 stated needs. After reporting their needs, the affected person is asked to prioritize the three most serious perceived needs. A total sum score can be calculated by adding up the total number of "serious problem" ratings [8]. The HESPER Web is a newly developed web based, self-administrated survey version of the original HESPER scale [11]. The HEPSER Web could be accessed by a web link by a mobile phone, tablet or computer. In addition to the 26 questions regarding certain needs and the prioritizing question in the HESPER Web, study specific feasibility questions were added to the HESPER web survey. These questions were: how long did it take for you to fill in the survey? were the questions easy to understand? what mean did you use for answering the survey? did you experience any technical problems when answering the survey? did you suffer from any harm by filling in the survey? could you answer the survey in privacy? and how did you get the invitation for this survey? --- Alternate Forms Evaluation The alternate forms reliability between the original HESPER scale and HESPER Web was evaluated using a voluntary convenience study sample of 50 study participants from the camp. Based on a power analysis that indicated a need for a minimum of 19 participants in both data collections in order to detect a statistically significant correlation and a power of 90%, and previous experiences from conducting alternative forms evaluation [11,14,15], a sample size of 50 was chosen. Inclusion criteria were that the person should be at least 18 years old, have access to the Internet by mobile phone, tablet or computer and be able to participate in the interview using the English language. For all participants, the HESPER interview was made prior to the web survey, due to practical reasons. The HESPER interviews were conducted by two male and two female volunteer assistants trained during a six-hour training session in accordance with the HESPER manual. Using a cluster random sampling method, four square areas within the Hagadera camp were first selected by lot to be included in the study [13]. Thereafter, the households asked to participate were selected using a kind of random walk method [8], where every second house in a direction pointed at by spinning a pencil was visited. In the first household, the first person to approach the interviewers was asked to participate. In the second household visited within the cluster, the second person seen by the interviewers was asked to participate, and so on. If any of the persons selected could not participate for any reason, the next person in the household was asked. The interviewers estimated that in every third household, there was no person eligible for participation. If so, the interviewers continued to the next household. A code list was used to group the HESPER scale and HESPER Web answers. Both data collections were answered anonymously, using the specific code only as reference in the web survey. The participants got a personally written reminder note from the interviewer, with the code and the link to HESPER Web, asking them to complete it within 48 h. The time between the HESPER interview and that taken to answer HESPER Web varied from a few hours up to three days. --- Feasibility Evaluation The sample for the feasibility evaluation of HESPER Web was conducted with 289 voluntary study participants who were recruited by advertising the study in the adult training centre, secondary schools and internet-and communication centers in Hagadera. Additionally, digital advertising on Facebook and on three specific pages aimed at people living in Hagadera or other Dadaab camps was used. Inclusion criteria were that the study participant should be at least 18 years old, have access to the Internet by mobile phone, tablet or computer and be able to participate in the survey using the English language. The data collection period lasted for seven days (see Figure 1). The web survey was anonymous, and there were no limitations on answers from the same IP address, in order to allow several responders to use the same computer, tablet or mobile phone to answer the survey. Data were saved in a secured research database at Orebro University in Sweden. --- Analysis For the alternate forms reliability between the HESPER scale and HESPER Web, intraclass correlation coefficients (ICCs), two-way mixed and absolute agreement [16], of the total number of reported serious needs was calculated. To assess agreement on an item level and the percentage match between first priority needs in the HESPER scale and HESPER Web, Cohen's <unk> was used. Additionally, descriptive statistics for analyzing the feasibility questions and the reported needs were used. SPSS software (IBM Corp. Released 2016. IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY: IBM Corp) was used to conduct the statistical analysis. A significance level of p <unk> 0.05 was used. --- Ethical Considerations Informed consent was obtained by each study's participants before participating in the interview and/or web survey. The study was approved by the Regional Ethical Committee in Sweden (ID 2017/481) and the National Commission for Science, Technology and Innovation (NACOST) in Kenya. Permission to develop and evaluate the HESPER Web was obtained from the WHO. --- Results In total, 308 individuals participated in the study: 50 in the alternate forms evaluation and 289 in the feasibility evaluation. Table 1 shows the demographics. There was no significant difference between the HESPER interview sample (n = 50) and the HESPER Web feasibility evaluation sample (n = 289) regarding gender (Chi 2 test, p = 0.33), age (Chi 2 test, p = 0.78) or present location (Chi 2 test, p = 0.99) but for country of origin (Chi 2 test, p <unk> 0.001). There was no significant difference between the participants who participated in both the HESPER interview and the web survey, and those who dropped out from the survey (gender (Chi 2 test, p = 0.61), age (Chi 2 test, p = 0.50) or present location (Chi 2 test, p = 0.99) and country of origin (Chi 2 test, p = 0.97). --- Alternate Forms Evaluation Of the 50 participants recruited for the alternate forms reliability evaluation and who participated in the HESPER interview, there were 19 dropouts who did not answer the HESPER Web. The alternate form results were therefore based on 31 participants. The ICC was 0.88 (CI 0.60-0.91) between the HESPER scale and HESPER Web. For the item-by-item evaluation between the HESPER scale and HESPER Web, Cohen's <unk> was calculated, and it varied between 0.43 (for the item concerning safety) and 1.0 (for the item relating to law and justice in the community and other serious problems), see Table 2. Regarding the first priority need rating, an overall match of 81% was found between the HESPER scale and HESPER Web. 1.00 1 Items presented in the HESPER Web order; 2 Kappa value could not be calculated due to zero answers in one or more samples. --- Feasibility Evaluation Answering the HESPER Web survey was quicker than being interviewed for many of the study participants (p <unk> 0.001, see Table 3). The questions asked in HESPER Web were considered to be easy to understand, and no participant reported experiencing harm caused by the survey. About 86% of all study participants could answer HESPER Web in privacy (see Table 4). An absolute majority of the participants used their own mobile phones to answer the survey (60%), followed by a significant number who used someone else's computer or tablet, including the school's or ICT center's (19%). About 13% used someone else's mobile phone or their own computers or tablets (4%). Chi 2 test, p = 0.00, Cramer's V = 0.503. --- Differences in Demographics between the Randomized Study Sample and Self-Selected Sample No significant difference in the total reported numbers of needs could be observed between the HESPER scale and HESPER Web study samples (p = 0.067, paired t-test) or when comparing HESPER Web (alternate forms) and HESPER Web (feasibility evaluation; two-sample t-test, p = 0.132). No significant difference in gender (Chi 2 test, p = 0.670) or age (two-sample t-test, p = 0.810) between the HESPER interview sample and the HESPER Web self-selected sample was observed. --- Perceived Needs When reporting results on their perceived needs, a total sample of 320 people was used, including all study participants who answered HESPER Web (as part of the feasibility evaluation (n = 289) or the alternate forms evaluation (n = 31)), and not the ones who only participated in the HESPER interviews. When reporting results on their perceived needs, a total sample of 320 people was used, including all study participants who answered HESPER Web (as part of the feasibility evaluation (n = 289) or the alternate forms evaluation (n = 31)), and not the ones who only participated in the HESPER interviews. The mean number of reported needs among the study participants in the HESPER scale was 4.52 (SD 3.2, range 1-15). The frequency of reported needs in total and sorted on gender is shown in Table 5. There was no significant difference between males and females regarding the mean of the total number of reported needs (Student's t-test: male mean 5.88, SD 3.9, (95% CI: 5.27; 6.48), range 0-21), female mean 6.43, SD 3.9, (95% CI: 5.50;7.0), range 1-19, p = 0.765), but there were some differences in what kind of needs were reported (see Table 5) --- Discussion HESPER Web was found to be reliable and usable for assessing perceived needs among refugees living in a large-scale humanitarian context such as the Dadaab refugee camp. The use of a web-based survey was positively experienced by the study participants, and the voluntary self-recruited study sample reported similar levels of needs and similar demographics regarding gender and age to the walking methods randomized study sample. The participants reported several unmet needs, and there were some differences in the kinds of needs identified depending on gender. The alternate forms evaluation showed overall good correspondence between the HESPER scale and HESPER Web in general (ICC 0.88) and on an item by item level (Cohen's <unk> from 0.43 to 1.0) [17]. The item with the lowest consistency was the question on perceived problems caused by security issues. The reason for this might be that the current level of security varied a lot from day to day and from location to location within the Dadaab camp. Additionally, there was an observed difference when reporting on educational needs for children. However, this item was frequently reported in the larger sample (Table 5) and therefore, we could not explain the difference noted in the comparison between the HESPER and the HESPER Web. The association for the first priority rating was very good (81%) [17], showing that HESPER Web reliably can be used to assess the most serious perceived needs instead of or as a complementing data collection method to the HESPER interview. However, it should be noted that the timeframe between first and second data collection was short (from a few hours up to 3 days), and that might have influenced the results. It would have been preferred with a longer timeframe between the two measurements, but due to security regulations, repeated visits could not be conducted. The short timeframe may have resulted in that participants remembered their answers from the first data collection, which may have contributed to a slightly overestimated alternate forms reliability coefficient. Even when taking this into account the alternate forms reliability between the two forms of administration of the HESPER is good. In the HESPER manual, strategies to perform data collection in order to ensure a proper study sample are described. When using web-based methods, the same procedures may be used with the difference that the study participant answers the web-based survey instead of taking part in an interview. If advertising the survey on social media or physical locations, the study sample will be a convenience sample. This study suggests that the study samples from the walk-around sampling method and the self-selected sample were similar, regarding both their demographics and the mean number of reported needs. However, it should be noted that the number of study participants differed between the samples, and the exact number of study participants needed for generalization of a web based, not randomized data collection cannot be concluded from this study. When conducting HESPER interviews face to face, the interviewer could interact with the person and, if needed, provide specific advice or refer to, for example, psychosocial support. When using a self-administrated web-based survey, this is no longer an option. Therefore, it is of extra importance for a survey provider to state the limitations and to provide practical support and to state where the study participants should turn for help in case of an immediate need for such support. In addition, a web-based survey may offer new possibilities to direct people who report need of support, and guide them on where to turn for available support. HESPER Web has shown potential in reducing several challenges that are common in disaster or humanitarian emergency health research related to the practical possibilities of physically reaching or visiting an area, security concerns and ethical considerations, such as the possibility of being anonymous [18][19][20]. HESPER Web can offer possibilities for conducting assessments and research studies that include populations that are rarely included in such evaluations, such as people who constantly move around, people evacuated from the study area or those who do not have access to a fixed address [20]. In addition, the tool may be used for longitudinal studies on perceived needs [11]. However, not all study populations or contexts are suitable for web-based needs assessment or research. The reasons may be several, including limited access to the Internet or a means for answering the survey, limited privacy when answering the survey or illiterate or severely traumatized populations where personal contact may be necessary to assess mental health or provide support. The responsibility of using a valid and proper instrument and data collection procedure and considering the context and study population is always the researcher's or the head of the organization's responsibility, and not that of the affected population. In this survey, the study participants reported several needs, although, they were settled in a long-lasting state of displaceability. Web-based methods for assessing mental health have been suggested to provide a better picture of the actual situation while offering anonymity and reducing stigma in the interview situation [19]. Higher levels of perceived needs can significantly predict psychological distress and lower levels of functioning [6]. It has been suggested that further emphasis should be put on developing tools for community mental health providers to enhance reach and effects from mental health interventions in low-and middle-income populations [21]. To assess perceived needs and plan for mental health interventions also in populations with long lasting displaceability seems therefore reasonable. Additionally, it has been suggested to further explore the use of self-help digital mobile applications used in community based mental health interventions in for example refugee camps [21]. For such purposes, the HESPER Web could be a feasible tool, but need to be further evaluated. This study had several limitations. It would have been preferable to let half of the study participants in the alternate form evaluation answer the HESPER scale first, and then HESPER Web, and the other half in the opposite order. Due to security reasons, that could not be done. Additionally, such a strategy was however considered to increase the risk of dropouts between the two data collections and therefore dismissed. The use of "random walk sampling" is usually not the preferably choice of the sampling method for research studies. However, it was considered as the best possible option, given the security environment and practical possibilities. The way the "random walk sampling" was used in this study can be described as a combination of a "spin the pen" sampling and a clustered sampling method and is recommended for research in humanitarian emergencies when other, traditional methods are not possible or suitable [22]. When conducting research in humanitarian emergency settings, the research needs to be done with, and for populations affected in order to determine interventions that are feasible and appropriate for the context [1]. In this study, several actors with extensive knowledge and involvement in local processes were involved in planning, practical data collection and the analysis of this study, including local UN agencies, NGOs and academic partners. Partnerships with local individuals ensure a local perspective and add value to the interpretations of the results [23,24]. However, the study participants themselves were not actively engaged in parts other than the data collection. The active engagement of the people affected is essential to ensure that the response is based on their actual needs and supports their recovery [23]. Asking the refugees themselves for their perceived needs may, therefore, contribute to both community engagement and individual recovery [25]. However, little is known about refugee participation in the development of policies and programs that matter to their health and well-being. Such participation is fundamental for more sustainable and responsive projects [4], and a plan for the dissemination of the results should, therefore, be considered in future projects. --- Conclusions HESPER Web was found to be reliable and usable for assessing perceived needs among a population affected by a humanitarian emergency. The use of a web-based survey was positively experienced by the study participants, and the voluntary, self-recruited study sample reported similar levels of needs and similar demographics regarding gender and age to the randomized study sample. HESPER Web offers a reliable and feasible tool for assessment of needs in situations where web-based surveys are considered as practical and suitable. It offers new possibilities for conducting remote assessments and research studies that include humanitarian populations that are rarely included in such evaluations. --- Data Availability Statement: The datasets analyzed during the current study are not publicly available due to the Swedish law on ethical approval for research but are available from the corresponding author on reasonable request. --- Funding: This research was funded by Elrha's Research for Health in Humanitarian Crises (R2HC) Programme, which aims to improve health outcomes by strengthening the evidence base for public health interventions in humanitarian crises. The R2HC programme is funded by the UK Government (DFID), the Wellcome Trust, and the UK National Institute for Health Research (NIHR). The funder did not participate in the design of the study, data collection, analysis, or in writing the manuscript. --- Institutional Review Board Statement: The study was approved by the Regional Ethical Committee in Sweden (ID 2017/481) and the National Commission for Science, Technology and Innovation (NACOST) in Kenya. Permission to develop and evaluate the HESPER Web was obtained from the WHO. Informed Consent Statement: Informed consent was obtained by each study participants before participating in the interview and/or web survey. --- Conflicts of Interest: The authors declare no conflict of interest.

Needs assessment is essential in the humanitarian response, and perceived needs can be associated with the levels of health in populations affected by humanitarian emergencies. This study aimed to evaluate the reliability and feasibility of The Humanitarian Emergency Settings Perceived Needs Web (HESPER Web) in a humanitarian context and to compare perceived needs of a random walk study sample with a self-selected study sample recruited though social media. The study context was the Dadaab refugee camp in Kenya. An alternate forms reliability evaluation and a feasibility evaluation was conducted. In total, 308 refugees participated in the study. HESPER Web was found to be reliable and usable for assessing needs, with an intraclass correlation coefficient (ICC) of 0.88, Cohen's κ between 0.43 and 1.0 and a first priority need rating match of 81%. The HESPER Web was positively experienced, and the self-recruited study sample reported similar levels of needs and similar demographics as the randomized sample. The participants reported several unmet needs. HESPER Web offers a reliable tool for needs assessment in humanitarian emergencies where web-based surveys are considered as practical and suitable. It offers new possibilities for conducting remote assessments and research studies that include humanitarian populations that are rarely included in such evaluations.

Introduction Over the past 2 years, the COVID-19 pandemic, as a global health emergency, has been exacerbating uncertainty and insecurity around the world. The major infectious disease also posed a great risk to sojourners. Sojourners refer to a group of people, including foreign students, foreign traders, diplomats, anthropologists doing research abroad, international journalists, and the like, who are willing to live in a country or region for some period of time out of their work arrangements or certain contracts [1]. As the pandemic became rampant, COVID-19 vaccines have been advocated as the most effective response against the SARS-COV-2 virus [2][3][4]. In this regard, the Chinese government gave mass vaccination priority and has been encouraging foreigners to receive the vaccination. On 26th March 2021, the Foreign Affairs Office of the People's Government of Beijing Municipality released an announcement encouraging foreign nationals in Beijing to get vaccinated, voluntarily, of course. With valid certificates, such as a passport and residence certificate, foreign nationals who have joined Beijing's social security and medical insurance service could receive free vaccination, while others would be charged CNY 93.5 per dose [5]. By the end of July 2021, more than 500,000 doses of Chinese vaccines have been administered to over 300,000 foreign nationals [6]. However, a poll in one report found that South Koreans showed the least willingness to get vaccinated: only 14% of respondents admitted that they wanted to receive the vaccination quickly [7]. In addition, more than half of South Korean diplomats in China were unwilling to receive Sinopharm or Sinovac COVID-19 vaccines (two China-developed vaccines) [8]. Despite the measures the Chinese government and health representatives have taken to advocate the benefits of receiving Chinese COVID-19 vaccines, the government still had to face South Korean sojourners' vaccination hesitancy. Vaccination hesitancy refers to the phenomenon that individuals are reluctant or unwilling to receive certain vaccines, although the vaccines are available. It is a complex issue and varies depending on time, place, and the type of vaccine [9,10]. Previous research showed that vaccine policies, cultural factors, levels of trust in institutions and media could affect individuals' vaccination intention dramatically [11][12][13][14][15]. Therefore, the goal of this study is to investigate how trust in institutions and media affects South Korean sojourners' vaccination intention of Chinese COVID-19 vaccines. The primary contributions of the present study are as follows: (1) it contributes to a comprehensive understanding of South Korean sojourners under a national context; (2) trust in Chinese institutions and trust in China's media are employed as a theoretical framework to explore South Korean sojourners' vaccination willingness; (3) this study clarifies the correlation between South Korean sojourners' reluctance to get China-developed COVID-19 vaccines and their trust in China's institutions and media. --- Literature Review and Theoretical Background --- The Influence of Institutional Trust on Vaccination Intention Previous research has found that three elements would determine the public's trust in vaccination: knowledge and expertise; openness and honesty; and concern and care. For the public, apart from basic information and knowledge, they also want to make sure that the people who are conveying the information and clarifying concerns are trustworthy and their communication manner is acceptable [16]. Institutional trust refers to the assessment the public performs to measure whether risk management institutions (e.g., the government, the health systems, science, and the medical establishment) have the willingness or ability to help them avoid risks [17]. Some studies have tested the importance of institutional trust during the global pandemic across the globe. The more trust individuals show toward the institutions, the more willing they are to get relevant health services like the vaccination [18]. During the COVID-19 pandemic, people who trust their institutions can be more easily convinced by the official information in terms of the safety and efficacy of the vaccines, and therefore, they are more open to receiving the vaccine [19,20]. Also, the public often relies on the vaccine's country of origin to evaluate its safety and efficacy [21,22]. A study revealed that in countries where people showed higher trust to a central government, including China, South Korea, and Singapore, the vaccination was more acceptable among the public [23]. On the contrary, lower trust in institutions always led to vaccination hesitancy [24]. Notably, the majority of people who have high institutional distrust were racial/ethnic minorities and populations having lower socioeconomic status; their life experience demonstrated that distrust often came from injustice, which ultimately made health inequality worse [25,26]. Recent studies have found that excessive pro-research propaganda easily leads the public to believe the primary purpose of the government-encouraged mass vaccination was shifting from disease prevention to financial gains by healthcare providers and vaccine manufacturers, thus causing institutional distrust [27]. Many studies have found a link between institutional distrust, vaccine hesitancy and conspiracy beliefs. One study conducted in England demonstrated that people holding conspiracy beliefs showed less trust to guidelines released by the government as well as the institutions [28]. Interestingly, conspiracy beliefs tend to be enhanced by individuals' behavior that only searches for and accepts information favoring their position and beliefs while ignoring unfavorable information [14]. In terms of vaccine hesitancy, conspiracy beliefs play a negative role in deteriorating one's trust in the government, health system, and medical industry [29]. Empirical evidence has showed that conspiracy beliefs were associated with people's perception of low socio-political control, politically-driven commands, and distrust in science [30][31][32]. Therefore, when people believed that the COVID-19 pandemic was driven by conspiracy beliefs, they were suspicious about the motivation behind every relevant measure put forward by the institutions, thus showing little willingness to get vaccinated [14]. --- The Influence of Media Trust on Vaccination Intention Media trust is another key factor regarding people's healthrelated behavior, and may further influence the public's health choices during a social crisis [33]. The information from different media channels varied according to their function and emphasis. Traditional media outlets, such as TV and newspapers, tended to provide more expert information [34]. During a health crisis when the public was in urgent need of authoritative information, traditional media outlets were regarded as a reliable source to both transmit knowledge and improve public awareness [35]. With the development of digital society, social media has been a breeding ground for health information regarding vaccines [36]. As an inclusive platform, social media, with vast unfiltered and latest information, was employed as a useful tool by the public [37]. Apart from media, interpersonal communication also mattered regarding health issues [38]. During a health crisis, interpersonal communication provided a network full of important information for people within it to communicate and deliver [39]. Extant research has shown that high trust in traditional media can significantly improve the public's vaccination intention during a public health crisis. For example, in China, traditional media outlets, both national and local, were used as tools for encouraging the public to get vaccinated massively. Hence, the public's high trust in traditional media would naturally contribute to lower vaccination hesitancy and higher vaccination willingness [40]. Surveys of nearly 2500 Americans during a measles outbreak demonstrated that users of traditional media were less likely to come across misinformation about the vaccines, thus showing higher vaccine acceptance [41]. Another study found that channels of information were used by the public to make vaccination decisions: individuals who received information from traditional media, such as national TV, national newspapers, and local newspapers, could receive more information about the vaccine, thus leading to a higher vaccination willingness [42]. Past studies have found that in addition to traditional media, people's trust in social media played a vital role in their vaccination intention. The more trust people showed to social media, the more likely they were to get vaccinated [43]. Various sources of information, including websites and mobile applications, enhanced the public's awareness of the virus, thus making the vaccination campaign easier to roll out [44][45][46]. For example, in terms of the HPV vaccination, the information from the media that advocated the benefits of HPV vaccines could help the public to understand the value, safety, and efficacy of those vaccines [47]. Nevertheless, several studies have demonstrated that social media was not an authoritative platform of health information, and in professional fields like vaccination, some information was even misleading [11,48]. In terms of COVID-19 vaccination, unverified vaccine posts were ubiquitous, causing real damage to public health [49,50]. Previous research has also shown that interpersonal communication among family members, friends and colleagues played a significant role in shaping one's health intentions and behaviors [51,52]. People having higher interpersonal trust were less likely to hesitate about terms of vaccination [53]. However, the dark side was that information from interpersonal communication was not always trustworthy because sometimes rumors might get a higher momentum than true information [54]. For example, a study found that the extremely high trust in interpersonal communication could affect people's COVID-19 vaccination motivation in the future; specifically, the extremely frequent person-toperson communication might make it harder to roll out the vaccination campaign, because some people might find it special and eye-catching if they did not get vaccinated [40]. While studies have investigated the correlation between the public's trust and their vaccination result, few of them have combined trust in institutions and trust in media as a theoretical framework to analyze their influences on people's vaccination willingness. Studies that have explored the importance of trust in institutions or trust in media mainly focused on local residents, while few have examined how the sojourners' trust in institutions and media influenced their vaccination willingness in the host country. Besides, no study has shed light on South Korean sojourners living in China in this respect. To address this research gap, this study proposes the following research questions: 1. RQ1: How does South Korean sojourners' trust in China's institutions influence their willingness to get Chinadeveloped COVID-19 vaccines? 2. RQ2: How does South Korean sojourners' trust in China's media influence their willingness to get Chinadeveloped COVID-19 vaccines? 3 Materials and Methods --- Study Design The purpose of this research is to investigate how South Korean sojourners' trust in institutions and trust in media influence their intention to get Chinese COVID-19 vaccines. A qualitative research method was utilized to achieve research goals. Qualitative research can provide researchers a more unrestrained, profound, and more flexible understanding of the target group's experiences, with higher levels of openness and the potential to adapt to changes as the inquiry goes deeper [55]. In terms of individuals' vaccination willingness, qualitative studies are rare, and due to the exploratory nature of the current study that focuses on meaning-making [56], a qualitative design should be more appropriate to explore abundant and unconstrained information about this topic. This study adopted one-to-one semi-structured interviews. Semi-structured interviews were appropriate because of its two features. First, to explore respondents' perceptions of and attitudes toward complicated and even sensitive subjects and search for as much information as possible, semistructured interviews are necessary. Second, considering the fact that respondents have different educational, professional and personal backgrounds, semi-structured interviews are more suitable than standardized interviews [57]. Therefore, with open-ended questions, the semi-structured interview is ideal to explore how trust in institutions and trust in media influence South Korean sojourners' willingness to receive Chinese COVID-19 vaccines. --- Participants This research focused on South Korean sojourners in Beijing. All respondents were informed of the purpose of this study, and they participated in semi-structured interviews voluntarily. The initial interview list was provided by the South Korean Chamber of Commerce in China, and the researcher conducted a purpose sampling, forming a preliminary interview list. Later, a snowball sampling was conducted to enlarge that list. The snowball sampling was based and expanded on the following selection criteria: all interviewees (a) have been living in Beijing for at least 1 year; (b) having South Korean nationality; (c) have gone through the COVID-19 pandemic; and (d) have not been infected. Finally, a list with 25 interviewees was formed. Among them, 10 were female and 15 were male. These interviewees were aged between 18 and 69 years old. Their sojourn period varied, and they had different levels of Chinese proficiency. --- In-depth Interviews Since the outbreak of the COVID-19 pandemic, to defeat the virus as soon as possible, China has been imposing strict regulations, including wearing masks, keeping social distance, limiting big gatherings, and regulating travel. Against this background, peer-to-peer online in-depth semi-structured interviews were conducted, and each one of the 25 interviewees was interviewed for 1 h or so. The interviews began with basic questions ("How long have you been living in Beijing?; What's your profession?; Have you been received Sinopharm or Sinovac COVID-19 vaccines?), and then moved to unstructured open-ended questions about South Korean sojourners' viewpoints on the influence of Chinese institutions and China's media on their vaccination intention. During the interview, the researcher asked questions in Chinese, and interviewees could respond either in Chinese or Korean, whichever they were more comfortable with. Of course, to make sure that the two sides could understand each other well, another research team member, Dr. Liu, a university lecturer in Korean, was present in every interview. Dr. Liu majored in Korean translation, and she has been teaching Korean in China for more than 10 years, indicating that her assistance was credible. With the consent of the interviewees, every session was recorded. --- Data Analysis Upon the completion of each interview, Dr. Liu would compile the recording data into transcripts, and then, both the transcripts and original recording would be sent to a professional translation firm to proofread, thus guaranteeing the accuracy of the transcripts. Later, the proofread version would be sent back to Dr. Liu for final check and confirmation to ensure consistency between the original interview content and the transcripts. This research employed thematic content analysis to analyze all data from interviews. Thematic content analysis is a frequently used qualitative descriptive approach that requires the data analyst to immerse himself/herself into the data to see the whole picture [56,58]. Therefore, it is the perfect choice for this study which analyzes a lot of interview data and categorizes them into different themes. Based on the research questions, we conducted a thematic analysis of the trust in institutions and trust in media based on interview transcripts. In terms of trust in media, a thematic analysis of traditional media, social media and interpersonal communication was conducted. --- Results China was among the first countries that promoted nationwide vaccination, with foreign nationals included. After the COVID-19 vaccination campaign rolled out in Beijing in March 2021, different districts subsequently released notifications to inform foreign nationals of the latest vaccination information. However, despite the convenient services, all interviewees in this study reported no COVID-19 vaccination history of China-develop vaccines. Among the 25 participants, 5 who had been vaccinated received the vaccine in South Korea. Some unvaccinated interviewees mentioned that they would consider going back to South Korea to get COVID-19 vaccines if policies in China forced them to get vaccines and their normal life would be disrupted otherwise. --- Influence of Trust in Chinese Institutions Nearly 80% of interviewees said that the over-hyping of COVID-19 vaccination campaigns by the Chinese government led to their doubt about the motivation behind it. Most of the respondents (n = 22) thought that such regulations were out of the pursuit of political and economic interests. Interviewee 3, an employee in a Cafe, believed that although vaccination charges for each person were not that high, it would still be lucrative if foreigners in China all got vaccinated. Moreover, some interviewees (n = 20) worried that the Chinese government would never make foreign nationals a priority and the latter might even be vaccinated with inferior vaccines. Interviewee 10, owner of a South Korean restaurant, claimed that foreign nationals might get different vaccines from what the Chinese people got considering the fact that it was hardly possible to produce a large number of vaccines in a short period of time. Even worse, some participants (n = 16) believed that the COVID-19 vaccination campaign toward foreign nationals was a conspiracy. Interviewee 13, manager of a feed company, thought that the South Korean government's refusal of Chinese COVID-19 vaccines could perfectly explain South Korean sojourners' reluctance. Interviewee 20, sales executive of an electronic enterprise, mentioned, "The conspiracy theory has been ubiquitous since the outbreak of the COVID-19 pandemic. Honestly, I also suspect that there is some kind of conspiracy behind the mass vaccination campaigns, including the great support toward foreigners, advocated by the Chinese government." Many interviewees in this study expressed their concerns about vaccines developed by China. Therefore, they were reluctant to get Chinese COVID-19 vaccines. Some respondents (n = 20) claimed that China's vaccine R&D was too fast to be reliable. On one hand, they could not find much information on vaccines developed by China. On the other hand, they said that current vaccines had not gone through abundant clinical tests to prove their efficacy and possible side effects. Interviewee 22, a lawyer, worried that the current clinical tests could not deny future risks of the vaccination since Chinese scientists had developed those vaccines in a very short period of time without informing the public of the R&D process and ingredients of vaccines. Interviewee 17, a housewife doubted the efficacy of Chinese COVID-19 vaccines because of the sporadic outbreak of the virus now and then under the context of mass vaccination. Moreover, some interviewees (n = 18) were uncertain about the capability of Chinese medical establishment to deal with adverse reactions of the vaccination. Interviewee 21, a HSK trainer, mentioned, "Vaccination always brings adverse reactions, such as headache, fever, secondary infection, and the like. I am not sure that China's medical system is capable of solving these potential problems. China has been developing rapidly in recent years, but the fact is that China remains to be a developing country with limited medical capacity." --- Influence of Trust in China's Media Over half of the participants said that they had no trust in the information from China's media, indicating that China's media outlets failed to persuade them to get vaccinated. Many participants (n = 17) complained that the excessive promotion of the advantages of the COVID-19 vaccination by China's traditional media, such as broadcasts and TV, added to their doubts. Interviewee 8, manager of a trade company, admitted that his doubts about China-developed COVID-19 vaccines came from various news on TV where many scientists and doctors were encouraging the vaccination. Interviewee 25, teacher of an international elementary school, mentioned that the overwhelming information advocating benefits of the vaccination on TV was far from persuasive because almost all vaccines, regardless of their type and function, could cause certain side effects. Moreover, other respondents (n = 16) were suspicious about the vaccination information from China's traditional media since they believed that those traditional media outlets were regulated and even controlled by the Chinese government. Interviewee 24, a housewife, thought that all information on TV programs in China reflected the government's will. In her eyes, it was natural for Chinese people to trust the Chinese government, but South Koreans were inclined to trust their government and were suspicious about information in China's TV programs. Apart from traditional media, social media also influenced respondents' attitudes toward the vaccination of Chinadeveloped COVID-19 vaccines. Misinformation went viral on Chinese social media, adding to South Korean sojourners' sense of insecurity. Some respondents (n = 18) admitted that they did not know which vaccination information was reliable on social media since they were flooded with vast amounts of information every day. Interviewee 5, manager of a machinery company, complained that the majority of information on social media remained unchecked, making him more anxious rather than helping him understand the virus and vaccine better. Interviewee 2, a Korean translator, believed that the information on China's social media was not qualified to act as vaccination guidance. Besides, most of the interviewees (n = 22) said that China's social media platforms exaggerated the risks of not getting vaccinated, which was neither accurate nor convincing. Interviewee 1, a student said, "My Chinese fellows always forwarded me some information about the vaccination of China-developed vaccines, as well as reports of some foreign media praising China's mass vaccination campaign, but I only regarded them as the over-hyping and fake information." Lastly, the interpersonal communication between South Korean sojourners and their South Korean peers also influenced their vaccination behaviors in China. Most respondents (n = 16) claimed that the behaviors and suggestions of their peers affected their intention to get China-developed COVID-19 vaccines. According to interviewee 19, an IT engineer, he was unvaccinated because his South Korean friends in China expressed concerns about Chinese vaccines. Interviewee 15, a former employee in a supermarket, who had little contact with local Chinese and got most of the virus-related information from his South Korean friends, was unvaccinated because those friends regarded China's mass vaccination campaigns as a conspiracy. Those respondents who have been vaccinated (n = 5) all got their jabs in South Korea at the suggestion of their peers. Interviewee 11, a government representative, mentioned, "Due to the nature of my work, the unvaccination may cause considerable inconvenience. Therefore, I accepted the suggestion of my South Korean colleagues and got vaccinated back in South Korea." --- Discussion This study explored the impact of South Korean sojourners' trust in institutions and trust in media on their vaccination intention as well as concerns about China-developed vaccines during the COVID-19 pandemic through qualitative research based on in-depth semi-structured interviews. After analyzing the interview data of 25 interviewees, it was clear that South Korean sojourners' mistrust in Chinese institutions and China's media lowered their willingness to get Chinese COVID-19 vaccines. Consistent with previous studies [9,10], South Korean sojourners did have vaccine hesitancy. The World Health Organization (WHO) declared vaccine hesitancy as one of the top ten global health threats. The high vaccine hesitancy might lead to the failure of immunization efforts. The current study found that the attitude of South Korean sojourners toward China-developed COVID-19 vaccines was even worse than vaccine hesitancy: they completely refused to receive Chinese vaccines. South Korean sojourners' vaccine hesitancy related to their distrust of Chinese institutions, which verified the past research [24]. The high institutional trust was an important factor facilitating the public's COVID-19 vaccination willingness, and institutional trust had the greatest impact on the public's vaccination decision [14,19,20]. However, in this research, South Korean sojourners showed mistrust in Chinese institutions. On one hand, South Korean sojourners distrust the Chinese government. Since the beginning of the mass COVID-19 vaccination campaign, the Chinese government has been mobilizing every sector in society to improve vaccination rates by educating the public about the vaccine's safety and efficacy, disseminating information about the risk of the virus and non-vaccination, organizing vaccination programs, and reporting real-time vaccination progress, in the hope of convincing people to get vaccinated as soon as possible. Nevertheless, studies have reminded us that excessive pro-research propaganda easily leads the public to doubt the primary purpose of government-endorsed mass vaccination [27]. Most participants in this study thought that the Chinese government encouraged mass vaccination among its public and foreign nationals to pursue economic interests, and they even regarded it as a conspiracy. Conspiracy beliefs further led to South Korean sojourners' vaccine hesitancy and exacerbated their distrust of the Chinese government. On the other hand, South Korean sojourners were suspicious about the safety and efficacy of China-developed vaccines, indicating that they distrusted Chinese medical establishment and science. Those who had more trust in science were inclined to show positive attitudes toward the COVID-19 vaccination and were more willing to get vaccinated, while the distrust in science always led to vaccination hesitancy and negative attitudes toward vaccines [59,60]. Also, as other researchers have noted [21,22], the public relied on the vaccine's country of origin to evaluate its safety and effectiveness. In this study, those five participants who have been vaccinated received the vaccines in South Korea, meaning that they trusted vaccines developed by their home country more. As for those participants who had no intention to get Chinese COVID-19 vaccines, their mistrust in China's medical system and science was the main reason. In this research, China's media failed to improve South Koreans' willingness to get China-developed COVID-19 vaccines, which was somewhat different from those of previous findings [33,36]. The reason behind it was that the sojourners did not trust the information from traditional media in China. Most South Korean sojourners believed that traditional media outlets were controlled by the Chinese government and acted as the official mouthpiece. Therefore, they were suspicious about the vaccination information conveyed by China's traditional media. In addition, the current study highlighted the influence of sojourners' identity traits on their trust in traditional media in terms of the COVID-19 vaccination. Different from this finding, previous research demonstrated that as a high-credibility message source, traditional media could promote the vaccination campaign by reporting the potential risk of a certain virus [40][41][42]. In this study, however, for those South Korean sojourners who held no trust toward China's institutions, traditional media, which represented the Chinese government, was no longer a highly credible information source. The past literature has concluded that social media's influence on vaccination intention is a double-edged sword. For one thing, social media helped to increase vaccination rates during the pandemic. The employment of social media, including text messaging, smartphone applications, targeted websites and portals, was effective in disseminating information, increasing vaccination intention, and improving vaccination uptake [44][45][46]. For another, misinformation and fake news were common on social media during the pandemic, which weakened its credibility and caused vaccine hesitancy [49,50]. However, contrary to earlier studies, this study found that Chinese social media did not exert direct positive influence on South Korean sojourners' vaccination intention. Vast amounts of information on social media platforms failed to educate South Korean sojourners about the importance of the COVID-19 vaccination; on the contrary, they did not trust the information on China's social media. Moreover, many online comments and praise from major foreign news agencies on China's vaccination efforts were regarded as delicately designed propaganda. Media trust is one of the key factors influencing people's health decisions and behaviors. Existing research has found that people who had a higher education level, who were married, and who had not been vaccinated against the SARS-COV-2 showed greater trust in traditional media, while people who had a higher income and had not been vaccinated showed greater trust in social media [40,61]. However, the current study found that South Korean sojourners interviewed did not trust China's traditional media nor social media even if some of the participants had received higher education, were married, had higher incomes, and had not 1 3 been vaccinated. Interestingly, this study found that in terms of vaccination, South Korean sojourners showed more trust in interpersonal communication, which, of course, happened among South Koreans, not between South Korean sojourners and local Chinese. Most respondents perceived Chinese COVID-19 vaccines as unsafe and ineffective through communication with their peers, while those respondents who have been vaccinated all got information about the necessity of the vaccination from their peers and went back to South Korea to get their jabs. This finding conforms to the research results that interpersonal communication with families and friends could facilitate individuals' vaccination behaviors [51][52][53], but diverges from those findings that attributed the lower vaccination rate to interpersonal communication [40,54]. Notably, there is a deeper reason for South Koreans to mistrust the Chinese institutions and China's media while showing absolute trust to the interpersonal communication among their peers-South Koreans' unique national pride and their collectivism-oriented culture. Previous research pointed out that the racial minority usually showed greater mistrust to the government [25,26]: and the vaccination willingness was influenced by cultural factors [15] with South Koreans having a high vaccination rate out of their high trust in a central government [23]. The finding of this study verified and expanded these results. In South Korea, national spirit education was a key part of its education system, and adolescents were imparted with patriotism and national pride from an early age. With a strong sense of national pride, South Koreans are quite loyal to their country and the government. Since the beginning of China's vaccination campaign, South Korea has been rejecting China-developed vaccines. Then, naturally, South Korean government's disapproval of Chinese vaccines has deteriorated South Korean sojourners' trust in Chinese institutions. Additionally, South Koreans grow up in an environment advocating collectivism, which champions the belonging of an individual to a large group, such as a family, a religion, and a country, and emphasizes the importance of mutual communication and help among a group. Hence, South Korean sojourners showed great trust to their peers rather than China's media, which further lowered their willingness to get China-developed vaccines. --- Conclusions This study focused on the influence of the trust in institutions and trust in media on the willingness to get Chinese COVID-19 vaccines of South Korean sojourners living in Beijing. Through in-depth semi-structured interviews with 25 participants, this study found that South Korean sojourners hold an extremely low intention to get China-developed COVID-19 vaccines due to their mistrust in China's institutions and media. Meanwhile, South Korean sojourners' interpersonal communication with their peers also affected their vaccination willingness. Although the Chinese government encouraged foreign nationals to receive COVID-19 vaccines, South Korean sojourners showed low vaccination intention, and their choice to not get vaccinated in China added a burden to China's pandemic prevention and control. Therefore, it is of great importance for the Chinese government and media to deal with the vaccination hesitancy of South Korean sojourners, which should also be the emphasis of relevant research in the future. --- Limitations This study has the following limitations. First, the current study was conducted during a certain period of the pandemic, and it was possible that South Korean sojourners' attitudes toward Chinese COVID-19 vaccines could change as the pandemic evolved. Second, this study focused on South Korean sojourners in Beijing, while those living in other cities in China were not covered, indicating that South Korean sojourners in other cities may have different levels of vaccination intention and concerns. Third, there is no discussion on specific measures that the Chinese government and media should take to enhance South Korean sojourners' trust, which can be a starting point for later studies. 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/. --- Data Availability Data are not publicly available because the information may compromise the privacy of research interviewees. --- Author Contributions Conceptualization, YL methodology, YL investigation, YL, XL resources, YL data curation, YL writing-original draft preparation, YL writing-review and editing, YL, XL. Funding None. --- Declarations --- Conflict of Interest The authors declare no conflict of interests. Ethics Approval Ethical review and approval were not applicable for this study, due to no clinical trials or medical practice involved. --- Consent for Publication Not applicable. 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(s) 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

People's willingness to get vaccinated determines whether the campaigns against the COVID-19 pandemic can be successful in part. Considering the fact that both foreigners and its nationals are exposed to the risk of infection in China, the Chinese government has taken measures favorable to foreigners in terms of the vaccination, yet South Korean sojourners were reluctant to get China-developed COVID-19 vaccines. This study employed the trust in institutions and trust in media as a theoretical framework and seeks to analyze how these two affect South Korean sojourners' intention to get Chinese COVID-19 vaccines. 25 South Korean sojourners living in Beijing participated in semi-structured interviews. The results showed that the mistrust South Korean sojourners have in China's institutions and media, both traditional and social media, led to their reluctance to get Chinese COVID-19 vaccines. In addition, South Korean sojourners' higher interpersonal trust in their peers also influenced their willingness to get vaccines. This study further interpreted such results from the perspective of cultural traits and national properties.

INTRODUCTION Japan is the fastest ageing society on earth and the first large country in the history to have its population start shrinking rapidly from natural causes. The life expectancy of Japanese people (mean age: men, 79.4 years; women, 85.9 years) is at the highest level in the world. The population of Japan, which currently stands at 127 million, is expected to fall to just under 100 million in the next 40 years. By 2050, 4 of 10 adults in Japan will be older than 65 years of age. Japan implemented the national social long-term care insurance (LTCI) system on 1 April 2000. Every Japanese person aged 65 and older is eligible for benefits based strictly on physical and mental frailty or disability. 1 In June 2006, the Japanese government implemented a major LTCI reform that focused on preventive benefits for the --- ARTICLE SUMMARY Article focus <unk> Measures of physical performance may identify older persons with a preclinical stage of disability. <unk> However, it is unclear which performance test and cut-point are the most useful to screen for risk of functional dependence in older Japanese people. <unk> The purpose of this study was to identify appropriate clinical tests for determining the risk of functional dependence in older Japanese people. --- Key messages <unk> Clinical tests of physical performance were associated with a functional decline in older people. <unk> Preventive strategies to avoid personal care may be required in older adults who show a gait speed slower than 1 m/s. --- Strengths and limitations of this study <unk> Strengths of this study include a large sample size and performance-based assessment, which could determine actual physical capacity and predict subsequent physical disability in older people living in the community. <unk> We analysed cross-sectional data. Therefore, further investigation of the validity of these tests in predicting the risk of disability in older people using a prospective study design is recommended. population at high risk of disability (ie, physical and/or cognitive frailty), to contain the skyrocketing costs of the LTCI. 2 Physical frailty increases with advancing age and is a major risk factor for dependency, institutionalisation, and mortality. [3][4][5] People with a disability have higher healthcare needs and use compared with those without a disability. 6 Although the biggest risk factor for future frailty is advancing age, other factors that are possibly modifiable through interventions should not be ignored. For the purpose of targeting risk factors for future frailty, adequate assessment of individual people may be required. One of the main characteristics of the elderly population is its heterogeneity, with elderly people in the same age range showing a wide variance with regard to their risk of disability. To prevent frailty or disability, population-based intervention programmes should be targeted at the population at risk. A feasible and valid screening tool available for research and clinical settings is required to identify target populations. The Interventions on the Frailty Working Group developed recommendations to screen, recruit, evaluate and retain frail older persons in clinical trials. 7 They reported that most researchers focus on the following domains for identification of physical frailty: mobility, such as lower extremity performance and gait abnormalities; muscle weakness; poor exercise tolerance; unstable balance and factors related to body composition, such as weight loss, malnutrition and muscle loss. 7 In an effort to select tailored preventive programmes in the Japanese LTCI system, those at high risk for subsequent disability are identified by a basic functional status questionnaire. Although the questionnaire is relatively quick to administer, a performance-based assessment could determine actual physical capacity and more accurately predict subsequent physical disability in communityliving older people. Guralnik et al 8 reported that measures of physical performance may identify older persons with a preclinical stage of disability who may benefit from interventions to prevent the development of frank disability. A previous study identified that a rapid gait test was more likely than other mobility performance tests to discriminate older women at high risk of frailty based on the Japanese LTCI system. 9 However, which performance tests including upper and lower limb muscle functions and which cut-points are the most useful to screen for the demand for personal care are not clear. This study investigated the relationships between performance-based physical assessments and demand for personal care in older people using two large sample cohorts in Japan. --- METHODS --- Participants We performed a national study of 10 351 individuals aged 65 years and older who had received personal care (n=6791) and those who had not received personal care (n=3560). The study included individuals who were enrolled in the Obu Study of Health Promotion for the Elderly (OSHPE) and the Tsukui Ordered Useful Care for Health (TOUCH) programme. To enrol in the OSHPE, an individual was recruited from Obu, Japan, which is a residential suburb of Nagoya. Inclusion criteria required that the participant was aged 65 years or older at examination in 2011 or 2012, lived in Obu, and had not participated in another study. Exclusion criteria stipulated that participants be certified as needing support or care by the Japanese public LTCI system, had disability in basic activities of daily living, and could not carry out performance-based assessments. To enrol in the TOUCH programme, an individual had to be 65 years or older and certified as needing support or care from the Japanese public LTCI system. Detailed information was provided in a previous study. 10 In brief, TOUCH sites (241 day-care centres) are located throughout Japan and provide comprehensive, facilitybased day-care services (eg, bath, lunch, physical and cognitive recreational activities and physical exercise). Most TOUCH clients have some physical disability and frailty, defined as the presence of weakness, low physical activity and/or slow gait speed, in accordance with the widely accepted definition of frailty. 7 A total of 10 351 older participants (mean age, 78.8<unk>8.0 years) underwent performance-based assessments. Informed consent was obtained from all participants prior to their inclusion in the study, and the Ethics Committee of the National Centre for Geriatrics and Gerontology approved the study protocol. --- Performance-based assessment The assessment measures were conducted by welltrained staff who had nursing, physiotherapy, occupational therapy or similar qualifications. Prior to start of the study, all staff received training from the authors in the correct protocols for administering all of the assessment measures. The assessment included several physical tests. Upper and lower limb muscle functions were assessed with the grip strength (GS) and the chair stand test (CST), respectively. 11 Gait function was assessed with walking time tests conducted at a comfortable pace (comfortable walking speed, CWS) and with the timed up-and-go (TUG) test. 12 GS was measured in kilograms in the participant's dominant hand using a Smedley-type handheld dynamometer (GRIP-D; Takei Ltd, Niigata, Japan). The CST involved sitting down and standing up five times, using a chair without an armrest. The score was the time taken to complete the task in seconds. Participants were asked to exert their maximum effort in GS and CST. CWS was measured in seconds with a stopwatch. Participants were asked to walk on a flat and straight surface at their CWS. Two markers were used to indicate the start and end of the path, and a 2 m and over approach was allowed before reaching the start marker so that participants can walk at their comfortable pace within the timed path. They were instructed to continue walking past the end of the path for a further 2 m and over to ensure that the walking pace was consistent throughout the task. The TUG test involved rising from a chair, walking 3 m, turning around, walking back to the chair and sitting down. 12 The TUG test is one of the most frequently used tests of balance and gait, and is often used to assess fall risk in older people. 13 The time to complete the TUG test was measured, in seconds, at each participant's usual pace. Both walking tests were measured once, and if a walking aid was nomally used inside the home, this aid was used during the tests. --- Cognitive function The Mini-Mental State Examination (MMSE) 14 for the OSHPE population and the Mental Status Questionnaire (MSQ) for individuals enrolled in the TOUCH programme were used to measure cognitive functioning, and were used as potential confounders in the association between performance-based physical assessments and functional dependence. 15 Individuals with 23 or fewer points on the MMSE and three or more errors on the MSQ were considered to have cognitive impairment. 15 16 Statistical analysis Demographic and clinical variables were compared between the participants with and those without personal care using Student t tests for continuous variables and <unk> 2 tests for categorical variables. To compare the predictive ability of the study measures, receiver-operated characteristic (ROC) curves were inspected to determine cut-points for each test that best discriminated between the individuals with and those without personal care. Cut-points for maximising the sensitivity and specificity for each test were determined using the Youden index. 17 The area under the curve (AUC), sensitivity and specificity were then calculated for the cut-points. We used multivariate logistic regression analyses to determine ORs and 95% CIs, and to assess independent associations of the cut-points of physical performance measures for demand for personal care. The participants were divided into two groups according to the cut-point of the performance-based physical assessments. Covariates were added sequentially to the logistic model to evaluate the associations at different levels of adjustment. Model 1 included each performance-based physical assessment, and model 2 included the model 1 variables plus age, sex and cognitive impairment as determined by the MMSE or MSQ. Model 3 included all performance-based physical assessments plus age, sex and cognitive impairment. The participants were then divided into five groups as follows: individuals with no risk and those with 1, 2, 3 or 4 risks, according to the number of risks identified by the cut-points of the performance-based physical assessments. The ORs and 95% CIs for the number of risks were calculated adjusted for age, sex and cognitive impairment. All statistical contrasts were made at the 0.05 level of significance, and all data management and statistical computations were performed using the IBM SPSS Statistics V.19.0 software package (SPSS Inc, Chicago, Illinois, USA). --- RESULTS --- Comparison between participants with and those without personal care Table 1 shows the characteristics of the participants. The participants with personal care were significantly older ( p<unk>0.001), included a higher number of women ( p<unk>0.001) and a higher number of persons with cognitive impairment ( p<unk>0.001) than those without personal care. For the comparison of performance-based assessments, the participants with personal care had significantly lower scores on all physical tests ( p<unk>0.001) compared Cut-points between participants with and those without personal care ROC curve analysis results, showing the performance cut-points for each test and associated statistics, are shown in table 2. The Youden index determined the cutpoints for the demand for personal care as follows: GS in men and women was <unk>26 and <unk>17 kg, respectively; CST was <unk>10 s, CWS was <unk>1.0 m/s and TUG was <unk>11 s for both sexes. The CWS score had the highest AUC for discriminating the demand for personal care and displayed good sensitivity and specificity (85-90%). High AUCs were also found for GS and TUG, as well as fair to good sensitivity and specificity (74-80%). --- Relationships between cut-points and risk of disability The multiple logistic regression models showed significant relationships between physical performances and the demand for personal care (table 3). The demand for personal care was most closely related to CWS in model 1 (OR=34.7; 95% CI 30.9 to 39.0). These results remained essentially unchanged after controlling for age, sex, cognitive impairment and other physical performance tests. In the final model (model 3), the highest OR of factors related to the demand for personal care was for CWS (OR=5.9; 95% CI 5.0 to 6.9). Figure 1 shows the distribution of CWS for participants with personal care. Participants who walked 1.1 m/s and faster had the lowest amount of personal care (20%). The rate of participants with personal care increased rapidly with a CWS slower than 1.1 m/s, and 90% of participants with a CWS slower than 0.8 m/s had personal care (figure 1A). The rate of functional decline increased rapidly for individuals walking slower than 1 m/s in women (figure 1C) rather than men (figure 1B), and with the rate of functional decline reaching 90% when CWS was slower than 0.8 m/s in both sexes (figure 1B,C). There was a significant relationship between the number of risks based on the physical performance tests and the demand for personal care. The ORs and 95% CIs for personal care in participants with 1, 2, 3 and 4 risks were 3.1 (2.6 to 3.8), 10.6 (8.7 to 13.1), 35.6 (28.6 to 44.5) and 141.3 (103.6 to 192.7), respectively, compared with participants without risks ( p<unk>0.001). Figure 2 shows the distributions of the number of risks for demand for personal care. The rates of participants with personal care who had no risk, 1, 2 and 3 or more risks were 8.7%, 38.5%, 75.6% and 90.0%, respectively (figure 2). --- DISCUSSION Neuromuscular function, including muscle strength, balance and gait, and cognitive function are important risk factors for disability. Performance-based assessment of these factors can be used to identify people at an increased risk of future functional decline. We examined the use of various measures to identify the most useful measure for screening the demand for personal care. --- Cut-points of demand for personal care In the current study, univariate analyses identified all physical tests as being able to discriminate between participants with and those without personal care. When performance was dichotomised for cut-points, GS, CST, CWS and TUG retained statistically significant relationships with personal care. The CWS test (cut-point, 1 m/s) displayed the highest OR in the final model, with good sensitivity and specificity with respect to identifying participants with personal care. At identified cut-points, GS (men, 26 kg; women, 17 kg), CST (10 s) and TUG (11 s) could also significantly discriminate participants with personal care with sensitivities and specificities of 67-89%. This result highlights what can occur when dichotomised rather than continuous data are used. There is an associated loss of information and reduced predictive accuracy as a trade-off for ease of scoring and test interpretation. These results, however, are consistent with previous findings that showed associations between measures of muscle strength and mobility and functional decline. 18 Gait speed and personal care Gait velocity, as measured by the CWS test in this study, has been consistently reported to differentiate between participants with and those without personal care, with frail older persons walking significantly slower, 10 19 and has proved to be a strong predictor of adverse events, such as disability, 18 20-25 mortality, 21 22 26 27 hospitalisation 21 22 24 28 and falls. 28 29 Gait slowing, which occurs in the latest stages of life, suggests that mobility is so central to life that energy is shifted away from walking activity only when other vital activities are threatened, 30 which may lead to increased functional independence. In addition, a slower walking speed is an associated factor for subsequent dementia. 31 Dementia is one of the most important factors of health problems for functional decline in the aged population. For our study sample, the cut-point for CWS was 1 m/s, which is the critical point for future functional decline in community-dwelling older people determined by previous studies. 18 21 22 24 25 These results suggest that walking speed may be the most crucial measurement to determine the demand for personal care in older adults. Measurement of walking speed is reliable, valid, sensitive, inexpensive, safe, quick and a simple tool. Therefore, measurement of walking speed is suitable to use in community settings as a screening tool and evaluation for the effect of a care prevention programme. --- Muscle strength and mobility and personal care In the current study, higher ORs were found for GS and TUG, as well as CWS. Hand GS is an estimate of isometric strength in the upper extremity, but also correlates with strength in other muscle groups, 32 and therefore, is considered an estimate of the overall strength. In addition, GS has proved to be a strong predictor of physical functioning and disability, 33 34 morbidity 35 and mortality. 36 37 Our findings support previous evidence and add cut-points of <unk>26 kg in men and <unk>17 kg in women that discriminate those at high-risk for disability in community-living older people. The TUG has been recommended as a screening tool for identifying older people who are at risk for falling. 38 39 Bischoff et al 40 proposed a normative cut-point of 12 s for community-dwelling elderly people between 65 and 85 years of age. In daily clinical practice, elderly persons who perform the TUG in >12 s should receive early evaluation and intervention. Our results regarding TUG cut-points are in line with these previous studies. --- Strengths and limitations Strengths of the present study include a large sample size and we used performance-based assessment, which could determine actual physical capacity and predict subsequent physical disability in community-living older people. However, the present study has a number of limitations. One of the limitations is that we analysed crosssectional data. Therefore, further investigation of the validity of these tests in predicting the risk of disability in older people using a prospective study design is recommended. Another limitation is that many frail older people using healthcare services cannot walk because they have multiple diseases or geriatric syndromes. Non-ambulatory participants were excluded from our study. Therefore, we acknowledge that the study findings may not be generalised to this frailer group. --- CONCLUSIONS This study provides preliminary evidence that clinical tests of physical performances can predict the risk of disability in older people. Logistic regression analysis selected CWS as the best independent correlate of disability, with good sensitivity and specificity. Further investigation is required, and future research should include a prospective measurement of the risk of disability to more accurately determine the validity of screening tests for this population. --- Contributors HS designed and organised the study, analysed and interpreted the data, and drafted the manuscript. TS and MS made substantial contributions to the conception, design, analysis and interpretation of the data, and critically revised the draft. All authors took responsibility for the accuracy and integrity of the study. All authors gave the final approval of the version to be published. Funding This work received financial support from a grant from the Japanese Ministry of Health, Labour and Welfare (Project for optimising long-term care; B-3) and a grant from the National Center for Geriatrics and Gerontology (Research Funding for Longevity Sciences; 22-16). The funding source played no role in the design or conduct of the study; collection, management, analysis or interpretation of the data; or preparation, review or approval of the manuscript. --- Competing interests None. Patient consent Obtained. Ethics approval Ethics Committee of the National Center for Gerontology and Geriatrics. Provenance and peer review Not commissioned; externally peer reviewed. Data sharing statement No additional data are available.

Objectives: To identify appropriate clinical tests for determining the demand for personal care in older Japanese people.

Introduction Recent meta-analyses have suggested that psychosocial work characteristics, such as job insecurity [1] and long working hours [2], are associated with a moderately higher risk of diabetes, while the health effects of highly adverse social work stressors, such as bullying and violence at work, are far less well documented. The prevalence of workplace bullying, defined as persistent, repeated harassing, offending and socially excluding behaviours of psychological nature over a long period [3], ranges from 5% to 24% depending on the country [4]. The prevalence of violence and threats of violence also vary between countries, but are generally more common in occupations with client contact; for example, 19.9% of nurses in the USA have reported exposure to violence and threats of violence at work [5]. Bullying and violence can adversely affect personal resources, such as self-esteem and coping capacity [3]. They have also been linked with an increased risk of chronic conditions, including type 2 diabetes [6], which is characterised by insulin resistance in liver and muscle and progressive beta cell failure [7]. Induced negative emotions, such as depression and anxiety [8,9], may contribute to diabetes risk [10] through prolonged activation of the hypothalamic-pituitary-adrenal axis and sympathetic nervous system, or indirectly through impaired sleep, for example [11]. Furthermore, stress-related coping strategies, such as comfort eating behaviour with an increased preference for energy and nutrient dense foods [12], may result in weight gain or an increase in waist circumference [13], which are both pivotal risk factors for diabetes [14]. However, it is unknown whether bullying or violence at work affects diabetes risk in the general population. To the best of our knowledge, the only existing study on negative interpersonal relations and diabetes was a cross-sectional study of 8499 male and 9025 female employees in the USA in 2015. It postulated a higher prevalence of type 2 diabetes among employees exposed to workplace bullying, threats or other kinds of harassment compared with those who were not exposed to these social stressors [15]. However, due to the cross-sectional study design, the temporality of the association is unclear. Therefore, the aim of the present study was to assess the prospective relationship between bullying and violence at work and the risk of incident type 2 diabetes using longitudinal individual-level data from four large Nordic cohort studies involving more than 40,000 participants. --- Methods --- Study baseline The study population was derived from four prospective cohort studies including the Swedish Work Environment Survey (SWES) [16], the Swedish Longitudinal Occupational Survey of Health (SLOSH) [17], the Finnish Public Sector Study (FPS) [18] and the Danish Work Environment Cohort Study (DWECS) [19]. For more details on the individual cohorts, please see the electronic supplementary material (ESM Study populations). Ethical approval was obtained from the Regional Ethical Review Board in Stockholm for SWES and SLOSH, and the Ethics Committee of the Hospital District of Helsinki and Uusimaa for FPS. In Denmark, questionnaire-and register-based studies do not require ethics committee approval. DWECS was approved by and registered with the Danish Data Protection Agency. Our baseline sample was restricted to participants with information on either workplace bullying or violence at baseline (Fig. 1). We only included those who were employed and aged between 40 and 65 years. Participants younger than 40 years of age were excluded in order to minimise outcome misclassification of type 1 diabetes and polycystic ovary syndrome [20]. Those who were previously diagnosed with diabetes or had previously used glucose-lowering medications or insulin were excluded. Our final sample included 19,280 men and 26,625 women. --- Workplace bullying and violence Being bullied or targeted by violent actions or threats of violence was measured using questionnaires with similar wording in all cohorts. We defined workplace bullying as reporting having been bullied at the workplace at least once during the past 12 months (in SWES, SLOSH and DWECS). In FPS, the time frame was slightly different as participants were asked whether they were currently being bullied. Workplace violence was measured as the experience of having been the target of violent actions or threats of violence in the past 12 months at the workplace (in SWES, SLOSH and DWECS). Violence was not measured in FPS; therefore, this cohort was not included in the analysis of violence. Participants were further identified as 'frequently exposed' if the bullying/violence occurred at least once a week. Detailed information on measurements and definitions of workplace bullying and violence can be found in ESM Table 1. Type 2 diabetes Using unique personal identification numbers in Sweden, Finland and Denmark, all participants were linked to nationwide health registers. We used available information for each country and at different historical time points to capture all incidences of diabetes. Type 2 diabetes was identified with codes ICD-8/9 250 and ICD-10 E11 (www.who.int/classifications/icd/en/) from inpatient registers only (in SWES95-01 and FPS) or both inpatient and outpatient registers (in SWES07, SLOSH and DWECS) and combined with information from death registers (in all cohorts). This was supplemented with information on prescription medication using Anatomical Therapeutic Chemical codes A10A, A10B and A10X (in SWES07, SLOSH and FPS). In DWECS, individuals were identified in the Danish Diabetes Register, which combines information from the national patient register on the use of insulin or oral glucose-lowering drugs, registration for chiropody for treatment of diabetesrelated complications and individuals with more than five blood glucose measurements within the period of a year [21]. For the individuals receiving insulin treatment, this register includes both type 1 and type 2 diabetes. However, as all participants were free from diabetes at baseline at an age of at least 40 years, type 2 diabetes probably represents the majority of these individuals. Other variables Potential confounders were identified based on prior knowledge and using directed acyclic graphs [22]. In addition to age and sex, potential confounders included country of birth, educational level and marital status. Information on educational level was obtained from the social registers in each country and was categorised as <unk>9 years, 10-12 years and <unk>13 years. Marital status, as a proxy for social support outside work, was also obtained from the population registers (in SWES, SLOSH and DWECS) or self-reported questionnaires (in FPS). It was categorised as unmarried, married/cohabiting, divorced/separated or widowed. Country of birth was selfreported and classified as 'Nordic countries', 'other European countries' and 'other continents' (in SWES, SLOSH and DWECS). Country of birth was not measured in FPS, but the vast majority of hospital employees in the cohort are from Nordic countries. We assumed that mental illness, excessive alcohol consumption and obesity would be on the causal pathway from bullying or violence to type 2 diabetes (and thus should not be controlled for). However, mental illness, excessive alcohol consumption and obesity may also be causes of workplace bullying and violence, and thus confounders. Therefore, we chose to include adjustment for BMI, alcohol consumption and mental illness in sensitivity analyses. BMI (in SLOSH, DWECS and FPS) was calculated using self-reported height and weight, grouping according to the WHO categories: underweight (<unk>18.5 kg/m 2 ), normal (18.5-24.9 kg/m 2 ), overweight (25-29.9 kg/m 2 ) and obese (<unk>30 kg/m 2 ). Information for alcohol consumption was available in FPS, DWECS and SLOSH. Alcohol consumption was divided into 'risky' or 'non-risky' based on exceeding or not exceeding 16/24 (in FPS) or 14/21 (in DWECS and SLOSH) alcohol units (12 g of alcohol per unit) per week for women/men, or weekly consumption of six or more units per occasion (in SLOSH). Mental illness was identified using only inpatient registers (in SWES95-01 and FPS) or both inpatient and outpatient registers (in SWES07, SLOSH and DWECS), and dichotomised into having at least one mental illness problem and having no mental illness. Furthermore, workplace violence is very likely to be clustered in occupations with frequent client contact. Thus, we separated personal and protective service workers, healthcare professionals, social work professionals and teaching professionals from other occupations using the current Swedish and Danish adapted version of the International Standard Classification of Occupations (ISCO-88) following suggestions from Madsen et al [23]. Statistical analysis Data from the individual cohorts were analysed separately. The cohort-specific results were later combined using meta-analyses. Two datasets with slightly different numbers of participants were created for each cohort for the analyses of bullying and violence, excluding participants with information missing for any of the covariates (Fig. 1). For the sensitivity analyses including SLOSH, FPS and DWECS, participants with missing information for BMI (in the BMIadjusted analyses) or both BMI and alcohol consumption (in the BMI-and alcohol-adjusted analyses) were excluded. For the main analyses we applied a marginal structural Cox model estimated by using inverse probability (IP) weights [24]. This approach is based on a counterfactual framework. Given the properly identified confounders, the method provides an estimate of the marginal HR. This corresponds to comparing the risk of diabetes in a pseudo population where everyone is bullied with the same population where everyone is not bullied. Our main analyses were conducted in three steps. In step one, the stabilised IP weight was obtained for each individual included by fitting a logistic model for the conditional probability of being exposed based on relevant confounders in each analysis [24]. In this step, the positivity assumption was verified in all of the analyses. Step two was to fit weighted Cox proportional hazard models using age as the underlying timescale. The IP weights varied depending on whether the Cox proportional hazard models were ageand sex-adjusted or fully adjusted. The proportional hazards assumption was tested using log (-log(survival)) curves. If a result was doubtable, a stratified Cox model was performed to confirm the violation. In our study, none of the violations were of major concern. In step three, the robust confidence interval was calculated using standard errors generated from bootstrapping steps one and two a total of 500 times. Sensitivity analyses based on specific IP weights suitable for each analysis were conducted on: (1) sex-stratified analyses; (2) analyses adjusting for BMI; (3) analyses adjusting for both BMI and alcohol consumption; (4) analyses adjusting for mental illness, in addition to the main adjustments; (5) analyses only including the first 4 years of follow-up in order to test whether the effect was dependent on length of follow-up; (6) analyses considering a one year washout period to address the possibility of reversed causality; (7) analyses based on different definitions of diabetes (inpatient plus death vs inpatient plus medication plus death vs inpatient plus outpatient plus death); (8) analyses to test a potential dose-response relationship between frequency of exposure to bullying/violence and risk of type 2 diabetes in cohorts with available information on frequency of exposure (in SLOSH and SWES); and (9) further stratified analysis for violence towards individuals in occupations with frequent client contact (in SLOSH, SWES and DWECS). To adjust for the small number of studies included, the risk estimates from each cohort were combined in the fixed-effect meta-analyses [25]. The I 2 statistic was used to test for heterogeneity between the study-specific estimates. All tests of statistical significance were two-sided and the significance level was set at 0.05 using SAS version 9.4 (SAS Institute, Cary, NC, USA) and R package'meta' version 4.8-2. --- Results Baseline characteristics Nine per cent of the participants reported exposure to bullying at work (Table 1). The prevalence of bullying varied between studies, with the highest prevalence observed in SLOSH (13%). The combined analysis on workplace bullying included 45,647 participants (26,396 women and 19,251 men). Twelve per cent of the participants had experienced violence or threats of violence at work within the past 12 months (Table 1). This proportion varied between the cohorts, with the highest prevalence reported in SLOSH (17%) and the lowest in DWECS (7%). There was no information on workplace violence reported for FPS. Therefore, the analyses on workplace violence were based on the 39,574 participants from the remaining three cohorts, including 21,023 women and 18,551 men. The prevalence of violence varied greatly between different occupations. The highest prevalence of violence or threats of violence were found among occupations with frequent client contact, including social work professionals (>46%), personal and protective service workers (>29%), healthcare professionals (>25%) and teaching professionals (>16%). Only 2-4% of participants reported exposure to both workplace bullying and violence, and there was very little statistical agreement between the two measures across all the included cohorts (Cohen's <unk> <unk> 0.20). Workplace bullying and type 2 diabetes During a mean follow-up of 11.7 years, we identified 1223 incident cases of type 2 diabetes. After adjustment, being bullied at work was associated with a higher risk of type 2 diabetes (HR 1.46 [95% CI 1.23, 1.74]; Fig. 2). The risk estimates varied across the studies but, apart from a very unstable estimate from SWES07 (based on only 131 incident type 2 diabetes events), they were all in the same direction and there was no indication of heterogeneity SWES07 [17] SLOSH [18] FPS [19] DWECS [20] Summary Total (n) Events (n) also did not change the risk estimate (ESM Fig. 2). Further, when limiting our analysis to the first 4 years of follow-up the risk estimate remained similar to that of the main analysis (HR 1.37 [95% CI 0.99, 1.88]). In addition, a one year washout period or differences in case ascertainment across the studies did not change the risk estimate (Fig. 2). The two Swedish cohorts (SWES and SLOSH) provided information on the frequency of bullying. Of those who were bullied, 10% reported being bullied frequently (i.e. at least once a week) and 90% reported being bullied occasionally. Compared with those who had not experienced bullying in the past 12 months, those who had experienced bullying occasionally had a higher risk of type 2 diabetes (HR 1.57 [95% CI 1.23, 1.99]), whereas those who experienced bullying frequently did not have a clear excess risk of type 2 diabetes (HR 1.24 [95% CI 0.58, 2.64]) (based on only seven incident cases of type 2 diabetes in the frequently bullied group). Workplace violence and type 2 diabetes During a mean follow-up of 11.4 years, we identified 930 incident cases of type 2 diabetes in the three cohorts used for the analyses on violence. After adjusting for age, sex, educational level, marital status and country of birth, workplace violence or threats of violence were associated with a higher risk of type 2 diabetes (HR 1.26 [95% CI 1.02, 1.56]) (Fig. 3). The risk estimates varied slightly between the different studies but there was no indication of heterogeneity across the studies (I 2 <unk> 0.1%) (Fig. 3). In addition, no obvious differences between men and women were revealed. Additional adjustment for BMI slightly changed the risk estimate from 1.33 (95% CI 1.00, 1.78) to 1.27 (95% CI 0.96, 1.70). Further adjustments for alcohol consumption and mental illness did not affect the risk estimate (ESM Figs 1,2). Moreover, the risk estimate was not influenced by the follow-up length (Fig. 3). The test for different outcome ascertainment showed no difference using various case ascertainment methods (Fig. 3). In the analysis of the dose-response relationship (based on SLOSH and SWES), among those experiencing violence at work, 17% were targeted by workplace violence frequently (at least once a week; 12 incident cases of type 2 diabetes in this group) and 83% were occasionally exposed (68 incident cases of type 2 diabetes in this group). Neither occasional (HR 1.18 [95% CI 0.94, 1.50]) nor frequent exposure to violence (HR 0.95 [95% CI 0.52, 1.76]) was clearly associated with an increased risk of type 2 diabetes. In addition, we observed a similar excess risk for occupations with (HR 1.20 [95% CI 0.90, 1.60]) and without frequent client contact (HR 1.36 [95% CI 1.03, 1.81]). --- Discussion In this large multinational, multicohort study, approximately one in ten employees reported being exposed to bullying or violence/threats of violence at work. Both men and women who were exposed to these severe social stressors were at a higher risk of developing type 2 diabetes. The higher risk was consistent across cohorts and independent of follow-up length or the method of case ascertainment. When we adjusted for BMI, the associations were attenuated but remained statistically significant in the case of bullying and suggestive in the case of violence. These findings add to the previous crosssectional study by Khubchandani et al that reported a risk estimate (OR 1.48 [95% CI 1.03, 2.15]) that is similar to our findings but is based on a less specific definition of bullying that included being harassed, threatened or bullied any time in the past 12 months [15]. We did not find a dose-response relationship trend for workplace bullying. The point estimate for those being frequently bullied was lower than those being occasionally bullied. This may be due to a very limited number of incident diabetes events in the frequently bullied group generating a very wide confidence interval. However, we did not find a dose-response relationship for workplace violence either. Occupations with frequent client contact can be a proxy of frequent workplace violence [23]; however, the point estimate was similar between occupations with frequent client contact and those without. A possible explanation for the lack of a dose-response relationship here could be a protective effect from general expectations and/or training prior to workplace violence among employees in occupations with frequent client contact [26]. It may also be attributable to cognitive adaption including finding meaning, enhancing self and gaining mastery [27] following frequent exposure. Bullying vs violence Our results suggest that while both bullying and violence represent negative interpersonal relationships they most probably constitute different concepts, with only 2-4% of participants reporting being exposed to both, and very poor statistical agreement indicating that bullying and violence are two distinct social stressors. Harassment and bullying refer to psychological aggression, including behaviours such as unfair criticisms, humiliating work tasks, isolation, ignorance and spreading rumours [3]. Violence or threats of violence on the other hand are more likely to be understood as physical violence or verbal threats relating to physical violence [28], and the actions included can be exemplified as pushing, kicking and screaming [29]. There can be situations where behaviours displaying bullying and violent characteristics overlap, especially when the negative behaviour of concern has been observed to persist and be repeated over a long period [3]. However, in most situations, bullying at work is often characterised by negative behaviours from colleagues and supervisors, sometimes also from clients [30], whereas the overwhelming proportion of violence at work is derived from clients, students, customers, patients, etc [29]. Hence, workplace bullying and workplace violence seem to be distinct behaviours and, consequently, their induced emotions can be different. However, the associated behavioural appraisals and physiological reactions may be similar and may explain the comparable associations with risk of diabetes observed in the present study. Plausible pathways Being bullied is regarded as a severe social stressor that may activate the stress response system and lead to a range of downstream biological processes that may contribute towards the risk of diabetes [31]. In agreement with this, bullying at work has, for example, been found to be related to a higher level of saliva dehydroepiandrosterone sulphate [32], although no increase in saliva cortisol has been documented [33] and there is no clear finding from longitudinal studies on stress biomarkers [34]. Nevertheless, these hormones may work together in affecting cellular activities and metabolic, cardiovascular and immune variables [15,35]. Metabolic changes and obesity are also possible mechanisms underlying the observed higher risk of type 2 diabetes associated with both bullying and violence, as stress responses may be related to the endocrine regulation of appetite [36]. In the present study, the relationships between bullying and violence and type 2 diabetes attenuated after adjustment for baseline BMI, which can either be due to the fact that obese employees are more likely to be targets for workplace bullying or violence or that exposed employees are more likely to gain weight and become obese. The first explanation (i.e. that obesity leads to bullying) was, however, not supported in a sensitivity analysis presented in a previous paper on bullying and cardiovascular disease based on data from the FPS study, where the authors did not find an association between baseline BMI and incident workplace bullying [37]. On the other hand, it is likely that both workplace bullying and violence can induce comfort eating behaviour [12] or increase the risk of experiencing negative emotions [8,26,38], and further contribute to weight gain and subsequent development of type 2 diabetes, making the causal pathway very plausible. --- Methodological considerations The exposure measurements differed slightly between the studies, especially the measurements from FPS, as in this study bullying is defined as Study SWES95-01 [17] SWES07 [17] SLOSH [18] DWECS [20] Summary Total (n) Events (n) Fig. 3 (a) Main analysis using a fixed-effect model on the association of workplace violence with type 2 diabetes, after adjustment for age, sex, educational level, marital status and country of birth. (b) Sensitivity analysis using a fixedeffect model by sex, adjustments (based on SLOSH and DWECS), follow-up time (based on SLOSH and SWES07) and case ascertainment 'currently being bullied' rather than bullying experienced in the past 12 months. Furthermore, workplace bullying was self-reported without providing a definition, which did not reflect the formal definition of persistent or repeated events [3]. According to a meta-analysis on measurement of workplace bullying in Nordic countries [4], the provision of a definition did not impact prevalence, presumably because the concept of bullying at schools and workplaces is well established. However, in this study, given the subjectivity of solely using a self-reporting method, it is possible that our results are, to some extent, affected by exposure misclassification. Furthermore, workplace bullying and violence were only measured at baseline, ignoring the possibility of changes in exposure status over time [39], which may have diluted our results. The incidence of type 2 diabetes was smaller in SWES95-01 compared with the other cohorts. This difference may be mainly ascribed to the fact that SWES95-01 could not be linked to the medication register. However, a meta-analysis is a robust tool with which to incorporate such variations. It is reassuring that although the association in FPS is possibly diluted by using a narrower exposure window it showed a similar risk estimate and direction as the other cohorts. Type 2 diabetes was ascertained differently across countries and at different historical time points, leading to some degree of misclassification. We chose to use the most comprehensive definition in each study to reduce outcome misclassification, which came at the cost of direct comparability in incidence rates across studies. However, when standardising case ascertainment across the studies in the sensitivity analyses, we found no obvious heterogeneity for the risk estimates dependent on case ascertainment method, suggesting that such misclassification is not a major source of bias. In addition, we have considered several important confounders. However, in observational settings, unmeasured confounders are unavoidable, e.g. personality and genetic factors. Thus, our results should be interpreted with caution. To the best of our knowledge, no previous longitudinal studies have addressed the relationships between workplace bullying and workplace violence and type 2 diabetes. All the analyses were done under the counterfactual framework, ensuring a more straightforward interpretation of the results. Further, we applied the best available outcome measurement, linking to nationwide registries to allow for a nearly complete follow-up and to minimise misclassification of other diseases. Our large sample size and long follow-up period provided sufficient statistical power to assess total and sex-specific effects, and to conduct relevant sensitivity analyses adjusted for different variables and case ascertainment. In conclusion, we have shown a moderate and robust association between workplace bullying and violence and the development of type 2 diabetes. Both bullying and violence or threats of violence are common in the workplace. Research on bullying and violence prevention policies with workplaces as the target are warranted to determine whether these policies could be effective means of reducing the incidence of type 2 diabetes. Further study of possible pathways, for example through weight gain, negative emotions and the physiological stress response, will be crucial in providing an understanding of the causal mechanisms, as well as developing more costeffective interventions with surrogate outcomes.

Aims/hypothesis The aim of this multicohort study was to examine whether employees exposed to social stressors at work, such as workplace bullying and violence, have an increased risk of type 2 diabetes. Methods The study included 45,905 men and women (40-65 years of age and free of diabetes at baseline) from four studies in Sweden, Denmark and Finland. Workplace bullying and violence were self-reported at baseline. Incident diabetes was ascertained through national health and medication records and death registers. Marginal structural Cox models adjusted for age, sex, country of birth, marital status and educational level were used for the analyses. Results Nine per cent of the population reported being bullied at work and 12% were exposed to workplace violence or threats of violence. Bullied participants had a 1.46 (95% CI 1.23, 1.74) times higher risk of developing diabetes compared with non-bullied participants. Exposure to violence or threats of violence was also associated with a higher risk of diabetes (HR 1.26 [95% CI 1.02, 1.56]). The risk estimates attenuated slightly when taking BMI into account, especially for bullying. The results were similar for men and women, and were consistent across cohorts. Conclusions/interpretation We found a higher risk of incident type 2 diabetes among employees exposed to bullying or violence in the workplace. Further research is needed to determine whether policies to reduce bullying and violence at work may reduce the incidence of type 2 diabetes in working populations. Research on the mechanisms is also highly warranted.

Introduction "PTSD does not exist in Nepal," declared an expatriate humanitarian worker in Kathmandu. This statement can imply that the seventeen symptoms comprising the current diagnosis of posttraumatic stress disorder (PTSD) cannot be identified among Nepalis. Another interpretation is that a PTSD-like construct is not present in local languages and conceptualizations of suffering, emotions, behavior, and perception. If the person were referring to the first interpretation-that one could not find the symptoms comprising PTSD among Nepalis-then he would have been wrong. Marsella and colleagues have commented, "We are not aware of any ethnocultural cohort in which PTSD could not be diagnosed, although prevalence rates have varied considerably from one setting to another" (Marsella, et al. 1996:531). Similarly, every study in Nepal that has looked for PTSD has found it: 53 percent prevalence among internally displaced persons (Thapa and Hauff 2005), 55 percent among child soldiers (Kohrt, et al. 2008), 60 percent among torture survivors (Tol, et al. 2007), and 14 percent among Bhutanese refugee survivors of torture (Shrestha, et al. 1998). However, the mere identification of PTSD symptoms in a population has questionable therapeutic utility. Observation of PTSD symptoms does not, in-and-ofitself, immediately reveal optimal clinical care or other intervention. Understanding personal or social meanings and experiences of distress associated with traumatic events also are necessary for effective treatment. The second interpretation of the humanitarian workers' statement has greater potential to be of value to those in the healing professions. The question of whether or not a local construct for PTSD or psychological trauma exists is crucial to identify persons who are locally recognized as needing support, employ existing healing practices, understand the social significance for persons labeled locally as having psychological trauma, and determine if persons with the label are stigmatized and deprived of care. All of these factors become important when considering how to alleviate suffering for survivors of traumatic events. Hinton and Lewis-Fernandez (in press) have identified seven forms of clinical relevance for idioms of distress: indicators of psychopathology, indicators of risk for destructive behavior, indicators of life distress, indicators of past exposure to trauma, indicators of psychosocial functioning, causes of distress, and targets for therapeutic intervention. Compared to the number of epidemiological studies of PTSD, there is a dearth of studies of these local conceptual frameworks influencing diagnosis and treatment psychological trauma in non-Western settings (Lemelson, et al. 2007). Such studies are not trivial. When trauma-healing programs do not address local psychological frameworks, there is risk of unintended consequences such as degeneration of local support systems, pathologizing and stigmatizing already vulnerable individuals, and shifting resources from social and structural interventions (Argenti-Pillen 2003;Kleinman and Desjarlais 1995;Summerfield 1999;2002). Proper identification and understanding of local concepts and idioms not only reduces risk of harm, it also enables improved treatment such as selection of culturallyappropriate therapeutic interventions (Hinton and Lewis-Fernandez in press). Investigating the presence of local PTSD-like constructs within non-Western frameworks of emotions, behavior, and perception is additionally important because of the history of changing psychological trauma concepts within Western culture. Young's (1995) description of the evolution of the category of PTSD in Western culture reveals that PTSD was not a universal and ever-present way of framing specific experiences and psychological consequences. Rather, PTSD is only the most recent incarnation of describing, identifying, and treating a manifestation of psychological suffering in Western ethnopsychiatry. The Diagnostic and Statistical Manual of Mental Disorders (DSM-III) diagnostic construct of PTSD, which only first appeared in 1980, is not simply the progression of scientific knowledge, but rather the confluence of political, economic, legal, and medical institutional forces. 1 Furthermore, the release from military duties and compensation of those psychological affected by combat exposure have played a significant role in PTSD's current place in Western psychiatric nosology (McNally 2003;Summerfield 2001;Young 1995). Since the DSM codification of PTSD, trauma psychiatry has burgeoned and extended into work with non-Western populations. In the 1980s, mental health researchers began to use the PTSD label in their work with Southeast Asian refugees (Kinzie, et al. 1990;Mollica, et al. 1987;1990). PTSD epidemiological studies now appear standard practice in settings of war and natural disasters as illustrated by the numerous Centers for Disease Control and Prevention (CDC) studies documenting widespread PTSD, depression, and other mental health problems in Bosnia, Pakistan, Afghanistan, Burma, and Thailand (Lopes Cardozo, et al. 2000;2003;2004a;2004b;Thienkrua, et al. 2006). And psychological trauma intervention programs are now commonplace ranging from targeted clinical interventions to psychosocial programs for large numbers of affected populations (Bass, et al. 2006;Bolton, et al. 2003;2007;Jordans, et al. 2009;Layne, et al. 2000;Mohlen, et al. 2005;Neuner, et al. 2008;Tol, et al. 2008;Verdeli, et al. 2008). One explanation for the global proliferation of PTSD as a way of framing suffering is that it is a useful lens in many settings to identify those in need of care and to select types of intervention goals. For example, Tol and colleagues (2007) have shown that PTSD has a direct relationship with severity of disability among torture survivors in Nepal. Another argument for the spread of the PTSD trope is legal advocacy for survivors of human rights violations. PTSD has been used as medical and scientific evidence of human rights violations (Physicians for Human Rights 1996;2001;2008); diagnoses of PTSD is evidence that an individuals rights have been infringed upon through torture and other forms of political violence. In the political asylum system, PTSD is nearly requisite as "evidence" of torture (Bracken, et al. 1995;Silove 2002;Smith Fawzi, et al. 1997;Summerfield 1999). Drawing attention to the psychological suffering of specific groups also is used to mobilize funding for non-governmental and humanitarian organizations. Breslau (2004) in a special issue of Culture, Medicine and Psychiatry (Vol, 28, Iss 2) on the "Cultures of Trauma" critiques how researchers, humanitarian aid workers, and local political forces dictate the application of the PTSD label. He claims applying PTSD labels in non-Western settings reflects the politics of humanitarian aid and local political forces attending to the global resource procurement. Breslau argues, for example, that research into the prevalence of PTSD among Bhutanese refugees in Nepal (Shrestha, et al. 1998) is a political endeavor, which by highlighting torture misses the broader suffering in refugee camps. This all results from the nature of medical labels tied to privileged resources and spotlighted by the humanitarian community. 2 Ultimately, what has been missing in many of these debates for or against the use of the PTSD label is the relation between the psychiatric category and local experience of distress. Without research on local idioms and frames of understanding trauma, one cannot effectively assess how importing or rejecting the PTSD label impacts well-being, treatment seeking, and stigma. 1 Young (1995) argues that the psychiatric lens of posttraumatic stress disorder (PTSD) originally was developed primarily to understand the mental health problems and frame treatment of returning American veterans from Vietnam For example, Young ties the growing interest in psychological trauma during the 1860s to increasing use of railway transport and British Parliament's passing in 1864 of legislation to financially compensate victims of railway injury. In addition, Summerfield (2001) correlates the growing interest in psychological trauma with changes in personhood related to entitlement, expected success, and absence of suffering. PTSD legitimizes victimhood, moral exculpation, and disability pension. Bremner (2002) also suggests that personality has changed from a warrior class where life was filled with violence and people endured severe hardship with great courage and strength to a "postwarrior" class where trauma overwhelms individual resources. From automobile collisions to overhearing sexual jokes in the workplace, identifying with the psychological trauma discourse provides the foundation for compensation Summerfield (2001) views PTSD as the product of compensation pursued on the basis of individual rights: "An individualistic rights conscious culture can foster a sense of personal injury and grievances and thus a need for restitution in encounters in daily life that were formerly appraised more dispassionately." 2 Argenti-Pillen (2003) describes the how the new breed of "fearless women" in Sri Lanka who engage in trauma intervention programs are also the women who traveled throughout the country pursuing claims for disappeared husbands and sons, also an example of specific political context in which the construct becomes employed. An outspoken critic of the trauma and violence literature, Summerfield (1999) states that (1) trauma is not necessarily widespread in conflict settings, and (2) war is a social experience requiring healing through social means, not Western mental health intervention. Summerfield refers to the work of Somasundaram (1996) in Sri Lanka, which states "none of the subjects considered themselves psychiatrically ill, and just saw their symptoms as an inevitable part of the war." Idioms of distress, ethnopsychology, ethnophysiology, and stigma Nichter (1981) defines idiom of distress as "an adaptive response or attempt to resolve a pathological situation in a culturally meaningful way." These idioms can include somatic complaints, possession, and other culturally significant experiences. Idioms of distress typically necessitate culturally prescribed healing and resolution. Idioms of distress related to frightening experiences include susto in Latin America (Rubel, et al. 1984), kesambet in Bali (Wikan 1989), rus in Micronesia (Lutz 1988), rlung in Tibetan medicine (Clifford 1990), yardargaa in Mongolia (Kohrt, et al. 2004), and soul loss in many cultures. In Thailand, lom, wuup, and rook klua lai tai are idioms representing cultural conceptualization associated with panic (Udomratn and Hinton 2009). Hinton and Lewis-Fernandez (in press) describe a range of anxiety-related idioms of distress in Cambodian and Caribbean Latino cases. Bolton (2001), employing a constellation of methods, indentified a number of local idioms that clustered into a "mental trauma" syndrome comprising symptoms of PTSD, depression, and other local idioms in Rwanda. The limitation of idioms of distress is they may easily become a vocabulary list without a cohesive framework detailing significance of these terms and how they fit into a broader understanding of suffering and wellbeing in a given cultural context. One solution to this is conducting rich ethnography as exemplified by Wikan's (1989;1990) work on the term kesambet, which positions the fright idiom within broader expectations of the selfexpression, emotions, and social relations. Her ethnographic work thus places the idiom within an ethnopsychological framework of understanding life and well-being. White (1992) defines ethnopsychology as the study of how individuals within a cultural group conceptualize the self, emotions, human nature, motivation, personality, and the interpretation of experience. It is a local lay psychology used to understand experience. A complement of ethnopsychology is ethnophysiology. Hinton and Hinton (2002) describe ethnophysiology as "the culturally-guided apperception of the mind/body rather than actual biological differences," and they provide examples of the hydraulic model of anger in Western culture and the pneumatic model of wind to explain panic in Cambodian culture. Thus, the way a cultural group interprets experience is related to how the group conceptualizes the body and its processes. Hinton and colleagues (2006) find that cultural concepts related to the mechanisms of mental and bodily events contribute to the types and experiences of anxiety disorders. Young's (1976) work in Ethiopia describes the experience of depression as an overheating problem related to the heart's control of jimat (cords) that move and motivate the body. Much of Tibetan, Mongolian, and Chinese concepts of mental health can be understood in terms of the humoral energies: rlung, khii, and qi, respectively (Clifford 1990;Janes 1999;Kohrt, et al. 2004). Among the Mandinka, Fox (2003) identifies four post-trauma syndromes, two of which are disorders of the heart, one affects the mind, and the last affects the brain. These syndromes operate cumulatively with fear leading to disturbance of the heart. If the heart problems are severe, this leads to dysfunction in the mind and ultimately the brain. Ethnopsychology helps to reorient understandings of traumatic events. Kenny (1996) using the example of susto suggests that cultural conceptualization and understanding of the self determine which types of events are considered misfortune. Ethnophysiological understandings have important implications for intervention as well. Hinton and colleagues (2009b) outline a framework for "concentric ontology security" explaining how interventions can engage different elements of the self depending on the presentation and primary complaints of the individuals, specifically in relation to nightmares. Idioms of distress reflect and influence the stigma associated with illness. Stigma worsens the experience of an illness and often leads to lack of access to care, social isolation, and internalized feelings of shame, inferiority, and fatalism (Weiss, et al. 2006). For some conditions, the impact of stigma may be more devastating than the condition itself. Stigma is understudied in relation to PTSD in any cultural setting. For example, both HIV and PTSD are relatively new diagnostic entities introduced to medicine in the early 1980s. Both have approximately 13,000 journal publications (Medline search, 14 Feb 09). However, whereas there are 199 publications with 'HIV' and'stigma' in the title, there is only one with 'PTSD' and'stigma' in the title. Summerfield's (2001) comment exemplifies this conception of PTSD not being stigmatized: "[it is] rare to find a psychiatric diagnosis that anyone liked to have but post-traumatic stress disorder was one." Despite a poverty of stigma literature related to PTSD in Western cultural settings, one cannot assume that psychological trauma is not stigmatized in non-Western contexts. In Cambodia, French (1994) describes the difficulty mobilizing support for amputees injured by landmines. The trauma resulting in amputation is seen as the result of poor karma, the sins of an individual in past lives. Similarly, Miller's (2009) film Unholy Ground dramatically demonstrates how traumatic experiences are seen as an individual's fate transmitted through karma. Therefore, one of the objectives of this study is to examine the meaning of trauma idioms and perceived causation in relation to assess the relationship of psychological trauma with stigma. --- Psychological trauma in Nepal The timing of research on local psychological trauma concepts is particularly important in Nepal. The country recently emerged from a decade-long civil war that resulted in the death of 14,000 people, thousands of disappearances, the displacement of over 150,000 people, the conscription of greater than 10,000 children into armed groups, and greater than 100,000 incidences of torture (Human Rights Watch 2007;INSEC 2008;Singh, et al. 2007;Tol, et al. 2010). In response to this, psychosocial programming has grown rapidly to address a putative burden of psychological trauma (Kohrt 2006). In addition, thousands of ethnic Nepali Bhutanese refugees, who are survivors of political violence, are emigrating from Nepal into the United States where they will encounter a medical system in which PTSD is a dominant clinical trope. A recent systematic multi-disciplinary review of psychosocial, mental health, and political violence literature in Nepal revealed that the dearth of social science understandings of trauma is a major impairment to epidemiological studies and intervention for Nepalis and ethnic Nepali Bhutanese refugees (Tol, et al. 2010). Previously, research on idioms of distress in Nepal has shown that jham-jham (numbness and tingling) and gyastrik (dyspepsia) are common complaints associated with psychological distress particularly depression (Kohrt, et al. 2005;2007); however, the idioms are not interchangeable with a specific psychiatric disorder. Sharma and van Ommeren (1998) have catalogued idioms of distress among ethnic Nepali Bhutanese refugees in Nepal by drawing upon clinical experiences and case note surveys. Their list includes dukha laagyo (sadness), dar laagyo (fear), jharko laagyo (irritation), metaphors in relation to being uprooted from one's homeland, and somatic complaints such as jiu sukera gayo (drying of the body) and kat kat khanchha (tingling and burning sensations). All of these idioms overlap with psychological trauma, but none exemplifies one-to-one synonymy with PTSD. Ethnopsychological/ethnophysiological models of the self have been outlined for Nepali frameworks of mind-body relations (Kohrt and Harper 2008). In Nepal, the self is considered an assemblage of the man (heart-mind), dimaag (brain-mind), jiu/saarir (corporeal body), saato/atma (spirit/soul), and ijjat (social status/honor), all of which are connected with samaaj (the social world). Within this ethnophysiological framework, afflictions of the brain-mind are highly stigmatized. To date, research has not examined the interplay of idioms, these ethnopsychological models, and psychological trauma. This paper addresses this gap in the research through an exploration of local constructs of psychological trauma, Nepali idioms of distress, and vulnerability to suffering. Nepali ethnopsychology and ethnophysiology provide schemata on which to formulate employing idioms in intervention, with attention to stigma reduction and prevention. The findings are incorporated into recommendations for humanitarian psychosocial workers, psychologists, psychiatrists, and other professionals in the mental health and humanitarian fields. --- Methods This study was conducted in Nepal using semi-structured interviews, free-lists, comparison tasks, and observant participation to identify types of events considered traumatic, idioms of distress related to these traumatic events, emotions associated with psychological trauma, and perceptions of causation and vulnerability related to traumatic events. All participants were Nepalis. The first author and a Nepali research assistant conducted all interviews in Nepali. All participants completed an informed consent form. The study was approved by Emory University Institutional Review Board and the Nepal Health Research Council. Free-list and emotion comparison tasks were conducted in 2005. The sample composition for the research tasks was limited to persons residing in Kathmandu because it was unfeasible to travel in rural areas during the time of the People's War, which did not conclude until the end of 2006. Thirty-five individuals participated in the free-list and emotion questionnaire portion of the study. Thirty-three participants had complete data for emotion comprehension ranking, 32 for emotion-term similarity analysis, and 30 for free lists. The sample included nineteen women and sixteen men, including eighteen Newar, eight high-caste Hindu, four low-caste Hindu, and five from other ethnic groups (Gurung, Magar, Rai, Tamang). Twenty were married, and fifteen were single. Age ranged from 18 to 72 years, with a mean of 33 years. Education level ranged from 0-18 years, with a mean of 10 years (School Leaving Certificate (SLC)-pass). This sample was selected for heterogeneity of ethnicity, age, and level of education to allow for an assessment of terms and concepts that were mutually intelligible and salient across Nepali demographic groups. All participants completed the surveys in Nepali regardless of their ethno-linguistic group. For participants who spoke another ethnic language, such as Newar or Magar, we inquired about their proficiency in Nepali. All of these respondents said they had been educated in Nepali and that they were as proficient or more proficient in Nepali than in their ethnic language. The survey comprising free lists and an emotion questionnaire were administered to the lay public participants to identity local concepts of psychological trauma. In free list tasks, participants were asked to list types of traumatic events, the effects of traumatic events on one's life, terms used to describe suffering and traumatic events (idioms of distress), emotions associated with trauma, and gender differences in the impact of trauma. The emotion questionnaire, which included comprehension ranking and dyadic comparisons, was administered to produce a general model of similarity and understanding of terms identified in the free list tasks. This enabled us to identify commonalities of understanding to complement the idiosyncratic individualized responses provided in free-lists and semistructured interviews. The emotion questionnaire includes a list of emotions derived from free lists. To identify how confidently people understood emotion terms, participants rated emotion terms on a scale of one to five: five signifying "I completely understand the emotion term," and one signifying "I completely do not understand the term." Intermediate numbers represent degrees of partial understanding. After rating their level of understanding of emotion terms, participants rated similarity among emotion term pairs to determine the conceptual similarity. To assess overall consensus on these ratings, we fit a cultural consensus model using an ordinal data type in UCINET 6 (Hruschka, et al. 2008). We used correspondence analysis to represent visually the conceptual similarity between trauma-related emotion terms in a two-dimensional space (Moore, et al. 1999;Romney, et al. 1997;2000). In such a graph, those items that participants consistently rated as more similar are closer together in the graph. In the graph, 95% confidence ellipses reflect the certainty about each emotion term's average location in the two-dimensional space (SYSTAT 2007). Semi-structured interviews were conducted to gain more in-depth information on concepts of psychological trauma. This was done with practitioners to elicit narratives of care, the practitioners' beliefs surrounding psychological trauma, and their perceptions of clients' beliefs and attitudes. Semi-structured interviews with clients were used to understand the traumatic events, models of causation, psychological and social sequelae, and terminology used to describe the event and related emotions. Semi-structured interviews were conducted with 32 individuals from 18 organizations involved in psychosocial programs. The organization participants included program administrators, counselors, other health professionals, and clients. Semi-structured interviews also were conducted with religious leaders regarding causality of traumatic events. Semi-structured interviews lasted approximately two hours. Semi-structured interviews were conducted in Kathmandu and rural areas of Western Nepal from 2005-2008. Observant participant was conducted in Nepali nongovernmental organizations (NGOs) and clinical health settings in both Kathmandu and rural areas from 2006 through 2008. Observant participation was used to identify language of clinical discourse, which may not have been reported in individualized interviews and free list tasks. Observant participation also elucidated common presenting symptoms and referral patterns. --- Findings Traumatic Events Participants were asked a range of questions to elicit categories of traumatic events such as (i) the worst events that happen in people's lives, (ii) the worst events they have heard about through conversations with others, television, radio, news, and other media, and (iii) other events that dramatically alter a person's life in negative ways. Probes included violence, accidents, natural and man-made disasters, and other threats to life. Four general themes dominated the categories of events described. (i) Counselors, program administrators, and other health professionals focused on events involving a threat to one's life or a family member's life. (ii) Clients placed more attention on chronic, ongoing events that impair livelihood, specifically failure or lack of success in specific domains of life. (iii) Spiritual and supernatural events were considered major traumatic events. (iv) Lastly, conflict-related events featured prominently; clients of one NGO felt that trauma only resulted from political violence. Threat to life and limb-Mental health professionals endorsed a range of events related to loss of life and well-being including domestic violence, road accidents, marital rape, burns and fires, plane crashes, snake bites, falling from heights, animal attacks, witnessing accidents or violence, and death of a loved one. Physical illness, even without the threat of death, was related to psychological trauma. Physical wounds were thought to cause suffering and impairment in life. Major diseases like HIV/AIDS and cancer were thought to be traumatic. Respondents stated that drugs and alcohol can cause mental trauma, and traumatized people are more likely to drink alcohol. Both male and female participants reported that the loss of a husband or son causes psychological trauma for women, but participants did not report the converse of losing a wife or daughter as traumatic for men. This gender difference was explained in part by the restrictions placed upon widows versus widowers (Bennett 1983;Kohrt and Worthman 2009). Widows cannot own property or remarry. Thus, they suffer significant livelihood instability. In contrast, widowers are able to maintain their wealth and property and free to remarry after losing their wives. Failure to meet expectations-The majority of clients responded to inquiries about the worst experience in life as those events where "reality does not meet expectations." Failed expectations in life domains included being unsuccessful in love, failing the School Leaving Certificate (SLC) exams at the end of 10 th grade, major loss in business, unemployment, poverty, and generally not being satisfied in one's desires. Too much ambition or too high expectations caused suffering in one's life. If someone works too hard, he/she will suffer. For those suffering from failure in some area of life, it was recommended to not think about it or "forget it." For women, central issues are producing male children and maintaining good relations among household members. For men, being unsuccessful in professional life or having one's social status damaged are examples of the most negative experiences in life. Men returning from work in Persian Gulf countries reported that they did return at the level of success and prosperity for which they had hoped; they identified this as the worst experience of their lives. These migrant workers had intended to return home as wealthy businessmen. However, they often found that they had been deceived by international manpower agencies, and they returned even more impoverished than when they left. Furthermore, homecoming is a very important process for most Nepali groups. Individuals are welcomed back from journeys with ceremonies that may include animal sacrifices, public festivals, and other activities (Kohrt forthcoming). The lack of a proper return and reintegration into the community has also been identified as a major source of distress for returning child soldiers in Nepal (Kohrt and Koenig 2009;Kohrt, et al. 2010). Similar distress arising from negative homecoming experiences was observed for returning U.S. military from Viet Nam (Fontana, et al. 1997;Johnson, et al. 1997). Spiritual causes-A man working as a psychosocial counselor at an NGO explained that spiritual events, such as cursing, possessions, and 'evil eye', are thought to play a dominant role in some forms of trauma (e.g. bigar, naag, devtaa, tunamuna, and aakhaa laagyo). Boksi and bhut laagyo (struck by witches or ghosts) can cause trauma. Food and water can cause suffering if it is cursed (tunamuna). Conflict related trauma-Conflict events included having family members 'disappeared', being detained, witnessing the brutal killing of a family member, bomb explosions, watching one's house and land being burnt, destruction of livestock, loss of money and savings, being kidnapped or witnessing relatives being kidnapped, being forced to join armed groups, being forced to leave one's children, having one's family dispersed across the country, lacking food and shelter, lacking access to healthcare or education, being forced to move repeatedly, and having constant feelings of insecurity. One NGO worker said that war is the only true cause of psychological trauma, thus demonstrating a tight linkage of the trauma narrative and political violence for some humanitarian workers. --- Trauma Idioms of Distress Free list responses from laypersons, NGO workers, and mental health professionals revealed a range of terminology and idioms of distress. Table 1 includes the idioms participants most commonly reported as responses to traumatic events. Idioms included fear responses, sadness, suffering, and worrying. Idioms associated with supernatural phenomenon were saato gayo (soul loss) and aitin laagyo (spirit induced sleep paralysis). NGO and mental health professionals used the English terms 'depression', 'trauma', and 'PTSD' as responses to these events. No lay participants mentioned these English terms. Participants explained that these idioms differed by phenomenology and triggers. Fearrelated idioms ranged from terror to anxiety: Dar is fear of something specific that lasts only briefly such as upon seeing a snake or ghost. Traas is terror. It is feeling insecure about something that is there all time, but you cannot do anything about it. It is something that cannot be controlled. Bhaya is also a fear, but it can be addressed by working hard or preparing, such as for a test. (24-year-old female NGO worker) Pir is constant, generalized fear, whereas chinta refers to something specific, such as fear of leaving the house to go to work. Chinta is momentary, brief, and related to something specific. (40-year-old female Nepali language instructor) With regard to depression and sadness, the 24-year-old female NGO worker added, Dukha is considered sorrow, but 'depression' (stated in English) is a state where one cannot do anything. They are unable to get motivated to do work. Riis (anger) is something that comes from small things, but traas (terror) and aaghaat (shock/ trauma) come from big things. NGO workers and mental health professionals routinely employed the term aaghaat, which refers to shock, impact, or clash. Some qualified the term as maanasik aaghaat (mental/ psychological shock) to distinguish mental shocks from physical shocks (aaghaat). However, counselors explained that clients and the public often are not familiar with the term aaghaat, especially in the psychological context. Counselors felt more comfortable employing piDaa (pain, anguish, agony, torment, suffering) or maanasik piDaa (mental anguish) and other idioms when speaking with clients and the public. A client at a center for Maoist victims stated, We have the most piDaa (suffering) and the most maanasik samasyaa (mental problems) of anyone in the entire world. Some participants, both lay and professional, reported that dramatic negative life events could lead to symptoms of madness, psychosis, and mental illness. These events could make someone paagal or baulaahaa (mad, psychotic, crazy). Some participants added that it is not the event per se that is the problem but rather individuals who "think too much" about the events and bad life experiences who are most likely to become paagal or baulaahaa. Health professionals employed the term maanasik rog (mental illness) rather than paagal. Maanasik rog was distinguished from maanasik piDaa. A woman working as a psychosocial counselor for an NGO explained, For maanasik rog (mental illness), these individuals can disclose things to counselors. They often have difficulty sleeping, their appetite is decreased, and they have problems studying. For maanasik aaghaat (mental shock) or piDaa (suffering, torment), individuals do not trust counselors. They do not get relief after two or three sessions. They do not want to talk. They act like statues. They act as if they are not in the real/physical world. In contrast, a psychiatrist at a mental hospital explained that maanasik piDaa could present as psychosis: People with piDaa (trauma/suffering) are paagal (mad). If a loved one dies, then their relatives may go mad. People believe that others should avoid utensils and water they have touched or else they will contract the madness. A client at one of the treatment centers recounted an anecdote of a neighbor who became crazy: This woman had a husband who was always unfaithful. She began to have strange behavior. She spoke to herself without stopping but would not speak when spoken to. She did not eat. She was pura baulaahaa (totally crazy). She ran away and was naked in a village. Her hair was unkempt. This lasted eight or nine years. One day she was startled by the loud noise of a jeep driving nearby. She suddenly realized how in disarray she was and that she was naked and so she became embarrassed. She then returned to her normal behavior. Although this woman recovered, it was generally felt that madness from psychological trauma was permanent. A psychiatrist at the mental hospital explained, Maanasik aaghaat has a long period of effect, often life-long and permanent. He/ she will be abnormal for the rest of his/her life after the traumatic event. Counselors reported that clients had symptoms of uncontrollable irritation and angered easily at small things. Counselors thought that those with maanasik aaghaat (mental shock) could do anything in anger; they can kill others or kill themselves. Memory is a central feature of presentation. Birsane nasakne darghaatnaa (unforgettable events) and manko ghaau (wound/sore/scar on the heart-mind) both refer to inability to forget such events. Manko ghaau refers to a scar/sore/wound on the heart-mind. A 24-yearold NGO worker explained how hearing stories of rape left such a scar on her heart-mind: I was 'depressed' ['depressed' stated in English] after hearing [the stories of rape survivors]. I could not sleep. Things kept racing in my mind, and I could hear our conversation over and over again. It left manko gaau (a scar on my heart-mind) that will never heal. In this instance, a distressing experience scars the heart-mind, which is the organ of memory, just as the body becomes physically scarred through physical trauma. This scar in the heart-mind then prevents the individual from forgetting about the

In the aftermath of a decade-long Maoist civil war in Nepal and the recent relocation of thousands of Bhutanese refugees from Nepal to Western countries, there has been rapid growth of mental health and psychosocial support programs, including posttraumatic stress disorder (PTSD) treatment, for Nepalis and ethnic Nepali Bhutanese. This medical anthropology study describes the process of identifying Nepali idioms of distress and local ethnopsychology and ethnophysiology models that promote effective communication about psychological trauma in a manner that minimizes stigma for service users. Psychological trauma is shown to be a multifaceted concept that has no single linguistic corollary in the Nepali study population. Respondents articulated different categories of psychological trauma idioms in relation to impact upon the heart-mind, brain-mind, body, spirit, and social status, with differences in perceived types of traumatic events, symptom sets, emotion clusters, and vulnerability. Trauma survivors felt blamed for experiencing negative events, which were seen as karma transmitting past life sins or family member sins into personal loss. Some families were reluctant to seek care for psychological trauma because of the stigma of revealing this bad karma. In addition, idioms related to brainmind dysfunction contributed to stigma while heart-mind distress was a socially acceptable reason for seeking treatment. Different categories of trauma idioms support the need for multidisciplinary treatment with multiple points of service entry.

in anger; they can kill others or kill themselves. Memory is a central feature of presentation. Birsane nasakne darghaatnaa (unforgettable events) and manko ghaau (wound/sore/scar on the heart-mind) both refer to inability to forget such events. Manko ghaau refers to a scar/sore/wound on the heart-mind. A 24-yearold NGO worker explained how hearing stories of rape left such a scar on her heart-mind: I was 'depressed' ['depressed' stated in English] after hearing [the stories of rape survivors]. I could not sleep. Things kept racing in my mind, and I could hear our conversation over and over again. It left manko gaau (a scar on my heart-mind) that will never heal. In this instance, a distressing experience scars the heart-mind, which is the organ of memory, just as the body becomes physically scarred through physical trauma. This scar in the heart-mind then prevents the individual from forgetting about the event, as a physical scar is a reminder of other types of trauma. The analogy extends even further: just as severe physical trauma can lead to not only a scar but also impaired functioning of an arm or leg, severe psychological trauma damages the heart-mind leaving a scar that prevents normal functioning of the organ of memory. Thus, when one tries to use heart-mind to recall other events, they only come up with remembering the traumatic event. An 18-year-old girl from rural Nepal whose family was attacked by Maoists explains that the event is constantly replaying itself in her heart-mind: I can't believe that it happened. The incident manmaa raakhiraakhne (is always on my heart-mind). I can't forget; I remember it all the time. Similarly, other survivors of trauma reported manmaa kura khelne (having worries playing in their heart-minds) that caused insomnia. Some clients ruminated about killings they had witnessed. Caregivers and psychological trauma survivors all endorsed insomnia and nightmares. A few trauma survivors describe jhajhalko aauchha (flashbacks, flashing/ flickering memories), feelings of the event happening again. One psychosocial counselor for torture victims finds that none of his clients reports the PTSD symptom of "difficulty remembering parts of the event." Most survivors have no problem remembering the event, on the contrary, they report being unable to forget the event. An Ayurvedic doctor described how his patients would think of the past event then subsequently "go into a panic condition." One young man who had been beaten along with his family by Maoists described intrusive memories of his traumatic event: I worship and pray and while I am doing those activities, I do not think about the incident. Nevertheless, as soon as I go out into the street with office work if I see an accident, the exact scene of my beating repeats itself over and over again. I cannot concentrate on the work that I have to do and what is going on in front of me. I do not want to remember my tragedy but seeing bad scenes forces me to remember. Memories of the event impair recall of other issues and decrease concentration. One NGO program administrator described a teacher who would forget what he was teaching in the middle of lessons; he stated, "Psychological trauma causes people to lose memory power." Somatic complaints were common. Health professionals identified somatic complaints as the primary presentation of psychological trauma: headaches, back pain, gastric problems, painful urination, visual disturbances, palpitations, high blood pressure, and fainting attacks. An Ayurvedic doctor commented that persons with psychological trauma have, "a poor immune system, weak nervous system, runny nose, and low blood pressure." One psychosocial counselor described a connection with psychological trauma causing physical problems; "maanasik tanaab (mental tension) has an effect on the body. One is more susceptible to illness." Counselors reported that clients appear comfortable discussing physical health problems. However, they do not typically self-disclose about traumatic events until it is uncovered through the medical history. Other common symptoms included not having enough energy to work. A last category of idioms referred to loss of social status through traumatic events. Participants explained that social honor is lost, ijjat gayo, or that social shame is incurred bejjat. The idiom naak khatne refers one's nose being cut as a metaphor for social shame. Participants explained that social status could be lost from traumatic events to oneself or one's family because it demonstrated poor karma-and thus was the person's or family's fault-or was related to loss of honor such as through being victim of rape or assault. The role of karma will be discussed further below. --- Trauma and Emotions The second phase of the research examined people's conceptions of emotion terms related to suffering from trauma, specifically their comprehension of terms and how they viewed the relationships of these terms. The comprehension results of the 22 emotion terms are presented in Table 2. Terms such as dukha, chinta, dar, riis, and pir are well understood by the general population. PiDaa is also relatively well understood. Aaghaat, however, was the least intelligible emotion term for the general public sample. The general population also poorly understood two English terms used by Nepali health professionals, 'depression' and 'trauma'. A cultural consensus model fit the judged similarities between common emotion terms (eigenvalue ratio = 13.1, mean first factor loading = 0.61 (SD = 0.15), all factor loadings > 0). Figure 1 represents the first two dimensions of a correspondence analysis of similarity ratings. Emotion terms related to fear and fright form a distinct cluster (bhaya, dar, and traas). Riis (anger) and paagal (madness) form a distinct and moderately overlapping cluster. A third cluster encapsulates negative emotional states related to suffering: chinta (worry), pir (anxiety/sorrow), piDaa (anguish/suffering), dukha (sorrow/hardship). The final four terms lie somewhere between the anxiety/suffering cluster and the madness/anger cluster. These are laachar (helplessness), aaghaat (trauma), tanaab (mental tension) and niraashaa (despair). Notably, the common clinical term, aaghaat, lies somewhere between its presumed synonym for trauma, piDaa, and the more stigmatized term, paagal (madness). Indeed, the judged similarity between paagal and aaghaat is significantly greater than the judged similarity between paagal and piDaa (mean similarity = 3.44 vs. 2.59, Wilcoxon matched pairs signedrank test, p <unk> 0.05). This suggests that the term (maanasik aaghaat) identified by mental health professional and NGO workers as synonymous with psychological trauma (maanasik piDaa) is viewed by the general public as having some similarity to madness/psychosis (paagal). --- Vulnerability to Traumatic Events Mental health professionals and clients discussed vulnerability at two levels: vulnerability to negative life experiences and vulnerability to express emotional and psychological symptoms after experiencing negative events. Previously, Sharma and van Ommeren (1998) found that many Bhutanese refugees explained their suffering in terms of karma ko phal, the fruit of sins in prior lives. Respondents in our study explained that karma dictates vulnerability to negative life events. Karma is the concept of good or bad deeds determining prosperity or suffering. Karma thus dictates who does and who does not suffer from traumatic events. A 56-year-old male Hindu priest in central Nepal explained the relationship of karma with traumatic events: Ghatanaa bhaeko chha, yo karma ko phal ho. [An accident has happened; it is the result/fruit of karma.] Naraamro nagarnu hai, paap ko phal dekhalaa hai.[Do not do bad things; you will see the consequences/fruit of your sins.] Karma is related to the experience of traumatic in two ways. First, an individual may have committed sins in a prior life and suffer the consequences as traumatic events in this life. When asked specifically whether sins committed in this current life can lead to traumatic events in this same lifetime, multiple Hindu priests explained that this did not happen. Instead, the sins committed in this life lead to traumatic events in future lives or in the lives of family members. Second, sins committed by one's family members in this life and sins committed by one's deceased ancestors can also result in the experience of traumatic events for other family members (pariwaarko karma, the family's karma). One Hindu priest explained that sins of the mother or father lead to the family's suffering. For example, a sinful parent will have children who are mute, deaf, experience accidents, and die at an early age. The Hindu priest explained, Mero purkhaa ko paap maile bhogdaichhu. [I must bear the sins of my ancestors.] Therefore, the experience of traumatic events can be highly stigmatizing because people think that the trauma survivor must have been a bad person in a prior life or that his/her family members and ancestors were bad people. Not surprisingly, traumatic events are often covered up within the family because the events reflect badly upon them as well as the individual who experienced the event. Women are considered to have been less pious than men were in their previous lives, and consequently they are born female and at greater risk of traumatic events. Similarly, persons from low castes are considered to have poorer karma, and thus they are fated to experiences more negative events during their lifetime. Regarding vulnerability to expression of psychological distress after negative life events, an Ayurvedic doctor referred to constitutional differences: Individuals are most susceptible to effects of trauma during the third trimester of pregnancy. If the mother is traumatized during this time, the child will be affected for his or her lifetime. To avoid this, pregnant women should only pay attention to the fetus, and ignore what is happening around them. Mothers who are strong and dedicated will not have traumatized children. Mothers with naasaa kaamjor (dry or weak nerves) are at risk of having children with psychological trauma. If trauma occurs after birth, it has much less effect on a person's life. However, if the trauma is severe, such as torture, then the person will have psychological difficulties for their lives that cannot be cured. Trauma from graha (unseen/astrological causes) can have lasting effects. Emotional, artistic, and mentally active people are most at risk of psychological trauma because they are of 'nerve-type' and get frustrated easily. Participants also described susceptibility based on how strongly one's saato (spirit or soul) is attached to the corporeal body. Souls are seen as less securely attached among children. Children lose their saato from frightening events, and this causes behavior changes, such as piDaa and nightmares. The loss of soul makes them more susceptible to illnesses ranging from diarrhea to respiratory infections. Traditional healers can treat saato loss. It is thought that adults lose their souls less easily than children do, and thus typically do not suffer from saato loss. Among adults, women are considered more vulnerable than men are to saato loss. One adult woman described an incident when she suffered saato loss. I was walking along a street. Then I saw men driving on motorcycles. They were driving around a man and beating him. I saw blood everywhere. Then saato gayo (my soul left). I do not know how I got back to where I was going. I could not do work that afternoon. My saato returned later that evening, but I had problems sleeping for more than a week. --- Model for Ethnopsychology of Trauma, Idioms of Distress, and Therapeutic --- Implications Figure 2 is a schematic attempting to represent psychological trauma within the context of this Nepali sample and other ethnographic work in different communities throughout Nepal (Kohrt and Harper 2008). At the left of the schematic is vulnerability to negative life experiences. Psychological trauma is rooted in concepts of vulnerability wherein individuals with poor karma who did not fulfill their religious path in prior lives or have sinful ancestors are more likely to suffer from traumatic events in this life. This vulnerability is important to keep in mind when considering therapeutic interactions because karma attributes a degree of culpability to people with negative life experiences. Traumatic experiences occur along a continuum. At one end of the continuum are chronic stressors and repeated burdens such as poverty, lack of education, and failure to achieve ones' goals. At the other end of the spectrum are specific acute events that an individual may describe as "unforgettable." We find that the term birsane nasakne dargatana (unforgettable accidents/events) captures individualized experiences that produce psychological distress. In an epidemiological study, Thapa and Huff (2005) also found that the role of lasting memory was central in eliciting traumatic experiences; they reported that nearly all persons with high PTSD symptomatology endorsed the phrase dukha lagne ghatanako gahiro chap baseko (having deep and long-lasting impression of the terrifying event). However, they did not report what percentage of people endorsed the phrase but did not have high PTSD symptom severity. In the model, individuals who suffer either chronic stressors or acute unforgettable events have differential susceptibility to psychological symptoms. Because children have saato (spirit/soul) that are attached weakly, they are more likely to have soul loss and fright. Women are more susceptible than men are because of being considered constitutionally weaker. Lastly, individuals with small or weak man (heart-minds) are also more likely to suffer psychological distress (McHugh 1989). The emotional and psychological consequences of negative life experiences, both chronic and distinct, can be mapped on the ethnopsychological framework of man (heart-mind), dimaag (brain-mind), jiu/saarir (physical body), saato (spirit/soul), and ijjat (social status/ social face). Referring back to the emotion model in Figure 1, the three clusters map onto these different elements of the self. The anxiety/suffering cluster (chinta, pir, piDaa, and dukha) reflects disturbance of the man (heart-mind). The fright cluster (dar, bhaya, and traas) are strong sudden emotions that disrupt the saato's attachment to the body. The cluster of madness and anger (paagal and riis) reflect disturbance of the dimaag (brain-mind). There are two categories of symptoms related to heart-mind disturbance. One category is the general emotions of suffering, sadness, and worry. The other category is memory related phenomenon because the heart-mind is the organ of memory. As reported above, memory symptoms are described through the ethnophysiology of scars on the heart-mind. This scarring of the organ of memory leads to flashbacks, rumination, and other intrusive recollections of the event. Desjarlais describes this among the Yolmo ethnic group of central Nepal: Among the Yolmo [people] significant events, hurtful ones especially, can leave a mark, a trace, at times a maja or'scar'... within a person's sem [heart-mind], much as words can be carved into stone or sounds engraved within a phonographic record. (Desjarlais 2003, p. 148) In urban and rural clinical settings, patients often focused on the inability "to forget" certain life experiences as one of their main complaints. The constant remembering of specific events in the heart-mind (i.e. intrusive memories) caused physical and psychological distress. Common to these complaints was a sense that one could not control when these memories would intrude and that once the thoughts started, individuals could not stop thinking about them nor prevent or abate physical and psychological sequelae. Patients consistently expressed the desire "to forget" these events. One patient stated, "my goal is not to remember the suffering." A male patient beaten by Maoists recounts, When I saw it all happening, I did not think it was real. I did not think this was happening to me. Even now, I cannot believe that it happened to me. All I want to do is forget that it happened. But, I cannot forget. Pettigrew and Adhikari (2009) describe this phenomenon among the Tamu-mai (Gurung) ethnic group in western Nepal: [Forgetting fear] should also be described as 'not remembering' or more specifically, no longer having intrusive memories. Since the end of the insurgency people have talked a lot more about forgetting fear. While people sometimes talked about past fears, fear was not a currently experienced emotion in the sense that it was during the insurgency. "Forgetting fear" was also a coping strategy that allowed villagers to put the past behind them. (Pettigrew and Adhikari 2009: 420) Disturbances of the dimaag (brain-mind) represent a more severe tier of problems. Brainmind disturbance symptoms include madness, psychosis, aggression, and unsocial behavior. The association of madness and anger is common in Nepal, with the inability to control one's anger seen as a symptom of madness and mental illness (Kohrt and Harper 2008). The community stigmatizes brain-mind problems more than heart-mind problems because of a fear of contagion and consideration of these brain-mind problems as permanent, whereas heart-mind disturbances may be transient. One of the key findings from the cultural consensus analysis and the semi-structured interviews was the interpretation of aaghaat. Whereas NGO and health professionals used this term to describe trauma as mental shock, the public often did not understand the term. More importantly, laypersons identified aaghaat in the paagal and riis (madness and anger) cluster of brain-mind disturbance. Thus, aaghaat when used by health professional has a potentially negative effect for those seeking care and their families because the term is associated with highly stigmatized conditions. To summarize, a Nepali ethnopsychology and ethnophysiology of trauma comprises vulnerability to traumatic events, with blame placed upon sins in past lives and sins of family members for the experience of these events. Events that cause suffering cover a range of experiences from chronic inability to meet ones needs to discrete 'unforgettable' events. The events produce suffering based on different elements of the self. Events that disturb the heart-mind cause sadness, despair, anguish, and worry. Moreover, disturbances of the heartmind lead to intrusive memories of the event. Events can disturb the brain-mind causing anger and/or madness. The frightening component of the event leads to rupture of the soul from the body with the concomitant effects of soul loss such as vulnerability to physical illness. The physical body may be damaged by a traumatic event leading to headaches, body pain, gastrointestinal problems, vision changes, and other somatic complaints. The experience of these events threatens one's social self or social honor. Because traumatic events typically imply poor karma, experiencing traumatic events is associated with a loss of social status. Idioms of distress and ethnopsychology/ethnophysiology can be incorporated into care for psychological trauma in a number of ways. Mental healthcare professionals, psychosocial workers, and other care providers will encounter survivors of trauma in a variety of settings. The framework described below fosters a multi-disciplinary approach in which individuals address needs at the point of service entry and make referrals to other sources of care. First, when trying to elucidate exposure to traumatic events, one of the most effective pathways may be focusing on the inability to forget certain events. In clinical settings, the phrases tapaaiko kunai birsane nasakne dargatana bhaeko chha? (have you had any events/accidents that you cannot forget?) and dukha lagne ghatanako gahiro chap baseko (having deep and long-lasting impression of the terrifying event) are effective for eliciting traumatic experiences. The type of symptoms most distressing to the individual should dictate initial treatment and referrals. Figure 3 presents treatment entry points for psychological trauma based on presenting idioms of distress. For example, if an individual or his/her family is most concerned about physical problems such as headaches, vision changes, or other somatic complaints, they are likely to first present to a health post worker or physician. In these cases, it would be beneficial to adequately address the possibility of underlying physical pathology as well as identify acceptable referrals for additional support. One of the dangers is ignoring possible treatable physical illnesses that are underlying or comorbid with psychological distress (Kohrt, et al. 2005). If a person presents with psychotic complaints or signs of mental illness, it would be more appropriate to highlight a combination of psychiatric care and counseling for treatment. If a survivor of trauma has a substance abuse problem, as was described by counselors in this study, then psychiatric inpatient substance abuse treatment may be necessary. Among Cambodians, anger may be a better indicator of trauma exposure than PTSD (Hinton, et al. 2009a), and therefore screening for anger may be an effective approach to begin considering intervention to alleviate trauma-related distress. In this Nepali sample, anger was the most proximal emotion to psychotic behavior. Thus, anger, should similarly be a potential screen for trauma, especially among those presenting to psychiatrists. In contrast, disturbances of the heart-mind are not seen as physical or mental pathology but rather a form of natural distress in response to certain types of events. Therefore, heart-mind problems may best be served by encouraging support from friends and family, supporting traditional healing if desired, and provision of psychosocial counseling. One form of traditional healing done by dhami-jhankri (shamans) in Nepal is man badne, which means to bind the heart-mind. This is seen as a way to control the distressing processes or bad memories, and has been reported as a form of healing sought out by former child soldiers (Kohrt forthcoming). Counseling in Nepal has been developed around the construct of the heart-mind, rather than a treatment for mental illness related to brain-mind dysfunction (Kohrt and Harper 2008). The literal translation of the Nepali term commonly used by psychosocial NGOs for 'counselor' is manobimarshakarta (person who advises on matters of the heart-mind). If the main complaint after a traumatic event is soul loss, it may be appropriate to recommend a traditional healing ceremony. Because the soul brings vitality to the body, the return of the soul is necessary before one's body, heart-mind, or brain-mind can begin to heal. Traditional healing for return of the soul can be done alongside medical treatment to heal the body (Kohrt and Harper 2008). Traditional healing for many forms of suffering, not only fright or traumatic events, has been described in many ethnic groups in Nepal (Desjarlais 1992;Maskarinec 1995;McHugh 2001;Nicoletti 2004;Peters 1981). If a primary concern is loss of social status related to experiencing traumatic events, this could be remedied through a number of mechanisms. One option is traditional or religious healing. Religious healing can be seen as a way to account for poor karma. Performing offerings or rituals for sins in a past life may prevent the experiences of future traumatic events in one's current life. These rituals also improve social perception and increase social support of sufferers (Desjarlais 1992;Maskarinec 1992;Peters 1981). Often if a child is identified as having bad karma or having been born in unfavorable astrological context, rituals can be performed to prevent a life of repeated traumatic events. Loss of social status also can be addressed through community-based social inclusion activities. Many sufferers of traumatic events are afraid they will be excluded from interactions with other. If social inclusion activities can be facilitated for trauma survivors, this could help ameliorate fear of exclusion. Thirdly, counseling can focus on the feelings of low self-worth and perceived interpersonal disruptions. Counselors, through supportive listening, can present a stable social connection for trauma survivors, as well as provide psycho-education and problem solving skills (Jordans, et al. 2007;Tol, et al. 2005). Ultimately, regardless of whether a trauma survivor presents to a psychosocial counselor, physician, traditional healer, or psychiatrist, it will be beneficial to support pluralistic healing for psychological trauma. Individuals may present with somatic complaints, but treating these uncovers evidence of heart-mind disturbances or soul loss. As Figure 3 illustrates, the practitioner can refer the individual and their family to other types of healers, in accordance with their religious, spiritual, and health-behavior beliefs. Likely, the individual and his/her family already will be seeking multiple forms of support. One of the crucial aspects of care is reducing and preventing stigma. Psychosocial nongovernmental organizations (NGOs) in Nepal and elsewhere have cautioned against employing terminology that may stigmatize program participants (IASC 2007). In Nepali, the term kalanka refers to a physical blemish or spot, but also to slander, vilify, or misrepresent to harm another person (Turner 1931). In Nepal, mental health problems historically have been associated with kalanka. This is particularly true for psychological problems perceived to be the result of brain-mind dysfunction (Kohrt and Harper 2008). Thus, it is crucial to consider how interventions inadvertently may exacerbate stigma and lead to barriers to care. One specific conclusion that can be drawn from this research is the problematic use of the idiom maanasik aaghaat (mental shock/trauma). While healthcare providers perceived this as a possible linguistic corollary for psychological trauma, the lay public did not always understand the term, and when they did, it was perceived as referring to the highly stigmatized condition of brain-mind dysfunctions such as madness and uncontrolled anger. If a healthcare professional or NGO worker was to label someone as having maanasik aaghaat, the individual and his/her family may perceive this as an incurable condition. Therefore, we would recommend language such as maanasik piDaa, which is more informative and less pathologizing than aaghaat. Another aspect of stigma is the cultural model of trauma causality in Nepal, which may differ from Western conceptions. Hinton and Lewis-Fernandez (in press) point out that effective evaluation and treatment of disorders necessitates understanding the stigma linked with idioms of distress. Summerfield (2001) and others have suggested that PTSD is a desirable label because it removes culpability from the individual and often is associated with receipt of monetary or other support through government policies or lawsuits. In contrast, traumatic events in Nepal can be perceived as the result of personal infractions in past lives and sins of one's family and ancestors, which are transmitted through karma into this life. Thus, people may be ashamed to report experiencing traumatic events. In addition, families may be reluctant to acknowledge or seek help for traumatized family members because it reflects badly on the family's and ancestors' religious piety. In addition, the experience of a traumatic event may differentially stigmatize certain family members. We found that mothers often blamed themselves or were blamed by others for traumatic events occurring to family members. Dahal's (2008) work among war widows in Nepal finds that families and communities often blame and ostracize widows for the deaths of their husbands. The issue should be considered in both public health approaches and individualized care. Despite claims of PTSD's desirability in Western context, most mental health workers in Western settings likely will have seen clients who also blame themselves for traumatic events without having to invoke the concept of karma. Western therapeutic approaches for these clients may be a worth considering for Nepali survivors of trauma as well. As mental healthcare and psychosocial support resources grow in Nepal, it will be important to address stigma hand-in-hand with development of care systems. If stigma is not a major focus of the mission to make mental healthcare and psychosocial services available to trauma survivors, there is a risk that newly established services will go unused or that interventions could make trauma survivors more vulnerable through stigmatization. --- Conclusion This research revealed that psychological trauma is not a unitary construct in Nepal. Rather, psychological sequelae of negative life events are expressed through an array of idioms of distress. The Nepali ethnopsychology/ethnophysiology both overlaps and diverges from DSM psychiatric classification of PTSD. In Nepal, the response to negative life events includes multiple possible manifestation such as general distress and inability to control memory reflected as a disturbance in the heart-mind, behavioral manifestations of anger and madness reflecting dysfunction of the brain-mind, a fright response resulting in loss of the soul, physical impairments with damage to bodily systems, and a threat to personal and familial social status. Our findings suggest that exploring a range of idioms of distress incorporated into a broad understanding of local ethnopsychologies and ethophysiologies can foster effective communication with clients. This is more helpful than trying to identify single terms to approximate the concept of psychological trauma. Similarly, others also have advocated that trauma treatment across cultural settings should comprise strong holistic interventions that keep PTSD in mind but also intersect more broadly with issues of guilt, somatic complaints, and other manifestations of distress (McKenzie, et al. 2004). A primary limitation of this study was the presentation of only one ethnopsychological/ ethnophysiological model. Ethnic composition likely influences ethnopsychological models significantly. The analysis of emotions presented here is based on interviews primarily with a sample representative of the ethnic heterogeneity of the capital Kathmandu. Therefore, the model produced illustrates how psychological trauma is communicated within this specific demographic. It is not representative of rural, low caste, or non-Newar ethnic minority (Janajati) groups who more commonly reside outside the capital. Therefore, the model does not speak to all ways of viewing psychological trauma in Nepal. Pettigrew (2009) provides a description of the understanding of fear among Tamu-mai (Gurungs) in western Nepal that sheds light upon possible ethnic diversity in traumatic experience. Ultimately, there is not one single totalizing ethnopsychology of trauma in Nepal because of the range of ethnic, linguistic, and socioeconomic groups. Rather than relying solely on the model provided here, we offer this study as a framework for practitioners working with trauma survivors in Nepal and elsewhere. We suggest that they develop their own ethnopsychological models when caring for specific populations. Another limitation is the rapidity of change and trends in the health and development fields. In both Western and other settings, the terminology for psychological trauma is shifting continuously. For example, Dwyer and Santikarma (2007) have written about the labeling transition from 'crisis clinics' to 'trauma clinics' in Indonesia. Public mental health campaigns should address the public interpretation of the evolving terminology for psychological trauma. The public interpretation of terminology may have unforeseen consequences in a world of rapidly changing vocabulary for distress. We found that clinicians' attempts to employ a linguistic corollary for PTSD led to use of a relatively unfamiliar term (aaghaat) that the public found stigmatizing. Rapidly shifting idioms, both within lay and professional language, further necessitate grounding within ethnopsychology/ ethnophysiology, which likely follows transformations that are more gradual. From an intervention and clinical standpoint, the alleviation of suffering starts with local frameworks, needs, concerns, and processes of identification. Ultimately, healing begins with understanding and understanding begins with listening to local models as expressed through idioms of distress. Ethnopsychological model with idioms of distress for psychological trauma Treatment entry points for psychological trauma based on presenting idioms of distress. (Gautam 2001;Schmidt 2000;Singh and Singh 1991).

In the aftermath of a decade-long Maoist civil war in Nepal and the recent relocation of thousands of Bhutanese refugees from Nepal to Western countries, there has been rapid growth of mental health and psychosocial support programs, including posttraumatic stress disorder (PTSD) treatment, for Nepalis and ethnic Nepali Bhutanese. This medical anthropology study describes the process of identifying Nepali idioms of distress and local ethnopsychology and ethnophysiology models that promote effective communication about psychological trauma in a manner that minimizes stigma for service users. Psychological trauma is shown to be a multifaceted concept that has no single linguistic corollary in the Nepali study population. Respondents articulated different categories of psychological trauma idioms in relation to impact upon the heart-mind, brain-mind, body, spirit, and social status, with differences in perceived types of traumatic events, symptom sets, emotion clusters, and vulnerability. Trauma survivors felt blamed for experiencing negative events, which were seen as karma transmitting past life sins or family member sins into personal loss. Some families were reluctant to seek care for psychological trauma because of the stigma of revealing this bad karma. In addition, idioms related to brainmind dysfunction contributed to stigma while heart-mind distress was a socially acceptable reason for seeking treatment. Different categories of trauma idioms support the need for multidisciplinary treatment with multiple points of service entry.

Introduction Access to safe and clean housing is a human right fundamental to the health and wellbeing of refugees. 1 Refugees, forced to flee their homes, families and livelihoods, often struggle to find decent housing in host countries. 2 Access to quality housing is particularly difficult for the most economically disadvantaged refugees who are reliant on humanitarian assistance to cover the cost of rent and other basic necessities. 3 These conditions describe the reality for many Syrian refugees living in Lebanon persevering through great adversity. In an effort to better understand and advocate for the needs of refugees, this paper explores the relationship between housing, social wellbeing, access to services and health among Syrian refugee children living in Lebanon. --- Housing and health The connection between housing conditions and physical and mental health is well established in the literature. 1,4,5 Poor housing conditions like cold, dampness and poor ventilation are associated with respiratory diseases and lung infections and are linked to excess winter mortality. 6 Other housing characteristics (e.g. type of waste disposal, water service, food storage, pests) may result in a multitude of health problems, including other chronic and infectious diseases. 7 Housing Fig. 1 Housing and public health framework 6. environments presenting safety risks have been linked with physical injuries, such as those resulting from fires, falling and accidents, particularly among children and the elderly. 8 While housing directly affects health, it also functions as a powerful social determinant of health withdownstream effects. 5,6 Shaw 6 developed a conceptual framework (Fig. 1) that explains the ways in which housing affects health. The framework presents health as affected through both 'hard' (physical/material) and'soft' (social/meaningful) pathways. Hard pathways include aspects of the physical environment (e.g. heating, ventilation, crowding, access to services, public sanitation). Soft pathways are mediated by perceptions and social experience (e.g. not feeling at 'home', social status, community and social capital) that may promote psychological strain and poor mental health. Shaw's framework identifies spatial dimensions to these pathways. Specifically, there are 'direct' (at the individual/household level) and 'indirect' (in the neighbourhood/area) routes through which hard and soft pathways influence health. This study applies Shaw's framework as an analytical lens through which to understand the relationship between housing and the wellbeing of Syrian refugee children. --- Syrian refugees in Lebanon In 2015, Lebanon became the country with the highest number of refugees per capita as it hosts more than 1 million Syrian refugees living among a Lebanese population of over 4 million. 9 Approximately 952,562 Syrian refugees reside in Lebanon, with more than half (55.4%) under 18 years of age. 10 These numbers released by the United Nations High Commissioner for Refugees (UNHCR) include only the refugees that have registered with the organization, but some sources estimate that the total number of Syrians in Lebanon is nearer to 1.5 million. 11 The population of Syrian refugees surged in Lebanon from the start of the Syrian conflict in 2011 through 2016. 10 The Lebanese context is more complex than that of other countries hosting refugees since it is not a signatory to the 1951 Geneva Convention and its 1967 Protocol, which provide basic rights for refugees. Instead, the Lebanese government works with the UNHCR through a Memorandum of Understanding that grants asylum seekers temporary residence with minimal social protections. 12 The influx of refugees into Lebanon put a strain on the country's economic, social and political landscape. 13 This has led to a climate of hostility towards refugees whom Lebanese nationals view as competition for jobs and social services. 14 Echoing these views, the Lebanese government has instituted policies limiting refugee access to work opportunities and other forms of social support. 13 Moreover, it was hesitant to build official refugee camps because of fears that Syrian resettlement may be long-term and affect the balance of power within the country. 13 This resulted in the proliferation of informal settlements that do not meet the requirements of safe and healthy housing. The living conditions in the informal settlements have uniquely affected Syrian children who experience multiple forms of deprivation, including poverty, food insecurity, social marginalization and hazardous living conditions. 14,15 These conditions undermine the health of vulnerable Syrian children residing in Lebanon. --- Research objectives Housing has been identified as one of the main factors affecting the physical and mental health of refugees. 1,16 For Syrian refugees in Lebanon, housing and health conditions are evidently tied to their deprivation and intimately reliant on their status as refugees. Among Syrian refugees residing in Lebanon in 2017, 34% live in makeshift communities known as informal tented settlements or non-residential structures (like garages). 17 The highest concentration of Syrian refugees in Lebanon reside in the Beqaa region in the eastern part of the country (340,512 individuals). 10 About 77% of refugees living in Beqaa, the site of the present study, are housed within informal tented settlements in remote and deprived areas. 18 These refugees are among the most vulnerable in the country as they are subject to harsh weather conditions and other lifethreatening hazards. Research on the health of refugees and children living in poor quality housing is scarce, as most of the literature focuses on resettlement efforts in wealthier Western nations. 1 This paper offers a contribution to the scholarly discourse at the intersections of refugee studies, housing and health. Using Shaw's framework, it explores the relationship between housing, social wellbeing, access to services and health among a population of Syrian refugee children in Lebanon. --- Methods This research is based on a cross-sectional survey conducted in 2017 with Syrian refugees living in informal tented settlements in Lebanon's Beqaa Valley. The 2017 study investigated child labour among Syrian refugees in Lebanon. More information on the methodological approach of the study can be found in a previously published paper. 15 Out of 3748 informal tented settlements in the Beqaa, a random sample of 153 informal tented settlements (see Fig. 2a) was selected from the Interagency Mapping Platform, a database for information on Syrian refugees living in informal tented settlements across Lebanon. 19 We recruited 33 fieldworkers, who attended a 7-day training workshop, and selected 27 for fieldwork. Data collection commenced on 14 August 2017 and was completed on 27 November 2017. Field staff coordinated with a local'shaweesh', a community gatekeeper who connects migrant workers with employers, 20 in each area to identify households in the sampling frame with working children between the ages of 4 and 18. The female homemaker in each household was interviewed in the survey since previous research experiences in Lebanon found that female homemakers are generally more knowledgeable about household affairs than other residents. 21 Information on children's (0-18 years) life experiences in Lebanon was collected. We surveyed 1902 households where 8284 children (0-18 years) lived. --- Questionnaire and measures A household questionnaire was developed and administered through face-to-face interviews with the female homemaker in colloquial Arabic. It included sections on the health status of all household members, household income, housing conditions, food security, awareness about presence of social services, being cautious when dealing with others, trusting merchants and feeling happy in the neighbourhood. We used a number of indicators that reflect the components of Shaw's model (Fig. 1). To describe the characteristics of housing, an index of housing problems was developed using 14 items including insect and rodent infestation, water leakage from the tent, holes and cracks in the walls, humidity and others describing housing conditions 4,22,23 (see Appendix). Each item was given a score of one, so the maximum score was 14. We categorized the housing problem index into three categories: six problems or less, seven to eight problems and more than eight problems. Monthly household income and food security were used to describe income and wealth. Monthly household income was used as a continuous variable. Food security was measured through (i) the Reduced Coping Strategies Index (RCSI), 24 which includes 5 individual coping behaviours, and (ii) an adapted version of the Livelihood Coping Strategy Index 25 using 10 longterm coping strategies including selling household goods, borrowing money to purchase food, withdrawing children from school and others (see Appendix). We categorized food security into three categories: food secure, food insecure and severely food insecure. Awareness about social services (yes, no) was used as a proxy measure of availability of services in the neighbourhood in Shaw's model. The indicators describing caution and trust in dealing with others (yes, no) were used as proxies for social capital and cohesion. 26,27 In addition, the indicator describing being happy in the neighbourhood (yes, no) was used as a proxy for the sense of belonging to the neighbourhood. 6,28 This study assessed health through self-reports of chronic and acute illness (3-month prevalence) and disability. Homemakers were asked about the prevalence of health problems among household members. When a health problem was reported for each household member, the homemaker was asked to specify which illness was present based on a list provided by the interviewer. For the purposes of analysis, we categorized health data such that each household member was identified as having zero to one, only two or three or more health problems (since the counts of health problems were minimal above three). Independent variables were adapted from Shaw's housing and health framework (Fig. 1), 6 which described household/neighbourhood and physical/social pathways through which housing affects health. The following indicators were utilized: • Physical/household: household income per capita, food insecurity, housing problems • Physical/neighbourhood: awareness of available social services • Social/household: feeling happy with the neighbourhood • Social/neighbourhood: cautious when dealing with people; trusting merchants --- Statistical analysis We performed a descriptive analysis of all household members 18 years and younger. The analysis reports included data on demographics, health, housing characteristics and wealth indicators. Frequencies and percentages were reported for categorical data and means and standard deviations (SD) for continuous data. The relationship between these measures and health was assessed using cross-tabulations and six logistic regression models (one for each health indicator as per Tables 2 and3), adjusting for the district, sex and age of the household member and for the effect of clustering at the household level. We considered an alpha value of 0.05 as statistically significant and conducted all analyses on Stata 15.0 (StataCorp). --- Results Among the 1902 households surveyed, the average household had 6.7 members including 4.36 children. The monthly household income per capita was 50.7 USD compared to a monthly per capita expenditure of 119.7 USD, which puts households in debt and makes them dependent on cash assistance provided by the UN and other international organizations to cover their expenses. 17 The mean age of household heads was 42.8 years. About 40% were reading illiterate, while about 21% had completed any formal education beyond primary school (data not tabled). Of the 8284 children, the mean age was 9.8 years, 50.9% were male, and 55.5% were actively working. Their sociodemographic characteristics are presented in Table 1. A third (n = 2736, 33.0%) of the children were reported by homemakers to have had a health problem. The prevalence of chronic illness was 11.9% (n = 985), while the prevalence of acute illness in the last 3 months was 24.2% (n = 2005). The most common chronic illnesses were allergies and asthma (n = 198), eye problems (n = 116) and anaemia (n = 95). The most prevalent acute illnesses were cold and flu (n = 1364), fever (n = 898) and diarrhoea (n = 260) (data not tabled). Around 98% of the households lived in makeshift tents, primarily made of fabric (see Fig. 2b). A considerable number of households (43.1%) had greater than eight housing problems, 39.6% had seven to eight housing problems, and 17.3% had zero to six housing problems. The most common housing problems were having insects inside the home (95.1%), humidity (92.7%) and having rats or mice inside the home (91.3%). A full list of housing problems can be found in Appendix. Nearly three-quarters of households (74.3%) were severely food insecure, 21.1% were food insecure, and 4.6% were food secure (data not tabled). Tables 2 and3 present associations between health and hard/physical and soft/social housing indicators. All four dimensions of Shaw's framework were associated with health, but this was inconsistent across health indicators. Regarding hard/physical pathways, households reporting increased housing problems had higher odds to have children with an acute illness in the last 3 months. Housing problems were also connected to multi-morbidity, and the magnitude of association was larger with more health problems. For example, children living in dwellings with more than eight housing problems had higher odds to have two health problems (adjusted odds ratio [AOR], 1.75; confidence interval [CI], 1.25-2.47) and to have three health problems (AOR, 2.39; CI, 1.50-3.81) than those living in households with six or less housing problems (Table 3). The prevalence of adverse health problems also increased with worsening food security status; this was statistically significant only in one model presented in Table 3. Here, children living in households experiencing severe food insecurity had higher odds to have one health problem than those living in food secure households (AOR: 1.75, CI: 1.11-2.76) (Table 3). Neighbourhood aspects of the physical environment were also associated with health. Lower rates of chronic illness, but higher rates of acute illness, were associated with being unaware of local social services for Syrian refugees. For example, children living in households that were unaware of local social services were at lower odds to have a chronic illness (AOR, 0.80; CI, 0.66-0.97) than those who were aware of such services (Table 2). Multi-morbidity returned mixed results, where children had higher odds to have two health problems when their households were unaware of services (AOR, 1.30; CI, 1.01-1.66) (Table 3). Lack of awareness of these services did not significantly affect multi-morbidity otherwise. Soft/social pathways were significantly associated with adverse health conditions. At the household level, not feeling 'happy in the neighbourhood' was associated with having any health problem (AOR, 1.54; CI, 1.32-1.81), a chronic illness (AOR, 1.31; CI, 0.10-1.54) and an acute illness (AOR, 1.63; CI, 1.34-1.97). A similar trend was present in the analysis on multi-morbidity. Children living in households that reported being unhappy with the neighbourhood had higher odds to have two health problems (AOR: 1.63, CI: 1.31-2.01) and higher odds to have three or more health problems (AOR, 1.70; CI, 1.30-2.22) than children living in households that reported being happy with the neighbourhood (Table 3). The neighbourhood social environment was also significantly associated with individual health and multi-morbidity. Children living in households who reported not being cautious when dealing with people outside of their home had lower odds of having any health problem (AOR, 0.53; CI, 0.38-0.74), having a chronic illnesses (AOR, 0.61; CI, 0.43-0.85) and having a recent acute illness (AOR, 0.43; CI, 0.26-0.69) than those living in households who reported being cautious (Table 2). --- Discussion --- Main findings of this study Our findings support the use of Shaw's framework as an analytical tool to understand and interpret the relationship between housing and health. Shaw's framework operationalizes the effects of proximal and distal factors within the lived environment and their effects on health. Our findings showed that elements of the social environment at both the household and neighbourhood levels were strongly associated with poor child health. The material/physical aspects of housing were also associated with poor health. The most common chronic and acute illnesses were respiratory illnesses. Also, severe food insecurity was significantly associated with poor health. --- What is already known on this topic The percentage of Syrian refugees living in informal settlements across Lebanon has increased from 17% in 2017 to 19% in 2018. 17 Shelters housing these refugees have been described as hazardous, structurally dangerous and lacking in basic needs such as water, electricity and appropriate sanitation facilities. 18 Previous studies 29,30 have reported on the housing and health conditions of Syrian refugees living in informal tented settlements; however studies adopting a multidimensional approach that includes the social and physical living environment and its effects on health are lacking in the literature. --- What this study adds The health profile of this sample is fairly inconsistent with what might be expected among other populations of similar age. Children in our sample had lower reported rates of chronic illness (11.9%) than found elsewhere in the literature. For example, the prevalence of chronic illnesses was 16.6% for children living in the USA, 31 18% among adolescents in New Zealand 32 and 19% among school-aged children in Portugal. 33 This likely relates to a lack of access to comprehensive health services that might detect and diagnose chronic illnesses at an earlier point over the life course of responding children. If in fact the prevalence of chronic illness is largely underreported, then multi-morbidity is also likely understated. The main hard factors affecting health were physical housing conditions, with the strongest associations to the prevalence of acute illnesses and multi-morbidity. These results confirm previous research linking poor housing to acute illnesses. 34,35 Further, a higher prevalence of acute illnesses within households with more housing problems may explain why these households reported higher rates of multi-morbidity among child residents. A recent review of studies evaluating the effect of housing improvements on health found that minor housing improvement, such as modifications to household heating and ventilation technologies, significantly affected the prevalence of acute illnesses and mortality. 36 Numerous studies have documented cost-effective and suitable improvement strategies that may be applicable in this context, for example, installing a deployable, lightweight, insulating floor inside refugee tents, 37 providing households with improved water supply, 38 introducing urine diverting toilets 39 and adopting labyrinth path design for improved visibility inside camps. 40 One challenge to the practicality of housing improvements is the temporary nature of informal tented settlements. Interventions must consider types of housing improvements that are sustainable, appropriate and acceptable by the host communities given current political tensions in the country. Recently, some 1500 Syrian refugees living in informal tented settlements on the banks of the Litani River, the longest river in Lebanon and a principal water resource in the country, were evicted from their tents for dumping sewage directly into the river, which is used for irrigation. [41][42][43] Although refugee camps are only one of many sources of pollution in an area housing a number of industries in the Beqaa, 44 these evictions reveal how refugee housing conditions and poor sanitation infrastructure could become intertwined with political, social and environmental factors that increase the precariousness of vulnerable populations. The analysis found significant associations between physical factors in the neighbourhood environment and health. A seemingly counterintuitive finding is the lower rate of children with chronic illnesses among households unaware of local refugee services. A possible explanation for this result is that households with chronically ill children are more likely to seek out services for treatment than households without chronically ill children. The findings also point to the salience of social factors both at the individual and neighbourhood levels. Children who lived in households that reported feeling unhappy with the neighbourhood experienced poorer health across the board. This result may reflect a dynamic where members in households with children experiencing health problems are less likely to report being happy about their situation. Future studies may want to employ qualitative research to explore the difference between place-related happiness and the psychological strain of illness and the degree to which illness affects the quality of life. The finding linking poor health in children with households' perceptions of the need to approach neighbours with caution has many potential explanations and implications. This indicator serves as a proxy for an aspect of the social neighbourhood environment, namely, social cohesion/fragmentation. 26,27 This may be related to previous incidents where informal tented settlements were targeted for removal because of tensions or security concerns. 45 Children living in households reporting this indicator may have experienced trauma that has ramifications for their wellbeing. Similarly, perceiving the need for caution may reflect living in areas where discrimination is common enough to affect children's engagement in health protective behaviours, such as outdoor physical and social activity. If residents are reluctant to leave their homes, they may stay within overcrowded domiciles, increasing their exposure to environmental health risks. --- Limitations of this study As a cross-sectional study, this research cannot make causal inferences about the associations identified. Further, the study had design limitations that may have affected the findings. Specifically, self-reports of health, particularly of chronic illness, may be unreliable given the study population's limited access to adequate health-care services that identify health problems in children. Moreover, all health conditions were given equal weighting in the construction of health indicators not accounting for the seriousness of the health problem. A potential area of sampling bias was the focus of this research on households with working children. This likely biased the sample towards the most socioeconomically vulnerable households among the Syrian refugee population in the country, thus limiting generalizability. In fact, the pervasive conditions of deprivation among responding households presented a challenge in the analysis, as many households were quite homogenous in the types of housing problems and poverty conditions they reported. --- Conclusion In the context of the protracted Syrian war, children are among the most vulnerable refugees. This is especially the case for refugees living in informal makeshift shelters. Such housing is inadequate and dangerous thus negatively affecting the health of residents and threatening their lives. Moreover, this research confirms the value of recognizing the physical and social pathways and proximal and distal factors mediating the relationship between housing and health. Without multidimensional interventions that consider improvements to housing and living conditions, the health of young Syrian refugees will continue to worsen. Such interventions could target the living conditions of refugees on multiple levels: the physical housing and infrastructure, access to services, food security and the neighbourhood environment. --- Supplementary data Supplementary data are available at the Journal of Public Health online. --- Appendix Additional methodological details --- Housing problems A housing problems index 4,22,23 was developed using the following 14 items: --- Food security Food security was measured using two indices: 1. The RCSI, which includes five individual coping behaviours (eating less preferred foods, borrowing food/money from friends and relatives, limiting portions at mealtime, limiting adult food intake and reducing the number of meals per day). The RCSI is used to measure the presence and severity of food security across different contexts and settings. 24 --- An adapted version of the Livelihood Coping Strategy Index 25 --- Quality control Following recruitment and survey completion, each household was revisited by a Quality Control Team to check whether (i) the tent was actually visited by the data collector, (ii) the female homemaker was interviewed whenever possible or was replaced by another member of the household, (iii) every working child (aged >8 to <unk>18 years old) in the household was interviewed, and (iv) the month and year of birth for family members were reported according to legal papers. The Quality Control Team detected and corrected inconsistencies in 8% (n = 152) of the households, mainly relating to the year of birth of household members.

Background This cross-sectional study explores the relationship between housing, social wellbeing, access to services and health among a population of Syrian refugee children in Lebanon.We surveyed 1902 Syrian refugee households living in informal tented settlements in Lebanon in 2017. Logistic regressions assessed relationships between housing problems, socioeconomic deprivation, social environment and health.Of the 8284 children in the study, 33.0% had at least one health problem. A considerable number of households (43.1%) had > 8 housing problems. Children in these households had higher odds to have three or more health problems compared to children in households with < 6 housing problems (adjusted odds ratio [AOR], 2.39; confidence interval [CI], 1.50-3.81). Nearly three-quarters (74.3%) of households were severely food insecure. Children in these households had higher odds to have one health problem than those in food secure households (AOR, 1.75; CI, 1.11-2.76). There was a significant positive association between households that reported being unhappy with their neighbourhood and the number of children with health problems in those households.This study highlights the association between the physical and social living conditions and refugee children's health. Without multidimensional interventions that consider improvements to living conditions, the health of young Syrian refugees will continue to worsen.

Introduction Do men and women authors differ in their use of tables and graphs in their research articles? In the 1960s it was generally thought in the West that men were more spatially and mathematically oriented than women and that women were more verbal (Maccoby and Jacklin 1974). This division was reflected in the major intelligence tests of the time that typically came in two halves-a verbal and a spatial one (e.g., see Heim 1970). And, in addition, males performed better than females on spatial rotation tasks (Maeda and Yoon 2013). Further, it was thought (and often still is) that girls were better at writing than boys (Peterson and Parr 2012). More recently, however, these divisions have been less subject to debate than they were. It is now generally thought that men and women are equally adept at both spatial and verbal tasks (e.g., see Hyde et al. 2008), with perhaps men having wider tails in the distributions of their scores (Robinson and Lubienski 2011). Similarly, some researchers argue that the differences between boys and girls in terms of their academic writing have also lessened (Hartley 2008). Indeed, popular thinking in education in the UK has generally switched from 'helping the girls' in the 1960's to currently 'helping the boys' to achieve their potential (Strand et al. 2006). However, the matter is still debated. There are still gender gaps in certain disciplines, with many more men than women studying subjects like mathematics, and engineering, and many more women than men studying subjects like psychology and biology (Good et al. 2012;Hartley 2006). And there are still the remnants of the earlier differences. 1 Wai et al. (2012) recently examined differences between men and women university students at the top end of the mathematical and verbal distributions of test scores (where one might expect to find potential academics). Here the men still scored higher than the women on tests of mathematical ability (although not as highly as in the past), but there were now fewer differences between them on tests of verbal ability. Similarly, the findings are mixed in terms of students' academic writing. Here a number of studies have looked to see whether or not men write differently from women students in a number of different situations [e.g., using e-mails, writing examination articles, and in academia: see (Hartley 2008, pp. 161-162)]. Hartley (2008, pp. 163-164) reported the results obtained from comparing the writing styles of men and women in six genres-academic book reviews; academic articles; student essays; tabloid newspapers; novels; and magazine fiction. He found that the writers (of either sex) tailored their writing to the style required by the genre: academic texts were hard and difficult to read, students essays were a bit clearer; newspaper articles a bit more clear; and novels and magazine fiction easier to read. But this pattern of performance was the same for the men and the women: there were no significant differences between their writing styles, whereas there were significant differences between the results found for each genre. Hartley (2008) thus argued that men and women did not differ in their writing styles-they just adapted them to fit the situation. Perhaps the same might be said of men and women using tables, figures and graphs, and thus we would not predict differences between them in using these academic tools? Nearly all of the studies cited above have, of course, involved English speaking participants. Studies carried out in other cultures may reveal similar or more varied findings. Isiksal and Cakiroghi (2008), for example, found a similar decline over time in the differences between boys and girls in mathematics in Turkey, but Park (2005) reported that male Korean students achieved significantly higher mathematical scores than did their female counterparts. --- Previous research on tables and graphs in academic journals There are several famous books and articles on the different functions of tables and graphs in academic text (e.g., Carter 1947;e.g., Cleveland 1994;e.g., Gelman et al. 2002;e.g., Smith et al. 2002;e.g., Speier 2006;e.g., Tufte 1983). However, we have been unable to find any articles on the use of graphs and tables by men and women. Hegarty and Walton (2012) did, however, include these variables in their study of factors in Psychology journals that affected citation rates and impact factors, but no important sex differences were found with respect to tables and graphs. In this article we ask whether men and women academic writers differ in their use of tables and figures in academic articles. In Study 1 we ask this question with respect to science journals. In Study 2 we ask the same question for the social science. In Study 3 we ask if there are differences between men and women in the use of what we might call 'text' tables (containing mainly verbal summaries, etc.) and 'data' tables (containing mainly data). Following Hanson et al.'s (2011) recommendations, we release the data retrieved and used in this study as an Electronic Supplementary Material. Arguing from the standpoint that men might be more mathematically and spatially oriented, and women verbally so, we derived the following hypotheses: H1 Men will have more proportionally graphs and figures in their scientific articles than women. H2 Women will have proportionally more tables in their scientific articles than men. H3 Men will have proportionally more graphs and figures in their social science articles than women. H4 Women will have proportionally more tables in their social science articles than men. H5 Men will have a higher ratio of 'data' tables to 'text' tables than women in their social science articles. H6 Women will have a higher ratio of 'text' tables to 'data' tables than men in their social science articles. --- Study 1. Science articles Do male and female researchers use the same proportions of tables and graphs in their scientific articles? Hypotheses H1 Men will have proportionally more graphs and figures than women in science journals. H2 Women will have proportionally more tables than men in science journals. --- Procedure To test these hypotheses we needed to: 1. Select acknowledged peer-reviewed journals from a wide range of scientific domains (for generalization purposes). 2. Consider the journals' contents published during a fixed time window (in order to avoid any bias arising from developments in information technology). As it is now easier to produce graphs and figures than it was in the 1980s, we needed to set a recent time window in order to report topical findings. 3. Select all single-authored research articles included in these journals. 4. Retrieve the full-text version of these articles. 5. Count the number of pages (P), the number of tables (T), and the number of figures (F) in each article. 6. Filter the articles so that we retain those with F? T [ 0 (i.e., at least one figure or one table). 7. Normalise F and T to avoid the confounding variables that different articles have different lengths, and longer articles are more likely to contain more tables and figures than shorter articles. To avoid this we need to normalize T and F by P for an average 10-page long article. We obtain T 0 1<unk>4 10 <unk> T=P and F 0 1<unk>4 10 <unk> F=P. 8. Annotate each article as ''M'' or ''W'' according to the author's gender, as inferred from the author's: (a) first name, or (b) online picture, or (c) by deducing it from his/ her online website (looking for pronouns ''his''/''her''). 9. Group the articles by gender and define four samples: M and T 0 W to check if the difference in the use of tables among men and women is statistically significant. Visual inspection is performed with notched boxplots (McGill et al. 1978). A notch is drawn in each side of the boxes. If the notches of two plots do not overlap this is ''strong evidence'' that the two medians differ (Chambers et al. 1983, p. 62). Then, we further the visual analysis by running the non-parametric Mann-Whitney U test on the two samples (two tailed). The null hypothesis H 0 assumes no difference between the ranks of the two samples. 11. Repeat this exercise for the case of figures: compare F 0 M and F 0 W. --- Data In this section we review each point of the Procedure section above to indicate how we implemented and operationalized our research design. • Point 1. We considered the 8,336 science journals listed in the Thomson-Reuters Journal Citation Reports (JCR), 2 Science edition 2011. We then selected the 752 science journals published by Wiley-Blackwell because the full-text of these articles were available online in valid HTML format (suitable for data extraction). We then tagged these journals with their JCR categories from the 176 categories spanning the whole range of domains in the Sciences (e.g., Agronomy, Logic, Pathology). For each category, we retained the journal with highest number of published articles per year. If this journal had already been selected we picked the following one, if it was available. In the end, we considered 148 distinct journals: one per category (when available). The rationale behind this choice was (a) to consider journals from all scientific domains, while (b) maximizing the number of articles considered, and (c) having various levels of impact factors to foster journal diversity. • Points 2-4. We studied a total of 3,576 single-authored articles published in 2011, for which we downloaded the full-text in HTML. • Points 5-6. We extracted P from article metadata. Computing T and F was achieved by extracting and counting specific HTML tags. A total of 1,682 single-authored articles with at least one figure or were further considered. • Points 7-10. We manually identified the gender of 1,403 authors (83 %). We thus ended up with 1,143 articles by men and 260 articles by women (23 %). Significance tests were computed with the SOFA statistical package. --- Results H1 Men will use proportionally more figures and graphs than women in scientific articles. In a typical 10-page science article there were a mean of 4.85 figures (Mdn = 4.00), as showed in Fig. 1. When grouping articles according to their author's gender, we found that men (M = 5.04) used 26 % more figures and graphs than women (M = 4.00). Moreover, this difference is statistically significant (men Mdn = 4.17 vs women Mdn = 3.33, U = 125,863.5, p <unk> 0.001). As a result, our data positively support Hypothesis 1. In a typical 10-page science article there were a mean of 1.83 tables (Mdn = 1.11), as showed in Fig. 2. When grouping articles according to their author's gender, we found that women (M = 1.99) used 11 % more tables than men (M = 1.79). However, this difference is not statistically significant (men Mdn = 1.11 vs women Mdn = 1.36, U = 139,086.0, p = 0.102). As a result, our data do not support Hypothesis 2. --- Study 2. Social science articles Do men and women researchers use the same proportions of tables and figures in their social science articles? Hypotheses According to our Introduction above we hypothesized: H3 Men will use proportionally more figures and graphs than women in social science articles. H4 Women will use proportionally more tables than men in social science articles. --- Data We replicated the procedure used for Study 1, only this time we analysed the presence of tables and graphs in social science journals. Thus, in more detail: • Point 1. We considered the 2,966 journals listed in the JCR Social Sciences edition 2011. We selected 384 journals published by Wiley-Blackwell in HTML format. We tagged these journals with their JCR categories, leading to 56 such categories spanning the whole range of domains in the social sciences (e.g., Anthropology, Linguistics, Sociology). For each category we retained the journal with the highest number of published articles per year if it had not already been selected, whereupon we picked the following journal. We ended up with 54 distinct journals, one per JCR category (when available). • Points 2-4. We studied a total of 2,091 single-authored articles published in 2011, for which we downloaded the full-text in HTML. • Points 5-6. We extracted P from the article metadata. Computing T and F was achieved by extracting and counting specific HTML tags. A total of 662 single-authored articles with at least one figure or one table were further considered. • Points 7-10. We identified the gender of 655 authors (99 %). We thus ended up with 441 articles by men and 214 by women. Significance tests were computed with SOFA statistics. --- Results H3 Men will use more figures than women in social science journal articles. Analysis of the data showed that a typical 10-page long social science article contained a mean of 1.33 figures (Mdn = 0.714), as showed in Fig. 3. When we grouped the articles according to gender, we found that men (M = 1.32) used 3 % less figures than women (M = 1.35). However, this difference is not statistically significant (men Mdn = 0.71 vs women Mdn = 0.77, U = 46,793.5, p = 0.861). So, our data do not support Hypothesis 3 for social science journals. --- Women (214) All ( 655) Men ( 441 In a typical 10-page long social science article there were on average 1.94 tables (Mdn = 1.48), as showed in Fig. 4. When grouping articles according to their authors' gender, we found that women (M = 1.93) used 1% less tables than men (M = 1.95) and that this difference was clearly not significant (men Mdn = 1.50 vs women Mdn = 1.43, U = 46,566.5, p = 0.784). As a result, our data do not support Hypothesis 4. --- Study 3. Social science articles Do men use more 'data' than 'text' tables in their articles in the social sciences, and women more 'text' tables? There are different sorts of tables, of course, as well as different sorts of figures. In the light of our introductory remarks we wished to see whether or not women would use more'verbal' or 'text' tables than men, and whether or not men would use more numerical or 'data' tables than women. We hypothesized that women would use proportionately more 'text' tables than men, and that men would use proportionally more 'data' tables than women. Hypotheses H5 Men will have a higher ratio of 'data' tables to 'text' tables than women in their social science articles. H6 Women will have a higher ratio of 'text' tables to 'data' tables than men in their social science articles. Nonetheless, before we conclude that there is only a limited difference in the use of tables and graphs by men and women, we need to consider some of the limitations of these studies. There are at least four points that should be considered: • Problems with journal sampling. The number of published articles per year is not uniform across all journals. This means that some journals (and thus JCR categories) are more represented than others. For example the Journal of the American Society for Information Science and Technology has 12 issues per year and Psychology Teaching Review only two (not included in this study). Similarly, the number of single-authored (and single-sex articles in Study 3) is not uniform across all journals. And this is also true of the distribution of genders (i.e., although the number of males predominates there are more females in some domains than others). To overcome these problems it might be possible to standardize the contributions in some way, but we did not do this in the present study. • Problems in deciding what to count as a table or as a figure. In Studies 1 and 2 the numbers of tables and figures were determined automatically according to the caption labels (e.g., Table 1; Fig. 2). However, some figures can contain a lot of text, or even (rarely) be labeled in some journals as a figure even when the data are set out in a tabular format, but these differences are not apparent from an electronic database. Nor, too, are whether or not figures contain numbers (we classified them as data tables), graphs, bar charts, maps, or even structural equations. The methods used in Study 3 can obviate this problem to some extent, but (a) they are time-consuming and (b) sometimes we found it hard to categorize a table that contained mainly text, but with one or two numbers in it. Here we classified these as data tables. • Problems with determining the sex of the authors. Determining the sex of the authors from many different countries can be a difficult and error-prone task for people like us who only speak one or two languages, and where-sometimes-only the initials of the first names are given. Although Google Scholar proved very helpful here, in some cases (17 % in Science, 1 % in Social Sciences) we had to give up because it proved too difficult for us to find out whether an author was a man or a women. • Problems with different disciplinary approaches. Finally, what might account for the differences found in Study 1? Do they simply reflect disciplinary habits? Are there generally more graphs in scientific journals than social science ones? Do more women than men write in the social sciences (and the arts) than in the sciences (Schucan Bird 2011)? And is this sex difference now declining with the increase in female researchers, see (e.g., see Kretschmer et al. 2012a, b;e.g., see van Arensbergen et al. 2012)? These articles show that the answer to most of the questions above is in the affirmative, but they also show that there is more than just disciplinary differences to account for our results. In this article we have equated different disciplinary approaches that might in fact in themselves make different uses of tables and graphs. Although we have distinguished between science and social science journals, we have not distinguished between the different methods used within these disciplines. Different theoretical approaches to doing research lead to different types of tables and graphs. Ashwin (2012) for instance differentiated between the use of studies in the social sciences using quantitative, qualitative and mixed methods in the USA and elsewhere. He reported that 73 % of US journals in his sample used quantitative methods, 19 % qualitative and 1 % mixed methods, whereas the percentages were 38, 39 and 12 % for non-US journals. It is likely, of course, that such different methods also use different styles of tables and graphs. In terms of male:female distributions in different disciplines we may note Schucan Bird's (2011) finding that female academics publish proportionally less than male ones (for various reasons) in the sciences, but publish at a comparable level with men in the social sciences. In Studies 1 and 2 we found similar results, in that the male:female ratio was much higher in the sciences (4:1) than it was in the social sciences (2:1), but in our study the male:female ratio in the social sciences was double the 1:1 implied by Schucan Bird (2011). --- Conclusions Our research suggests that there are few, if any, statistically significant differences so we conclude that men and women sometimes do things differently, but most of the time there are large overlaps in their accomplishments. Study 1, in fact, was the only one that provided support for the hypothesis (H1) that men would use significantly more figures and graphs than women in the sciences. Today, the proportion of men and women entering higher education and contributing to research is increasing (van Arensbergen et al. 2012). It would indeed be interesting to try to replicate once again some of the older studies into sex differences in academic writing to see if things have changed. --- Method In this study we had to count and classify the tables produced by men and women by hand. Our previous electronic analyses of journals only counted the number of tables and figures by these designations in the text: they did not report on the kinds of tables (or figures) used. Accordingly what we did was: 1. Select journals that we had to hand (in the social sciences) or that we could read online. 2. Select those articles written by men or by women (in singles or multiples) and discount any articles written by members of both sexes. 3. Examine whether or not there were tables in each of these articles and record a single 'Yes' appropriately for each article if it contained text or data tables, or both. We then totaled the numbers of articles using text and data tables for male and female authors overall. Table 1 gives the results. It can be clearly seen that although a greater proportion of female authors than men use text tables there are no significant differences between them. As a result, hypotheses H5 and H6 are not supported. --- General discussion The main findings from these three studies suggest that there are few, if any, differences between the use of tables and graphs by men and women, except in the Sciences. Study 1, in fact, was the only one that provided support for the hypothesis (H1) that men would use more figures and graphs than women in scientific articles. We wonder, therefore, if this is just typical of the disciplines and that graphs are more common in the sciences or that the use of tables in social sciences might be more effective if they were to be turned into graphs more often (Gelman et al. 2002).

In psychological research there is huge literature on differences between the sexes. Typically it used to be thought that women were more verbally and men more spatially oriented. These differences now seem to be waning. In this article we present three studies on sex differences in the use of tables and graphs in academic articles. These studies are based on data mining from approximately 2,000 articles published in over 200 peer-reviewed journals in the sciences and social sciences. In Study 1 we found that, in the sciences, men used 26 % more graphs and figures than women, but that there were no significant differences between them in their use of tables. In Study 2 we found no significant differences between men and women in their use of graphs and figures or tables in social science articles. In Study 3 we found no significant differences between men and women in their use of what we termed 'data' and 'text' tables in social science articles. It is possible that these findings indicate that academic writing is now becoming a genre that is equally undertaken by men and women.

Background Parents play a significant role in enabling participation of children with a physical disability at home, at school, and in the community. Participationthe involvement in life situations to fulfil social roleshas a positive impact on children's health and well-being [1][2][3]. Children with a physical disability often come across restrictions in their everyday participation [4][5][6]. The most important factor influencing successful participation of children and adolescents with disabilities is the interplay between the child's environment and activities [7][8][9][10][11][12]. Parents have knowledge and experiences how to involve child's social environment (e.g. peers) to support engagement in preferred and desired activities [13]. Family-Centred Service (FCS) is considered best practice in service for children with a physical disability [14]. Its effectiveness is related to the understanding professionals have about needs and actions of both child and parents, including what parents do to support their child with a physical disability [15,16]. A recent scoping review of the literature on this topic [17] revealed 14 studies, which identified several parents' actions, challenges, and needs that were summarised in a preliminary thematic framework. The framework includes two major themes: "parents enable and support performance of meaningful activities of their child at home, at school and in the community" and "parents enable, change and use the environment" [17]. Connected to the major themes, the framework includes three categories (actions, challenges, and needs) and several subcategories with a total of eight actions, eight challenges, and four needs, as presented in Fig. 1. However, the scoping review [17] also underlines how little information is available on what parents actually do every day to enhance their child's participation. Timeuse diaries are particularly relevant and suitable instruments for studying the daily lives of families, including families with a child with a physical disability, as they give insight into what they as a family actually do in context [18]. Participants who fill out the diaries are both observers and informants, providing the researcher a "view from within" [19,20]. The aim of this study is to gain insight into parents' own daily actions, challenges, and needs while supporting their child with a physical disability at home, at school, and in the community. Additionally, the study results will be used to refine the preliminary thematic framework. --- Methods Between May 2012 and July 2012, a qualitative research inquiry was performed based on using a diary over a 7day period. Direct content analysis was used as theoretical framework to systematically organise data into a structured format [21][22][23]. --- Study population Potential participant families of children with a physical disability were selected from the electronic database of the Dutch Association of People with Disabilities and their Parents (BOSK). All potential families had a child aged between 4 and 12 years with a physical disability that is neurological and non-progressive in nature (e.g. cerebral palsy, spina bifida) and who was living at home. Persons identified in the database of the BOSK as the contact person for each family (parent or primary caregiver) were considered eligible for this study; in total 47 parents participated in this study. --- Recruitment procedure In May 2012, 559 (17.2% from all the BOSK members) eligible parents (or primary caregivers) of children who fulfilled the criteria received an information letter from the BOSK inviting them to participate in the study. A stamped return-envelope, an informed consent form, information about the study, and a diary covering a 7-day period, were included. This study was conducted jointly with a quantitative study aimed to give a wide-ranging portrayal of the number, domains, and priority of needs expressed by parents using the Family Needs Inventory -Pediatric Rehabilitation [24,25]. Parents (or primary caregivers) had the choice to refuse cooperation, to fill in the diary or the questionnaire, or both. Parents (or primary caregivers) who did not respond to the first invitation received a reminder from the BOSK after three weeks. --- Data gathering Participating parents (or primary caregivers) returned the completed diary and/or questionnaire as well as the informed consent form in the return envelope directly to the Centre for Data and Information Management (MEMIC) in Maastricht. Solicited diaries, with optimum length between 1 and 2 weeks, provide a rich source of data and are often utilised to elicit specific information [18,26,27]. Participants were asked to complete the diary each day over a 7-day period, reflecting on issues that are of interest to the studyactions, challenges, and needs while supporting their child with a physical disability at home, at school, or in the community. A pilot study including fifteen parents (February 2012) revealed that the 7-day period solicited diary was feasible and took about 15 min per day to complete. Participants used a printed diary (A4 booklet template) with worksheets for each day of the week, and an example of the worksheet with participant instructions. On each sheet columns were created for Time of start of the activity, Activity the child was engaged in, Actions parents performed, Challenges parents came across, Needs parents experienced, and the possibility for additional remarks (for an example of a spreadsheet see Table 1). Participants were asked to write down all activities their child did over 24 h, name the actions they did to facilitate those activities, and mention challenges and unmet needs they experienced. Next, an option was given for possible additional information. --- Data analysis For the data-analysis of the diaries both deductive and inductive reasoning was used [28,29]. Directed content analysis [23,27], using deductive reasoning, was conducted to validate or conceptually extend the existing Fig. 1 A preliminary thematic framework [17] preliminary thematic framework as described above [17]. In this deductive analysis, the existing framework guided development of an initial coding and categorising scheme, and operational definitions for the codes were used [30]. Inductive reasoning was also used. In this way, potentially new major themes, categories and/or sub-categories could be identified from the data through the researcher's careful examination and constant comparison [31]. After transcription of all diaries into a wordprocessing package, a coding scheme was developed and subsequently applied by means of manual coding. The first author (BP) prepared the coding scheme and the second author (AJHMB) inspected it to ensure congruence with the elements of the preliminary thematic framework. Then, the first author (BP) applied the identical coding and categorising scheme to all data using techniques of memoing, constant comparison, and questions. NVivo software (v.9, QSR International, Cambridge, MA) was used to organise the data during analysis. Two debriefing sessions took place between the user as co-researcher (BC) and the first author (BP) to discuss the findings leading to small changes in the wording of the codes of actions, challenges, and needs. In this study Lincoln and Guba's [32][33][34] four criteria (credibility, transferability, dependability, and confirmability) for evaluating interpretive research work were applied in accordance with Elo's et al. [35] aspects of trustworthiness in relation to qualitative content analysis. Credibility was ensured with organising debriefing sessions between the first researcher (BP) and a coresearcher (BC) and by using random sampling. Transferability was guaranteed by passing information to the reader about the boundaries of the study and providing characteristics of the study population. Dependability is assured by reporting in detail the processes within this study, thereby enabling a future researcher to repeat the work, as is confirmability by reporting findings that are solely the result of the experiences and ideas of the participants. --- Results In total, 51 diaries were returned. Three diaries were largely unfilled and one completely empty, and so these were excluded, leaving 47 eligible for the analysis. Participants were all biological parentsmainly mothers (92%)who have a child with a disability who is between 4 and 12 years of age (for more characteristics see Table 2). In comparison with the original framework, two previously identified major themes, "parents enable and support performance of meaningful activities at home, at school and in the community" and "parents enable, change and use the environment", remained the same. Additionally, the categories "actions", "challenges", and "needs" did not change. However, in all three categories new sub-categories were identified: two in the category actions, two in the category challenges, and three in the category needs. On the contrary, the previously named challenge regarding financial burden, and the need for service and information were not identified in the current analysis. Figure 2 portrays the presentation of all the major themes, categories, and sub-categories that were identified in both studies. Previously identified major themes with categories and sub-categories of the Fig. 2 Preliminary framework of parent's actions, challenges and needs based on the results of the diary study preliminary framework [17] that remained the same are marked "italic"; new sub-categories that emerged from the inductive analyses are marked "bold". Two subcategories that were not identified in this study are "underlined". Major theme 1: Parents enable and support performance of meaningful activities at home, at school, and in the community This theme concerns actions, challenges, and needs of parents while supporting their child with a physical disability to engage in meaningful activities at home, at school, or in the community. --- Category: Actions to support meaningful activities One new action was identified (role taking) and four previously identified actions (choosing for, structuring, educating, and modifying) emerged in this study. The new action "role taking" implies how parents are taking up a new role that allows supporting meaningful activities of their child. Parents in this study act as a volunteer in scouting to help their child to complete tasks like cooking or assisting a swimming teacher. Parents also gave assistance to teachers at school during physical education. The action "choosing for" refers to parents making choices regarding the kind of activities in which their child will be socially engaged. A number of examples were described in the diaries: deciding on joining a birthday party with peers without a disability, playing at a friend's home, or playing outside the house with neighbours. Parents mentioned that they chose a team sport, such as judo or grass-hockey, as a leisure activity for their child to enable contact with peers without a disability. Parents also described their choice for doing "normal" family activities outside their house, such as visiting an open-air museum by bus or going to a playground. "Structuring" describes the parents' action to organise daily routines at home in such a way that their child can engage in meaningful activities. One parent explained the change of a daily routine of eating a warm evening meal at 6 p.m. to be in time for horse riding. The action "educating" applies to teaching the child a new strategy to be able to participate in activities. Parents described this action with different examples, such as showing their child an easier way to use the computer; showing their child how to perform a domestic task such as preparing cutlery for seven people using a strategy to count aloud one, two, three...; and by using modelling while practicing with a flute. Examples related to school were also described, such as teaching their child how to reach his classroom independently and showing an alternative way of doing mathematics. The action "modifying" stands for adaptations of activities to support the child's independence and social interaction. One example in this study was about a family walk in a forest. A mother described that she walks next to her bicycle while her child with a physical disability sits on it. Another example is buying clothes with Velcro tape that enables a child to dress himself at school. --- Category: Challenges while supporting meaningful activities Parents reported several challenges that emerged while supporting their child's meaningful activities. One new challenge (choosing the right type of activity at home) and four formerly identified challenges (being supportive in a correct manner, coping with child safety, choosing the most appropriate leisure activities, and selecting the best type of education) have been identified. The challenge "choosing the right type of activity at home" refers to the struggle over how to decide what kind of activities their child can engage in with other children at home. An example is that parents go through great effort to find an appropriate activity suitable for their child to play together with a brother or sister without a disability. The challenge "being supportive in a correct manner" is about deliberating whether the child is capable of doing an activity (at home or outside) alone and estimating what would be appropriate support. Several examples were given in this study. Parents explained that it is a dilemma during breakfast whether or not to cut bread; the child had the ability to do it alone but it might slow down the morning routine, resulting in being too late for school or the school-taxi. Other examples were parents struggling with whether they should tie their child's shoes or to explain how to do it, and how much support should be given during bathing, showering, dressing, or cycling. "Coping with child safety" is about parents having difficulties coping with situations with no parental control. Activities provoking these challenges were: learning how to make a fire at scouting, playing at a friend's home, and going with another family to a playground. The challenge "choosing the most appropriate leisure activities" indicates problems finding a leisure activity that fits the child's abilities and brings a sense of accomplishment. Parents reported challenges finding suitable team sports and appropriate swimming lessons for their child. "Selecting the best type of education" is a challenge parents also faced in this study. For example, one parent explained that it was difficult to make the choice for the type of education for their child, as they got dissimilar advice from different professionals. --- Category: Needs while supporting meaningful activities Parents provided information about unmet needs while supporting meaningful activities of their child. In this study one new need (instructions or coaching) emerged and one existing need (identifying and obtaining information) was confirmed. The need "instructions or coaching" is about receiving written or verbal support. One parent wrote in the diary that having instructions or coaching in how to put their child into the car and how to put her in the seat would be helpful. Other parents expressed a need for instructions to support independence during meals. One parent showed a need for coaching on how they can support their child while playing in a play garden. In addition, parents expressed a need for coaching in understanding laws and regulations (e.g. the 'Social Support Act' -WMO) or in choosing a new type of educational program. "Identifying and obtaining information" is a need for more information about meaningful activities. For example, one parent expressed the need for an overview of suitable leisure activities for children with a physical disability. --- Major theme 2: Parents enable, change, and use the environment This theme relates to actions, challenges, and needs of parents while using, enabling, and changing the social and physical environment at home, at school, and in the community to make participation of their child with a physical disability possible. --- Category: Actions to enable, change, and use the environment One new action (changing the physical environment) was identified and four previously discovered actions (networking, educating, advocating, and creating opportunities) were confirmed. The new action "changing the physical environment" demonstrates how parents take the initiative to change the physical environment into a more suitable one to support their child's participation. One example of such action is an adaptation of a bathroom at home to aid the child in becoming independent. Another example of this kind of action is that some parents made a school entrance accessible to their child. "Networking" refers to the establishing of connections with people with similar experiences. Parents in this study explained their attempts to find people through their social network to connect and share experiences, and their attempts to connect with people through social media, like forums. The action "educating" is defined as the giving of instructions to others on how to support the participation of their child. Examples in this study are parents educating employees of the day care centre or educating a judo trainer to optimise assistance of their children. "Advocating" refers to the competing of resources, supports, and services within the system. One parent reported in the diary about her way of going up against a taxi service to make sure their child would get extra support and be taken on board as a passenger. "Creating opportunities" means the creation of events by parents in order to shape opportunities for their child to get acquainted with other children. One example in this study concerns parents organising a "get together" meeting where their child could connect to other children. Additionally, this action is also about making and keeping connections at school. One parent described in the diary that she organised extra after-school lessons to make sure her child will be able to stay in the same group of children next year. --- Category: Challenges to enable, change, and use the environment One new challenge (customised products) and three previously recognised challenges (attitudes of others, insufficient system support, and barriers in the natural and physical environment) emerged during the analysis. "Customised products" pertains to a new challenge faced by many parents in this study. Several examples were described about difficulties in trying to find appropriate clothes that support independence and the challenge of finding fitting shoes, mainly for children that need to wear a splint. Parents also noted that it is difficult to find shoes with Velcro tape, particularly in bigger sizes. The "attitudes of others" refers to the experience of parents facing negative attitudes from other children or adults towards their child with a physical disability. Examples in this study are negative attitudes of friends during play activities and negative remarks about disability from friends' parents. Similar attitudes were experienced with professional services, like taxi service employees (drivers) reacting negatively towards a child with a disability and not being willing to help a child to get in and out of the taxi. The challenge "insufficient system support" concerns unsupportive social structures. Parents in this study described that teachers and sports instructors lack knowledge of how to support children with a physical disability at school and during swimming lessons. Parents mentioned that support assistants (paid from personal budget funded by the Dutch Exceptional Medical Expenses Act -AWBZ) are not always capable of supporting their child, such as while eating a meal or engaging in play activities with their child at home. Others wrote that it is difficult to get a support assistant between 6 a.m. and 8 a.m. One parent described the complication of getting assistance for an empty wheelchair tire when visiting a museum. A "barrier in the natural and built environment" refers to the physical accessibility of buildings and public places. Parents' examples of these challenges concern the built environment: non-user-friendly public toilets, as at the Zoo; inaccessible walking paths in an open-air museum; and family cars that are unable to fit in a special swing-auto chair or a wheelchair. Similar challenges were described for the natural environment, like the impossibility to use walking paths in the forest with a wheelchair. --- Category: Needs to enable, change, and use the environment Two new needs (accessible products and environments and tailored advice about equipment, devices, and adaptations) and two previously identified needs (equipment and adaptations, social and system support) emerged. "Accessible products and environments" shows a new need for products and environments designed to be usable by all people. Parents gave examples of this need in relation to the built environment (e.g. shops and other public places, houses, playgrounds) but also to shoes and clothes. "Tailored advice about equipment, devices, and adaptations" illustrates a need for personalised advice about equipment, devices, and adaptations. Examples from this study describe a need for advice about adaptations in the house or in the day care centre. Furthermore, parents expressed a need for advice about the availability and appropriateness of devices for their child and about suitable clothes, shoes, and adaptive cutlery. "Equipment and adaptations" refers to the need for equipment that is designed to support independence and participation in activities. Examples from this study demonstrate a need for equipment to support and facilitate independence while eating. Furthermore, parents expressed a need for equipment that can be used in a normal car, a need for easy-to-modify adaptations for a toilet or a shower, and a safe bicycle seat for bigger children. "Social and system support" refers to the needs of parents for more expansive social networks to enable their child's participation. In this study parents' examples referred to extra support from grandparents and support assistants to bring their child to leisure activities, assistance during physical education at school or during church services, engagement in play activities at home or in the play garden, and to do educational activities on a computer. --- Discussion The purpose of this study is to gain insight into parents' own daily actions, challenges, and needs while supporting their child with a physical disability at home, at school, and in the community. In one week, on every single day, all participating parents described several efforts to enhance the participation of their children with a physical disability by using, enabling, or changing the social and physical environment, or by supporting their child to perform or engage in meaningful activities. Fascinatingly, those actions were primarily a result of challenges caused by restrictions in social and physical environments. Needs described by parents notably spotlight environmental aspects, like a need for environments designed for all people. Additionally, this study intended to refine the existing preliminary thematic framework arising from a scoping review [17]. Two major themes, all categories and subcategories, except two, of the preliminary thematic framework [17] were consistent with the actions, challenges, and needs mentioned by the parents in this study. One challenge regarding financial burden, and one need for service and information were not identified in the analysis. However, the analysis reveals two new actions (role taking, changing the physical environment), two new challenges (choosing the right type of activity at home, customised products), and three new needs (instructions and coaching; accessible products and environments; tailored advice about equipment, devices, and adaptations). The previously named challenge regarding financial burden, and the need for service and information were not identified in the current study; this might be related with socio-economic status and level of education of our sample. However, at this point this is a speculation. The previous scoping review [17] showed that parallels exist between the parents actions, challenges, and needs described in the preliminary framework for children with a physical disability and studies done with parents of children with Down syndrome, young people with epilepsy, and young adults with a physical disability [36][37][38]. Studies and policy reports from the Netherlands support the newly found actions, challenges, and needs related to environment. For example, free access to public buildings and places for all citizens in the Netherlands was not taken into the Equal Opportunities Act; there is no obligation to guarantee access for persons with a disability [39]. Consequently, only 29% of people with a severe physical disability in the Netherlands can enter shops [40]. Parents in this study expressed a new need for "accessible products and environments", like playgrounds, and showed their own initiatives to change the physical environment into a more suitable one to support their child's participation (action: "changing the physical environment"). In the Netherlands there are no specific regulations for leisure and sports clubs concerning children with a disability [41]. The inclusion of children with disabilities in mainstream primary education has been arranged by law in August 2014 [42]. Therefore, it might be valuable to all educators to construct learning experiences that are meaningful for all young children, including those with diverse abilities, by applying universal design principles for learning [43]. The 47 included diaries indicate that parents of children with a physical disability carry out many actions, face numerous challenges, and have several unmet needs while supporting their child's participation at home, at school, and in the community. It is interesting that parents' actions, challenges, and needs are mainly directed towards the social or/and physical environment. Therefore, paediatric rehabilitation using FCS may need to address the importance and the impact of the environment on the participation of a child with a physical disability, rather than only focus on the child's abilities. Care professionals might need to ask themselves whether the real world of children with a physical disability and their parents is central in their approach and whether they involve the knowledge of parents in shared-decision making. In order to effectively support parents while enabling the participation of their children in daily life, tailored approaches are compulsory. These approaches may also contribute to stress reduction and better health and well-being of parents. Raina et al. [44] found that health and well-being of parents of children with cerebral palsy seem strongly influenced by child disability; Parkes et al. [45] showed that a quarter of parents of children with cerebral palsy experience very high stress. However, any approach needs to be based on parents' wishes for support. The goal of most qualitative studies is not to generalize but rather to provide a rich, contextualized understanding of some aspect of human experience [46]; however, the analysis showed that several actions, challenges and needs of the Dutch parents resemble with the actions, challenges and needs from other studies in Europe or outside [17]. Future research into the meaning parents ascribe to their experiences regarding actions, challenges, and needs while supporting participation of their child in different cultural contexts is warranted. --- Study limitations Some possible limitations in this study are noteworthy to discuss. Selection bias may have occurred as we used the database of the Dutch Association of People with Disabilities and their Parents (BOSK) to draw our participants from, and only 9% of eligible parents decided to participate. The participating parents, 92% mothers, had a higher educational level than average in the Netherlands, and, considering their membership in BOSK, may have been parents who are very involved and motivated in enhancing the participation of their child with a physical disability. However, the data set is rich in nature and there is a large variety among data; some parents reported more actions, challenges, and needs at home, and others at school or in the community. Further, mothers are usually the respondents in other similar studies, like Almasri et al. [47]. They are known as caretakers of children with a disability, and are therefore most involved in enhancing the participation of their child. In a study about parents of children with intellectual disabilities, Rowbotham et al. [48] found that mothers undertake more daily care-giving tasks than fathers, but the range of tasks is similar. However, it is unclear how such differences influence actions, challenges, and needs of fathers and mothers. Additionally, it can be argued whether data collection of just one week is sufficient. As the optimal length for this type of study is one to two weeks [27] and the data received were rich in nature, the period of time probably has not been a limitation. --- Conclusions Actions, challenges, and needs of Dutch parents' of children with a physical disability are mainly directed towards the social or/and physical environment. The presented thematic framework can offer practitioners knowledge of how to support parents and promotes relevance for future research investigation. With the intention of supporting parents, further work is necessary to supply tailored approaches. --- Availability of data and materials The authors obtained approval for this study from the Atrium Medical Centre, Orbis Medical and Healthcare Centre and Zuyd University of Applied Sciences under the condition that raw data not be open to the public. Participants were also recruited under the same conditions. Therefore, the data used in this study or a direct log-in in the NVivo database cannot be shared by the authors. The data can be viewed by researchers, with permission in aggregate anonymous form only. --- Competing interest The authors declare that they have no competing interests. The authors alone are responsible for writing this article, and for its contents. --- Consent for publication Not applicable. --- Ethics approval and consent to participate The local medical ethics committee (Atrium Medical Centre, Orbis Medical and Healthcare Centre and Zuyd University of Applied Sciences) approved the study protocol (12-N-37). Participating parents (or primary caregivers) returned the completed diary, as well as the informed consent form in the return envelope directly to MEMIC (Centre for Data and Information Management) in Maastricht.

Background: Parents have a vital influence on the participation of their child with a physical disability. The aim of this study is to gain insight into parents' own daily actions, challenges, and needs while supporting their child with a physical disability at home, at school, and in the community. An additional objective of this study is to refine the preliminary thematic framework previously identified in a scoping review. Methods: A qualitative research inquiry was performed based on using a diary over a 7-day period to gather data. To systematically organise data into a structured format, content analysis has been applied using both inductive and deductive reasoning guided by the existing preliminary thematic framework. Results: Analysis of the eligible diaries shows that the actions mentioned by the 47 parents describe several efforts to enhance participation of their children with a physical disability by using, enabling, or changing the social and physical environment, or by supporting their child to perform or engage in meaningful activities. Those parents' actions are primarily a result of challenges caused by restrictions in social and physical environments. Parental responses highlighted, above all, the need for environments designed for all people. Based on the findings a redefined thematic framework is presented. Conclusions: Parents' actions, challenges, and needs are mainly directed towards the social or/and physical environment. The presented thematic framework can offer practitioners knowledge to support parents. More work is necessary to provide tailored approaches. Paediatric rehabilitation may need to address the importance of the environment on the participation of a child with a physical disability.

Introduction Processes of first-and second-generation integration vary strongly between migrants of different origin countries and between the countries they have migrated to. One domain of integration concerns the educational attainment of the children of immigrants: are second-generation children of migrants disadvantaged in the education system compared to children of the majority population in the host country? Some groups, such as Asian migrants to the USA, are known to have exceptionally good performances compared to the majority populations, while other groups, such as children of the Mediterranean labor migrants to Europe of the 1960s-1970s, do much worse (Heath et al. 2008). One potential source of these stark differences among immigrant groups across different countries has been hardly addressed in the comparative literature: the selectivity of the migrant population. A number of single-country studies have shown that emigrants are positively selected on educational attainment compared to the homestayers (e.g., Feliciano 2005a;Bertoli et al. 2013;Ichou 2014). Others have examined selectivity of migration for a large number of countries, but have not examined the consequences for the integration into the educational system of the children of immigrants (Bru <unk>cker and Defoort 2009;Belot and Hatton 2012). Given that parental education is one of the most important predictors of children's educational attainment, it is likely that positive selection has implications for immigrants' integration into the host society. Moreover, selectivity of migration may partially explain the significant differences in the educational outcomes across ethnic groups in various destination countries. Especially, second-generation children's educational achievement and attainment may benefit from a community that is 'positively selected' in terms of education, over and above the positive effects generated by the educational attainment of parents. Ethnic communities are defined here as the combination of origin country and destination country (Van Tubergen et al. 2004). It is likely that positively selected communities generate particularly high aspirations for children. These aspirations may translate into more ambitious educational choices and achievements, as argued by Kao and Tienda (1995). Furthermore, while selectivity of migration may contribute to our understanding of variations in the educational disadvantage of different immigrant groups, it also is plausible that its importance varies across destination countries. Positively selected migrant communities may be less disadvantaged compared to neutrally or negatively selected communities, but the magnitude of these differences are likely to be modified by destination country institutions. In some destination countries, low-educated communities may be at a greater disadvantage than in other countries, depending on the types of integration policies or the nature of the educational system. In this paper, we take up the challenge to study these comparative research problems. We present an index of selective migration into ten Western host societies, across 34 distinct ethnic groups, totaling 81 'communities.' We examine whether selectivity of migration is related to the extent to which second-generation groups are disadvantaged across various educational outcomes, in comparison with majority populations. We then examine whether the educational disadvantage of negatively selected migrant groups is particularly found in destination countries with unfavorable integration policies or strongly diversified educational systems. Doing so, our approach follows a research line that emphasizes that integration and assimilation of migrants (and their children) is a function of the migration context (Van Tubergen et al. 2004;Van Tubergen and Kalmijn 2005;Crul and Schneider 2010;Hillmert 2013). Integration is affected by characteristics of the destination country (such as migrant integration policies and educational systems) and of the community of migrant groups in a particular destination country. Moreover, we study whether destination country institutions have differential effects on different groups of migrants. These interaction effects between the selectivity of ethnic communities and destination country policies have not been studied before. We use a harmonized dataset comprising the best available national datasets to study ethnic educational inequality, brought together for a large comparative project (Heath and Brinbaum 2014). For the ten destination countries, we harmonized existing data that cover three outcomes in various crucial stages of the secondary school career: (1) the performance in standardized test scores in the first stage of secondary education; (2) whether a vocational or academic route is chosen at upper secondary education; and (3) whether upper secondary education has been completed. Our data thus enable the study of multiple outcomes in the school career, instead of only test scores which are the major object of study in the existing literature. The study of multiple educational outcomes has been called for by Alba et al. (2011). We can control for parental education and occupation in order to find 'net effects' of ethnicity in comparison with'similar' children of majority populations on all these outcomes. Moreover, a unique feature of our data is that we construct a 'net difference index' of selectivity of migration with regard to human capital for all ethnic groups in ten host countries and associate it with racial/ethnic educational inequality. While following the approach of Feliciano (2005a), our measure is unique because it covers multiple host societies instead of one, which enables us to study migrants in a multiple-origin multiple-destination design. 2 Selective Migration in Context: Theory and Hypotheses on Second-Generation Immigrants' Education --- Selective Migration A number of studies have examined whether immigrant communities are positively or negatively selected in comparison with the non-migrants in the origin country (Borjas 1987;Bertoli et al. 2013;Chiswick 1999;Ichou 2014). Migrants are reported to be positively self-selected on educational attainment, motivation, ability, and effort, compared to the homestayers. Even if the (positive) returns to education are equal between origin and destination countries, economic theory would predict positive selection on education given a fixed amount of relocation costs for all migrants (Chiswick 1999). Yet, societal circumstances (of both the origin and destination countries) can have effects on the extent to which there is selectivity of migration. If the returns to human capital are higher in destination countries, selectivity will be stronger, and selectivity is stronger if there are wealth or income constraints (as, for instance, imposed by larger geographical distance or other obstacles between origin and destination) (Bertoli et al. 2013;Borjas 1987). In line with this economic model, Van Tubergen et al. (2004) indirectly studied selectivity of migration for each immigrant group in a great number of destination countries by comparing the level of income inequality between the destination and origin countries. A comparatively high level of inequality in the destination country would be indicative of positive selection on income-generating characteristics (skills, education), as the returns to human capital are higher in high-inequality countries. Essential of this literature is that, both between and within destination countries, there is a wide variety of selectivity between migrant groups. Observed levels of selectivity of migrant groups in different host societies result from two distinct processes. First, some people may be more likely to migrate abroad than others (selectivity), and second, migrants choose countries based on the expected returns to their skills (sorting, Grogger and Hanson 2011). A wide variety in educational selectivity has been demonstrated for the USA (Feliciano 2005a). For almost all of the 32 immigrant communities that were studied, selection in terms of educational attainment was positive (except for Puerto Rico), but much less positive for central American immigrants to the USA than for Europeans and Asians. Factors associated with selectivity were geographical distance to the USA (more selectivity if the distance is larger) and the average level of education in the home country (less selectivity in more highly educated origin countries). Selective migration can be expected to be associated with immigrant integration and assimilation. For the USA, it has been demonstrated that positive selection (indicated by low inequality in the origin country and larger geographical distance) was related to higher earnings of first-and second-generation migrants (Borjas 1993). Moreover, the integration outcomes are not restricted to migrants themselves, but can be transmitted to the next generation of children born in the destination country. Feliciano (2005b) found that college enrollment of second-generation students was higher for more strongly educationally selected communities. Moreover, differences in selectivity accounted for differences in college completion rates between ethnic groups, particularly for the high rates among Asian immigrants. Kao and Tienda (1995: 5) explained the high achievement of secondgeneration Asians by ''their parents' optimism about their socioeconomic prospects [which] leads youth to behave in ways that promote educational success.'' Thus, selectively migrated groups are likely to have resources in their communities beyond parental educational levels, which are likely to help the structural integration of the second generation. Much less is known about the relevance of migration selectivity on integration in cross-national comparative perspective. In the study of Van Tubergen et al. (2004), the ratio of income inequalities in destination and origin countries was positively associated with immigrants' integration in the labor market, suggesting that highinequality countries attract especially high-ability migrants. Also, academic achievement of children of immigrants to traditional migration societies with strict immigration policies (such as the USA and Canada) is higher than of immigrants to more recent host societies with more relaxed policies regarding the human capital of migrants (Levels et al. 2008). That same study also showed that (parental) educational disadvantage of first-generation immigrants relative to natives of the destination country harms academic achievement of school-aged immigrants' children. Thus, poorly educated immigrant communities are harmful to children's educational performance, independent of the socioeconomic status of the family. Another indirect approach to assessing selectivity is to compare student test scores of migrant groups with the test scores of non-migrants in the origin country (Dronkers and De Heus 2010). For many immigrant groups, it appeared that their children performed worse than the majority children in the origin country, a finding that persisted after controlling for social class differences. However, it is uncertain whether such differences result from selective migration, or from difficulties in completing achievement tests in a non-native language (Heath and Kilpi-Jakonen 2012). It is somewhat worrying that most immigrant groups are, according to this approach, negatively selected, while educational selectivity is typically positive in other approaches. We expect that more positively selected immigrant communities have lower disadvantages in education (relative to the majority populations in the destination countries) in comparison with more negatively selected immigrant communities (hypothesis 1). --- Destination Country Institutions While comparative studies have shown some indirect evidence for the relevance of selective migration on integration, to the best of our knowledge, no earlier comparative study has directly measured the implications of selective migration for children's educational attainment. A comparative approach is highly relevant, because destination countries vary significantly in how immigrants are incorporated into host societies. It allows us to investigate whether selectivity has different effects in migrant-friendly societies compared to more restrictive host societies. Various scholars have drawn attention to the way in which host country policies and practices may affect the 'warmth of the welcome' afforded migrants and their prospects for successful integration, particularly focussing on economic integration (Reitz 1998;Portes and Zhou 1993;Heath and Cheung 2007). Such policies include anti-discrimination legislation, access to employment, citizenship and long-term residence, and practical assistance for recently arrived migrants and can be thought of as policies which tend to favor the inclusion rather than exclusion of migrant communities. A special form of integration policies are multicultural policies, an overlapping but conceptually distinct set of policies which facilitate recognition of ethno-religious groups and their distinctive cultures (Koopmans et al. 2005;Koopmans 2010;Banting and Kymlicka 2006;Wright and Bloemraad 2012;Bloemraad and Wright 2014). One perspective stresses that multicultural policies would improve the integration of migrants into host societies (Banting and Kymlicka 2006;Wright and Bloemraad 2012). However, others have argued that multicultural policies may in fact stigmatize migrant groups and hence deteriorate their integration in society (Duyvendak and Scholten 2012;Koopmans et al. 2005). Maybe both effects are in operation, leading to a nonexistent association between multicultural policies and the level of disadvantage. It is clear that comparative studies disagree on the relevance of host country institutions such as integration policies for educational integration (Rothon et al. 2009;Hillmert 2013). To make progress in this discussion, it is relevant to study the relation between integration policies and the relevance of selectivity of migration. It is likely that the extent to which selective migration is associated with educational disadvantages depends on migrant-friendly integration policies. Warmth-of-welcome policies may be particularly important for children of 'negatively selected' migrant communities, i.e., with lower levels of human capital than the non-migrating homestayers. These groups are often targeted by integrationfriendly policies. Based on this reasoning, we can formulate the hypothesis that the association between educational selectivity of immigrant communities and children's reduced educational disadvantage is weaker in societies with inclusive migrant integration policies (hypothesis 2). In those migrant-friendly societies, strongly positively selected immigrant communities are less helpful for children's educational attainment than in destination countries more hostile to immigrants. A second destination country institution relevant for ethnic educational inequality concerns the tracking of the educational system in the host societies (Cobb-Clark et al. 2012;Crul and Vermeulen 2003;Entorf and Lauk 2008;Griga and Hadjar 2014). Education in the destination country is one of the key institutions that determine how context helps or hinders migrants' children's integration (Cobb-Clark et al. 2012;Crul andSchneider 2009, 2010). Educational systems vary strongly in the timing of selection into different school types. In Germany, students are selected into different school types, often in different school organizations, at the age of 10. In the Netherlands, this happens at the age of 12. These are early selecting countries. On the other hand, in countries like Sweden and Canada, selection into different school types occurs much later. 1Early tracking magnifies inequalities in educational achievement and school type enrollment because second-generation pupils are given 'little time to pull themselves out of their disadvantaged starting position' (Crul and Vermeulen 2003: 979). Tracking is likely to have different effects for different groups of migrants, depending on their human capital selectivity. Through mechanisms of peer effects, resource allocation, and teacher quality, early tracking is particularly harmful for the educational opportunities of students placed in the less demanding tracks. Students in the higher tracks may benefit from early selection. How does tracking affect students of different levels of selection on the basis of human capital? A first line of reasoning is that tracking is particularly harmful to disadvantaged groups, so negatively selected immigrant communities (hypothesis 3a). More positively selected migrant groups may, according to this perspective, be less harmed by early tracking; also in early tracking systems, they may find their way to higher levels of achievement and attainment. Support for this reasoning may come from the consistently high levels of performance of Asian-origin students across many societies (Heath and Brinbaum 2014). However, an alternative perspective is that children with a lot of learning potential and aspirations (i.e., coming from highly motivated and positively selectively migrated communities) will be particularly harmed by early tracking. They are the ones for whom early tracking limits the time to demonstrate their learning potential. For unselective, more disadvantaged students, tracking may be less influential on their educational career, as they would have had a lower performance independent of whether they are tracked early or not. Tracking would then effectively constrain people with high aspirations but initial low achievement, whereas non-selective systems allow them to continue. An alternative hypothesis therefore is that tracking reduces the positive influence of positive selection, as it will prevent positively selected groups from following their high aspirations (Waters et al. 2013) (hypothesis 3b). Although we do not explicitly formulate hypotheses on differences in the statistical relationships across the educational career, it is possible to speculate on this issue. From one perspective, it can be expected that various indicators of disadvantage have stronger effects early in the school career. Many studies have reported reduced effects across the career, for two reasons: because there is an increased homogeneity of students after rounds of selection in the education trajectory and because children are more independent of their parents' structural position at higher ages (Breen and Jonsson 2005). If the effects (which are our dependent variables) are smaller, this translates into lower variances in our dependent variable, potentially leading to lower statistical relationships. However, given that we study unconditional models, meaning that there is no increased homogeneity on unobservables across the outcomes, it could also be that institutions affecting test results at the beginning of secondary education have persistent effects across the career. --- Research Design --- Data Our analysis concentrates on gaps in educational outcomes between the second generation and majority groups. The second generation is defined as the group that is born in the destination country, but with at least one parent born abroad. If one parent was born abroad, the origin country of that parent was used to identify ethnic group. This means that we define ethnic background based on country of origin of the parent(s), which, we realize, can be a simplification if multiple ethnic groups can originate from one origin country. We focus on ten host countries in Europe and North America: Belgium, Canada, England, Finland, France, Germany, the Netherlands, Sweden, Switzerland, and the USA. We brought together expertise from these countries in a European research project funded under the EQUALSOC Network of Excellence (a Sixth Framework Programme funded by the European Commission) and reported in Heath and Brinbaum (2014). We included countries where a sizeable second generation has already gone through the complete school system. The ten countries have all become increasingly diverse in the past decades, and the set of countries includes important variations with regard to selective migration, socioeconomic origin and destination country institutions. For reasons of data availability, for Switzerland, we only included the first outcome (test scores). We maintained Switzerland because our team included expertise on Switzerland, and because including it is relevant to study the impact of school tracking, one of our contextual variables. We selected nationally representative datasets to study ethnic educational inequalities at various stages in the educational career and focused on what the national experts considered the best available datasets to study ethnic educational inequality in their country. We focus on three outcomes in the secondary school phase: (1) test scores at the first stage of secondary education (roughly at the age of 13-16), (2) enrollment in vocational or general/academic tracks in upper secondary education, and (3) the completion of upper secondary education. Based on available data and national expertise to analyze the first step in the analysis (see below), we were able to collect data on ten host countries, including traditional receiving societies (e.g., Canada, USA), countries with immigration from former colonies (e.g., France, England), countries with a major 'guest worker' source of immigration (e.g., Germany, the Netherlands), and countries with a large refugee population (e.g., Sweden). Moreover, these countries differ in substantial ways with regard to their institutions that may be influential on the role of selective migration. The data we brought together include cohort studies of educational or birth cohorts, register data, and cross-sectional data. This approach has the advantage that the best datasets are chosen that have been used for national studies of ethnic and social inequalities in education, many of which are longitudinal and collected by national statistical agencies. A disadvantage is that the data were initially not collected for comparative purposes. Datasets varied in some notable respects between countries. For Sweden, Finland, and Belgium, population-level data were obtained from official registers, while for other countries, cross-sectional data were combined with educational cohort studies. Nevertheless, all these datasets were nationally representative. Given that the data came from various sources (including national registers), there was no possibility to bring the individual-level data together. Not all educational outcomes are available for all ten host countries. See ''Appendix 1'' for an overview of the dependent variables that have been studied per country and of the ethnic groups that have been compared to majority populations. ''Appendix 2'' lists all the datasets that have been used. --- Estimation Strategy We use a two-step procedure to assess the relationships between the contextual variables and ethnic educational inequality. Ethnic educational inequality is measured by the net regression coefficient of ethnic group, for each ethnic group and each destination country separately, on the outcome under study. The first step consists of country-by-country regressions on the three outcomes, controlling for family situation (single parenthood), social background (parents' education and social class), and gender. These models are unconditional models; for instance, the chance to obtain a full secondary-level qualification is estimated for the whole sample, not just for people that have successfully completed lower secondary education. 2Of these regression models, the coefficients displaying the difference in the outcome of a particular ethnic group with children from the majority population are saved, as well as the standard errors of these coefficients. With regard to test results, these coefficients are taken from ordinary least squares regression models. For the other two outcomes, which are dichotomous, we have taken the probit coefficients. Negative net coefficients would thus indicate a disadvantage of that particular ethnic group for that particular educational outcome compared to similar children of majority populations, and positive coefficients represent net advantages. This paper reports about the second step of the regression analysis, which relates the strength of the (dis)advantage of second-generation students to contextual variables such as integration policies that are described below. The dependent variables are, thus, the coefficients of ethnic background predicting the educational outcome, which indicate the level of (dis)advantage of the particular ethnic group relative to the majority population, controlled for socioeconomic background (occupational group and both parents' education). These models are known as slopes as outcomes models or two-step multilevel models (Achen 2005;Bryan and Jenkins 2016;Gebel andGiesecke 2011, 2016). These coefficients measure the ethnic gap of ethnic group coming from origin country j in destination country k (relative to the majority population), which we call d jk. We estimate the size of the gap as a function of variables at the level of destination country k and of the combination of origin and destination countries <unk>j,k<unk>, and of an interaction term between these two different types of variables. Equation (1) describes the second-step multilevel regression model, with ethnic groups nested in host societies (estimated separately for each educational outcome): d jk 1<unk>4 a <unk> bX k <unk> cZ jk <unk> sX k Z jk <unk> f k <unk> e jk<unk>1<unk> Since our dependent variable d jk is an estimated coefficient rather than the true population-level parameter, we take into account the degree of uncertainty of the coefficients by weighting. We apply the weighting scheme proposed by Borjas and Sueyoshi (1994), which takes into account the two error components that exist in two-stage models (see Huber et al. 2005). The first component is the variance of the coefficients estimated in the first-stage regressions. This variance is easily derived on the basis of the standard errors of the first-stage coefficients, in such a way that more precise estimates get greater weights than more imprecise estimates. The second component is the residual variance from the second-stage regressions before adding weights-that is, the residuals of this model net of the variance resulting from the first-stage coefficients. This second component is necessary because the macro-level variables do not explain the whole variance between countries. Due to the relatively low number of cases (i.e., ethnic communities), we are careful not to include too many explanatory variables at once. Therefore, each analysis is built up in steps. The main predictor variables are the selective migration index (see below), which is subsequently interacted with two destination country variables: an indicator for multicultural policies, and an indicator of early tracking in the educational system. We control for language similarity of the origin and host societies, as it is shown to be related to both the selectivity of migration (Bru <unk>cker and Defoort 2009) and educational attainment (Heath et al. 2008). The dataset has ethnic group by destination country as its unit of analysis. With 32 ethnic groups in ten countries (including an 'other' and'mixed' category in each of the countries), the total N lies between 57 and 80 (depending on the specific educational outcome). Figure 1 shows the kernel distributions of the net effects on the three educational outcomes. It can be seen that both negative and positive net effects are found in the data. In ''Appendix 4'', we show models on the subset of the countries that are available for all three outcome variables. In the conclusion and discussion, we reflect on the robustness of our findings in light of these replications on smaller datasets. --- Selectivity of Migration We follow the approach of Feliciano (2005a) in measuring selectivity of migration. A selectivity index is constructed to indicate how the educational level of migrants compares to their origin countries' populations from the same birth cohort (in order to take account of the fact that educational levels have been rising rapidly in many of the 'origin' countries just as it has in the Western destination countries). Data on (first-generation) fathers' completed education are readily available in the datasets used in the first-step analysis. 3 The host country data are taken from nationally representative data among children, so first-generation migrants without children are not represented. For the non-migrant population of origin countries, data were gathered from several (cross-) national surveys. Like Feliciano, we use the net difference index (NDI) introduced by Lieberson (1976Lieberson (, 1980) ) to calculate a measure of selectivity. This index enables us to compare the entire frequency distribution of completed educational level by migrant fathers to that of non-migrant males of a comparable age-group in the origin country. 4 Its logic is to sum up the percentages of migrants in the destination country that completed a similar, lower, and higher educational level in comparison with their non-migrant counterparts in the origin country. The sum score represents how often the educational level of a migrant will 123 exceed that of a non-migrant, or the other way around. The more positive the sum score, the more the educational level of a migrant group exceeds that of a nonmigrant as opposed to the other way around. A negative sum, in turn, indicates that migrants more often have a lower educational level than non-migrants. The selectivity index ranges from -1 (all migrants are less educated than non-migrants) to? 1 (all migrants are more educated than non-migrants). A score of 0 means that the educational distributions of migrants and non-migrants are equal. Selectivity cannot be calculated for the'mixed' and 'other' ethnic groups since these categories have no specific origin country. For the 'other' category, the mean selectivity index of the destination country is imputed. For the'mixed' category, we impute a selectivity of 0, equating this category to the majority group. It should be noted that this measure only assesses selectivity with regard to educational attainment. Other forms of selectivity, for instance on the basis of intelligence, motivation, or economic resources, are not explicitly measured, while these could be influential on the level of disadvantage in immigrant communities. Figure 2 shows the selectivity index calculated for each origin group in each of the destination countries. The figure shows that selectivity varies strongly between ethnic groups within countries, and within ethnic groups between countries. For instance, the Turkish first-generation migrants living in Germany are more positively selected than Turkish migrants in the Netherlands or Belgium (while Zuccotti et al. 2017 showed overall positive selection of Turkish migrants in six European destination countries). Moreover, in line with Canadian policy, it is evident that all migrant groups that were examined are positively selected (i.e., more highly educated than the homestayers in the origin countries). All other countries have both positively selected and negatively selected migrant groups. The control variable language similarity between origin and destination countries is measured as a dummy variable where a score of 1 signifies similarity of languages. Ethnic groups subsumed under the 'other' category are given a 0. The'mixed' category is scored 1. --- Destination Country Institutions: Integration Policies and Early Tracking At the level of the destination country, one central variable measures the extent to which the country has policies aimed at promoting migrant integration. We use the Migrant Integration Policy Index of 2010 (MIPEX III, MPG 2011), which can be thought of as an overall index of the inclusiveness of host country institutions; the MIPEX index aims to 'create a rich, multi-dimensional picture of migrants' opportunities to participate in society by assessing governments' commitment to integration' (MPG 2011: 6). On the MIPEX Web site, the aims of the index are more elaborately described as follows: 'MIPEX measures policies that promote integration in all societies. Integration in both social and civic terms rests on the concept of equal opportunities for all. In socio-economic terms, migrants must have equal opportunities to lead just as dignified, independent and active lives as the rest of the population. In civic terms, all residents can commit themselves to mutual rights and responsibilities on the basis of equality. When migrants feel secure, --- Selectivity of migration Note: 'Other' ethnic groups have been assigned the mean selectivity score. Mixed ethnic origins have been assigned the score 0 Fig. 2 Selectivity of migration by ethnic group and host country confident and welcome, they are able to invest in their new country of residence and make valued contributions to society. Over time, migrants can take up more opportunities to participate, more rights, more responsibilities and, if they wish, full national citizenship.'5 Using expert surveys and policy assessments, seven dimensions of integration policies are quantified: labor market mobility, family reunion, education, political participation, long-term residence, access to nationality, anti-discrimination. Countries are scored according to the extent to which migrants have equal opportunities as the rest of the population. It can be thought of as a measure of the overall 'warmth of the welcome.' It should be noted that some elements of MIPEX may be less relevant for migrants within the European Union, as they are entitled to free movement and residence. Nevertheless, other elements such as entitlement to full national citizenship and rights to participate can also vary with regard to European migrants and their families, depending on the host country. It is important to note that the explicit focus is on policies, not on factual integration levels. We have included the total MIPEX score across all dimensions, rather than focus on education policies alone, because the reduction in educational disadvantages is likely affected by broader sets of policies to integration of families, including their labor market, political and civic integration. 6 The MIPEX index correlates strongly with multiculturalism indicators (Koopmans et al. 2012). 7The tracking of the educational system is measured by a standardized scale based on age of first selection, the length of the differentiated curriculum, and the number of school types available for 15-year old students (taken from OECD reports and Brunello and Checchi 2007, see Bol and Van de Werfhorst 2013;Bol et al. 2014). Factor scores of an underlying factor were calculated and standardized with a mean of 0 and a standard deviation of 1, taken over the maximum number of countries available in the source data. Summarizing different indicators of tracking into one index is important, as together they indicate the timing (i.e., at which age), duration (which proportion of the compulsory years of education), and form of tracking (i.e., in how many tracks students are separated). Regression diagnostics of the association between tracking and the slope of family background on student achievement are more supportive of the composite index than of the separate underlying indicators [Authors]. It should be noted that the tracking index classifies educational systems particularly with regard to differentiation in the first stage of secondary education, thus before the outcome that we study of whether children enroll the academic or vocational track in upper secondary education. Table 1 shows the scores on all the contextual variables, averaged across ethnic groups within host countries. Regarding selectivity of migration, we see that the average selectivity is larger than zero in all ten destination countries, indicating positive selection on educational attainment of the first-generation migrants relative to the homestayers. As one would expect based on selective immigration policies, the highest value is found in Canada, where the average selectivity is 0.406. A number of countries have an intermediate selectivity index of around 0.10, including England, Finland, Sweden, and the USA. Then, countries that adopted many 'guest workers' in the 1960s and 1970s (Belgium, Germany, the Netherlands) have lower selectivity scores. Language similarity is highest in France and England, countries where many migrants from former colonies moved to. The migrant integration policy index is highest for Sweden and also comparatively high in Finland, the Netherlands, and Belgium. In France, Germany, and Switzerland, migration integration policies are less inclusive. Educational tracking is highest in the German educational system, where students are typically selected around the age of 10 for three separate school types. Also Belgium and the Netherlands score high on tracking, with selection happening around the age of 12, for separate school careers for multiple years. Less tracking is found in the comprehensive schooling systems in Scandinavia, England, Canada, and the USA. --- Results The outcomes of our two-step multilevel models with additional clustering of migrant groups within countries are shown in Table 2. The table lists five models for each of the three educational outcomes. First, we start with a model with only selective migration and language similarity. Then, the migration policy index is added (model 2) plus interaction terms with selectivity of migration (model 3). Then, in models 4-5, the migration policy index is replaced with tracking; first the main effect of the variable at the destination country level is inserted, after which interaction term is added. The first outcome is ethnic educational inequality in student test scores. Table 2 shows that selectivity of migration is positively associated with the coefficient of ethnic background relative to the majority population, in line with hypothesis 1. Given the value of the constant, the average disadvantage of unselective groups coming from countries with a different language is -0.173 standard deviations in test scores. If selectivity increases, the gap is reduced and could turn even into an advantage (given the maximum score of selectivity of, roughly, 0.7). Language similarity is also associated with reduced disadvantages, although the coefficient is not statistically significant. In model 2 where the MIPEX migrant integration policy index is added, the coefficient of language similarity is increased and turns into significance, while the positive coefficient for selectivity of migration gets weaker, but also stays significant. The MIPEX is positively associated with the ethnicity coefficient, meaning that disadvantage in student test scores is lower in countries with more favorable migrant integration policies. Model 3 adds the interaction term between MIPEX and selectivity of migration, but that turned out to be irrelevant and statistically insignificant (falsifying hypothesis 2). So, while migrant integration policies are associated with lower disadvantages among migrants relative to majority populations, this relationship is highly similar across ethnic groups of different levels of selectivity. Model 4 replaces the MIPEX index with the tracking index of the educational system. The overall effect of tracking is negative; in more strongly tracked educational systems, the average level of disadvantage of ethnic groups is larger than in countries with comprehensive schooling systems. Model 5 shows that the negative effect of tracking gets even more negative for more strongly selected migrant groups-a finding in line with hypothesis 3b. The interaction term is pretty strong, but the interaction effect is not replicated if we only focus on countries for which we study all three dependent variables (Table 6 of Appendix). It is replicated, however, if only European countries are analyzed (Table 7 of Appendix). The second panel of Table 2 shows results of an analysis of ethnic differences in the choice for general/academic forms of upper secondary education, relative to vocational education. As already seen in Fig. 2, the overall level of disadvantage is smaller than with test scores, and in fact many ethnic groups have a net advantage over children of the majority population with

Selectivity of migration varies significantly between ethnic/origin country groups, and between the destination countries which these groups have migrated to. Yet, little comparative research has measured empirically how selective different migrant groups are in multiple destination countries, nor has research studied whether the selectivity of migration is related to the magnitude of ethnic inequalities among the children of migrants in Western societies. We present an empirical measure of educational selectivity of migrants from many different origin countries having migrated to ten different destination countries. We examine whether selective migration of a particular ethnic group in a particular destination country is related to the gap between their children's and native children's educational outcomes. We find that the disadvantage in educational outcomes between the second generation and their peers from majority populations is smaller for ethnic groups that are more positively selected in terms of educational attainment. We also find some evidence that the effect of selective migration is moderated by the integration policies or tracking arrangements in the educational system in the destination country.

ifying hypothesis 2). So, while migrant integration policies are associated with lower disadvantages among migrants relative to majority populations, this relationship is highly similar across ethnic groups of different levels of selectivity. Model 4 replaces the MIPEX index with the tracking index of the educational system. The overall effect of tracking is negative; in more strongly tracked educational systems, the average level of disadvantage of ethnic groups is larger than in countries with comprehensive schooling systems. Model 5 shows that the negative effect of tracking gets even more negative for more strongly selected migrant groups-a finding in line with hypothesis 3b. The interaction term is pretty strong, but the interaction effect is not replicated if we only focus on countries for which we study all three dependent variables (Table 6 of Appendix). It is replicated, however, if only European countries are analyzed (Table 7 of Appendix). The second panel of Table 2 shows results of an analysis of ethnic differences in the choice for general/academic forms of upper secondary education, relative to vocational education. As already seen in Fig. 2, the overall level of disadvantage is smaller than with test scores, and in fact many ethnic groups have a net advantage over children of the majority population with similar socioeconomic backgrounds. The intercept has a positive value, indicating an average advantage for children from non-selective communities from non-native-speaking origin countries (i.e., overrepresentation of second-generation migrants in the general/academic programs). Model 1 shows that selective migration is positively associated with the size of the regression coefficient of ethnic background. More selectively migrated communities have higher likelihoods to be enrolled in the academic tracks in upper secondary education. This is in line with hypothesis 1. The MIPEX index is positively associated with the ethnicity slope, indicating higher levels of ethnic advantage (or lower disadvantage) in societies with more migrant-friendly policies. Model 3 shows that selective migration is even more positively associated with minorities' opportunities in education in countries with favorable migrant integration policies. In other words, migration integration policies are particularly helpful for communities that are strongly positively selected, which goes against hypothesis 2. One interpretation of this finding may be that favorable integration policies are particularly helpful for positively selected groups with high aspirations. It is easier for them to take advantage of the opportunities available in favorable integration regimes. Model 4 shows that early tracking is associated with larger disadvantages (or smaller advantages) of ethnic minorities relative to majority populations with regard to upper secondary academic enrollment. So, in countries where tracking happens earlier and more rigidly at the first stage of secondary education, there is a weaker overrepresentation of second-generation immigrants in the academic schools at the upper secondary level. This finding corresponds to the multination comparison of student achievements at age 15 (Cobb-Clark et al. 2012), but is now further The unit of observation is migrant group within destination country supported using post-harmonized educational cohort data for an outcome at the upper secondary level. Model 5 shows that tracking is particularly harmful to students coming from more strongly selected migrant communities. Like with student test scores, this is in support of hypothesis 3b, which argues that tracking is particularly harmful to ambitious and motivated migrant communities. The third panel of Table 2 shows results on the completion of full upper secondary education (in any track). Overall, the results are less strong than in the previous analyses, possibly because different datasets have been used, and fewer countries could be included (see ''Appendix 2''). For instance, for the Netherlands, we now have to rely on the educational attainment of young adults in a crosssectional survey, while for previous analysis, we used prospective educational cohort data. Also for Britain, other data are used. There is no significant association between selective migration and secondary degree completion, although the coefficient is positive as in previous analyses. Model 2 shows that migrant integration policies are not associated with secondary degree completion. So, while the previous panel showed a positive association with the enrollment into academic forms of secondary education, we do not see that this translates into lower inequalities in terms of the completion of secondary education. The coefficient for the migrant integration policy index is unrelated to the selectivity of migrant communities (model 3). 8Model 4 shows that early tracking magnifies the ethnic educational inequalities in secondary degree completion. So, like with test scores and the choice for the academic track, tracking is associated with larger ethnic gaps. Model 5 shows that the interaction term between tracking and selectivity is negative, as with the previous two dependent variables, but in this case, the association is not statistically significant. --- Conclusion and Discussion We studied ethnic educational inequalities among second-generation migrants in ten destination countries, concerning three crucial outcomes in educational careers: test results, choosing the academic or vocational route in upper secondary education, and the completion of upper secondary education. With our career perspective on ethnic inequalities in education, holding constant for socioeconomic differences between groups, we examine whether 'context matters' for the integration and assimilation of migrant children in the secondary educational system. Building on the perspective that ethnic communities are defined based on the combination of the country of origin and the country of destination of migrants (Van Tubergen et al. 2004), our focus was on the question whether selective migration of a migrant group relative to the non-migrating 'home stayers' is related to the level of disadvantage in education. We found support for the baseline hypothesis that more positively selected migrant communities have lower levels of disadvantage (or a higher level of advantage) relative to the majority population. This effect was only found to be statistically significant for the standardized tests taken at the first stage of secondary education and for the choice of academic versus vocational education. For secondary school completion, the effect of selectivity of migration was more modest and had higher levels of statistical uncertainty. Possibly, differences across educational outcomes imply that the ethnic community seems more important for educational outcomes in the early school career than for outcomes later in the career. We also examined whether selective migration was differentially associated with ethnic disadvantage in education depending on host country institutions, in particular migrant-friendly policies and the early tracking of the system. First of all, it appeared that 'context matters' in that the minority-majority education gaps were associated with these institutional characteristics. Gaps were less negative for minority students (or more positive, as the net ethnic gaps are often favorable for the second generation when socioeconomic differences are taken into account) in societies with migrant-friendly policies-again with the exception of the likelihood to complete upper secondary education. Gaps were more negative (or less positive) in societies with strongly tracked educational systems, and consistent so for all studied outcomes. However, the evidence that these institutional environments had differential associations with ethnic inequalities depending on the selectivity of migration is less clear. If anything, we see that tracking in education is particularly harmful to positively selected immigrant communities, but the effect is not robust to specific country selections. In terms of policy implications, the 'warmth of the welcome' and the tracking of the system seem to matter, but there is, based on our findings, little reason to think that these institutional characteristics matter differently depending on the migrant group's selectivity. Our focus on tracking does not mean that this is the only form of school segregation that may impact children of immigrants disproportionately. Other forms of segregation may be manifested through residential segregation, or through private schools as apparent in the USA and England. Nevertheless, the private sector only includes around 10 percent of primary and secondary school children in the USA (McFarland et al. 2017) and 9 percent in England (Department for Education 2016), while academic/general tracks typically have an intake of around 40-50% of students. Nevertheless, it is worthwhile to further study the impact of private education on students of different migration backgrounds. This study's findings are important for at least three fields of investigation. First of all, while there are many studies that have aimed to explain the extent to which migrant communities are positively selected based on human capital, few studies have examined the implications for the level of ethnic educational inequality, let alone inequalities at various stages of the educational career. Showing the implications of selectivity of migration further emphasizes the relevance of studying the migration process itself. As a follow-up to our study, a multi-origin-multidestination design can also examine the educational distribution in the home country in a different way. As Feliciano and Lanuza (2017) show, the educational attainment of immigrants may also be seen as a positional good, where a medium-level qualification may in fact represent a comparatively high achievement level, depending on the origin country. Second, our study has contributed to the understanding of ethnic educational inequalities, by emphasizing the relevance of the context of reception. The observed level of selectivity of migration is a consequence of two joint processes: (1) selectivity of who migrates and (2) the choice of country to migrate to. This makes that we have to be careful to think of the associations of selective migration as causal. Nevertheless, given that the destination of migrants is at least partially a non-rational random process, it is worthwhile to study context in relation to selectivity of migration. Moreover, if an optimal sorting process would have occurred, no differential effects would have been found as each community would have ended up in the best host country. Nevertheless, a possible weakness of our design is that our findings may result from ethnic communities being conducive to educational attainment of children on top of individual socioeconomic background, while selective migration has no causal effect on a counterfactual interpretation of what would have happened had migration not taken place. A third field of enquiry to which our study speaks is concerned with the impact of the educational institutional structure on various inequalities. While most of the literature on educational systems is concerned with social, rather than ethnic, inequalities (Brunello and Checchi 2007;Van de Werfhorst and Mijs 2010), some studies have pointed to larger ethnic inequalities in early tracked educational systems (Cobb-Clark et al. 2012, Crul andVermeulen 2003). Our data found evidence for this as well, studying multiple educational outcomes. For all three educational outcomes, ethnic minorities were further disadvantaged in societies with early tracking systems (cf. Griga and Hadjar 2014 for tertiary education). Moreover, early tracking was particularly harmful to strongly positively selected ethnic communities. Tracking thus seems to harm the educational potential specifically of highly motivated and ambitious ethnic groups. In summary, selectivity of migration appeared a relevant correlate of the educational performance differentials between the second generation and majority populations. Although our data do not contain measurements of aspirations and motivations, these offer relevant interpretations of this central finding. Positively selected migrant groups do better in education than migrant groups that have been less positively selected. Migrants are, according to the immigrant optimism thesis of Kao and Tienda (1995), anxious and motivated to integrate in their new country, and their optimism stimulates their children's schooling outcomes. This particularly affects ethnic communities that form a positive selection of their origin country's population in terms of human capital. Finally, it is worth mentioning some weaknesses of our study. With our approach to select the best possible data per country, each with sizeable immigrant communities, we had to harmonize the data after data collection, which always comes at a cost with regard to comparability. One disbalance is that the countries differed in the number of immigrant groups that were included. It is possible that our results are partly driven by the countries with many immigrant groups, most notably Sweden. Another disadvantage of post-harmonization is that some datasets are longitudinal and others are not. Also, our two-step design required many 'cases,' necessitating us to combine all countries and all origin countries. Robustness checks with fewer countries sometimes refuted the overall findings, but particularly with regard to the interaction of destination country institutions by selectivity of migration. Finally, our approach to measure selective migration conflates selection processes in the decision to migrate and the destination where to move. Our study can be seen as complementary to studies that more specifically address these joint decisions in the migration process (Grogger and Hanson 2011;Guveli et al. 2016). --- Appendix 2 See Table 4. --- Conflict of interest The authors declare that they have no conflict of interest. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. --- Appendix 1 See Table 3. --- 368 H. G. van de Werfhorst, A. Heath --- Appendix 3 See Table 5. --- Appendix 4 See Table 6. --- Appendix 5 See Table 7.

Selectivity of migration varies significantly between ethnic/origin country groups, and between the destination countries which these groups have migrated to. Yet, little comparative research has measured empirically how selective different migrant groups are in multiple destination countries, nor has research studied whether the selectivity of migration is related to the magnitude of ethnic inequalities among the children of migrants in Western societies. We present an empirical measure of educational selectivity of migrants from many different origin countries having migrated to ten different destination countries. We examine whether selective migration of a particular ethnic group in a particular destination country is related to the gap between their children's and native children's educational outcomes. We find that the disadvantage in educational outcomes between the second generation and their peers from majority populations is smaller for ethnic groups that are more positively selected in terms of educational attainment. We also find some evidence that the effect of selective migration is moderated by the integration policies or tracking arrangements in the educational system in the destination country.

the median age of 48 years. The major identified barriers for practicing COVID-19 prevention measures were the presence of strong cultural and religious practices, perceiving that the disease does not affect the young, misinformation about the disease, and lack of trust in the prevention measures. --- Conclusions Socio-cultural, religious, and economic related barriers were identified from the participant's narratives for the practice of COVID-19 prevention measures in south Gondar Zone. Our findings suggest the need to strengthen community awareness and education programs about the prevention measures of COVID-19 and increase diagnostic facilities with strong community-based surveillance to control the transmission of the pandemic. --- Background The world is closely watching the outbreak of respiratory illness associated with the novel beta coronavirus SARS-CoV-2 [1]. Initially termed 2019-nCoV, sequencing showed that the now officially named SARS-CoV-2 is 80-89% similar to bat severe acute respiratory syndromerelated coronaviruses found in Chinese horseshoe bats [2]. Prior to the global outbreak of SARS-CoV in 2003, HCoV-229E and HCoV-OC43 were the only coronaviruses known to infect humans. Following the SARS outbreak, 5 additional coronaviruses have been discovered in humans, most recently the novel coronavirus COVID-19, believed to have originated in Wuhan, Hubei Province, China. SARS-CoV and MERS CoV are particularly pathogenic in humans and are associated with high mortality [2,3]. The world is now suffering from this global pandemic which has killed millions of people worldwide [4,5]. As evidence shows, all people in the world are susceptible to this viral respiratory tract disease, although the impact is different depending on several socio-demographic, economic, and other infrastructural differences. There are no any proven pharmacological treatments developed for the disease until now, although a number of trials are currently underway by scientists across the globe. However, COVID-19 vaccines are developed within the shortest period in the history of vaccine production and now become available. Ethiopia also received AstraZeneca vaccines manufactured by Serum Institute of India (SII) on 6 March 2021 [6,7]. However, there are scientifically accepted and recommended prevention measures against COVID-19 infection such as social distancing, frequent hand washing, use of face mask, and proper use of alcohol-based hand rubs or sanitizers [8][9][10]. Ethiopia is at high risk for the introduction and spread of the novel coronavirus disease (COVID- 19) and the effect might be devastating because of the multiple health challenges the country already faces: rapid population growth and increased movement of people; existing endemic diseases, such as human immunodeficiency virus, tuberculosis, malaria and increasing incidence of non-communicable diseases [11]. Although some parts of the world are returning to pre COVID levels of engagement, the risk of the pandemic is continued and becomes a global challenge [12]. According to the September 2, 2021 Worldometer report, currently India and South Africa became the number one affected countries from Asian and African countries respectively [13]. Prevention practices are critical to combat the spread of Coronaviruses (CoVs) [9]. Poor prevention practices can lead to economic, social, and political crisis, and increased risk of death. The benefits of COVID-19 prevention, particularly in the developing setting like Ethiopia are invaluable because of the deprived health system [11,14,15]. To contain the pandemic, the country established a National Ministerial Committee and the government declared a state of emergency on April 8, 2020. On March 16, 2020, the committee released guidelines that included: 1) "A 14 days quarantine of passengers (travelers) in isolated centers; 2) Ethiopian Airlines to stop flying to 30 countries: 3) regulating the market to avoid unethical exploitation of the situation:4)stopping religious gatherings; 5)avoiding public overcrowding; 6) strictly refraining from harassing foreign nationals; 7) strict adherence to self-prevention and protection; 8) medical professional training;9) facilitating the acquisition of testing facilities; 10) temporarily closing of night clubs and bars; 11) regulating market to avoid unethical exploitation of the situation; and 12) supporting regions preparedness to contain the disease" [8,11,15]. In this study, healthcare workers and key informants in the community including local leaders were interviewed either individually or as part of focus groups to share their perceived barriers to COVID-19 prevention measures. This is helpful from the view of the following perspectives. Healthcare workers are the frontline in battling the spread infection as a whole [16]. Secondly, community leaders have a great role in the prevention of COVID-19 infection in the community [17]. The basic protective measures against COVID-19 including distancing, personal protective equipment, and handwashing as noted above are poorly practiced in the community of South Gondar Zone, along with the above scientific explanations, controlling the spread of infection is essential to preventing outbreaks of COVID-19 [18]. Basic protective measures against the new coronavirus include staying at home, washing hands with soap and water, frequently, using alcohol-based hand rub or washes, use of personal protective equipment, and keeping physical/social distance [9,15,19]. Thus, the implementation of these practices to reduce the spread of COVID-19 would be challenging in Ethiopia. Consequently, the perception of health extension workers and community leaders as a key informant are important to underpin the prevention of COVID-19 Infection [11,20]. In response, the present qualitative study was designed to explore the community risk perception about COVID-19 and the perceived barriers to practice recommended prevention measures among community key informants and other community member's in South Gondar Zone, North west Ethiopia. --- Methods and materials --- Study area and period This study was conducted on purposely selected Woredas (the lowest administrative areas called woreda's or districts in Ethiopia, government structure which form zone [21] of South Gondar Administrative Zone (Worata, Libokemkem (Adiszemen) and Guna Begemidir woreda (Kimirdingay) from May 15 to June 01, 2020. There are eight primary and one comprehensive specialized hospital and more than 187 health centers in the zone. According to the 2007 E.C population census report, the total population of the zone is around 2,578,906. Currently, there is one isolation center prepared at Atse Seyife Yared Health center and one quaratine center at Debre Tabour University. There is also one treatment center at the zonal level. However, COVID-19 treatment centers are not still established at the woreda level. --- Study design A qualitative study using a phenomenological approach was used to explore the perceptions, thinking, feelings, and experience of community members and health extension workers on a specific topic to explore the community risk perceptions and barriers related to the practice of COVID-19 prevention measures in the community. This study follows COREQ guidelines for reporting qualitative studies [22]. --- Inclusion criteria <unk> People resided for at least six months in the area <unk> People with age greater than 18 years <unk> Community key informants (health extension workers, Women, health development armies, and religious leaders) <unk> Health extension workers who have been working more than two years in the respective sites --- Participants of the study and sampling procedures The participants of this qualitative study were purposely selected community members and health extension workers. The community key informants (Health extension workers, women health development armies, and religious leaders) were selected for in-depth interviews with the assumption of having detail information about the issue in each community, while other community members were selected for focus group discussions in two rounds (7 participants in each round) at the selected woreda's. We determined the sample size based on a theoretical saturation point in data collection time when new data no longer bring additional information to the research question. Therefore, the interview process was continued until that point was obtained. Finally, a total of 9 community key informants for in-depth interview and 14 individuals from the community were participated in focus group discussions. --- Data collection tools and procedures In-depth interviews and focus group discussions were conducted using a semistructured indepth interview and focus group discussion guide to facilitate the interviews (see S2 and S3 Files). The guide was developed by the research team by reviewing similar literatures on the area. It was first developed in English and then translated to the local language (Amharic) to facilitate communication with participants. The guides were focused on the community's experiences with the practice of the recommended prevention methods of COVID-19, their risk perceptions on acquiring the disease, and the barriers to practice it in the community. All interviews and discussions were taperecorded and written notes were also taken during the interviews. Participants of the study were selected with the help of the health local leaders. After we obtained informed consent to participate, an appropriate place and time for an interview and focus group discussion was arranged in a private setting. Two public health professionals who had a Master of public Health in health education and Epidemiology collected the data with the assistance of two note takers. Interviewers and facilitators were fluent in written and spoken Amharic and English language. All interviews were conducted within the community setting and lasted 30-60 minutes. The transferability of the findings was established by collecting data on randomly selected woreda's in the administrative zone with community key informants who are representatives of the community and having detail information about the people in their surroundings. This helped to get more information from different perceptions of the community with different socio-demographic and cultural experiences. To maintain the validity of the findings, the investigators developed a rapport with participants who were participated in the study. Credibility was maintained through participant checking during in-depth interviews and focus group discussions, and through feedback of findings at the end of the study from whom the data was taken. Keeping a record with information about imitation was enhanced conformability. The dependability of data was maintained by taking the depth of information until reaching the data saturation point. --- Data quality assurance To maintain data quality, the supervisor and data collectors were trained for two days on the basic principles of data collection, study overview, and how to do other related activities during data collection by the principal investigator. Strict daily supervision of the data collection process were maintained throughout the data collection period. One supervisor was responsible for one study site. The data collectors were debriefed each day after the data collection by the supervisors. --- Ethical approval and consent to participate Ethical approval letter was obtained from Debre Tabor University. Official letter of co-operation was written to community leaders and health extension workers to obtain their co-operation in facilitating the study. Oral informed consent was obtained from each participant prior to data collection after information on the study was explained to them (see S1 File). This is because the study is just a qualitative research which did not have any intervention /clinical trial and did not have follow-ups. In addition, some of the participants were illiterate, so taking oral consent was sufficient after giving a detail explanation on the aim of the study to the participants (see the details of the study information sheet in S1 File). The ethical committee was approved the oral consent procedure. Confidentiality of information was assured by excluding names and identification in the interviews. In-depth interviews and focus group discussions were conducted by maintaining WHO recommendations for social distancing and use of appropriate personal protective equipment's. --- Results --- Socio-demographic characteristics of study participants The table below describes the socio-demographic characteristics of focus group discussions and in-depth interview participants for this particular qualitative study. A total of 9 in-depth interviews and 2 focus group discussions were conducted. The age of the participants ranged from 24 to 70 years with the median age of 46 years. The majority 14 (60.9%) of the participants were females. Among the participants of in-depth interview, 2 were health extension workers, 5 were women health development armies, and 2 were religious leaders. About 19 (82.6%) were orthodox religion followers while the rest were Muslims. Regarding marital status, most 14 (60.9%) of the participants were married while the rest were single. The majority 17 (73.9%) of the participants were from rural areas (Table 1). --- Risk perception and barriers for the practice of COVID-19 prevention methods Three main themes (Personal factors, Socio-demographic & economic related factors, cultural and religious related factors) and five categories (Lack of knowledge & awareness, Sociodemographic, economic, cultural, and religious related barriers) were emerged from the narrations of the participants regarding the risk of acquiring of COVID-19 and perceived barriers for the practice of prevention measures (see S4 File). --- Personal related barriers Lack of knowledge and awareness about COVID-19 infection. As most of the participants' described, there were various reasons mentioned for not practicing COVID-19 prevention measures in the community. One of the major reasons stressed by almost all participants for not practicing COVID-19 prevention measures was lack of knowledge and awareness about risk factors, signs, and symptoms of the disease. Most of the focus group discussion participants did not hear about the prevention measures as well as the clinical features of the disease. "I heard this new disease from the health extension worker, however, I did not understand about the disease. I do not know, it will infect me or not...." (A 32 -years-old female FGD participant) "I heard the name of the disease which is called "Corona'. I did not know about its characteristics. There was no such disease in our country." (A 52 -years-old female FGD participant) As some participants described, there is a lack of information access about COVID-19, particularly for the rural community, still some participants did not hear about the disease, while some others heard only the coming of new disease or epidemic only but they did not hear about the disease clinical features and prevention measures. --- "I did not hear about this new disease which you called Corolla or Corona. I heard it know from you. As you know, my house is far from the health post. Unless the health extension worker comes to our Kebele, I could not hear anything about health." (A 60 -years-old women health development army) "I heard the name Corona virus disease through a telecom message in my phone. However, I did not understand about the prevention methods" (A 45 -years-old FGD participant) As narrated by the participants, some heard about the prevention methods, but they did not understand how to practice it. Even some participants mentioned the recommended prevention measures like hand washing with soap and water, maintaining, physical distancing, and staying at home, but none of them have detail information about how to practice them. --- I have television and I heard the prevention methods of corona virus, but I am not aware and clear about its importance and how to practice it." (A 35 -year-old women, health development army) Socio-demographic and economic related barriers. Almost all participants explained that it is very difficult to practice all prevention measures of COVID-19, mainly it is difficult to stay at home, although it is essential to keep oneself from COVID-19 infection since almost all people need their daily consumption from their daily work. Therefore, it will cause hunger and another social chaos or distraction if it is obliged to stay at home. They explained as the stayat-home policy could not be wrought for low-income countries such as Ethiopia. --- "I myself could not stay at home because I could not get my daily foods because I am a daily worker." A 28 -years-old FGD participant "Ohhh.... How can we stay at home? For how many days? It is unthinkable, we all have work outside our homes." A 32 -years-old FGD participant Some of the participants relate the disease to demographic factors like age and residence. They perceived as the disease could not infect children and young's and some perceived as the disease will affect the urban population. As a result of these mentioned barriers, people fail to practice the prevention methods. "---I think the disease does not affect young people. Therefore, I will not be at risk of this disease. Thus I do not afraid and care about it too much" A 24 -years-old FGD participant "As I heard, the disease transmission will be fast in cold environment than hot areas. We are living in a hot climatic areas, so the disease will not affect us that much." (A 40 -year-old health extension worker) ".....I am 70 years old, and I have heart problems. My child is a college student and he told me as I am at risk of this disease if it comes to our village..." (A 70 -years-old FGD participant) As the participants explained, most of them did not have adequate money and other resources for their daily lives. As they narrated, those individuals who have money will buy their essential goods for a month or more than that and they can stay at home. However, those who get their income from their daily work could not afford that and could not stay at home. --- "I think staying at home is very difficult for most people in our country since most people could not get their daily consumption, rather the impact will be worse than the effect of the disease." (A 46 -year-old Women, health development army) As the participants described, there is no even access to get soap for practicing hand washing mainly for the rural people. They did not get also alcohol-based hand rub/ sanitizer/. --- "I could not get adequate soap for my family to regularly wash our hands. It is costly for me and I could not afford to buy it. I do not know hand sanitizer, I only heard from you now." (A 39 -years-old FGD participant) Among the participants, the majority of them were from rural areas. As they said, their being in rural and remote areas makes them to not to access updated information's about the disease since they are far from the health facilities and some life in hard-to-reach areas so that their frequent contact is very low to the health care providers. --- "I live in very remote area. Unless I come to a health facility, no one can come to my village and tell me about the disease since there is not transport access to our village." (A 46 years old FGD participant). "I lived in a rural area so I faced many difficulties when I came to this now because of the long distance. Since there is no road transport access." A 52 -years-old FGD participant The participants also explained that the absence of an educated person in the household would have a significant role for not practicing the prevention methods. Since it is difficult to understand the disease characteristics. Cultural and religious related barriers. The other important barrier mentioned by the participants was considering the disease as not that much serious and even some understand as it is the disease of those who eat wild animals which are not allowed to be eaten in the bible. --- "...As I heard from other people, corona will infect those who eat wild animals but not us." (A 38 -year-old health extension worker) The most frequent reason reflected by the participants for not practicing COVID-19 prevention measures was that they trusted in their religion and did not affect those who had a strong religious practice. As they described, if we follow the right principles of religion and strongly believe on GOD, they will not be affected by the disease. All participants believe that the disease comes due to our sin and GOD wants to teach his power to the world. "...I am sure that this disease is due to our sin and the GOD wishes to teach us by this disease and he just shows his supreme power to the world as no one can do anything without the will of GOD". (A 50-year-old religious leader) "...if you strongly believe in the power of GOD, this disease will not infect you." (A 65 year's FGD old participant) "There were several predictions that this kind of epidemic will occur from our ancestors. Therefore, the disease is due to in human acts like homosexuality, racism." (A 55 years-old religious leader) "If I virus infects me, I will go to use holy water and I will pray. I will not prefer to go to the health facilities since there is no any medical treatment till know." (A 45-year-old participant, FGD said). As the participants said, almost all participants were trusted with the use of traditional treatments prepared from different herbal plants. As the participants believed that the disease could be healed by common traditional treatment modalities. When they were asked also what they will do if they have the disease, they said as they will use these traditional treatments prepared by the local traditional healers from different plant leaves and roots and to use holy water. Most of the participants have a strong belief on the use of herbal medications and spiritual treatment options for treating corona virus than modern medical care. This all-mentioned reason may consider as barriers for practicing COVID-19 prevention measures. --- "I think the herbal medication which we used for common cold will be effective to treat corona since it has similar characteristics." (A 40-year-old participant, FGD said) "I think this disease is due to our sin, so praying and going to use holy water is the only option we have to do". (A 68 year-old woman, health development army) When the participants asked about their communities' perception about the disease, there are different beliefs that made to not practice COVID-19 prevention measures. One thing that was mentioned was that their perception that the disease does not come from their locality. They believe that the government should close all boarders and should test all those who came from abroad, otherwise they think as our prevention could not have any effect on the disease. "...people in my community believe that the disease does not come from our village and they strongly recommend the government to control those who come from abroad." (A 57 -year-old FGD participant) "In our community we have a strong social interaction and most of the time we live together and we meet for different social gatherings like mourn, wedding, Idir, Ikub, and even we drink coffee together. In addition, we do not want to eat alone, we eat together by sharing what we have in our hands, so this is very challenging to stop in a short period of time." (A 64 -yearsold FGD participant) "I think it is very challenging to practice prevention measures because our way of living is difficult to bring behaviour change. If someone tries to practice, some others are so careless. Therefore, one practice may not be enough and does not give a guarantee unless all other people surrounding you practice it." (A 63 -year-old women, health development army) "As I think, most of us try to practice, but sometimes we forget it and we do our normal part of life since we do not have such culture before." (A 50 -year-old religious leader) Some of the participants believe that the disease could not affect people living in very hot areas. As a result, some carelessness is observed in practicing COVID-19 prevention measures. "As you know, we live in hot area and the people in my village believe that the disease cannot come to hot areas, the disease cannot survive in hot areas..." (A 26 -years-old FGD participant) --- Discussion As evidence shows, the proper practice of the recommended prevention methods of COVID-19 is an effective and the only measure to control the spread of the pandemic [10,14,18]. However, the practice of such prevention measures is not consistent from place to place and peoples usually are reluctant to practice mainly in Ethiopia, although efforts are made by the government [11,15,23], so we tried to explore the perceived barriers for the practice of COI-VID-19 prevention measures and community risk perception on the disease focusing on both rural and urban settings. Our study revealed that most community members perceived as they are at risk of acquiring COVID-19 infection, but lack of knowledge and awareness, sociodemographic, economic, cultural, and religious factors affect them to consistently practice prevention measures. The findings are similarly reported in other similar studies conducted in different regions of Ethiopia [17,[24][25][26][27]. Regarding social distancing, our study revealed that most people are not that much aware of practicing social distancing. As most said, they did not understand the concept of social distancing, how much distance is allowed for protection in COVID-19. People's way of living has a significant role in the implementation of COVID-19 prevention measures since most people in Ethiopia are living together in closed environments. Thus, the peoples have difficulty to maintain their social distancing. This finding is in line with a study conducted in other parts of the country in which lack of awareness is a significant contributor for improper practice of COVID-19 prevention measures [17,25]. Lack of access to information is also a factor mentioned in our study as most people could not get updated information about the disease risk factors and prevention methods. This finding is consistent with a study conducted in Northwest Ethiopia which found the overall rate of information exposure about COVID-19 was less than 50 percent [24]. The other important recommended and effective prevention measure is stay at home; however, our findings showed that it is a very difficult preventive measure to practice since most people live in hand-to-to-mouth way of life so that they need to get their daily consumption through daily work otherwise they will face hunger. As a result of this, most people cannot stay at home. As Ethiopia is one of the low-income countries in the world with very young population and most of the people are living in the traditional way of life which is risky for contagious diseases like COVID-19. As a result of these, most people live together in a clustered environment, approximately 4-8 people in a single room. The finding is in line with other studies conducted in Ethiopia [11,15,28]. In our study, cultural and religious factors were mentioned as barriers for the practice of COVID-19 prevention measures. There are several strong religious and cultural practices in Ethiopia which interact with people in different social activities like mourning, wedding, Idir, Ikub, and coffee ceremonies. Therefore, it is a challenging task for the government to change people's behaviour to practice prevention measures. Culturally, there are a number of practices which need public gathering in Ethiopia. Therefore, the presence of such strong cultural norms, myths, and beliefs affect peoples to practice COVID-19 prevention measures properly [26]. Most people also do not have access to hand washing facilities. Some are unable to get soap due to financial limitations while some others do not have access to water supply. Our study also revealed that several religious barriers are mentioned as reasons for not practicing COVID-19 prevention measures. These barriers mentioned were strong religious believers which people spend most of their time at morning and evening at churches and mosques which make them to expose to each other. Peoples are reluctant to stop to go to public gathering areas like churches and mosques. There is a strong belief in the community that this disease could only be tackled through keeping religious activities since most of the society believe that the disease is caused due to people's sin as GOD's punishment. The findings are similar to other study findings in Ethiopia in which strong religious practices make people not to trust on prevention methods, rather prefer to pray together and strongly engaging in other religious activities to combat the pandemic [26,28,29]. This perception may also become challenges in the acceptance of COVID-19 vaccine if supplies become available in the area unless community sensitization to dispel myths and misconceptions. Still, the acceptance of the vaccine is low in Ethiopia (31.4%) [30]. Our study implies that, although community members perceived as they will be at risk of acquiring the disease, their practice of recommended prevention methods is nil, which is also a national problem encountered in the country [11,15]. Similarly as a study in Gondar city found, the overall prevalence of good adherence towards COVID-19 prevention measures was 51.04% [28]. --- Strength of the study Based on the researcher knowledge, this study is the first qualitative study in the area which explored the risk perceptions of the community and the perceived barriers to the practice of COVID-19 prevention measures from community key informants and other community member's perspective using in-depth interviews and focus group discussions. Limitations of the study. Since the majority of the participants were from a rural areas, the residential status may be a determinant of the perceived barriers. --- Conclusion Lack of knowledge and awareness about the disease, socio-demographic barriers, presence of strong cultural and religious practices, perceiving that the disease does not affect the young, misinformation about the disease, and lack of trust on the prevention measures were the major perceived barriers for the practice of COVID-19 prevention measures explored in this study. Our findings show the need for creating community awareness and education programs about the prevention measures of COVID-19, and it is essential to increase diagnostic facilities with strong community-based surveillance to control the transmission of the pandemic. Additionally, strict measures should be taken nationally to those who are reluctant to practice the methods to save the lives of the majority. --- All relevant data are within the paper and its Supporting Information files. --- Data analysis procedure The qualitative data collected from in-depth interviews and focus group discussions were transcribed verbatim and translated into English language and thematically analyzed using open code software version 4.02 by relistening the tape recorder several times and reading the field notes line by line. The transcripts were carefully read and entered in to open code software. Transcripts and translations were cross-checked for accuracy and consistency by two independent persons. First, repeated responses for each question were identified, and then similar responses and grouped into codes. The coding process was continued until the data was exhausted. Then, the codes were categorized and themes were merged.

According to the World Health Organization, viral diseases continue to emerge and represent a serious issue for public health. The elderly and those with underlying chronic diseases are more likely to become severe cases. Our study sets out to present in-depth exploration and analyses of the community's risk perception and barriers to the practice of COVID-19 prevention measures in South Gondar Zone, Northwest Ethiopia.A qualitative study was done in three districts of South Gondar Zone. Community key informants and health extension workers were selected purposely for in-depth interviews and focus group discussion. The interviews were conducted by maintaining WHO recommendations for social distancing and use of appropriate personal protective equipment. The sample size for the study depended on the theoretical saturation of the data at the time of data collection. The qualitative data generated from in-depth interviews and focus group discussions was transcribed verbatim and translated into English language and thematically analyzed using open code software version 4.02.Three main themes and five categories emerged from the narrations of the participants regarding the perceived barriers for the practice of COVID-19 prevention measures. A total of 9 community key informants (5 women development armies (HDA), 2 health extension workers (HEW), and 2 religious leaders participated in the in-depth interview, while two focus group discussions (7 participants in each round) were conducted among purposely selected community members. The age of the participants ranged from 24 to 70 years with

Background During the early stages of the 2020 COVID-19 pandemic, many governments and public health experts were focused on finding best practices in controlling the spread of the disease. Fact checking, source critique and precision in how data and statistics are presented are necessary preconditions for the sharing of cross-border experiences. As researchers we are also part of the media: we may be interviewed, we write opinion pieces and we share news on social media. We are reliant upon accurate information and we also have a responsibility for how we engage with the media. Using the example of international reporting on Sweden's COVID-19 response, in this debate I examine the ways in which the international media both constructs and represents a pandemic. One issue was that much of the media reporting was not fully accurate, hindering an in-depth discussion of best practice. Media reporting is also often at a superficial level and does not allow space for the details of policies or a consideration of regional or local experiences. While there was a desire for facts and evidence, the frantic search for best practice also led some to draw premature conclusions. All of this misleading media coverage led to wasted time: Swedish representatives were forced to put significant energy into combating misinformation, rather than dealing with domestic challenges. A second issue was over the use of 'culture' as an explanation for policy choices. Culture and context certainly play a role in any country's response; yet, a focus on culture can lead other countries to dismiss the lessons learnt from cultures that are seemingly unlike their own. A focus on culture can also lead to the exotification or simplification of a country's response. This article examines the first month of the 2020 COVID-19 pandemic. Sweden took the same strategy as most other countries, working to "flatten the curve," by slowing transmission so that that the healthcare system could cope with the disease. However, unlike most other countries, much of Sweden's implementation focused on voluntary and stepwise action, rather than legislation and compulsory measures (Table 1). This led to considerable attention in the international media, with the response gaining the moniker "the Swedish experiment." [1] The suggestion that Sweden's laissez faire response was synonymous with inaction became problematic enough for the Foreign Minister to dispute this on CNN [2]. In the book Good Sweden, Bad Sweden, journalist Paul Rapacioli describes how Sweden and Swedish values have been used as a weapon in a post-truth world [3]. Relatively few people have firsthand knowledge or experience with the country, and what they know tends to be positive and associated with so-called progressive values such as innovation, democracy, environmental concern and feminism. Sweden also ranks highly on most league tables. This means that negative stories are tantalizing, as they provide an exciting juxtaposition to the "lazy myth... that Sweden is some kind of paradise rather than a many-layered country of ten million people." Moreover, changes in how news is produced, circulated and consumed have also given rise to misinformation, disinformation and even so-called 'fake news.'Rapacioli was writing in the context of the migrant debate which characterized much of the 2010s, arguing that the far right was seeking to undermine progressive values by using selective reporting to 'prove' that migration was destroying Sweden. In international media, Sweden is often discussed as a "peculiar country in the north where everything is either perfect paradise or a collapsing hellhole." [4] While the context of COVID-19 is different, many of the same mechanisms are at work here. This polarized view of Sweden is exemplified by two articles: three days before The Guardian reported that "Swedish PM warned over 'Russian roulette-style' COVID-19 strategy", it had published a story covering southern Sweden's emerging wine industry [5,6]. In the media reporting of the "Swedish experiment" six main narratives emerged which I discuss in the sections that follow (1) Life is normal in Sweden, (2) Sweden has a herd immunity strategy, (3) Sweden is not following expert advice, (4) Sweden is not following WHO recommendations (4) the Swedish approach is failing and (6) Swedes trust the government. While these narratives are partially grounded in reality, in some media outlets, the language and examples used to frame the story distorted the accuracy of the reporting. --- Main text I carried out a rapid review of mainstream Swedish and international news, focusing on the month following the WHO's decision to characterize COVID-19 as a pandemic (11 March 2020) (Table 2). I started with the websites of three Swedish morning newspapers: Dagens Nyheter and Svenska Dagbladet are national papers, while Sydvenskan is a regional newspaper for Sk<unk>ne. I also reviewed the websites of the two main national evening papers: Aftonbladet and Expressen, and that of The Local, which produces Englishlanguage news in Sweden. I watched COVID-19 related news on Sweden's public service television channels (Sveriges Television, SVT), using the online streaming service SVTPlay.se. I watched the WHO's press briefings every other day and the Swedish Public Health Agency (Folkhälsomyndigheten, FHM)'s daily press briefings, as well as press briefings with the Prime Minster and other government • Stay home even if you feel the least bit sick. • Wash your hands frequently with soap and water for at least 20 s. • Keep away from others both indoors and outdoors. • Keep away from others on the bus, train, subway, tram and other public transport. • Avoid, funerals, baptisms, parties or weddings. • Keep away from others at sports venues, swimming pools and gyms and avoid changing in public changing rooms. • Do not travel during rush hour if you can avoid it. • Only travel if necessary. • If you are 70 years of age or older, it is of the utmost importance that you limit your social contacts and avoid places where people gather. Key restrictions and policy measures 11 March: Public events with more than 500 people banned (took effect 12 March). 16 March: Advice to work from home if possible. 17 March: Recommendation that senior high schools, universities and higher education move to distance learning (took effect 18 March). 19 March: Schengen-wide travel entry ban goes into effect. 24 March: Crowding in restaurant, cafes, and bars not allowed; table service should be utilized to minimize crowding at the bar. 29 March: Public events with more than 50 people banned (announced 27 March). 1 April: Further detailed guidelines and binding recommendations on physical distancing for individuals and for the public and commercial sectors released: for example, shops have a responsibility to rearrange their floor space to prevent crowding; public transport operators need to reduce crowding. representatives. I took notes and used SVTPlay to go back and transcribe key passages. The State Epidemiologist at FHM, Anders Tegnell, was one of the most visible faces of the official response and drew considerable ire online [7]. I used Twitter to follow discourse on the Swedish response and to identify which news articles were being shared. Hashtags I reviewed were: #Tegnell, #Coronasverige, #CoronaVir-usSverige and #SwedenInDenial. Through Swedish media and social media, I identified which international news articles were being discussed in Sweden and reviewed these. I also used Google Search to look for further articles, using the search string 'Sweden's COVID-19 Strategy.' Additionally, I read The Guardian and BBC News online every day. Most of the international reporting I discuss here comes from English-language sources, but I also reviewed articles from Danish, German and Norwegian sources. I created an archive of 88 screenshots, primarily from Twitter, 248 news articles and transcriptions from press briefings, which I organized by date. Although many of the media outlets I reviewed print physical newspapers, I reviewed only the online versions. My approach to studying the media was inspired by ethnography. As a process, ethnography is about immersing oneself in a community, whether online or in real life. This was not a properly ethnographic study because I did not interact with people; rather I took an ethnographer's view while consuming online media. A good ethnographer approaches the field with an open mind, and uses inductive and iterative approaches, akin to grounded theory; that is one does not go into data collection with a hypothesis, rather the themes, concepts and narratives are identified through immersion in the setting [8][9][10]. In this case, I immersed myself in online news media and social media, primary Twitter. As part of the analysis I drew upon narrative approaches from the social sciences: I examined how media outlets created an understanding of the Swedish approach through the ways in which they organized and presented facts and interview data. While much of the international reporting of Sweden was balanced and mostly accurate, misinformation and de-contextualized truths wove in and out of the coverage, giving rise to further misinterpretation on social media. In the sections that follow, I discuss six of the main narratives of media reporting about the Swedish response during the first month of the pandemic. This debate provides a snapshot; that is, as the pandemic progressed the dominance of certain narratives waxed and waned, and some of the content changed. However, the early reporting set the tone for later news stories, so it is worth exploring this first month in detail. [11]. International reporting painted a picture of Swedes eating at outdoor cafes and strolling in the sunshine, or even skiing [12,13]. Trains and busses were'still shuttling people all over the country' and cinemas were open [14]. 'Sweden also keep domestic flights going, 'despite the risk of spreading the coronavirus.' [15] On the 9th of April, Vox reported that: For months, the Scandinavian nation allowed large gatherings to form, schools for younger children to remain open, restaurants to serve late-night guests, and resorts to welcome thrill-seeking skiers [16]. The Vox article was striking, claiming that 'for months' Sweden had been acting normally. Yet, it was published less than a month after the WHO had characterised COVID-19 as a pandemic so most other countries had also been acting normal "for months." One the first English-language articles to break the Swedish story was the aforementioned "Swedish PM warned over "Russian roulette-style' COVID-19" strategy [5]. An early published version of the article stated that the Prime Minister, in a speech the night before had "urged Swedes to lunch at a local restaurant." The text was later changed: Stefan Lövfen, in fact, had specified a "takeaway" lunch during a speech broadcast on the 22nd of March. The narrative of Sweden as normal was misleading. The bulk of international media reporting focused on Stockholm, and most online videos showed the area around Kungsträdg<unk>rden and Drottninggatanthe very city center. There was some truth to this narrative: during the second full week of April, it had been widely reported in national media that Stockholmers were pushing the limits of good advice, by sitting in crowded outdoor seating areas; the city of Stockholm responded by carrying out extra checks on restaurants and cafes over the Easter weekend [17]. However, central Stockholm is not representative of Sweden. The main cinema chain in Sweden, Filmstaden, closed on the 17th of March so few people were attending the cinema when these articles were written [18]. After much discussion, SkiStar, the country's main ski operator, decided to close its facilities on the 6th of April; ski gondolas were already closed on the 24th of March [ 19]. Sweden's decision to keep a select few flights open was due to their designation as "societally important" routes: Visby is the main city on Gotland, an island. Kiruna is above the polar circle and the other cities: Lule<unk>, Ume<unk>, <unk>stersund, Skellefte<unk>, <unk>rnsköldsvik, Sundsvall are also in the north and relatively remote. The government decided that flights needed to run not for tourists, but to ensure essential supplies and workers could reach these cities [20]. The wider picture -Sweden as normalwas misleading, because life was not normal. CNBC reported on a report released by Swedbank that forecast the economy would contract by 4% and that unemployment would reach 10% by the summer [21]. Already during the course of March 2020, the numbers of business going into administration (konkurs) had increased, with the hospitality industry seeing an increase of 1400% [22]. Stockholmers drinking in the sun should not symbolize the various Swedish COVID-19 experiences. --- Narrative 2: Sweden has a herd immunity strategy Now they talk about Sweden, but Sweden is suffering very greatly. You know that, right? Sweden did that, The herd. They call it the herd. Sweden is suffering very very badly, it's a way of doing it. -Donald Trump [23] During this month, several Swedish researchers who disagreed with the government's approach stated, both in national and international media, that the governments was following a herd immunity strategy. The herd immunity theory also forms a large part of the "Russian Roulette" article, which served as a basis for much of the English-language reporting, and the concept came up in most international reporting [16]. On the 9th of April, Vox published an article with the title and subtitle: Sweden's government has tried a risky coronavirus strategy. It could backfire. Critics say the government seeks "herd immunity" from the coronavirus. That could lead to more deaths. The author tweeted the article with the text: It sure looks like Sweden is trying to achieve #herd immunity." The government and FHM denied multiple times that they were following a herd immunity strategy, with the Foreign Minister Ann Linde discussing this on Swedish television and CNN [2]. The herd immunity rumor has a complicated history. Mathematician Marcus Carlsson forwarded this theory in a series of YouTube videos and Annika Linde, the former state epidemiologist also promoted the theory that FHM's strategy was based on herd immunity; she did this in a Facebook post which was picked up in the media [5,24,25]. State epidemiologist Anders Tegnell also mentioned the term often on television which further led to public confusion [26]. At other times his quotes were taken out of context. In an article in Svenska Dagbladet, which was re-quoted in international media, Tegnell stated that the main strategy was to have a slow transmission of COVID-19 so that the healthcare system could manage, but that herd immunity was "not contradictory" with this. He also said that herd immunity was a "great concept." [24] Part of the herd immunity rumor had to do with how Swedish implementation was described in international media as being drastically different than other countries. Anders Tegnell argued that critics were emphasizing the differences, rather than similarities of the approach [27,28]. That is, Sweden was also trying to flatten the curve but without draconian measures. The government took this line not in order to promote herd immunity but because they did not think it realistic or healthy "to keep people inside for 4 or 5 months." [29] That is, the government and FHM considered the level of restrictions put in place in the first month of the pandemic to be sustainable over the long term. Similarly, schools remained open in part because of the negative impact of closure on mental and physical health, and also because it would create a childcare problem, not least for key workers [30]. There was no evidence that FHM had a "secret" herd immunity plan, and it was never the 'deliberate strategy' of the government to 'let the virus spread its course and spread through the population [31].'It is more accurate to state that there was a hope that some level of herd immunity would be a side effect of the approach, but it was not the main strategy. Narrative 3: Sweden is not listening to experts or data At the end of March, a group of over 2000 individuals signed an open letter to the government asking for more measures. Time described them as 'doctors, scientists and academics," the Guardian as "doctors, scientists, and professors," Vox as "academics and experts" and The Economist as'scientists and professors." [11,14,16,32] Both the Economist and the Guardian pointed out that Prof Carl-Henrik Heldin, chairman of the Nobel Foundation was on the list. To support this statement Time and Vox linked to The Guardian's live reporting from 24 March in which Jon Henley had written that "More than 1,500 medical and other academics in Sweden ranging from full professors to post-doc researchers have signed a petition calling on the government to change its coronavirus strategy." There are several problems with this letter. Although there are a number of well-regarded public health experts on the list, the six main authors, while professors, did not work in public health. This was something pointed out by Dagens Nyheter, while international media ignored this [33]. Moreover, there were many masters and PhD students on the list, as well as a handful researchers outside of Sweden who signed the letter. These are hopefully the experts of the future, but not necessarily current experts. The letter was not widely circulated. The copy of the letter and list I saw was tweeted by Sten Linnarsson on the 26th of March and then retweeted by Nils-Göran Larsson, both of whom were signatories. To my knowledge, the only media source which has shown the letter was Vice in a video [34]. The first paragraph of the English version, as displayed by Vice was: The Swedish government must act not (sic) to step up the fight against COVID-19 We have signed this open letter to request the Sweden Government immediatley (sic) take measures to follow the recommendations of the World Health Organization (WHO) to their full extent, including interventions to reduce the mobility and contact in the population and to support and incorporate the latest technologies to increase our capacity to test for Covid-10 (sic) infections nationwide. The letter's starting point was that the Swedish government had not followed WHO Recommendations (I discuss this narrative in the following section). The second paragraph of letter also perpetuated the rumor that Sweden was following a herd immunity strategy. A second example comes from media reporting of a piece in the Lancet. According to AFP and Euronews, respectively: "A study published last week in the medical journal The Lancet, titled "COVID-19: Learning from Experience," said that the "initial slow response in countries such as the UK, the USA, and Sweden now looks increasingly poorly judged." [35] "A study published in the medical journal, The Lancet, indicated that "the initial slowness of reaction from countries such as the UK, the US and Sweden now appears to be increasingly unwelcome" [36]. The'study' was in fact an editorial written by the Lancet's editor-in-chief Richard Horton [37]. Richard Horton is immensely influential in global health and no doubt could be described as an expert. However, this was his opinion, not a study, even though it was reported as such. These examples aside, there were other instances of researchers, including those with backgrounds in virology, epidemiology and public health who raised valid concerns about how FHM has handled the evidence. An interesting question revolved around modelling done by Imperial College. This included the infamous work that, according to media reports, pushed the United Kingdom to change its COVID-19 strategy, although Anders Tegnell has argued that the British media have overstated the importance of the research in its policy decisions [38]. Within the debate pages of Swedish newspapers, researchers discussed the strengths and weaknesses, scope and applicability of the Imperial College modelling to the Swedish context [39][40][41][42]. FHM also publicly discussed their consideration of the evidence [43]. Some Swedish researchers also accused the Swedish government of only looking at the economy, with one writing in a leaked email chain "How many lives are they prepared to sacrifice so as not to... risk greater impact on the economy? [5,11] Another researcher was cited several times as stating that the government was not using scientific evidence in their decision making, stating: "We don't have a choice, we have to close Stockholm right now." [16,32,44,45] Modelling is not evidence, and models are only as good as the assumptions and data that is fed into them. Professor David McCoy discussed eloquently in The Guardian how the health and socioeconomic impacts of various lockdown measures had not been modelled [46]. I would suggest that the Swedish government and authorities did not take into account "the economy" but were cognizant that there are long and short term health effects of unemployment and underemployment. Additionally, the choice to 'lockdown' an entire community was also an experiment taken in the absence of definitive evidence, and many types of 'lockdown' measures, such as the quarantining of a whole city are not possible under Swedish law [47,48]. Yet, international press focused on the conflict between FMH and other researchers, rather than fundamental questions about the limits and potential of mathematical modelling, and its role in policy-making. This was a missed opportunity to educate readers. I return to the scope of evidence in the conclusion, but offer a further example here: On the 10th of March, FHM recommended against 'unnecessary' visits to care facilities (äldreboende) and the government implemented a ban on the 1st of April. Care facilities are run by local municipalities and in practice many municipalities had already banned visits before the 1st of April. In Norway, municipalities implemented different types of bans at different times, many later than in Swedish municipalities, with a national ban coming into force on the 7th of April. Yet, already by the 7th of April, FHM had reported that COVID-19 was present in half of the care homes in the city of Stockholm [49]. Undertaking serious analysis in the middle of a pandemic is challenging, but an initial comparison between care homes in Norway and Sweden suggest that simply banning visits is not an easy solution because it did not work in Sweden. --- Narrative 4: Sweden is not following WHO recommendations The argument that Sweden is not following WHO recommendations came up in several debate articles in Swedish news [42,50]. Internationally, Time reported that "experts say the Swedish government is not following the World Health Organization." [32] CNN included a quote from a spokesperson from the WHO Regional Office for Europe who had said "it's imperative" that Sweden "increase measures to control spread of the virus, prepare and increase capacity of the health system to cope, ensure physical distancing and communicate the why and how of all measures to the population." They also included a link to a Tweet by WHO/Europe from 8 April which read "There is a fresh surge [of cases] in Sweden." It also found in the letter written by over 2000 individuals, as noted in the previous section. There are two issues with this line of argument. Firstly, there is a general misunderstanding about what WHO recommendations are and are not, and secondly, I argue that Sweden followed the WHO's'recommendations' regarding public health measures [51]. The WHO issues many types of documents, such as recommendations, advice, or technical guidance. Articles 19-23 of the WHO's constitution cover the legal basis for recommendations, conventions, agreements and regulations. What people think of as "WHO recommendations" are usually a mosaic of binding and mostly non-binding policy, advice, technical guidance and even international law and commercial law. A'recommendation' found in an Interim Guidance document is completely different than a'recommendation' endorsed or adopted by the World Health Assembly. Many commentators took Dr. Tedros's mantra "test, test, test" to be a WHO recommendation that all countries should test everyone. A more accurate interpretation of the WHO's Interim Guidance is that testing needs to be carried out with regard to a county's capacity, context and situation [52]. In general terms, the WHO provides guidance to 194 Member States, which means that WHO'recommendations' are rather vague. Most of the time, the WHO provides a menu of policy options from which countries can adapt or implement measures to take based on their national and local contexts. In Interim Guidance issued on the 7th of March, the WHO provided recommendations for public health measures, writing the countries should 'consider, based on local and/or global evaluation' the following: avoid crowing, school closures, public transportation and workplace closures and public health quarantine [53]. FHM did consider these, and adapted WHO recommendations in line with the Swedish context. Another example comes from the International Health Regulations (IHR). The WHO declared COVID-19 a public health emergency of international concern on the 30th of January and issued temporary recommendations under the IHR. Some temporary recommendations under the IHR are binding, such as the obligation to share information with WHO, but most are not binding. In general, the WHO prefers that countries avoid border closures, in part because the disruption they cause can prevent medical supplies from reaching populations in need; also border closures are not usually that effective. Specifically, under Article 43 of the IHR: States Parties implementing additional health measures that significantly interfere with international traffic are obliged to send to WHO the public health rationale and justification within 48 h of their implementation. WHO will review the justification and may request countries to reconsider their measures. WHO is required to share with other States Parties the information about measures and the justification received [54]. Sweden did not close its borders on its own but was forced into the Schengen-wide border closure. WHO advice, technical guidance and recommendations are extremely important, not least in a pandemic situation and countries should take seriously documents issued by the WHO. It is important to understand that most'recommendations' are not universal but are written in a way which allows countries to adapt them to their own contexts. Whether or not Sweden's interpretation of WHO'recommendations' in this instance was effective or appropriate is a different question; but Sweden followed WHO'recommendations' with regard to public health measures. This distinction matters: on the 21st of May 2020, the sustainability rating company Standard Ethics, downgraded Sweden's rating having assessed that: "during the first phase of the COVID-19 pandemic, Swedish health policy did not comply with World Health Organization recommendations." Narrative 5: the Swedish approach is not working; and Sweden has or will change course On the 3rd of April, Reuters led with "Sweden's liberal pandemic strategy questioned as Stockholm death toll mounts," in which they wrote about a "spike in novel coronavirus infections and deaths." [44] The Business Insider reported that "Sweden, which refused to implement a coronavirus lockdown, has so far avoided a mass outbreak. Now it's bracing for a potential surge in deaths." [55] On the 4th of April Bloomberg noted that the number of Swedish deaths rose to 373 on Saturday, up 12% from Friday [56]. On the 9th of April, Time Magazine reported that "Sweden's relaxed approach to the Coronavirus could already be backfiring," using the case fatality rate as part of the argument, although they recognized the challenges of this metric [32]. On the 4th of April, Deutsche Welle reported that "Sweden mulls U-turn on coronavirus restrictions." [57] Part of this reporting originated in a bill (proposition) proposed 7th of April (but which had been discussed in different forms during the previous days). This bill gave the government additional powers under the Communicable Diseases Act (Smittskyddslagen) to implement so-called 'lockdown' measures, such as closing harbors, train stations and shopping malls [58]. The interpretation in much of the international media was that Sweden was changing course. However, the Irish Times was one of the most accurate accounts: "Sweden has reached cross-party agreement on sweeping emergency powers to close shops, restaurants and schools in response to the coronavirus crisisbut the government has no immediate plans to use them." [59] That is, during the course of the pandemic it became clear that existing legislation did not allow the government to act quickly in taking more 'lockdown'-type restrictions, but there will be no immediate plans to use the measures [60]. In a press conference The Prime Minister referred to these new measures as 'tools in a toolbox.' The idea was not that Sweden was changing course but wanted to be prepared to face future uncertainties caused by COVID-19. The effectiveness of the Swedish approach is difficult to fully assess at the time of writing, which I return to in the conclusion. Already after the first month of the pandemic, Sweden had seen a decrease of seasonal influenza and calicivirus disease (vinterkräksjukan) which indicated that voluntary social distance measures were working [61,62] People's behavior had also changed, although perhaps not as quickly as it could have [63]. Additionally, the pandemic curve in Stockholm was ahead of the rest of the country, making it difficult to draw conclusions on Sweden overall. It is also challenging to compare cross-country data, especially when journalists are not precise about what numbers are being reported: case fatality rates, total deaths, deaths per 100, 000 or excess mortality. Other newspapers compared Sweden to the other Nordic countries, without noting that Sweden's population is roughly twice as large as Norway, Denmark, or Finland. Although these countries later diverged in the number of deaths per 100,000, during the first month of reporting Denmark and Sweden had similar rates, so this omission distorted the early picture. If more stringent measures are put in place, then they may be reported as 'Sweden changing course' or 'the Swedish approach is failing.' However, as early as the 22nd of March in an address to the nation, the Prime Minister was very clear that more measures should be expected; and the Public Health Agency stressed repeatedly that they were are taking a stepwise approach: "the right measure at the right time." Similarly, when some travel restrictions for the summer were lifted in June, the government was clear that restrictions would be put back in place if the epidemiological situation changed [64]. If the government puts in more measures, it will be in line with their actions and the advice of FHM. As a follow-up example, in June 2020 Anders Tegnell gave a short interview to Sveriges Radio in which he stated that there was 'room for improvement' in the Swedish approach and that, if the virus were encountered today, with the knowledge that we now have, that Sweden's approach would have been between what the country did and what other countries did [65]. This was picked up in the international media with headlines such as "Sweden chose a looser lockdown. The scientist behind the strategy now says the death toll is too high" and "Architect of Sweden's coronavirus strategy regrets not imposing tougher lockdown." [66,67] Tegnell clarified his statement the following day, as reported in The Local: We still think that the strategy is good, but you can always make improvements, especially when looking back. I personally think it would be rather strange if anyone answered anything else to such a question. You can always do things better," he said, adding that he did not necessarily think he had been misquoted, but that his comments had been overinterpreted [68]. --- Narrative 6: Swedes trust the government The final narrative returns us to some of the points raised in Good Sweden, Bad Sweden. Much of the early international coverage focussed on high-levels of trust in Sweden [32]. Historian Lars Träg<unk>rdh explained it accurately for several papers, noting that citizens trust public institutions and the government, the government trusts the people to do the right thing, and there is social trust amongst citizens as well [11,12,69]. As part of this is the cultural code of recommendations and advice. The government expects that Swedes understand that recommendations are to be followed, they are "not just loose tips," as the Prime Minister clarified in a speech. Representatives from FHM repeatedly noted that public health measures, including infectious disease control, are based on "voluntary measures, acceptance and understanding in the population -that the Swedes trust and follow recommendations from their authorities." [70] By most quantitative and qualitative measures, there are high levels of trust in Sweden. However, in describing the Swedish approach, national myths blend with truths. In addition to the "Swedes trust authority" discourse there are also a number of ideas circulating about how Swedes are. Writing for Politico Lisa Bjurwald, herself Swedish, argues that Swedes naturally socially isolate, writing "Skype-based relationships? No hugging? For Swedes, that's not social distancing. That's just life." [71] Carl Bildt, a former prime minister joked to the Economist that, "Swedes, especially of the older generation, have a genetic disposition to social distancing anyway [11]. On social media, people have questioned whether or not Swedes are responsible, citing the existence of the state-owned liquor store Systembolaget as evidence. This is the narrative that Swedes cannot control their alcohol intake and must have alcohol sales restricted. Even the term 'peace-damaged' (fredsskadat) was discussed in national media [72]. Simplified, this is the idea that, because Sweden has not been at war for over 200 years, it is difficult for people to imagine bad things happening on a large scale on Swedish soil, and that is why Sweden did not act as forcefully as other countries. A related narrative, found more in social media than in the news, is that Swedes blindly trust authority, or that "trust seems to be equated with not questioning." [73] Some of this has come from foreign residents, who are not used to the Sweden system or feel that Swedes are dismissive of their concerns [74]. Similarly, Swedes have been described as arrogant or even post-colonialist in their approach to non-Swedes living in Sweden [74,75]. Yet, a country cannot be reduced to a few cultural stereotypes, and distilling an entire country into a set of traits and values also ignores complexity. It is not simply the Swedes have high levels of trust and 'know' how to follow recommendations. Much of the government and authority's response also has to do with the legal system and history of Swedish administration, the latter of which gives a great deal of independence to agencies such as FHM [12,47,48,76]. There are other contextual factors that impact the spread of disease, which have been noted by some articles. Sweden has a low population density compared to most other European countries and more than half of households are single-person (https://ec.europa.eu/ eurostat/web/products-eurostat-news/-/DDN-20170905-1?inheritRedirect=true). Focusing on culture also leads to constrained thinking, in which commentators find it difficult to explain 'Stockholmers drinking beer in the sun' or people ignoring recommendations, without concluding that assumptions of 'Swedish responsibility' are false. One can also become stuck in a cycle of criticizing Swedish 'culture.' In setting health policy, there is a fine line between taking context into account and reducing behavior to 'culture,' as if it is a fact or a variable to be plugged into a model. Rather, history, law, socioeconomic factors and tradition have created a tangled 'web of significance'to borrow from anthropologist Clifford Geertzin which different Swedes make decisions and live their lives. --- Conclusions Not all reporting is misleading, and there were a number of balanced articles in the month following the WHO's characterization of COVID-19 as a pandemic

In the first month of the 2020 COVID-19 pandemic, Sweden took the same strategy as most other countries, working to "flatten the curve," by slowing transmission so that the healthcare system could cope with the disease. However, unlike most other countries, much of Sweden's implementation focused on voluntary and stepwise action, rather than legislation and compulsory measures, leading to considerable attention in the international media. Six main narratives emerged in the international media reporting on Sweden during the first month of the COVID-19 pandemic: (1) Life is normal in Sweden, (2) Sweden has a herd immunity strategy, (3) Sweden is not following expert advice, (4) Sweden is not following WHO recommendations (5) the Swedish approach is failing and (6) Swedes trust the government. While these narratives are partially grounded in reality, in some media outlets, the language and examples used to frame the story distorted the accuracy of the reporting. This debate examines the ways in which international media both constructs and represents a pandemic, and the implications for how researchers engage with news and social media. Cross-country comparison and the sharing of best practice are reliant on accurate information. The Swedish example underlines the importance of fact checking and source critique and the need for precision when presenting data and statistics. It also highlights limitations of using culture as an explanation for behavior, and the pitfalls of evaluating policy during a pandemic.

://ec.europa.eu/ eurostat/web/products-eurostat-news/-/DDN-20170905-1?inheritRedirect=true). Focusing on culture also leads to constrained thinking, in which commentators find it difficult to explain 'Stockholmers drinking beer in the sun' or people ignoring recommendations, without concluding that assumptions of 'Swedish responsibility' are false. One can also become stuck in a cycle of criticizing Swedish 'culture.' In setting health policy, there is a fine line between taking context into account and reducing behavior to 'culture,' as if it is a fact or a variable to be plugged into a model. Rather, history, law, socioeconomic factors and tradition have created a tangled 'web of significance'to borrow from anthropologist Clifford Geertzin which different Swedes make decisions and live their lives. --- Conclusions Not all reporting is misleading, and there were a number of balanced articles in the month following the WHO's characterization of COVID-19 as a pandemic. Several of the 'foreign' reporters were Swedish or had lived in Sweden for years, spoke the language and could contextualize the reporting. It is not uncommon that an editor changes text and headlines, so the blame for sensationalizing does not always lay on the journalist. Also, owing in part to changes in the news industry over the past two decades, working conditions can be very difficult, especially for freelance journalists. Moreover, a journalist or newspaper cannot always control how a story is interpreted and shared on social media or in the online comments sections. For instance, several newspapers covered a statement from the Italian ambassador in which he stated that the WHO ranked the Italian health system as the second best in the world, while Sweden's was 23rd [77]. This statement became truth on Twitter. Very few noted that this ranking was from the 2000 World Health Report, itself a controversial report [78]. Additionally, the interplay between social media and news media amplifyies bias, which feeds back into reporting. I also do not wish to criticize individual researcherspeople's complex arguments can be turned into soundbites when they are interviewed in the news; this is why it can be helpful to attend media training offered by many universities. However, while most articles are balanced, the negative ones are shared more widely [79]. This can have a negative impact upon public health action and can hinder cross-border comparisons. Both the head of FMH, Johan Carlson and the State Epidemiologist, Anders Tegnell argued multiple times that the Swedish implementation of COVID-19 measures was not that different, but rather it was the 'rhetoric and language' that differed [80]. Interviewed in The Local, science journalist and epidemiologist Emma Frans noted that journalists often look for conflict, which is problematic in a pandemic: "It's a different issue compared to other kinds of news reporting. You have to be very responsible and it's very important to get things right," she says. "In a democracy you should always be able to question authorities, but with disease outbreaks, it's also important that people trust the authorities and listen to their advice otherwise we get problems [81]. While I have focused on how international media has reported on Sweden's handling of the COVID-19 pandemic, these findings raise broader questions on how the media reports on all country responses. Firstly, fact checking and source critique in a pandemic is important. Two characteristics of the international reporting was that many of the articles cited previous articles without checking the source or that they reported 'expert' opinion as scientific truth. The reporting was also not always discerning about defining concepts such as expertise, facts, data and science. Secondly, some international media failed to communicate the complexities of science and policy, especially in the first month of the pandemic. For instance, terms like "surge" or "dramatic increase" are neither precise nor helpful. The line "the death toll keeps rising" is particularly useless because the number of deaths will rise until the disease is eliminated; it clarifies nothing about changes in the number of daily deaths. A'spike' in cases can be a data artefact; it can be a result of delays in reporting or cases getting'stuck' in the system before it can be updated. Other metrics, such as the number of people in intensive care, may also be useful. The idea of looking at excess mortality barely surfaced in mainstream media. Different countries are at different places in their pandemic curve, and it is difficult to draw immediate conclusions on cross-country data. Most science and health correspondents are well versed in interpreting findings and metrics. However, because there was such demand for COVID-19 context, some newspapers relied on journalists not used to interpreting public health policy interventions and graphical displays of data. Thirdly, country-wide statistics and narratives are misleading. In the first month of the pandemic, the majority of cases and deaths were in greater Stockholm. The pandemic played out very differently in other parts of the country, both in terms of statistics but also behavior. Fourthly, the use of 'culture' as an explanation also has its limits. When culture is used as an explanation, it can be easy to dismiss a country's response as too contextual. While policies cannot be copied and pasted between countries, the lessons learnt from policy failures and successes are useful. Another challenge is that when culture is reduced to a soundbite, it can be tempting to focus on exotic practices or to judge people's beliefs and traditions as negative. That is, culture can become a scapegoat for 'harmful behavior.' The reality is that culture is somewhat ephemeral, it exists alongside tradition, law and is shaped by socioeconomic forces. Media is how we know what we know; it constructs reality. Journalists and editors choose what stories are told and what stories are not told [82]. Fake news is rarely totally fake and this matters for all of us. As Paul Rapacioli has noted: "When it comes to news, truth is context. A lack of context doesn't make a story false but it greatly reduces the truth of it." [3] Cross-border comparison is reliant on accurate information. All countries, particularly those in which the pandemic has not yet peaked, look for information on best practice. How we can learn from each other when we do not have the facts straight? Finally, none of us are experts in all aspects of science and as researchers and lectures we depend on good reporting for understanding the world, and for communicating it to our students. As researchers we are also part of the news. We all have a responsibility to act on our conscious and speak out if we have a scientific basis for questioning authority. We have a responsibility to speak within the bounds of our expertise and to present facts, not suppositions. We have a responsibility to be critical of sources, both scientific and non-scientific. As many researchers are active on social media, we also have a responsibility to share accurate information and news. --- A final word on evaluating policy in a pandemic It is already clear that there were problems with the Swedish public health response to COVID-19, not least that the government and FMH, like many other countries, underestimated the risk of COVID-19 spreading outside of Asia. Early reporting also identified class-based differences in mortality and morbidity patterns [83]. Taking the wider perspective, the success or failure of the so-called "Swedish experiment" may never be definitely proven. Researchers will be analyzing the data on COVID-19 for years and evaluating policy measures is challenging. While it may be relatively easy to compare the number of deaths directly attributable to COVID-19 there are many other metrics to consider in calculating the overall burden of disease. One also needs to look at the demographics of the spread, and to consider regional differences when analyzing epidemiological patterns. There are also downsides, particularly in places with more stringent 'lockdowns.' Already in the first month of the pandemic, alarms were raised over the risks living with an abuser in 'lockdown' and of increased alcohol misuse. What will be the cost to health of the lost wages and unemployment affect health outcomes? What is the short and long-term impact of limited exercise and confinement, postponed health and dental care? Many of these challenges disproportionally affect already disadvantaged communities, and particularly low and middleincome countries. The response to COVID-19 may potentially exacerbate inequalities for generations to come. Public health is not about the individual case. Arguably it is not even about the specific disease. Rather, public health is a about the whole picture. Here we can invert the question about the Swedish 'experiment': What is the real experiment, and what do we have evidence for or against: living in a state of semi-physical distancing or locking down entire countries? The answer is that we do not know yet. --- Availability of data and materials All the online news sources are available online, although some are behind paywalls. Most television news from SVT is available for a week on SVTPlay.se. Researcher can request access to the television archives upon request. --- Author's contributions It is a single-authored paper. The author(s) read and approved the final manuscript. --- Ethics approval and consent to participate Not applicable; no research on humans was carried out. --- Consent for publication Not applicable. --- Competing interests None. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

In the first month of the 2020 COVID-19 pandemic, Sweden took the same strategy as most other countries, working to "flatten the curve," by slowing transmission so that the healthcare system could cope with the disease. However, unlike most other countries, much of Sweden's implementation focused on voluntary and stepwise action, rather than legislation and compulsory measures, leading to considerable attention in the international media. Six main narratives emerged in the international media reporting on Sweden during the first month of the COVID-19 pandemic: (1) Life is normal in Sweden, (2) Sweden has a herd immunity strategy, (3) Sweden is not following expert advice, (4) Sweden is not following WHO recommendations (5) the Swedish approach is failing and (6) Swedes trust the government. While these narratives are partially grounded in reality, in some media outlets, the language and examples used to frame the story distorted the accuracy of the reporting. This debate examines the ways in which international media both constructs and represents a pandemic, and the implications for how researchers engage with news and social media. Cross-country comparison and the sharing of best practice are reliant on accurate information. The Swedish example underlines the importance of fact checking and source critique and the need for precision when presenting data and statistics. It also highlights limitations of using culture as an explanation for behavior, and the pitfalls of evaluating policy during a pandemic.

INTRODUCTION A blockchain is a decentralized state machine, in its simplest form, a decentralized ledger for financial transactions. The machine is controlled by several distinct processes, called nodes or agents, and computes by packaging state transitions (e.g., the order of transfering tokens from Alice's account to Bob's) in batches, which are called blocks. Each block determines the next global state of the machine (e.g., the next state of the ledger, where Alice's account has tokens less and Bob's more). Each new block is appended to the list of older blocks, thereby determining a growing appendonly list of global states of the machine-its computation history. Crucially, each block points in a temper-proof way (via a cryptographic hash) to the previous block to which it was appended, thereby enforcing an immutable description of the history of the machine. The core of a blockchain is the protocol that the nodes follow in order to agree on which transitions to incorporate into the list, i.e., to achieve consensus on the state of the machine. Paper motivation. Blockchain was born as the backbone of the Bitcoin cryptocurrency. The consensus protocol behind Bitcoin, known as Nakamoto consensus [33,52], showed that such a decentralized consensus on an append-only list is possible even in large, open peer-to-peer networks. This was a significant breakthrough with respect to existing approaches to consensus, which worked specifically on systems of limited size with controlled access. This breakthrough relied on one key insight: nodes in an open system cannot be controlled, so their influence on consensus should be kept at bay by linking it to the ownership of a non-monopolizable resource-in the case of Bitcoin, computing power. In other words, nodes in an open network cannot be controlled nor trusted, but can be incentivized. Nodes that contribute to achieving consensus are rewarded with (they'mine') tokens, that is, units of currency. This is the key intuition behind the application to consensus of the Proof-of-Work (PoW) technique (originally developed to thwart email spamming [27]), which has proven extremely robust, maintaining the Bitcoin blockchain for over a decade (cf. for recent overviews [53,70]). So, blockchain consensus is the result of a rational response to incentives. Traditionally, research in blockchain has focused mostly on the cryptographic foundations and the distributed computing aspects (e.g., protocol correctness) of the technology. At the same time, a game-theoretic perspective on blockchain protocols has also been gaining attention: is behavior in accordance with the protocol economically rational, in some precise equilibrium-theoretic sense? In other words, are protocols strategy-proof? This game-theoretic perspective has historically been marginal in distributed computing [1], but has proven significant in blockchain. By now, it has been extensively applied-including by researchers in the AAMAS community-to Nakamoto consensus (e.g., [4,28,56,61]), as well as to other protocols (e.g., [2,5,17,45]). See [44] for an extensive recent overview. However, the economic issues that the designer of a blockchain system faces go well beyond incentive-compatibility alone, and reflect broad issues in the design of collective decisionmaking mechanisms, such as forms of equity and fairness. This interface with the economic theory of group decision-making, and specifically social choice theory, is the focus of the present paper. Paper contribution. The contribution of this paper consists in arguing how social choice theory, and its algorithmically-focused branch-computational social choice-have an important role to play in providing stronger foundations for the principled development of blockchain technology. This paper sketches a number of research challenges at the interface of computational social choice and blockchain. I claim there exists now a perfect match between the state-of-the-art in blockchain research on the one hand, and the state-of-the-art in on computational social choice on the other. Blockchain offers a wealth of novel questions that can push the boundaries of the existing body of results of computational social choice, and in doing so contribute concrete solutions to the challenges blockchain research itself currently faces. To substantiate this claim, this paper reviews how key mechanisms that have a long tradition within (computational) social choice find deployment in blockchain systems: randomized mechanisms, voting mechanisms, and trust mechanisms. In reviewing these mechanisms this paper identifies challenges that their deployment in blockchain gives rise to, phrasing them in terms of properties that, again, have occupied social choice theorists in the past: fairness of lotteries, manipulation of voting, and false-name proofness of trust mechanisms. Such an overview has no claim to be exhaustive and rather aims at illustrating, by means of examples, the potential for cross-fertilization between the two research areas. Given the nature of this paper I will try to limit technical jargon to a minimum, use informal language as much as possible and just convey the gist of my arguments without resorting to explicit mathematical details. --- RANDOMIZATION 2.1 Lotteries in Blockchain and Social Choice Randomization has a long tradition in distributed computing as a way to bypass impossibility results such as the so-called FLP impossibility theorem [31]. 1 Blockchain has further stressed the importance of randomization. Lotteries are at the heart of the main approaches to blockchain consensus such as proof-of-work (PoW, currently used in both Bitcoin and Ethereum [12]) and proof-ofstake (PoS, currently used for instance in the Ouroboros protocols [23,24] and Algorand [38]). At a high level, and leaving network latency issues aside, these protocols work as follows: nodes participate in a distributed lottery; the winner appends a new block to the chain (or, depending on the specific protocol, becomes part of a committee which will then vote on the block to be appended) and receives a compensation in the native currency-what is referred to as mining. Importantly, the chances of winning this lottery depend on a resource that is assumed not to be monopolizable-such as computational power (in PoW) or currency ownership (in PoS). This makes participation to the lottery (directly or indirectly) costly and prevents the manipulation of the lottery through identity forging (the so-called Sybil attack [26]). Based on this blueprint, several so-called 'proof-of-X' (PoX) [7] schemes have been proposed (e.g., proof-of-storage [51]). Like in distributed computing, also in social choice theory randomization is an established route to circumvent fundamental impossibility theorems of the deterministic social choice framework, such as the Gibbard-Satterthwaite theorem [36,57]. 2 In a famous later theorem Gibbard himself [37] showed how randomization can provide a possibility result that is out of reach in the deterministic setting: the random draw of one agent (the 'dictator', or 'leader') from the set of agents is the only decision mechanism that is 'lottery' strategy-proof (in the specific sense of stochastic dominance 3 ) and 'lottery' efficient (in the sense of never assigning positive probability to alternatives that are Pareto dominated). From a 1 The Fisher, Lynch, Paterson (FLP) impossibility result states that in an asynchronous system no deterministic consensus protocol exists which can tolerate even just one faulty node, where a faulty node is a node which stops interacting with the protocol. 2 The theorem states that no social choice function exists which is simultaneously non-manipulable and non-dictatorial. 3 I provide a sketch of the definition here: is the set of agents; is the set of alternatives; P = 1,..., | | <unk> P is a profile of preferences over. A randomized decision rule is a function : P <unk> <unk> ( ), that is, a function assigning a lottery over to each profile of preferences. A lottery stochastically dominates a lottery for iff <unk> : ( ) <unk> <unk> : ( ), for any <unk>. A randomized decision rule blockchain perspective the theorem can be thought to offer a justification for the use of randomization in PoW and PoS consensus: using a lottery to select the next block in the chain is economically efficient and elicits the true preferences of nodes about what block should be included in the chain. A string of works have further extended Gibbard's result (e.g., [6,37,55]) and randomization is recognized as an important toolbox in the design of economic mechanisms, from allocation to voting [11]. This body of theory offers a sophisticated framework to understand important economic properties of the sort of lotteries deployed in blockchain protocols. One such property is fairness. --- Challenge: Long-Term Fairness By linking winning chances to resource ownership and linking lottery outcomes to currency allocation, blockchain protocols make participation resistant to Sybil attacks (participation is costly) and at the same time monetarily rewarding ('no-show' is disincentivized). This scheme, however, has been shown to induce centralizing effects on several blockchain systems (see, e.g., [30,46]). Participants with more resources, all other things being equal, have higher chances of being selected by the lottery and thus accrue more resources in the long term. The resulting allocation of resources invested in the system becomes therefore increasingly unequal over time, and this inequality in resource ownership de facto biases the lotteries upon which consensus is based. In blockchain, such bias translates in a centralization effect, which makes the blockchain more vulnerable to failures or attacks. So, randomized blockchain consensus protocols need to implement, in the long term, allocations of winning chances which are fair towards participating nodes in two somewhat opposing senses: fair in the sense of being lotteries that are not too biased towards few participants, in order to preserve decentralization; and at the same time fair in the sense of being responsive, even proportional, to the resources invested so that participation is suitably incentivized. The analysis of such a trade-off is an inherently social choice-theoretic question: can randomized mechanisms for blockchain be designed which achieve participation incentivization and equitability at the same time, in the long-term? Extensions of existing work in computational social choice may provide the right stepping stone to approach this question, along the following lines. Perpetual lotteries. First, while randomized mechanisms in social choice are normally studied in one-shot interaction, fairness properties in blockchain should be conceptualized in the context of indefinitely repeating interaction: a lottery may satisfy forms of fairness in one-shot interaction (e.g., assigning winning chances proportionally to invested effort, like in PoW and PoS lotteries) but may fail fairness criteria in the long run (see [30]). This suggests the study of randomized mechanisms within the recently developed perpetual voting setting of [47] (see also [32]). A key feature of the perpetual voting framework that appears particularly relevant for applications in blockchain is, in particular, the historydependence of decisions. In blockchain this dependence manifests itself via the positive feedback between current and future winning chances. Functions capturing this positive feedback could be is strategy-proof, w.r.t. stochastic dominance, if for all <unk> it never selects a lottery which is stochastically dominated by another lottery for. approached axiomatically as well as via computer simulations (see [69] for recent work in this direction) in order to provide a framework to understand long-term fairness properties of blockchain protocols viewed as perpetual randomized mechanisms. Contest-based lotteries. Second, work on randomized mechanisms in social choice normally assumes uniform probabilities or is agnostic about the specific probability mass functions defining the lotteries. Furthermore, when lotteries are iterated, as in random serial dictatorships [11], probability functions do not vary over time. As noticed above, this is not the case in blockchain: an agent's chance to win equals its share of a total non-monopolizable resource. But there is another crucial element of lotteries in blockchain. The winning probabilities for can be defined as = ( ) <unk> ( ), where is's invested effort to acquire a non-monopolizable resource (e.g., currency, computing power), and is a function (here assumed unique for all agents for simplicity) mapping each agent's effort to its acquired amount of resource (see [25]). An agent's expected utility in this type of interaction is, therefore, -. That is,'s probability of winning times its reward (e.g., block reward plus transaction fees in Nakamoto consensus) minus the effort invested. In economics jargon this type of interaction is called a contest [21,22,65]. Linking contest theory to the make-up of the lotteries in randomized mechanisms would capture a crucial aspect of perpetual randomization in blockchain, and offer a parsimonious framework in which to study long-term fairness properties. --- VOTING 3.1 Voting in Blockchain and Social Choice Voting mechanisms have been at the heart of distributed computing since its early days (e.g., [34]). Still, almost no crossover has happened between the social choice literature on the analysis of voting mechanisms and the distributed computing literature (a recent exception is [50]). This is perhaps not surprising, as the role that voting plays in the two traditions is fundamentally different. In distributed computing, voting is a consensus mechanism producing agreement in contexts where what matters is agreement itself, and not so much the option upon which agreement settles. That is, the agents involved in consensus are assumed not to be invested in any specific option, but to just aim at reaching consensus. In social choice instead, voting is eminently a mechanism for preference aggregation. Given that blockchains are open systems and that, therefore, agents' interests cannot be assumed to align, the social choice theoretic perspective on voting becomes naturally relevant. There are several applications of voting in blockchain, but fork resolution is arguably the main one. 4 I will illustrate this application of voting with respect to a specific protocol proposed for Ethereum called Casper [13], although voting mechanisms for fork resolution are used in many systems (e.g., the Spectre protocol [60]). In a nutshell, the aim of Casper is to use voting among a randomly selected committee of agents (called validators) to resolve forks in the blockchain and guarantee consensus on a canonical chain (the so-called finality property). Whenever a fork occurs, agents vote on blocks occurring on different branches in the fork (this can be thought of as voting on the nodes of the tree of possible chains). Each agent's vote is weighted by the agent's stakes, that is, the agent's deposit in currency (so, a high deposit means greater voting power), and a block is considered to belong to the canonical chain whenever a weighted supermajority of 2 3 votes for that block. This voting procedure guarantees that the winning blocks identify a legitimate chain (i.e., without forks), provided that the agents submitting individually consistent votes (that is, not voting for blocks on different chains) own at least 2 3 of the total deposits. --- Challenge: Manipulating Byzantine Voting Voting procedures are vulnerable to manipulations of different kinds: strategic voting [62], lobbying [18], bribery and control [29], vote negotiation [41]. Given the incentives layer of blockchain protocols, all such forms of vote manipulation are potentially relevant in view of the deployment of voting mechanisms. For example, returning to Casper, although the protocol can be shown to be robust against a share of Byzantine voters worth 1 3 of the total deposit in stakes, to what extent is the protocol robust against strategic forms of vote manipulation of the types mentioned above? More generally: to what extent should voting mechanisms in blockchain consensus be robust against forms of vote manipulation? Given that voting in blockchain systems is conducted by computational processes, the computational complexity approach to robustness against manipulations appears especially natural: robustness to manipulations pursued through computational intractability [8,29]. In this perspective, one might argue more specifically that voting mechanisms for blockchain should satisfy two properties: have a tractable winner determination problem; have intractable manipulation problems with respect to forms of manipulation that, given the application, may be considered relevant. Results from the above mentioned literature offer an obvious starting point, but should be extended in order to incorporate the possibilitywhich is fundamental in the distributed computing perspective-of Byzantine agents, in a way akin to what has been pursued in applications of game theory to distributed computing (see, for instance, the so-called Byzantine-Altruistic-Rational fault-tolerance models [19]). In particular, the decision problems themselves concerning the existence of manipulation strategies would need to be adjusted to this setting, by allowing for the possibility of shares of the agents population to be Byzantine. --- TRUST 4.1 Trust in Blockchain and Social Choice Among the key limitations of PoW are its high energy demand, 5high latency and low transactions throughput [9]. An influential approach trying to address these limitations has been proposed by blockchain systems like Stellar [49] and Ripple [58] 6 or TrustChain [54]. The approach of these systems to safeguard consensus against Sybil attacks is not based on lotteries but on the idea of leveraging existing'real-world' trust relations in order to select the agents that may participate in consensus. In other words, the system remains open, but new participants are admitted only if trusted by existing ones. According to the proponents of these systems, this should make it possible then to use well-established consensus approaches proper of permissioned (i.e., closed) systems, like Byzantine fault-tolerant consensus (BFT, [16,48]). Nodes participate in consensus, but only in as much as they are trusted by others. This controls participation to the consensus process by restricting access. It does so, however, in a way that is in principle open and decentralized. Abstractly, the above systems work on the basis of an underlying trust network that links nodes according to who trusts whom. At any given time, each honest node broadcasts a truthful opinion about the state of the next block (e.g., whether a given block should be included or not) to the nodes that listen to it. But the system may contain Byzantine nodes, and such nodes may reveal different opinions to different honest nodes. A consensus protocol running on such a trust network can be viewed as a discrete time dynamical system generating a stream of opinion vectors of nodes. The final vector should be such that all honest nodes hold the same opinion, that is, the system does not fork-the so-called'safety' property. At least three strands of research in computational social choice appear relevant for the above class of systems. The first one is the axiomatic and computational analysis of trust and reputation mechanisms (e.g., [35,59,63]). The second one is the analysis of influence and power in structured groups and social networks (e.g., [40,42,43]). The third one is the analysis Sybil-proofness, or falsename proofness (e.g., [20,64,67]). In what follows I will outline how especially the latter two strands of research can offer insights into the fundamental tradeoffs between decentralization, safety and sybil-proofness that the above systems need to handle. --- Challenge: Decentralized Sybil-proof Trust Understanding how to deploy BFT consensus in a permissionless setting is explicitly recognized as an open problem in distributed computing [14,66]. Yet, little academic research exists on the trustbased approach to consensus used by systems like Ripple and Stellar. Computational social choice could provide a fruitful framework from which to study this approach: Can trust systems for blockchain consensus be designed that maintain safety (i.e., absence of forks) are decentralized and, at the same time, Sybil-proof? Decentralization & safety. Trust involves a form of influence: a node's voting decision on whether to validate a block depends on the decisions of the nodes it trusts and 'listens to'. At the same time, safety demands that no two honest nodes express dissenting (finalized) votes and this in turn imposes structural properties on the trust structure: if we do not want forks to occur, then trust structures should satisfy specific properties (see [10,49]). But how decentralized a system really is when such properties hold is unclear, because some nodes may end up accruing disproportionate influence in the network. De facto centralization is a well-known phenomenon in PoW and PoS blockchains (see the above discussion about randomized mechanisms) but a comparable understanding for consensus protocols based on trust mechanisms is still lacking. In particular it is unclear how to even quantify influence on the consensus process in such systems, although some proposals have been put forth using the theory of power indices [10]. Sybil-proofness of trust mechanisms. If consensus is based on trust systems like the ones sketched above, what kind of Sybilproofness guarantees can be achieved? Intuitively, trust relations should make the system harder to access for Sybils as they would be able to exercise influence on consensus only if trusted by honest nodes. But can this argument be made exact? PoW and PoS protocols implement a form of costly identity. They make participation to consensus costly by making it dependent on the investment in a resource (e.g., computing power). In a way nodes purchase identities on a continuum and proportionally to their ownership of a non-monopolizable resource. Costly identities have been investigated also in social choice and mechanism design as one possible approach to achieve the so-called falsename-proofness of a mechanism. A mechanism is said to be falsename-proof if no agent participating in it would benefit by using more than one identifier to interact with the mechanism. A number of routes to enforce false-name-proofness have been investigated in this literature and costly identities have been studied also in the context of voting [68]. 7 In this context, while creating Sybils may be costless, creating trustworthy ones can be made costly. The issue then translates into understanding the tradeoffs that manipulators face between identifiers' costs and their payoffs measured in terms of expected influence on the consensus process. --- CONCLUSIONS This paper has highlighted how the applications to blockchain of mechanisms such as lotteries, voting or trust systems give rise to challenges that have a natural social choice dimension. These challenges all revolve around understanding fundamental tradeoffs, often in the long run, among properties that are crucial to the correct behavior of blockchain consensus protocols and are linked to the incentive dimensions of such protocols. In pursuing such problems from a computational social choice perspective, one would need to adapt standard concepts, definitions and techniques to the blockchain setting. This adaptation opens up promising lines of research and this paper has tried to illustrate some such lines, which appear promising to the author. What covered, however, should not be considered exhaustive. In particular, this paper has focused on the problem of consensus, but looking more broadly at the blockchain field one finds many more points of contact with social choice, some of which have already been identified in the literature. For example, voting theory and coalitional games have been applied to the problem of the algorithmic governance of hard forks in blockchain [3], and epistemic social choice has been applied to the so-called oracle problem [15,39], that is, how to reliably link blockchain records to realworld events. For all these reasons I believe there exists now a deep cross-fertilization potential between the two areas of research, and one that would greatly benefit both.

One of the most innovative aspects of blockchain technology consists in the introduction of an incentive layer to regulate the behavior of distributed protocols. The designer of a blockchain system faces therefore issues that are akin to those relevant for the design of economic mechanisms, and faces them in a computational setting. From this perspective the present paper argues for the importance of computational social choice in blockchain research. It identifies a few challenges at the interface of the two fields that illustrate the strong potential for cross-fertilization between them.

Introduction Existing epidemiological evidence for alcohol use disorder (AUD) supports the notion that women who have ever had children (mothers) experience a motherhood advantage; that is, they exhibit lower rates of AUD than women who have never had children (nonmothers) [1][2][3]. A lifetime motherhood role provides a number of protections including becoming more careful with alcohol use and not having a lot of extra time to drink [3]. However, to date, no studies have examined if this advantage differs by race. The purpose of this study is to use cross-sectional national survey data to compare AUD among mothers and nonmothers, while at the same time disentangling any racial disparities. Additionally, we examine associations with individual-difference variables to illuminate potential general (e.g., perceived stress, social support, and discrimination) and race-related psychological factors (e.g., racial networks) that might explain different observed patterns by race [3][4][5]. In the biomedical literature, AUD is a condition impacted by genetic [6], metabolic (e.g., [7]), family history [8], and social contextual factors related to availability of alcohol [9] and access to services [10]. In the general-population epidemiological literature, the prevalence of AUD is considerably lower among women than men, which has been attributed to less access to alcohol and the different metabolic processes of women (e.g., [9]). Perhaps because of lower rates of AUD, gender specific life role patterns have been underexplored [11]. Family history of an alcohol problem presents a significant risk despite the fact that the risk is 2 ISRN Addiction lower for women than men [8]. Cultural norms and family processes are more significant for women than for men [12]. Racial disparities in the lifetime prevalence of AUD are also well recognized in the literature. Lifetime prevalence of AUD is lower among Black women as compared to White women [11]. Although the incidence of AUD is lower among Black women when compared with White women, it is more likely to be chronic among Black women as compared to White women [13]. That is to say that it persists through several life role transitions including marriage and through changing employment. To date, the role of motherhood for AUD has largely been neglected. Focusing on the effects of motherhood specific to Black women is significant because of the evidence that Black women have experienced high levels of chronic stress which has been associated with high rates of disease [14]. This chronic stress appears to manifest as a lifelong process with the worst outcomes experienced by Black women in their middle adulthood [14]. The intersection [15,16] of race with AUD in women has not been examined in the current literature. In the current study we addressed this gap by examining the intersecting or nonadditive effects of race and motherhood. The evidence supports a motherhood advantage for AUD. Women who have never had children have higher rates of AUD than women who were ever mothers [3]. Using a prospective epidemiological study design, Chilcoat and Breslau [2] found a protective effect of transition to parenthood for the development of AUD. Perhaps this is because women who are considering motherhood are less likely to be abusing substances in the first place [17,18]. Second, women who are considering biological parenthood are more likely to access medical screening which would enable them to find adequate resources for treating an existing AUD [18]. Paradoxically, in the face of social disadvantage, medical screening may not be protective because women may be forced to hide their drinking problem [19]. Furthermore, poor women from racially disadvantaged groups may not have access to medical screening. Motherhood and additional role responsibilities increase perceived stress which is a risk factor for AUD. Under these circumstances, motherhood may be a source of disadvantage for some women and exacerbate racial disparities. Motherhood can increase role responsibilities [18]. Further, it can increase the burden of scarce economic or social support resources [20,21]. Women who have experienced gender and racial discrimination in services are less likely to utilize services later [21,22]. Among treatment seekers, mothers worried that they would lose custody of their children if they sought treatment for alcohol problems [19]. For some racial groups, social disadvantages and racial disparities are increased through the effects of discrimination [21]. Black mothers may not utilize treatment adequately or may experience considerably more disadvantages including racial discrimination. Although transition to motherhood is generally considered protective, lifetime motherhood effects for AUD have not been considered. No study we came across examined differences between Black and White women. This is particularly significant as it relates to Black women, among whom different familial arrangements have been documented [23,24]. Because of these differential family arrangements, existing health disparities, and experience of chronic stress [5,14], the connections between lifetime motherhood status and AUD may exhibit different patterns among Black and White women, a point we considered in the current study. Figure 1 presents the conceptual model for our study. The underlying psychological processes which can explain racial differences in the links between motherhood and AUD have also been neglected in the epidemiological literature. Conceivably, the effect of mother status and race may intersect and cause differences in the way risks and resources influence AUDs [15,16,22]. Although no study compared mothers and nonmothers by race, a growing body of literature supports the notion that there are health disparities and being Black does indeed affect health. Among the psychological mechanisms, perceived discrimination has been heavily implicated for the poor health status of Black Americans [5,22,25]. For both Black and White women, gender discrimination may also be an intersecting factor that increases risks for psychiatric problems in general [26] and alcohol use disorders among some minorities [27]. Although there is evidence that discrimination is related to alcohol problems among Blacks [5,22], no study that we came across used DSM-IV AUD criteria. Differences in how Black and White women perceive stress have been implicated as another important mechanism which explains different health outcomes experienced by Blacks. Some studies suggest that Blacks are likely to appraise their lives as stressful at rates higher than Whites [19,28]. This is evidenced in the increased chronic stress experienced by Black women in particular when compared with White women and with men [14]. The types of social support are also important and must be considered for AUD [29]. However, the relationship between social support and AUD is currently mixed. Although the general expectation is that social support has a positive effect on AUD, some studies do suggest that social support may not be associated with AUD [30]. Race-ethnic identity and race-ethnic networks also present themselves as important psychological mechanisms that can explain racial differences in health [28,31], but their relationships to AUD are complex [32]. They may either have a positive or negative effect on AUD. Identification with one's group in the sense that one interacts with members of the group can represent a source of instrumental support [31,32]. Furthermore, this enables racial and ethnic group members to cope and recover from multiple everyday stressors, especially discrimination, which are known to pose increased risk for heavy drinking and alcohol abuse [25]. In epidemiological literature that uses DSM-IV alcohol use disorder criteria, psychological measures of racial networks are rare and few studies include an individual's own reported networking patterns. Instead, neighborhood measures are more widely studied. Unequivocally, these studies suggest that racial segregation exposes women to risk factors such as economic disadvantage [33][34][35] and increased availability of alcohol for coping [36]. Furthermore, structural factors, such as features of the built environment that foster isolation by race (e.g., residential segregation), may reinforce this effect. In one study, Black women in race-segregated neighborhoods were increasingly targeted by alcohol advertisements [36]. Isolation from other racial groups also appears to be a key stressor for Black women in many health areas [37]. Because racial networks are important factors but psychological aspects are understudied we include this measure in the current study. Overall, intersecting motherhood and racial group differences are not sufficiently explored in the etiology of AUD. Although absence of a motherhood effect may be one reason why a chronic pattern of AUD is observed among Black women, these differences have not been examined. The psychological resources and risks which may explain any racial differences have also received insufficient attention. Using national survey data, the current study disentangles racial group differences focusing on motherhood effects. We expected a motherhood advantage to be seen for White women but perhaps not among Black women. We also expected that the groups will differ on psychological risks such as discrimination and resources such as racial networks and that Black mothers would experience the worst disadvantages. --- Methods --- Sample. We utilized a subsample from the second wave of the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) conducted in the years 2004-2005. Data for the original survey was collected using face-to-face and computer assisted personal interviews in respondents' homes, yielding a cumulative survey response rate of 81% [38]. For the current study, adult female lifetime drinkers who were self-identified as either "White" (N = 11,017) or "Black" (N = 4,133) were included. This dataset was selected because it was rich in AUD and psychological measures. Psychological measures were only available in wave 2 data and hence only this wave was used in the current analysis. In order to maintain temporal consistency in measures, only past-year alcohol use disorder and psychological conditions at the time of interview were used. --- Measures --- Alcohol Use Disorder (AUD). NESARC used the alcohol use disorder and associated disabilities interview schedule-IV (AUDADIS-IV: 38) to measure alcohol related factors. For the current study, we used NESARC preconstructed measure (available in wave 2 data) of alcohol abuse and/or dependence anytime during the past year. Alcohol abuse and dependence criteria corresponded to the Diagnostic and Statistical Manual of Mental Disorders, 4th ed., alcohol abuse and dependence disorder symptoms [39]. These included recurrent drinking resulting in failure to fulfill major role obligations; recurrent drinking in hazardous situations; recurrent drinking related legal problems; continued drinking despite recurrent social or interpersonal problems caused or exacerbated by drinking; persistent desire to drink; and continued drinking despite problems [39]. --- 2.2.2. Demographics. These included a continuous measure of current age and dichotomous measures of current marital status (single versus married/in a relationship), family income (less than USD 20,000/greater than or equal to USD 20,000), nativity (i.e., born in the United States or not), and employment status (currently employed for over 35+ hours a week or not). A single item, preconstructed measure available in NESARC wave 2 of the total number of children women "ever had, " was also included. It accounted for any children or adopted/foster/step children. Although this measure gave us lifetime motherhood status, this measure could not distinguish between biological and nonbiological parenthood. Furthermore, this measure could not account for age of children. These remain limitations of the current study. --- Major Predictors (i) Family History of Alcoholism. Four items from wave 1 data were included to reflect an alcohol problem in a full parent or full sibling. (ii) Perceived Discrimination. Six items measured perceived discrimination (e.g., during past year how often did you experience discrimination in ability to obtain health care or health insurance coverage because of your race/ethnicity? During past year how often did you experience discrimination in how you were treated when you got care because of your race/ethnicity? During the past year how often did you experience discrimination in public on the street, in stores or restaurants, because of your race/ethnicity?). Three items assessed perceived discrimination because of race/ethnicity and three similarly worded items assessed perceived discrimination because of gender. The NESARC scales were originally modified fromKrieger [40]. Each discrimination question was measured on a scale from 0 to 4 (0) never, (1) almost never, (2) sometimes, (3) fairly often, or (4) very often. Consistent with some other approaches (e.g., [32]), the scores were dichotomized to indicate the presence of a discrimination experience in the past year. (iii) Perceived Stress. We used the Perceived Stress Scale which contains four items (PSS4: 41). The PSS4 measures cognitively mediated emotional responses to events. Items on this scale (e.g., able to control important things in their lives) were measured as follows: (0) never, (1) almost never, (2) sometimes, (3) fairly often, and (4) very often. Herd and Grube [31] reported high reliability for the AUDADIS version. Two of the four items were reverse coded such that high scores indicated greater perceived stress. A mean perceived stress score was computed for each respondent in cases in which an individual responded to at least two of the four items (mean = 1.99, SD = 0.74, and range = 1-5). (iv) Social Support Networks. The Social Network Index (SNI) developed byCohen and Hoberman [41] was used to measure social networks. This measure assessed participation in 12 types of social relationships/resource including spouse, neighborhood, and community. Participants indicated whether they had any contact with any of these relationships or resources in the past two weeks. If any of these networks was active, a positive score was assigned. A summary score was accordingly computed to reflect the size of an individual's network (mean = 8.78, SD = 0.93, and range = 0-10). (v) Perceived Social Support. The 12-item Interpersonal Support Evaluation List (ISEL 12 : 43) was used to measure availability of helpful social resources. Respondents reported items on a scale ranging from 1 to 4: (1) definitely false, (2) probably false, (3) probably true, and (4) definitely true. A high score was indicative of high social support (mean = 3.51, SD = 0.47, and range = 1-4). (vi) Mixed Racial-Ethnic Networks. Four items assessing composition of networks were used. Each of the items addressed a particular interpersonal domain: (1) close friends; (2) party visits; and (3) visitors. Participants responded on a five-point scale ranging from 1 to 5: (1) all from my race/ethnic group; (2) more from my race/ethnic group; (3) about half and half ; (4) more from other race/ethnic groups; and (5) all from other race/ethnic groups. Ruan et al. [42] noted high reliability of these four items. A sum of these items was obtained to compute a racial network score. Higher scores reflected mixed or nonsegregated racial networks and lower scores implied that the networks were racially segregated (mean = 6.85, SD = 2.61, and range = 1-15). (vii) Racial-Ethnic Identification. This was measured by the Multigroup Ethnic Identity Measure (MEIM: 45). Participants responded to each of eight items (e.g., you have a strong sense of yourself as a member of your race/ethnic group) on a six-point rating scale ranging from (1) strongly agree to (6) strongly disagree. A mean score was obtained by summing all the values across the 8 items if at least half the items (i.e., four items) were answered by the respondent. Otherwise the measure was treated as missing (mean = 2.33, SD = 0.85, and range = 1-6). --- Analytical Strategy. Wave 2 of NESARC used a multistage, stratified, complex sampling design in which primary sampling units (PSUs) were stratified according to certain sociodemographic criteria. Our variance estimation procedures accounted for this complex sampling design. We used wave 2 sampling weights as our analysis was restricted to wave 2 data. We used design-based analytic techniques to obtain estimates of population parameters. SAS 9.2 was used for data preparation; SAS-Callable SUDAAN version 10 [43] was used for all estimation processes because of SUDAAN's use of the Taylor series linearization estimation of standard errors. To examine effects of lifetime motherhood, we computed the prevalence of alcohol use disorders among women who had ever been mothers and women who had never been mothers and then compared "Black" and "White" women. In order to examine the role of psychological measures, associations between the predictors and AUD were estimated using logistic regressions. All continuous predictors were standardized to the entire NESARC population mean. --- Results The demographic characteristics of the sample are presented in Table 1. On average, White women were slightly older than the Black women. A larger proportion of Black than White women had migrated to the US. Almost half of the sample reported a personal income below 20,000 USD. Similar proportions of Black and White women reported ever having children suggesting that attaining motherhood was not skewed by race. The prevalence of alcohol use disorder (see Figure 2) was lower among mothers compared to nonmothers. However, when the prevalence of disorders was examined by race, different trends were observed. A motherhood advantage was observed for White women but not Black women. Table 2 lists psychological characteristics. Family history of alcoholism was greater among both Black and White mothers as compared to nonmothers. Curiously, mothers reported lower discrimination than did nonmothers. Black women reported slightly elevated levels of perceived stress but smaller social networks and less social support. Black women also reported networks that are more racially mixed than White women and low levels of racial identification. Finally, we performed logistic regressions to examine the associations of psychological risks and resources with the presence of an AUD. Table 3a presents the bivariate associations of the predictors and AUD. As expected, family history of alcoholism predicted an AUD. However, this association was significant for White but not Black women. Several psychological risks and resources also predicted AUD including perceived stress and racial networks. When all psychological risks and resources were included simultaneously and adjusted for demographic characteristics some differences in associations occurred between Black and White women by parenthood status (see Table 3b). Significantly, the associations between racial networks and AUD were seen only for Black women with children. Intriguingly, a similar association was also noted for White women without children. Perceived discrimination was associated with AUD among both White and Black women with children. No such associations were observed for women without children. --- Discussion In this study we examined motherhood effects on AUD, potential psychological mechanisms, and racial disparities. Using data from a large US national survey, we examined whether lifetime motherhood presented an advantage for current AUD for both Black and White women alike. Subsequently, we examined the role of differential psychological risks and resources for AUD at each racemotherhood intersection. Overall rates of AUD suggest a general motherhood advantage; that is, mothers reported lower AUD than nonmothers. However, when the racial groups were examined separately, motherhood advantage was only observed for White women. This is consistent with the notion that life-role-related AUD may be different for Black women [12-14, 19, 23-25]. This also calls for developing a better understanding of family and life role processes that may be affecting AUD specifically for Black women and begs for further exploration among other racial and ethnic minorities. Black women with no children (i.e., nonmothers) had lower rates of AUD compared to White nonmothers. Extending this finding, we speculate that Black mothers with AUD are developing AUD at a later age, yet the underlying processes and reasons for this process are not entirely clear. Underlying race specific patterns in both motherhood processes and AUD onset and progression need to be understood and leveraged to allow for the development of culturally sensitive and effective treatments for AUD. Furthermore, if motherhood advantage for White women is due to the availability of screening and treatment when White women become mothers, then Black women may not be receiving adequate screening at the time that they become mothers. Perceived discrimination further increased the odds of AUD among mothers. We speculate that these multiple factors cumulatively increase disadvantage especially among Black mothers. Potential areas in which women and especially mothers experience discrimination need to be identified to allow for the development of better treatment for women with AUD [10,19,21]. An additional race specific pattern that we observed related to the associations of family history of alcohol problems with current AUD. Family history was associated with past-year AUD only for White women among both mothers and nonmothers alike. Psychological predictors were more significant in explaining AUD among Black women. Notable among these were perceived discrimination, perceived stress, and mixed race-ethnic networks. Mixed race-ethnic networks were protective for AUD among Black mothers. Perhaps women with mixed race-ethnic networks also have access to many social and economic benefits available to advantaged groups. Conversely, Black mothers whose networks were racially isolated were disadvantaged [35,36]. This is consistent with the literature that social segregation may lead to a higher level of drinking as a form of coping with social stressors [36]. Quite surprisingly, we found a similar effect of mixed racial networks for White nonmothers. This is consistent with the literature suggesting that women's drinking is sensitive to peer group cultural norms [9]. Perhaps White women who were associated with other White women (where norms for drinking were supportive) had higher levels of AUD. This is also consistent with the sociological literature on friendship segregation in schools [32], where segregation has been found to be beneficial to racial groups whose drinking norms disfavored drinking. Higher perceived stress increased the odds of an AUD across both racial groups and among mothers and nonmothers alike. Therapies like mindfulness training could be combined with AUD treatment to reduce perceived stress. An important negative finding pertains to social support which was not independently associated with AUD. This is, however, consistent with some findings that social support has little effect on problem drinking when another social disadvantage is present [30]. Overall, our findings highlight the significance of disentangling life roles at the intersection of race among women. Even though AUD is a medical condition, social etiology and (1) All continuous predictors are standardized to the mean. --- ISRN Addiction intersecting risks from race and motherhood are significant and important. Future studies should disentangle and explore these processes in greater detail. Limitations. Our findings need to be interpreted in the context of several limitations. First of all, because wave 1 of the national survey did not contain psychological measures, our data were cross-sectional, and causality cannot be inferred. Rather, our study points to differential life role processes among Black and White women [23,24] and extends this to motherhood. Additional community-based studies using prospective study designs are needed to fully comprehend the underlying reasons for why motherhood advantage accrues only to White women. Second, our study was based on secondary data analysis of publicly available limited use data and community level measures could not be included. Third, most results were based on self-reported data, which may be subject to recall bias. No physiological measures pertaining to perceived stress and AUD were available. Fourth, our measures primarily focused on differential aspects of social context; thus race specific metabolic and genetic factors were not considered. Fifth, the biological relationship of mothers with their children is not known and could not be considered. Despite these limitations, the study's findings contribute to the growing literature on health disparities by extending a focus to AUD. In particular, AUD among women has received less attention relative to other health conditions. We highlighted heterogeneity in psychological processes underlying AUD for Black and White women. Future studies and prevention programs must consider these racial differences critically. --- Conclusion Using cross-sectional national survey data, we examined whether the notion of motherhood advantage for alcohol use disorders (AUD) was applicable to both Black and White women. This advantage did not accrue to Black women. We found that intersecting psychological risks and resources predicted a current AUD among Black and White mothers and nonmothers. Perceived stress was associated with AUD for all women. Discrimination and racial-ethnic networks were significant for Black mothers. Psychological factors had stronger effects on AUD than family history of AUD for Black women but not White women. Future studies should critically consider mechanisms by which race, motherhood, and psychological factors interactively affect AUD in women. --- Conflict of Interests The authors declare that there is no conflict of interests regarding the publication of this paper.

Rates of alcohol use disorders (AUD) are generally low among women who have ever had children (mothers) compared to women who have never had children (nonmothers), presenting a motherhood advantage. It is unclear if this advantage accrues to "Black" and "White" women alike. Using National Epidemiological Survey on Alcohol and Related Conditions (NESARC) wave 2 cross-sectional data that is rich in alcohol use and psychological measures, we examined the following: (a) if motherhood is protective for past-year AUD among Black (𝑁 = 4, 133) and White women (𝑁 = 11, 017); (b) potential explanatory psychological mechanisms; and (c) the role of race. Prevalence of a past-year DSM-IV AUD was lower among White mothers compared to White nonmothers, but this same advantage was not observed for Black women. Perceived stress was a risk for all women, but race-ethnic segregated social networks and perceived discrimination predicted current AUD for Black mothers. Unlike White mothers, current psychological factors but not family history of alcohol problems predicted AUD for Black mothers. Future prospective studies should address the mechanisms by which race, motherhood, and psychological factors interactively affect AUD in women.

Providing appropriate and culturally sensitive care to the rapidly growing number of Latinx older adults in the United States with neuropsychiatric conditions has been identified as a major public health challenge (Kim, 2018;Substance Abuse and Mental Health Services Administration, 2011) and a focal concern among mental health services providers (Fortuna, n.d.;Guzman et al., 2015). Latinx older adults use mental health services less than their same-age white counterparts (Cook et al., 2017;López et al., 2012), yet they are more likely to screen positively for depressive symptoms (Hooker et al., 2019). This is particularly alarming because the Latinx population has a longer average life expectancy than the general U.S. population (National Center for Health Statistics, 2019a), yet a higher risk of excess morbidity and mortality from preventable health conditions (Howard et al., 2014). The confluence of these factors suggests that Latinx adults live more years in poor health over their lifetimes, which places them at an increased risk for developing mental health issues in later life. For example, a scoping review of the literature and national statistics on Latinx health in the United States (Velasco-Mondragon et al., 2016) shows that Latinx populations, including the Mexicanorigin population, are particularly vulnerable to chronic health conditions that are associated with cognitive decline (Taylor et al., 2020) and depression (Downer et al., 2016), such as type 2 diabetes (González et al., 2020), hypertension (National Center for Health Statistics, 2019b), and cardiovascular disease (Samieri et al., 2018). Moreover, older Latinx adults are at least one and one-half times more likely to develop Alzheimer's disease or other dementias compared with non-Latinx white adults (Alzheimer's Association, 2020) and develop symptoms at younger ages (Clark et al., 2005). Other factors put some Latinx adults at increased risks for cognitive decline and poor mental health in older age. Socioeconomic status (SES; Phelan & Link, 2005;Phelan et al., 2010) and health insurance coverage (Healthy People 2020, n.d.) are two well-documented social determinants of health where the Latinx population falls short compared with the general U.S. population. Among the older than 65 years population, Latinx adults compared with non-Latinx white adults are more likely to be financially insecure and live below the federal poverty line (National Hispanic Council on Aging, 2016), live in linguistically isolated households (Burciaga Valdez & Arce, 2000) and neighborhoods (Ward et al., 2018), and to have faced discrimination throughout their lives (Krogstad & López, 2016). These factors have been shown to be associated with poor physical and mental health status (Braveman et al., 2011;Braveman & Gottlieb, 2014), including higher depressive symptoms (Bilotta et al., 2010;Ward et al., 2018) and trajectories of cognitive decline among older Latinx adults (Downer et al., 2016;González et al., 2020;Howrey et al., 2020). This body of research is consistent with earlier work showing that older Mexican American adults have a significantly higher burden of depressive symptoms, compared with non-Latinx white adults (Sorkin et al., 2009). Culture is another important factor associated with mental health and illness. According to Goodenough (1999), a culture consists of "principles or standards for deciding what is, what can be, how one feels about it, what to do about it, and how to go about doing it" (p. 85). In 2001, the Surgeon General issued a report (U.S. Department of Health and Human Services, 2001) that examined mental health, culture, race, and ethnicity. The report highlighted that culture affects the ways in which individuals view the causes, manifestations, and the course of mental disorders, including symptomatology and explanations for distress, help-seeking and coping, treatment adherence, and response (Gone & Kirmayer, 2010;U.S. Department of Health and Human Services, 2001). Thus, understanding the meaning that patients assign to their symptoms and circumstances is an important aspect of their well-being (Bignall et al., 2015) and could be one step toward improving prevention, the accuracy of diagnosis, and consequently, the relevance and completion of psychotherapy and other forms of treatment. For example, the perceptions of the cause of illness can lead to different help-seeking and self-care behaviors (Hedemalm et al., 2008;Lichtman et al., 2015;Simons et al., 2017). Research suggests that the effectiveness of psychotherapy and/or prescription medication for the treatment of disorders like depression may pivot on factors such as cultural acceptance of the treatment (Izquierdo et al., 2014). Cultural factors have been shown to play a role in the mental health of Latinx adults also and their use of mental health services (Barrera & Longoria, 2018). For example, religion in Latin American countries is closely aligned with Catholicism and has been associated with worldviews such as "fatalism" (fatalismo), which is characterized by the belief that misfortune is due to God's will or destiny, and that a person has no control over life events (Abra<unk>do-Lanza et al., 2007;Caplan et al., 2011;Flórez et al., 2009). The influences of fatalism are complex and contested (as reviewed in the study of Abra<unk>do-Lanza et al., 2007). For example, findings from a recent meta-analytic review show that fatalism is associated with health behaviors, even though the relationship is smaller than previously suggested (Cohn & Villar, 2015). Cohn and Villar (2015) found in their analysis of 46 studies using fatalism scales that adults with more fatalistic beliefs were less likely to seek preventive cancer screening. Another study showed that older Latinx adults may hold onto beliefs in a higher power that guide how they perceive the causes of their mental health problems, including depression (Anastasia & Bridges, 2015). These studies are consistent with an earlier study showing that Latinx older adults are more likely to adhere to traditional worldviews of mental health and their treatment than non-Latinx older adults (Lau & Gallagher-Thompson, 2002). A better understanding of symptom attribution can provide insights and guidance in the provision of mental health care for Latinx patients, yet few studies in this area have focused on Latinx populations. Izquierdo et al. (2014) analyzed data from qualitative telephone interviews with Latino adults (50-88 years) who reported a recent encounter with primary care seeking help for a personal, emotional, or mental health problem. The researchers found that the perceptions of the sources of depression and of the illness itself influenced views on treatment and willingness to take medications. Jimenez et al. (2012) analyzed data collected as part of a randomized trial of adults 65 years and older with depression, anxiety, or at-risk alcohol use. The researchers investigated beliefs of the causes of mental illness by race/ethnicity and found that Latinx older adults attributed mental illness to loss (in terms of death and social activities) and medical reasons more frequently than non-Latinx white, African American, or Asian American older adults in the sample. However, the analysis was limited to one question on attribution, preventing a more comprehensive examination of Latinx older adults' views of the causes of mental illness. Bignall et al. (2015) conducted focus groups with a multicultural sample to investigate the perceived attributions of 19 mental health disorders and found that racial/ethnic minority adults, including Latinx adults, tended to view mental health disorders as a normal part of life or caused by spiritual reasons. However, these results were based on a small sample of Latinx adults (n = 9). Moreover, there is a paucity of mental health research across and within the Latinx population, despite its documented heterogeneity in physical and mental health outcomes (Arias et al., 2020;Borrell & Crawford, 2009;Hooker et al., 2019). To move the literature forward on Latinx mental health and provide formative research for future studies, we analyzed administrative data from a psychiatric outpatient clinic to identify how older Latinx and non-Latinx older adults attribute the sources of their mental health problems. The primary aims of the study were to determine how these types of attributions differed by ethnicity and to identify the association of acculturation and depression symptom severity on Mexican-origin older adults' mental health attribution. Based on extant literature, we expected to observe substantive between-group differences and quantitative within-group differences in how Mexican-origin and non-Latinx older adults characterize the cause of their mental health symptoms. --- Method --- Data Collection The data for this secondary analysis came from a retrospective chart review of administrative data collected between 1995 and 2005 at a rural psychiatric outpatient clinic in Imperial County, CA. Imperial County is located north of the U.S. Mexico border with a predominantly Latinx population (82.7%), most of whom are of Mexican origin (DataUSA, 2018). The administrative data were collected at the first clinical visit for patients who voluntarily completed a 15-min self-administered questionnaire in their primary language (English or Spanish). The questionnaire included questions on depressive symptoms, views regarding the perceived sources of their mental health ailments, cultural orientation, and sociodemographics. All patients were informed at the time of intake that the information provided may be used for research purposes. Caregivers were used as proxies to help patients with a diagnosis of dementia complete the questionnaire. All records were deidentified by the clinic prior to analysis and the study was approved by the Oregon State University institutional review board. --- Participants The questionnaire was completed by 673 adults aged 55-95 (430 Mexican origin and 243 non-Latinx). Of the 673 respondents in this study, 64% were of Mexican origin and 62% were women. The majority (94%) of non-Latinx adults identified as white. The average age for the entire sample was 69.48 years (SD = 9.63). The average years of schooling was 6.01 years (SD = 4.89) for Mexican-origin adults and 12.48 years (SD = 3.99) for non-Latinx adults in the sample. Although participants' SES was not specifically assessed during the data collection, Imperial County estimated a 13.6% poverty rate for adults 65 years and older, 19.4% for all families, and a median family income of $35,226 in 1999(U.S. Census Bureau, 2000). Specific sample demographics and descriptive information for variables of interest are given in Tables 1 and2. --- Measures Attribution of mental health ailments Vahia et al. (2013) developed a 13-item questionnaire to identify four primary domains (genetic/biological factors, social/interpersonal factors, economic factors, and cultural factors) to which Mexican Americans attribute symptoms of physical and mental illness. The items were developed based on a literature review and finalized through a consensus process with clinical experts (Vahia et al., 2013). A three-point scale (0 = "no," 1 = "maybe," and 2 = "yes") was used to be accessible to the clinic's low-literacy population (Vahia et al., 2013), which were dichotomized for the analysis to ease interpretability, so that 0 = "no" and 1 = "maybe" and "yes" responses. Patients were asked, "What do you consider the cause of your ailments?" for each of 13 possible causes, grouped into substantive clusters by the current authors for analytic interpretability (Table 3). --- Acculturation Acculturation was measured using the Brief Acculturation Scale for Hispanics (BASH; Norris et al., 1996), which uses the language use subscale of the Short Acculturation Scale for Hispanics (SASH; Marin et al., 1987), and thus uses English language use as a proxy measure. BASH comprises four items related to the use of English versus Spanish languages for reading, speaking, and thinking: (a) In what language do you read and speak?; (b) What language did you use as a child?; (c) What language do you use at home?; and (d) In what language do you think? Each item had five response categories: 1 (only Spanish), 2 (Spanish better than English), 3 (both equally), 4 (English better than Spanish), and 5 (only English). The mean score from these four items was used as the total acculturation score in our analysis. Previous work using this scale found significant correlations between BASH score and generational status (r = 0.67) and length of time in the United States (r = 0.56; Norris et al., 1996). --- Depressive symptom severity Depressive symptom severity was assessed using the 20-item Center for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977). The CES-D is a commonly used measure of depressive symptom burden in community samples (Hahn et al., 2011), with higher scores reflecting greater depressive symptomatology. The CES-D appears psychometrically valid among Latinx adults, with one report of Cronbach's <unk> = 0.89 (Chung et al., 2003). --- Demographic characteristics Demographic information collected from each patient included age (continuous years), sex (female/male), ethnicity (Mexican-origin Latinx and non-Latinx), race (categorical), country of origin (categorical), years of formal schooling (continuous), and years living in the United States (continuous). Demographic characteristics of the sample are given in Tables 1 and2. --- Statistical Analyses Latent class analysis (LCA) is a cross-sectional, personcentered statistical technique that illustrates heterogeneity within a given population based on identifying patterns of responses (Masyn, 2013). This multivariate approach is useful in this context because it allows the current researchers to explore the mental health attribution beliefs of two different populations in a way that honors the complexity and multidimensionality of these groups. This is particularly important for the Mexican-origin group because Latinx communities have been aggregated to a whole historically and have been treated as cultural monoliths in both research and applied settings (Comas-D<unk>az, 2001). Additionally, LCA concerns itself with groupings based on unknown (latent) variables such as the mental health attribution beliefs investigated in this study, rather than observable variables, such as gender or age (Masyn, 2013;Nylund-Gibson & Hart, 2014). A separate series of LCAs was conducted for the two groups of interest in this study, Mexican-origin Latinx and non-Latinx older adults, using Mplus 8.1.5 (Muthén & Muthén, 1998-2018). The 13 indicators from the attribution of mental health ailments questionnaire (Table 3) were used as indicators of the latent class using class enumeration procedures recommended by Masyn (2013). An optimal LCA solution was chosen based on an overall evaluation across multiple fit statistics (Masyn, 2013;Nylund et al., 2007;Tein et al., 2013). The following approximate fit criteria were used, where lower values indicated superior fit: log likelihood, Bayesian Information Criterion (BIC), Adjusted Bayesian Information Criterion (ABIC), Consistent Akaike Information Criterion (CAIC), and Approximated Weight of Evidence Criterion (AWE). Additionally, the p values of the Lo-Mendell-Rubin adjusted likelihood ratio test (LMRT) and the bootstrapped likelihood ratio test (BLRT) were assessed to determine whether adding a class would significantly improve model fit. The Bayes Factor (BF) provided a relative comparison of fit between each successive model, and the correct model probability (cmP) provided an estimate of the "correct" model relative to all models tested; higher values for both the BF and cmP indicate support for a particular model. Overall classification quality was evaluated according to each model's entropy, an omnibus index where values greater than 0.80 indicate "good" classification (Clark & Muthén, 2009), and each model's average posterior probabilities (AvePP; Masyn, 2013), where values more than 0.70 indicate adequate class separation (Nagin, 2005). Lastly, overall substantive meaningfulness and interpretability of each model were evaluated using a visual inspection of conditional item probabilities and theoretical guidance (Muthén, 2003). After the final LCA solution was identified for each group (Mexican-origin and non-Latinx), an additional LCA was conducted with covariates added using the Bolck, Croon, and Hagenaars three-step method (Asparouhov & Muthén, 2014;Bolck et al., 2004;Vermunt, 2010). The rationale for this approach was to prevent the covariates from unintentionally biasing the class enumeration process (Nylund-Gibson & Masyn, 2016). Because the focus of this analysis was to better understand the potential impacts of acculturation and symptom severity on Mexicanorigin older adults' mental health attribution, we focused our covariate study on this population and did not run covariates with the non-Latinx sample. --- Results The fit statistics for the LCA series for each group are given in Table 4. The three-class solution was chosen for both the Mexican-origin and non-Latinx groups, as it provided the best LMRT p values and the second-best BIC, CAIC, AWE, and BLRT fit statistics, and defined a substantive meaningful third class for both groups. Additionally, although the BF and cmP values supported the two-class solution, this solution obscured the presence of this meaningful third class. The AvePP indicated that the models created distinct, well-separated classes for each group (Table 5), despite the slightly low entropy for each group's three-class solution (Table 4). The conditional item probability plots for the Mexicanorigin and non-Latinx LCA models are shown in Figures 1 and2, respectively. The Mexican-origin LCA (Figure 1) revealed a relatively small but distinct class of individuals (n = 52, 12%) who were more likely to attribute their depressive symptoms to spiritual and/or supernatural causes, labeled Spiritual. This class was unique to the Mexicanorigin model, as the non-Latinx LCA did not identify a similar class. The Spiritual class possessed a similar pattern of responses to the largest Mexican-origin class (n = 200, 46%), which was labeled Internal/Brain. Individuals in both classes were more likely to attribute their symptoms to internal ("the way you are") and brain-related ("function of the brain or mind") causes, but the Spiritual class was distinguished by its high endorsement of spiritual/ supernatural causes. In fact, individuals in the other two classes were very unlikely (.10-.20 probability) to attribute their symptoms to any spiritual/supernatural causes. Moreover, the Spiritual class was further delineated by its members' relative likelihood of attributing symptoms to social problems (e.g., "problems with family members... how you get along with people";.60-.65 probability). The third and second-largest Mexican-origin class (n = 178, 42%) was labeled the Low Attribution class, as these individuals were not likely to attribute their symptoms to any of the 13 listed causes. The lowest endorsed items across all three classes of the Mexican-origin group appeared to be the jobrelated and alcohol/drugs attributions. The non-Latinx LCA (Figure 2) revealed three classes with proportions similar to the Mexican-origin LCA classes, including a small class (n = 38, 15%). However, this non-Latinx small class, labeled Social/Internal/Financial, possessed item response characteristics very different to the small Mexican-origin Spiritual class; the individuals in the non-Latinx Social/Internal/Financial class were very likely (.70-.90 probability) to attribute their mental health symptoms to social difficulties (problems with significant others, family members, and getting along with people), financial problems, internal causes ("the way you are"), and brain-related issues ("function of the brain or mind"). This Social/Internal/Financial class was distinguished from the largest non-Latinx class, labeled Internal/Biological (n = 123, 51%), by its high attribution of social and financial difficulties. In contrast, the individuals in the Internal/ Biological class appeared to solely attribute their mental health symptoms to internal causes and brain-related issues. The second-largest class and approximately one third of the non-Latinx group (n = 82, 34%) was labeled the Low Attribution class, due to the individuals' low likelihood of attributing their mental health symptoms to any of the given 13 causes. This non-Latinx Low Attribution class possessed a similar probability plot to the Mexican-origin Low Attribution class. Overall, across all three non-Latinx classes, the lowest endorsed mental health attributions were the spiritual/supernatural causes, lack of nutrients/ vitamins, and alcohol and drugs. Years of formal education was found to significantly (p <unk>.01; Table 6) affect whether a Mexican-origin individual was assigned to the Internal/Brain class versus the Low Attribution class, meaning that Mexican-origin individuals with more years of school were more likely to attribute their mental health symptoms to Internal/ Brain causes rather than not. Years of school did not appear to significantly affect Spiritual class membership. No other statistically significant covariate effects were found (Table 6) for the Mexican-origin LCA model, suggesting that Mexican-origin individuals' class membership did not significantly differ by years lived in the United States, gender, depression symptom severity (CES-D), or acculturation (BASH). However, according to the covariate odds ratios, Mexican-origin women were twice as likely (odds ratio 1.93-2.37) than Mexican-origin men to be assigned to the Spiritual class rather than the Internal/Brain or Low Attribution classes, indicating a relatively important yet statistically nonsignificant effect of gender on Spiritual class membership. --- Discussion The current study sought to better illuminate betweengroup and within-group differences in how older Mexicanorigin Latinx and non-Latinx individuals attribute their mental health symptoms using comparable LCA models as hypothesized, which yielded notable findings across the three-class models for each group, demonstrating substantive differences between the two ethnic groups in how they attributed their mental health, as well as withingroup heterogeneity. First, the analysis provided additional empirical evidence for a group of Mexican-origin adults' attribution of mental health issues to spiritual and supernatural causes, while revealing a gender difference in this attribution. Second, the within-group differences found in the Mexican-origin Latinx LCA and lack of significant acculturation-related covariate effects challenged extant literature's conceptualization of the Latinx population as a homogeneous group and overemphasis of English language acculturation on Latinx worldviews, while highlighting the potential impact of education on mental health attribution. Third, the analysis revealed heterogeneity of mental health attribution within the majority white (94%) non-Latinx group, adding a more dynamic understanding of mental health and symptom beliefs among this group. For Mexican-origin older adults, a small class appeared to attribute their mental health issues to spiritual and/ or supernatural factors, and Mexican-origin women were twice as likely to be grouped into this class compared with Mexican-origin men. This class was unique to the Mexicanorigin LCA group and distinct from other Mexican-origin mental health attribution classes. Thus, the current study provides additional empirical support for the acknowledged importance of spirituality (espiritismo; Applewhite & Gonzales, 2012) and Latinx worldviews to older Latinx Mexican-origin individuals (Baez & Hernandez, 2010), and the need to consider these cultural values and their intersection with gender when working with older Mexican-origin clients. Passalacqua and Cervantes (2008) provide an exploration of the intersections between spirituality, culture, and gender and the utility of this integration into therapeutic work, recommending that clinicians recognize how healthy development may differ from a western male model of development, consider the intersectionality of clients' identities, and the importance of rituals for some religious clients. To this end, Moreira-Almeida et al. ( 2014) recommend taking a spiritual history with clients to improve patient compliance, satisfaction, and list validated and clinically useful instruments for this purpose. The clinician should also take care not to overemphasize the potential role of espiritismo on clients' self-perception of their mental health; the same Mexican-origin adults who attributed their ailments to spiritual/supernatural causes also attributed their mental health issues to interpersonal, internal, hereditary, and biological factors. Thus, exploring the potential role of spiritual beliefs on client self-perception of functioning is only one aspect of culturally informed care. Mexican-origin Spiritual class membership was unaffected by acculturation level (as measured by English language acquisition), depression symptom severity, and time spent in the United States. This provides support for challenging researchers' and clinicians' historical overemphasis on acculturation and westernized symptomology when seeking to understand and conceptualize Mexicanorigin Latinx clients' mental health experiences, which can lead to "othering" (Viruell-Fuentes, 2007, 2011), over-pathologizing, misdiagnosing, and inappropriate treatment (Cassano et al., 2012;Mischoulon et al., 2005;Moreira-Almeida & Carde<unk>a, 2011). Given the diversity within the U.S. Latinx population with respect to cultural practices and beliefs, and complex historical immigration and sociocultural relationships with the U.S. and dominant white culture (Miville, 2010), clinicians may wish to take a multicultural approach (Santiago-Rivera et al., 2002), using a tool such as the Cultural Formulation Interview (DeSilva et al., 2015) to learn more about a particular client's acculturation level, cultural values, immigration history, and language, sociopolitical, and ethnic identity issues. Doing this allows the clinician and client to better understand whether and how these issues may affect a client's self-perception or attitudes toward treatment approaches. The finding that Mexican-origin individuals were more likely to attribute their mental health symptoms to internal and brain-related causes rather than no other causes (Low Attribution) when they received more formal education suggests that schooling may have potentially impacted how these clients conceptualized their mental health attributions. Thus, clients' experiences with education, particularly westernized education, may provide a better understanding of their exposure to westernized internal and biological views of psychopathology rather than their English language acquisition or other markers of acculturation. In contrast to the Mexican-origin Latinx group, a large portion of the majority non-Latinx white older adults (51%) were most likely to attribute their mental health issues to solely internal and biological causes, and non-Latinx adults were unlikely to attribute their mental health symptoms to spiritual or supernatural causes across the three classes. These findings reinforce previous research that non-Latinx white adults are less likely to believe their ailments are caused by spiritual or supernatural forces compared with racial/ethnic minority (Bignall et al., 2015) and Latinx adults (Anastasia & Bridges, 2015). One contribution of our study is the finding that the majority white non-Latinx older adults too was a multidimensional group with diversity in how they conceptualized their own mental health difficulties. Our analysis identified a small portion of the group (15%) that was likely to attribute their symptoms to social and financial problems, in addition to the internal/biological attribution beliefs held by the largest class. These findings suggest that a certain portion of non-Latinx adults may perceive the origin of their difficulties as stemming from multiple domains in their life (internal, financial, and interpersonal). Clinicians may benefit from assessing the resources and areas of need in clients' various systems, such as using Bronfenbrenner's ecological theory framework when conceptualizing cases (Eriksson et al., 2018). In addition to these within-and between-group differences, both the Mexican-origin Latinx and non-Latinx groups of older adults shared a likelihood of attributing their mental health difficulties to internal, genetic, and biological issues, especially attributing their symptoms to "problems with the way you are," "hereditary factors," and "function of the brain or mind." Given the significant effect of education on Mexican-origin individuals' mental health attribution beliefs, it is likely that both groups of individuals are more likely to endorse beliefs in internal and biological causes when exposed to more formal education. It is also possible that there were cultural differences in how each group of older adults interpreted these internal attribution items, in that Mexican-origin individuals possibly used a fatalismo lens (Rosales & Calvo, 2017), whereas the non-Latinx individuals ascribed to western behaviorist/environmental beliefs (Bignall et al., 2015). An alternative explanation would be that mental health literacy or stigma influences the ways in which Latinx and non-Latinx individuals view their mental health symptoms and seek treatment (Benuto et al., 2019;Carpenter-Song et al., 2010;Lopez et al., 2018). Additionally, research has shown that primary clinicians approach mental health treatment and referral differently for racial/ethnic minority adults (Jones et al., 2018). A recent analysis of Medical Expenditures Panel Survey data on primary care and special mental health visits showed racial and ethnic disparities in mental care visits, suggesting that Latinx adults have lower access to mental health services compared with their sameage white counterparts (Jones et al., 2018). Taken together, these experiences and perceptions may lead to older Latinx adults' greater stigmatization and negative attitudes toward treatment of "less severe" pathologies such as depression or anxiety (Hantzi et al., 2019) and lower utilization of mental health services for these issues (Villatoro et al., 2018). --- Limitations and Future Research There are a few limitations to the current study that lend themselves to recommendations for future research. First, the generalizability of the results to a broader older adult population is limited by the study's majority Mexicanorigin clinical sample from a single facility in Imperial County, CA, in the United States. As this sample is a selfselected clinical population engaged in outpatient treatment, different mental health attributions may be found in nonclinical and untreated populations of older adults. Generalizability and a greater understanding of Mexicanorigin older adult mental health attribution may be further expanded by including data collection from both rural and urban area clinics with varying proximities to the U.S.-Mexico border. As is typical with secondary data analyses, the study was limited to the measures and items included in the original chart review data collection. The symptom attribution questionnaire was developed based on the extant literature and expert agreement, but it has not been psychometrically validated. However, psychometrically validated questionnaires assessing cultural health and mental health beliefs and attributions are currently limited. Thus, the current study's questionnaire serves an adequate purpose for assessing cross-cultural mental health attributions for a low-literacy population (Vahia et al., 2013), but could benefit from validation through exploratory and confirmation factory analyses. Future researchers may also wish to use a more involved measure that specifically targets Latinx health beliefs, such as the Cultural Health Attributions Questionnaire (Murguia et al., 2000), which has benefited from some evaluation and revision (Fox et al., 2015). The study also used the BASH (Norris et al., 1996) to measure acculturation, in which English language use is utilized as a proxy to measure acculturation. Although this proxy measurement is common (González et al., 2001), acculturation is a complex and multidimensional construct (Thomson & Hoffman-Goetz, 2009;Viruell-Fuentes, 2007), and future studies would be improved by using a more comprehensive measure, such as the full SASH (Marin et al., 1987), which measures acculturation beyond English language use. These more thorough measures may better account for Mexican-origin older adults' spiritual/supernatural mental health attributions, rather than the English language acquisition proxy. Differences in mental health attributions between the Mexican-origin and non-Latinx groups were limited to inferences based on comparing visual plots and item probabilities because LCAs do not allow for a direct statistical comparison of two models (Nylund-Gibson & Masyn, 2016). Furthermore, the cross-sectional and nonexperimental design of the study bars any causal inferences and limits the interpretations of the results to a single time point. This highlights the need for researchers to use longitudinal data and latent transition analysis (Nylund et al., 2007) to explore whether the classes identified are stable over time and if Mexican-origin Latinx and non-Latinx older adults' mental health attributions can be more directly linked to their service utilization, coping behaviors, and mental health diagnoses as distal outcomes. Additional research questions of interest include whether attribution beliefs are affected by specific mental health diagnosis (e.g., Are older adults with psychotic disorders more likely to attribute their mental health to spiritual or supernatural causes, even when controlling for Latinx ethnic identity?) and whether there are differences in sources of support sought by Mexican-origin and non-Latinx older adults for their mental health issues, based on class membership or attribution beliefs. Lastly, the data analyzed for this study are more than 15 years old, which raises the question of relevance to 2020. We argue that the data are indeed relevant for two reasons. First, the findings add to a literature that is not rich or well examined. Second, deficiencies persist with respect to culturally competent care in mental health care services (Barrera & Longoria, 2018;Izquierdo et al., 2014;Jimenez et al., 2012) as do disparities in mental health service utilization and outcomes for older Latinx adults (Cook et al., 2017;L

Background and Objectives: Providing appropriate and culturally sensitive care to the rapidly growing number of U.S. Latinx older adults with psychiatric conditions presents a major public health challenge. We know little about older Latinx adults' perceived causes of mental health problems, offering clinicians limited insight to guide successful and culturally congruent treatment. Moreover, there is a paucity of mental health research examining heterogeneity in how Latinx individuals may attribute mental health symptoms. The present study sought to identify how Latinx and non-Latinx older adults attributed the sources of their mental health problems and how these types of attributions differ by ethnicity.This study analyzed data collected from a retrospective chart review and survey of 673 adults aged 55-95 years (430 Mexican origin and 244 non-Latinx) from a rural psychiatric outpatient clinic near the California-Mexico border. We conducted stratified latent class analysis (LCA) by race/ethnicity to explore the mental health attribution beliefs of Mexican-origin and non-Latinx clinic patients. Results: Different LCA patterns for Mexican-origin Latinx versus non-Latinx groups were found. For non-Latinx adults, there was a class of individuals who attributed their mental health issues to social and financial problems. For Mexicanorigin adults, there was a class of individuals who attributed their mental health issues to spiritual and/or supernatural factors, unaffected by acculturation level, depressive symptom severity, and time spent in the United States, but differing by gender. We found within-group heterogeneity: Not all Mexican-origin or non-Latinx older adults were alike in how they conceptualized their mental health. Discussion and Implications: Mexican-origin Latinx and non-Latinx older adults attributed their mental health issues to different causes. More Mexican-origin older adults attributed their symptoms to spiritual causes, even after controlling for contextual factors. Further research is needed to determine whether attribution beliefs are affected by specific mental health diagnoses and other cultural factors not measured in this study.

attributions can be more directly linked to their service utilization, coping behaviors, and mental health diagnoses as distal outcomes. Additional research questions of interest include whether attribution beliefs are affected by specific mental health diagnosis (e.g., Are older adults with psychotic disorders more likely to attribute their mental health to spiritual or supernatural causes, even when controlling for Latinx ethnic identity?) and whether there are differences in sources of support sought by Mexican-origin and non-Latinx older adults for their mental health issues, based on class membership or attribution beliefs. Lastly, the data analyzed for this study are more than 15 years old, which raises the question of relevance to 2020. We argue that the data are indeed relevant for two reasons. First, the findings add to a literature that is not rich or well examined. Second, deficiencies persist with respect to culturally competent care in mental health care services (Barrera & Longoria, 2018;Izquierdo et al., 2014;Jimenez et al., 2012) as do disparities in mental health service utilization and outcomes for older Latinx adults (Cook et al., 2017;López et al., 2012). Thus, the data provide a unique opportunity to deepen our understanding of mental health attribution among older Latinx adults to inform efforts to improve treatment and outcomes for this population. --- Conclusions The current study's findings speak to a need for clinicians and researchers to better understand older adults' conceptualization of their mental health, especially Mexicanorigin and other Latinx adults because of their increased mental and physical health risks in older age (Bilotta et al., 2010;Howard et al., 2014;Ward et al., 2018) and lower mental health service utilization (Cook et al., 2017;López et al., 2012). Both culturally sensitive and individualized conceptualizations of and approaches to mental health treatment are likely to lead to vulnerable adults' greater access, utilization, and adherence of treatment and thus better treatment outcomes (Barrera & Longoria, 2018;Bignall et al., 2015;Izquierdo et al., 2014;Vaughn et al., 2009). Researchers and clinicians may develop and implement more effective treatments for both the historically underserved and dominant majority client populations by examining mental health service utilization and implementation with an intersectional lens that encompasses age, gender, culture, and other important identity factors (Villatoro et al., 2018) as well as a heterogeneous personcentered approach to studying these groups (Bauer & Shanahan, 2007). --- Funding None declared. --- Conflict of Interest None declared.

Background and Objectives: Providing appropriate and culturally sensitive care to the rapidly growing number of U.S. Latinx older adults with psychiatric conditions presents a major public health challenge. We know little about older Latinx adults' perceived causes of mental health problems, offering clinicians limited insight to guide successful and culturally congruent treatment. Moreover, there is a paucity of mental health research examining heterogeneity in how Latinx individuals may attribute mental health symptoms. The present study sought to identify how Latinx and non-Latinx older adults attributed the sources of their mental health problems and how these types of attributions differ by ethnicity.This study analyzed data collected from a retrospective chart review and survey of 673 adults aged 55-95 years (430 Mexican origin and 244 non-Latinx) from a rural psychiatric outpatient clinic near the California-Mexico border. We conducted stratified latent class analysis (LCA) by race/ethnicity to explore the mental health attribution beliefs of Mexican-origin and non-Latinx clinic patients. Results: Different LCA patterns for Mexican-origin Latinx versus non-Latinx groups were found. For non-Latinx adults, there was a class of individuals who attributed their mental health issues to social and financial problems. For Mexicanorigin adults, there was a class of individuals who attributed their mental health issues to spiritual and/or supernatural factors, unaffected by acculturation level, depressive symptom severity, and time spent in the United States, but differing by gender. We found within-group heterogeneity: Not all Mexican-origin or non-Latinx older adults were alike in how they conceptualized their mental health. Discussion and Implications: Mexican-origin Latinx and non-Latinx older adults attributed their mental health issues to different causes. More Mexican-origin older adults attributed their symptoms to spiritual causes, even after controlling for contextual factors. Further research is needed to determine whether attribution beliefs are affected by specific mental health diagnoses and other cultural factors not measured in this study.

Introduction Organizations are never neutral; rather, they become a means to crystallize specific socioeconomic interests. In the contexts of maximizing profit and exploiting centrality within work processes, workplace bullying might even be considered an event that can be expected to occur with a certain regularity and frequency. This reality drives researchers to pay special attention to the sources, means, and dynamics generated by power inequality in labor settings [1]. In this line of thought, the phenomenon of workplace bullying has a detrimental effect on both individuals and organizations (e.g., managerial costs and turnover escalate and productivity declines) as the number of witness distractions and the emotional/physical health of the victims increase. Such cases become exponentially worse when a potential lawsuit for unjust dismissal or work compensation/disability is added to an already unfavorable situation. Other economic pitfalls, with a significant negative impact on profits, can sometimes be difficult to measure and clearly define. These pitfalls may include a reduction in the quality, negative impacts on the organization's reputation, the escalation of mistakes and absenteeism, and the corrosion of customer relationships due to a lack of attention paid to their objectives and commitments, among others [2]. This assertion becomes even more significant for those organizations mainly composed of employees providing particular assistance in a close and direct way to patients (e.g., healthcare workers). This paper was written in accordance with previous studies that clarify how workplace bullying among healthcare workers has become a persistent phenomenon within organizations. In this sense, Rowell states that, at present, workplace bullying has particularly increased in the health and community care sectors and that such behavior is four times more prevalent in this sector than sexual harassment [3]. In line with these findings, DuHart reports that physicians and nurses are occasionally victims of workplace hostility [4]. The physical violence rates against doctors and nurses are 16.2 per 1,000 and 21.9 per 1,000, respectively. In the European Union, 52% of healthcare jobholders have experienced some sort of aggression at work, followed by 39% of social care workers and 25% of service workers [5]. In the scientific literature, several types of bullying have been studied [6]: intimidation, harassment, victimization, aggression, emotional abuse, and psychological harassment or mistreatment at workplace, among others. The variation in definitions may hinder the conceptualization of the workplace-bullying phenomenon in a more consistent way, inhibiting effective contributions among researchers and practitioners [7]. Bullying is commonly defined by its social manifestations, which are clearly classifiable under the same umbrella as aggressive behavior [8] that generally occurs during interpersonal interactions in work settings [9]. Similarly, there seems to be a consensus that bullying, as a behavior, can be defined in terms of intentionality, frequency (e.g., weekly) or duration (e.g., approximately six months), the targets' reaction(s), perceived imbalance and misuse of power between the perpetrator and target, inadequate support, and the target's inability to defend himself from such aggression [10][11][12][13], as well as having to cope with negative and constant social interactions [13], physical or verbal badgering, insulting remarks [12], and intense pressure [14]. Regarding the extent of its manifestation, there is a strong disagreement about the prevalence of bullying; estimates range from 4% to 5% in Northern European Countries [15,16] to 15% in Southern European nations [17]. Certain factors, such as cultural characteristics and social changes, seem to explain the variations in these prevalence rates, as do issues related to research methodology [15]. For instance, studies on workplace bullying have utilized a wide variety of measurement methods, instruments, and research designs [18,19], to the extent that it appears reasonable to consider certain methodological procedures to be biased with respect to their reported prevalence rates [20]. Statistics paint a bleak picture regarding the exposure of healthcare employees to hostility, mostly because bullying at work in the context of healthcare services includes interactions among such varied groups as co-workers, supervisors, patients, families, visitors, and others [21]. Although previous definitions shown a propensity to combine the persistence and duration of the bullying into the same key construct of this phenomenon, the present paper posits that workplace bullying involves a strong psychological component in its materialization. Thus, the main objective of this research was to identify the determinants of workplace bullying among healthcare professionals that emerge from personal variables, working conditions, and contextual factors. In fact, an essential condition of bullying is that the act itself must be perceived as a hostile situation by the target [10,22]. From this point of view, the pernicious effects of workplace bullying (e.g., anxiety, depression, absenteeism, and lack of organizational commitment) are externalized with a greater magnitude at the moment the victim perceives the unpleasant condition, independently of the persistence or duration of the bullying action. --- Factors Influencing Workplace Bullying Given the negative consequences of workplace bullying on the mental health and well-being of employees and, hence, on the performance of any organization, it is vitally important to understand the reasons that trigger the emergence and development of this social phenomenon [23]. In this respect, the psychologists currently leading this specific research approach have mostly focused on victim and/or bullying pathologies. From a humanistic perspective, this predominantly psychological scope has been utilized to address workplace bullying at an individual level, and many of the studies conducted have been clearly linked to emotional effects and therapeutic practices. This concrete research field has provided a sufficiently broad view and a group of scholars that study the influence of micro-organizational factors (e.g., role conflict, leadership, political aspects, or organizational culture) on individual conduct [12]. Given the above, it appears that research on workplace bullying has evolved towards a multi-causal understanding. In this respect, Hoel and Cooper identified five core areas of focus: subjects, social interaction, group dynamics, working environment, and organizational, societal and political levels [24]. Nevertheless, a considerable number of experts agree that workplace bullying happens as the result of specific interactions among the factors that influence the individual, organizational, or contextual milieu of the people involved [25][26][27][28]. --- Individual Factors for Workplace Bullying Some personal characteristics of the victims might constitute potential workplace bullying triggers. In fact, early studies on the subject indicated that employees who experienced conflict at work also experienced similar situations in other contexts, such as with their partners, family, and friends [29]. The perspective of individual antecedents related to workplace bullying has been a controversial topic, as the results are often interpreted as "blaming the victim" [30]. However, studies aiming to identify personality types that are specifically inclined to bullying are far from conclusive [31]; the majority of researchers believe that a personal predisposition for playing the role of victim or bully might not exist [32,33]. Nonetheless, some studies have attempted to identify a selection of individual factors (e.g., gender, age, and seniority) that may increase the risk of becoming a victim or bully [27,32]. The presence or absence of these variables could influence bullying ratios [34], i.e., when bullies weigh the potential costs and personal benefits of their actions based on the particular characteristics of their victims. In this sense, certain groups are considered more vulnerable than others (e.g., women or junior employees). One of the crucial factors that may be utilized to study bullying on the individual level is gender, although the current results of empirical studies do not quite seem categorical. Some authors have observed a higher frequency of bullying among women compared to men [26,[35][36][37][38], while other large-scale studies indicate that, except for sexual harassment, both men and women are equally prone to being bullied at work [19,[39][40][41][42]. In any case, Einarsen et al. suggest that the gender differences found by some researchers are in fact consequences of the discrimination that both genders can suffer as a result of their position within an organization [12]. From this perspective, one bullying action could perfectly correspond to a concrete behavior oriented toward a specific minority at work, regardless of the gender of this minority. In research concerning nursing staff in Norway, a profession in which men are underrepresented, Eriksen and Einarsen found that female bullying actions reached 4.3%, while this parameter among males escalated to 10.2% [43]. Findings related to a different personal factor, such as the victim's age, have not identified a clear association with workplace bullying. Referring to this lack of association, Rayner reports that bullying victims are normally under 25 [40]; similarly, Hoel and Cooper find that young people are more likely to experience a greater level of bullying in comparison with older employees [39]. The exact opposite findings are reported by Einarsen et al. and Einarsen and Skogstad, who observed a higher incidence of bullying among senior employees [22,44]. This conclusion is also reached by Vartia and Piirainen et al. in subsequent research [45,46]. --- Organizational Factors for Workplace Bullying The conceptualization of any organization as a whole entity is essential to understand the phenomenon of bullying; it seems quite complicated to imagine a labor context as excessively independent or as non-influential enough to impede internal workplace bullying. Therefore, although early studies have focused mainly on the psychological characteristics of bullies and their victims, several scholars have more seriously pondered the influence of the specific working and structural characteristics of organizations on people. In this paper, we present a brief bibliographical review of the core studies in which the relationships among a number of internal dynamics (e.g., job stability, job design, and human resources practices) and workplace bullying is analyzed. --- Job Stability The level of labor stability might influence the degree of vulnerability to bullying, not only because unstable and temporary jobs are frequently held by lower-status professional employments but also because insecurity reduces the perceived power of employees vis-à -vis their superiors. An empirical exploration among university employees at a specific academic center was conducted in a noteworthy effort to demonstrate that flexible working arrangements can contribute to the prevalence of bullying [47]. In fact, one of the reasons given to explain the increase of bullying within 21st century organizations is that the organizational restructuring processes and higher levels of outsourcing have enlarged the power gap between managers and employees [48,49]. In this scenario, we could take for granted that the bullying rates among employees with temporary contracts would be higher than rates registered among their colleagues with permanent contracts. However, Kivimä ki, Elovainio, and Vahtera do not observe any difference between these two groups, or between full-time and part-time employees [50]. In reference to these differences, the research results seem to be conflicting as well; while Baron and Neumann find a positive relationship between part-time employment and bullying, Hoel and Cooper report the same finding, but among full-time employees [39,51]. --- Intrinsic Characteristics of Job Position The empirical research on workplace bullying and the intrinsic characteristics of job position is also extensive. Previous studies have identified certain variables, such as workload [26,39,44], control [36,41,[52][53][54], role ambiguity [44,55], role conflict [44,56], leadership behavior [41,44], social support from co-workers and supervisors [36,57], social climate [36,39,41,44,53,58], and organizational change [39,[59][60][61], as the key elements predicting the occurrence of bullying within organizations. A large investigation conducted in the United Kingdom on 5,200 subjects reveals that victims of workplace bullying, compared to non-bullied individuals, are distressed by their workload, rarefied working environment, greater organizational change, unsatisfactory relationships at work, and a more consistent intention to resign [39]. Similarly, a study on Norwegian employees by Einarsen et al. reports a significant correlation between the variables described above and workplace bullying (i.e., workload, control, role ambiguity, role conflict, leadership behavior, social climate, and organizational change) [44]. Similarly, Salin finds that bullying appears to be correlated with politicized and competitive organizational climates and even slightly more strongly with workload [26]. Correspondingly, in a sample of 400 employees from five Swedish organizations, Hansen et al. observe a negative correlation between bullying and the support given to employees by their colleagues and their superiors [57]. Bowling and Beehr's meta-analysis, which reviewed over 90 studies published between 1987 and 2005, makes a remarkable contribution to the investigation of workplace bullying by compiling and organizing the extant empirical research [31]. Regarding the characteristics of a job position, these authors report that bullying tends to emerge in occupational settings where other stressors, such as role conflict (r = 0.44), role ambiguity (r = 0.30), overload (r = 0.28), and work limitations (r = 0.53), are often simultaneously identified. Likewise, they confirm that autonomy at work is negatively associated with bullying (r = -0.25). Further organizational variables that have been studied for their associations with bullying include monotony, complexity, and teamwork. Zapf et al.'s research makes evident that monotonous and repetitive tasks are more frequent among bullying victims [36]. Correspondingly, in a subsequent investigation, Zapf does not corroborate any association between bullying and work complexity [28]. Similarly, Zapf et al. realize that, during activities requiring teamwork, bullying among peers seems to be more likely to occur [36]. According to those authors, the social environment generated within these groups contributes to the search and selection of scapegoats among the less powerful members to redirect team aggressiveness. Moreover, numerous analyses have validated the connections between workplace bullying and individual perceptions within organizations, such as job satisfaction and commitment. The former has been amply studied by Vartia and Hyyti and constitutes an additional and plausible alternative variable related to bullying [53]. Job dissatisfaction, which causes victims emotional distress, can be considered a condition necessarily linked to affective commitment. However, quite a few authors have reported a negative relationship between these two variables [39,62]. Employees who are highly committed to their organizations may be more vulnerable to stressors in their working environment precisely due to their emotional ties their social structures [63]. --- Occupation and Bullying The academic literature is expanding with a prolific number of studies concerning bullying in specific types of occupations. On this subject, certain authors report that 44.0% of nursing staff members have been bullied at some point in their working lives [64]. Other occupations with high incidences of bullying include restaurant employees [65], teachers [66], university professionals [34,67,68], business professionals [69], transportation workers [68], and police officers [70]. Diverse investigations have identified multiple occupations within the same studies: blue-collar workers, clerks and service workers, associate professionals, managers and professionals, among others [71,72]. Related to this issue, Woodman and Cook report interesting results in the UK utilizing a sample of 512 managers; 39.0% of the respondents affirmed that they had been bullied in the past three years [73]. Bullying appears to have detrimental effects at all management levels; middle managers, as an example, appear only slightly more prone to suffer workplace bullying, representing 49.0% of the reported cases during the past three years. This figure may support the phenomenon known as "management squeeze", in which middle managers are subjected to the particular pressures of being required to implement unpopular policies as a result of the decisions made at more senior levels. In previous studies, Ariza, Morales, and Menor identify assorted factors that may contribute to the emergence of workplace bullying within managers [74]. Apparently, the likelihood of a manager being bullied increases when job insecurity is present, when people are dissatisfied with their work and salary, when employees are predominantly in the public sector, and when work activities are very emotionally demanding. --- Contextual Factors in Workplace Bullying In addition to the factors related to the internal dynamics of organizations, bullying may also be occur as a result of the context in which the organization operates. Research on this subject reveals that bullying is more frequent in the service sector than in any other industry, particularly in health, public service, education, and financial service [34,52]. Furthermore, Leymann argues that bullying most commonly occurs in the health care sector, especially among nurses, due to their work overload and the double supervision they are subjected to by doctors and chief nurses, which violates the Unity of Command Principle [42]. Supporting this argument, Yildrim and Yildrim affirm that 87.0% of nurses in Turkey have experienced some form of bullying, especially those in the public sector [75]. High levels of bureaucracy, the existence of very strict norms, and excessively high job security may generate environments amenable to the occurrence of bullying, as these settings make bullies invisible and victims less likely to resign [69]. In this sense, Zapf et al. provide a wide-ranging summary of European studies and conclude that the prevalence of bullying is higher for the public sector, service, health, education, and assistance fields than for private industry [19]. A similar conclusion is presented by Giorgi, Arenas, and Leon-Perez in Italy [17] and by Hoel and Cooper in the United Kingdom [39]. These authors report more significant bullying activity within public services (e.g., education or correctional assistances) and a lower prevalence of bullying in the retail and industrial sectors. Similarly, Soares' research shows that 4.4% of public education and health care employees have been bullied occasionally by their patients or students while completing their daily tasks [76]. Furthermore, LaVan, Katz, and Jedel suggest that public sector jobholders should manage their employment relationships differently than do workers in the private sector. This difference may lead to an alternative form of workplace bullying [77]. Although there are some research papers suggesting that bullying might be higher within the public sector [19,39,69,78], LaVan, Katz, and Jedel firmly believe that this actually occurs because countless jobs in the public sector entail a great deal of emotional labor rather than instrumental work [19,77,79]. Public sector employees enjoy a special type of employment status; they are protected by civil service rules and regulations, by unions with internal grievance procedures, and by statutes that provide protection against retaliation for whistle blowing. For the purpose of the present work, a comprehensive workplace-bullying model is proposed in the following section. Then, the most relevant empirical results obtained through a logistic regression analysis are presented, followed by the main conclusions and limitations of the study. --- Methods The data utilized for this research were obtained from the 5th European Working Conditions Survey, conducted in 2010 by the European Foundation for the Improvement of Living and Working Conditions. This survey provides insight into to the working environment and employment conditions of the 27 EU Member States, including Albania, Croatia, Kosovo, Macedonia, Montenegro, Norway, and Turkey. The target population includes workers aged 15 years and over (16 and over in the case of Spain, the UK, and Norway) who are employed and reside in the country being surveyed. This was a multi-stage investigation using a stratified random sample. Over 43,000 interviews were collected in 2010. The study found that the prevalence rate of workplace bullying was 11.3% among healthcare workers. Given the objective of this research, we gathered a sub-sample of 284 health professionals, including members of the International Standard Classification of Occupations (ISCO-08) subgroup 22 (e.g., medical doctors, nursing and midwifery professionals, traditional and complementary medicine professionals, dentists, ophthalmic opticians, and physiotherapists). In total, 41.2% of these health professionals claim to have experienced workplace bullying (N = 117), while 58.8% indicate that they have not (N = 167). The subjects in this sample are medical doctors (66.9%), nursing and midwifery professionals (21.5%), or other health professionals (11.6%). They are drawn from both the public (67.6%) and private sectors (32.4%), are 60.9% female and 39.1% male, and have an average age of 44.1 years. Finally, 11.3% completed secondary education, and 88.7% completed their university studies. The term "secondary education" is used to categorize individuals with either high school study or vocational/technical training. The dependent variable for this analysis is bullying at work. Respondents were asked to answer just one question based on their individual experience: Over the past 12 months, during the course of your work, have you been subjected to bullying/harassment? Bullied professionals are codified as 1, while those who claim not to have felt bullied are coded as 0. Two main approaches are used in the bullying research questionnaires were implemented: the self-labeling and operational approaches. The limitations and advantages of these methods are discussed in Nielsen et al. [80]. Workplace bullying is considered in this study to be a complex phenomenon that arises due to the dynamic interactions of labor environment variables and individual factors. Taking into account preceding studies on bullying at work, this study is arranged in three sets of independent variables grouped into three categories: personal and family factors, working conditions factors, and organizational/contextual factors. The codes and classification of explanatory variables are as follows: Individual characteristics: Gender (0: Male; 1: Female), Age (0: 15-24; 1: 25-39; 2: 40-54; 3: 55 or over), Level of education (0: University education, 1: Secondary education), Marital status (0: Partnered; 1: Single), and with Children at home (0: Yes; 1: No). Working conditions: Length of service (0: more than 10 years; 1: more than 5 up to 10; 2: more than 1 up to 5; 3: up to one year), Type of contract (0: A permanent contract; 1: A temporary contract), Working hours (0: More than 40 h The IBM SPSS 20 (Statistical Package for Social Science) software application was utilized to measure the variables. The methodology employed to accomplish the objectives was based on the binary logistic regression model, a specific type of regression model intended for dichotomous variables. This statistical technique is used to determine the probability that an event will happen (workplace bullying, in this case) compared to the probability that it will not. --- Results Table 1 lists some of the main sociodemographic characteristics of both the healthcare workers who reported that they were bullied (N = 117) and those who did not (N = 167). Furthermore, some statistical differences were observed regarding the array of variables. Workplace bullying emerges as even more acute among female healthcare workers (72.6% compared to 52.7%; <unk> 2 = 11.507, d.f. = 1, p = 0.001) young workers (46.2% compared to 27.3% between 25-39 years old; <unk> 2 = 17.107, d.f. = 3, p = 0.001), workers who did not attend university (21.4% compared to 4.2%; <unk> 2 = 20.301, d.f. = 1, p = 0.000), workers with a permanent contract (85.0% compared to 78.8%; <unk> 2 = 3.568, d.f. = 1, p = 0.042), workers not in management positions (84.2% compared to 66.9%; <unk> 2 = 10.485, d.f. = 1, p = 0.001), workers in the public sector (66.7% compared to 55.7%; <unk> 2 = 8.182, d.f. = 1, p = 0.004), workers who are not satisfied with their jobs (32.8% compared to 12.0%; <unk> 2 = 17.927, d.f. = 1, p = 0.000), and workers who experience stress in their daily work (89.7% compared to 77.2%; <unk> 2 = 7.408, d.f. = 1, p = 0.004), compared to healthcare professionals who do not consider themselves as bullied in their workplaces. Table 2 shows the results for the estimations calculated by logistic regression that were derived from the factors determining the level of workplace bullying within health professional settings. The statistical tests applied to assess the validity of the model (Hosmer-Lemeshow analysis; Chi-square test: 5.444; Sig. 0.709) largely suggested enough basis to acknowledge its validity; that is, they affirmed that the set of job-related variables considered for the general model of this research may potentially explain in a satisfactory manner whether a health professional is prone to experiencing bullying at work. We should also highlight that the chosen variables allow the model to be generalized, indicating its possible utility for predictive purposes. The logistic regression model integrates individual and organizational factors and estimating that the likelihood of workplace bullying is 80.8% (87.6% for bullied healthcare workers and 71.6% for non-bullied ones). In general, the model brings to light the extent to which the probability of a health sector employee considering him/herself bullied is higher among younger and less educated women who have small children at home, are unsatisfied with their working conditions (e.g., working in shifts or performing monotonous and rotational assignments), and suffer from work stress and lack of promotion opportunities within their organizations. As a consequence, the personal and organizational variables appeared reliable for predicting the development of certain attitudes, such as workplace bullying, and for identifying critical constructs for the understanding of the phenomenon under study. Each of the personal and job-related significant factors that measure the probability of workplace bullying have a different degree of impact, as indicated by the analysis of the confidence intervals obtained in the corresponding odds ratios (see Table 2). Accordingly, the most influential variables related to workplace bullying are the level of education (OR 5.507; CI 1.789-16.951), work stress (OR 4.962; CI 1.703-14.456), and expectation of career growth (OR 4.517; CI 2.089-9.765). These factors could prove to be strong predictors of whether an employee is exposed to bullying or not. The odds ratio coefficients for other variables (gender, children at home, shift work, monotonous tasks, rotating tasks, and working condition dissatisfaction) remain over 2.0, with the exception of age, with an odds ratio coefficient of 0.627 and a confidence interval ranging from 0.382 to 1.030. --- Discussion Advances in understanding the primary circumstances that precede workplace bullying take on great importance in the development of more effective prevention and intervention tools to remedy this social problem [81,82]. Workplace violence in healthcare settings occurs four times more often than in all other private-sector industries combined [83], with the highest incidences in psychiatric wards, accident and emergency departments, and high-dependency units [84]. The bullying prevalence varies significantly from one country to another and even within the same country. In Europe, for example, even though the inferences may vary depending on the measurement and estimation methods being utilized [15,85], studies of the occurrence of workplace bullying report rates of approximately 4-10% [19]. For the present research, the prevalence rate happens to be slightly higher: 11.3% of healthcare workers labeled themselves as bullied in their professions. Nevertheless, this ratio is still far from the results obtained in the U.S., where 38% of the healthcare employees report psychological harassment [86]. This relationship is similar to that described by Dellasega, who finds that 44.0% of nurses experience episodes of bullying at some point during their working lives. [64]. These results indicate that the rate of workplace bullying for health professionals is larger than the predicted average calculated from similar parameters for employees laboring in any other occupational sector. Regarding this point, Zapf et al. provide an extensive summary of European studies and conclude that the prevalence of bullying is higher in the public sector (e.g., service, health, education, and assistance) than in the private sector [19]. A similar conclusion is reached by Hoel and Cooper in the United Kingdom, who report a higher incidence of bullying within public services, such as education or correctional assistance, and a lower incidence in the retail and industrial sectors [39]. Similarly, Soares' research shows that 4.4% of public education and health care employees have been occasionally bullied by their patients or students while completing their daily tasks [76]. A body of literature has emerged describing the possible triggers of workplace bullying within healthcare staff and has focused primarily on two areas. The first area pertains to the personal and individual differences among those involved in the bullying incidents, while the second pertains to the characteristics of the surrounding organizational settings in which these circumstances occur. Similar to the present paper, many authors currently embrace the concept that both individual and organizational factors are important to understanding bullying behaviors. For instance, healthcare workers under 40 are the most frequent victims of violent events [87]. Researchers have also observed that older workers experience significantly less violence than young workers [88][89][90][91][92]. Other characteristics of healthcare workers that have been associated with an increased risk of workplace bullying include gender and marital status. Furthermore, a greater percentage of female physicians fear a potentially violent encounter at work compared to male physicians [93]. Lin and Liu's study reports that unmarried workers are significantly more likely to experience workplace violence compared to married employees [94]. In the European Union, these results suggest that there are specific sociodemographic features that may influence the phenomenon of workplace bullying. Regarding this concern, the current regression analysis outcomes indicate a tendency for young female healthcare workers with only secondary education and with children at home to suffer bullying. The group described above is certainly in a position of greater weakness in relation to the other groups that have greater power, for example, men with university degrees or higher seniority in their organizations. Additionally, regarding concrete cases related to female jobholders, persistent and predominant sexist attitudes should be noted, as well as structural barriers that inhibit women's careers to a certain extent compared to their male colleagues. These circumstances make these groups particularly more vulnerable; therefore, they are more likely to end up as victims of workplace bullying. Furthermore, some organizational factors are found to increase the odds of workplace bullying against healthcare workers. For instance, with respect to working conditions, McAneney and Shaw report that violent events in long-term care are more likely to take place during the evening and night hours [95]. It has been claimed in recent meta-analyses that there are some specific organizational variables (e.g., workplace bullying antecedents) that are worth noting, such as conflict and role ambiguity [31], work overload, stress, lack of autonomy, and absence of organizational fairness [56]. Zapf et al.'s research shows that performing monotonous and repetitive tasks is more common among bullying victims [36]. At the organizational level, this study emphasizes that the propensity for a healthcare worker to experience bullying escalates among those who work on a shift schedule, perform monotonous and rotating tasks, suffer from work stress, experience a lack of satisfaction due to their working conditions, and do not perceive any opportunity for promotion within their organization. This unpredictable environment, characterized by insecurity, role conflict, and tension, allows few opportunities for socialization and even less time for conflict resolution; both of these factors may indirectly contribute to the emergence of aggressive behaviors and bullying. Ultimately, a stressful social climate and precarious work atmosphere create a breeding ground for workplace bullying, as reflected in the results of the present empirical study. Finally, it seems accurate to say that the contextual variables of an organization do not influence workplace bullying. Given this finding, it is possible to deduce that workplace bullying is prevalent with the same degree of intensity in both public and private organizations, regardless of their size. --- Conclusions Workplace bullying has become a serious and growing problem that affects a significant proportion of healthcare professionals. As a result of its negative consequences on the mental health and well-being of employees, and hence on the performance of the organizations, the importance of understanding the factors that contribute to the emergence and development of bullying is vital [23]. In this regard, the present study aims to contribute to the development and implementation of measures to prevent bullying in the health sector. The multidimensional model created in the present research is intended to identify healthcare workers who are prone to being bullied at work as individuals; the study findings have valuable and pertinent implications for institutions that aim to thrive and to enhance organizational performance. This work provides reasonable evidence that could be of significant benefit in the implementation of human resource policies: responsible managers could reduce the organization-wide levels of workplace bullying by adjusting certain working conditions that negatively affect employees who are especially susceptible to being bullied, given their personal characteristics. This research paper offers an empirical basis for further studies related to health sector issues in Europe. Attracting and retaining the most qualified and experienced professionals has become essential for successful and competitive organizations in the healthcare industry; organizations are urged to implement strategies oriented toward reducing workplace bullying. Consequently, the labor force in this economic sector has specific traits that should not be ignored. From a practical standpoint, the present findings could assist practitioners in facilitating harmonious social relationships among healthcare workers. Particularly, the results suggest that limiting job demands and increasing job resources could reduce workplace bullying. Specific attention should be paid to young women who feel dissatisfied with their working conditions, as they constitute a group with an increased risk of experiencing bullying. Despite the significant findings of this study, its intrinsic methodological limitations must be considered. First, the phenomenon of bullying was measured by self-report, which might increase the risk of common method variance, forcing us to assume a corresponding bias in the key variables. Exploring the experiences of 385 self-identified Canadian nurses, we showed that when targets identify themselves as victims, they report decreased job satisfaction, escalation of their level of burnout, and greater psychological distress in comparison with those exposed to bullying but who do not label themselves as sufferers [96]. Second, by utilizing self-identification without a strict definition, bullying is measured in broad terms, and consequently, there is a risk of overestimating its prevalence, as the respondents could report incidents that would not qualify as bullying according to the researchers' understanding of the phenomenon [20]. Third, a related methodological problem could be social desirability; previous scholars have analyzed the repercussions of desirability in workplace bullying studies. Given the particular understanding of the phenomenon under investigation, it seems probable that any given prevalence rate would exceed the rates obtained in this type of research, as many of the present victims took a large amount of time to acknowledge and accept that they were subjected to aggression of this nature. This predicament is particularly acute among certain population groups that are considered more vulnerable, such as women, young people, and employees with temporary interrelations. To correct this problem, some authors propose to make use of multi-method data and utilize objective measures that may reinforce workplace-bullying research. Examples of this type of data include managerial reports and scores from third parties (e.g., researchers) [97]. It should be noted, however, that assessing third party scores of workplace bullying without trying to counteract these behaviors raises ethical concerns due to the many negative consequences of workplace bullying for the parties involved, as well as for the work unit and the organization [98]. Fourth, the observed correlations between bullying and the variables analyzed in this study should be assessed cautiously, as the data are cross-sectional and not experimental. Finally, this study represents only a partial perspective of this phenomenon: the point of view of the victim but not of the bully. --- Conflict of Interest The authors declare no conflict of interest.

This paper aims to assess consistent predictors through the use of a sample that includes different actors from the healthcare work force to identify certain key elements in a set of job-related organizational contexts. The utilized data were obtained from the 5th European Working Conditions Survey, conducted in 2010 by the European Foundation for the Improvement of Living and Working Conditions. In light of these objectives, we collected a subsample of 284 health professionals, some of them from the International Standard Classification of Occupations-subgroup 22-(ISCO-08). The results indicated that the chance of a healthcare worker referring to him/herself as bullied increases among those who work on a shift schedule, perform monotonous and rotating tasks, suffer from work stress, enjoy little satisfaction from their working conditions, and do not perceive opportunities for promotions in their organizations. The present work summarizes an array of outcomes and proposes within the usual course of events that workplace bullying could be reduced if job demands were limited and job resources were increased. The implications of these findings could assist human resource managers in facilitating, to some extent, good social relationships among healthcare workers.

of the victim but not of the bully. --- Conflict of Interest The authors declare no conflict of interest.

This paper aims to assess consistent predictors through the use of a sample that includes different actors from the healthcare work force to identify certain key elements in a set of job-related organizational contexts. The utilized data were obtained from the 5th European Working Conditions Survey, conducted in 2010 by the European Foundation for the Improvement of Living and Working Conditions. In light of these objectives, we collected a subsample of 284 health professionals, some of them from the International Standard Classification of Occupations-subgroup 22-(ISCO-08). The results indicated that the chance of a healthcare worker referring to him/herself as bullied increases among those who work on a shift schedule, perform monotonous and rotating tasks, suffer from work stress, enjoy little satisfaction from their working conditions, and do not perceive opportunities for promotions in their organizations. The present work summarizes an array of outcomes and proposes within the usual course of events that workplace bullying could be reduced if job demands were limited and job resources were increased. The implications of these findings could assist human resource managers in facilitating, to some extent, good social relationships among healthcare workers.

The prevalence of diabetes is rising globally, and an estimated 8.5% of adults are affected by the disease (1). About 90% of all cases of diabetes are type 2 diabetes, the disease related to physical inactivity and overweight (1). The prevalence of type 2 diabetes is particularly high among ethnic minorities living in Western societies (2). End-stage diabetes complications (3) and related mortality (4) are more common among ethnic minorities. Psychosocial well-being is an often overlooked aspect in people with type 2 diabetes (5)dregardless of ethnicityd and is of key importance to their ability to manage their diabetes. The emotional burden, stress, and worry associated with diabetes are collectively known as diabetes distress (6). The Distress and Depression in Diabetes Study (3D Study) has indicated that over 40% of all people with type 2 diabetes suffer from at least a moderate degree of diabetes distress (7), and almost 30% experience a high level of diabetes distress (8). In view of this, combined with the fact that ethnic minorities often experience migration-related stress, we hypothesized that ethnic minorities have a greater chance of suffering from diabetes distress than do patients of Caucasian descent. Indeed, three recent studies have shown a higher prevalence of diabetes distress among ethnic minorities (9,10). By contrast, the Amsterdam Health Monitor Study did not find diabetes distress to be more common among immigrants with diabetes than among native Dutch patients with diabetes (11). However, these studies had relatively small sample sizes and were restricted to a few ethnic minority groups (9)(10)(11). The question, therefore, is whether ethnic minorities are indeed at greater risk of suffering from diabetes distress relative to those of Caucasian descent, and whether lifestyle and metabolic and cardiovascular risk factors play an explanatory role. In this study we investigated the prevalence of diabetes distress among multiple ethnic minority groups of people with type 2 diabetes in the Netherlands, and whether the association between ethnicity and diabetes distress is explained by metabolic control, lifestyle factors, cardiovascular risk factors, and the presence of diabetes complications. --- RESEARCH DESIGN AND METHODS This study was part of the Parelsnoer Initiative, a partnership between all eight university medical centers in the Netherlands. The Dutch Diabetes Pearl is an observational cohort study within this partnership, consisting of 6,666 people with type 2 diabetes treated in various geographical areas and settings (i.e., primary, secondary, and tertiary care). For this study we selected people with type 2 diabetes for whom we had complete data for ethnicity, obtained from the Problem Areas in Diabetes Survey (PAID), and for covariates (n = 4,191 [63%]; see the flowchart in the Supplementary Data). Data regarding prior cardiovascular disease (CVD), neuropathy, and nephropathy were available for a smaller sample. The medical ethical committees of all eight university medical centers approved the study. All participants provided written informed consent. People with a language barrier were not able to participate in this study, as they could not fulfill the informed consent criteria. Details on the design of the Dutch Diabetes Pearl were previously published (12). Data were collected between 2009 and 2015, and included information on demographics, physical examinations, and laboratory measurements, and information from various questionnaires. --- Ethnicity The Netherlands has a long history of immigration and is home to many ethnic groups; this is described in the Supplementary Data. The largest groups originate from Morocco, Turkey, Asia, and Suriname. The latter group consists of Hindustani-Surinamese, a group that originated in the northern part of India but moved to Suriname in the 19th century and migrated to the Netherlands from 1975 onward. This specific group has different genetic and cultural backgrounds than Asian people (13). Ethnicity was estimated based on country of birth and/or judged by study nurses based on conversations during which patients were asked about their ethnicity. The latter was done to overcome misclassification of ethnicity for individuals who had been born in the Netherlands but who had a different ethnic background, as we did not register the country of birth of patients' parents (14). In the case of a discrepancy between the country of birth and the judgement of the study nurse, we used the latter to define ethnicity, applying the following categories: Caucasian, Asian, Moroccan, African, Latin American (commonly known as Hispanic in the U.S.), Turkish, and Hindustani-Surinamese (Table 1 and Supplementary Data). --- Diabetes Distress (PAID) We used the PAID questionnaireda 20-item questionnaire that measures the emotional impact of diabetes and various other aspects of diabetes-related quality of life (9,10)dto assess diabetesrelated distress. The Dutch version of the PAID scale has good factorial validity, good internal consistency, and good convergent and discriminative validity (15). The PAID applies a five-point Likert scale ranging from "not a problem" (score of 0) to "serious problem" (score of 4). Multiplying the summed score by a factor of 1.25 provides the PAID score, which can range from 0 to 100; higher scores indicate greater levels of distress. Participants with a score $40 were considered to have high diabetes distress, further referred to as diabetes distress (8). --- Covariates Data on age and sex were collected from hospital information systems. Education level was self-reported. A low education level was defined as no education, primary school not finished, primary education, or low vocational education. A moderate education level was defined as intermediate vocational education, high secondary education, or high vocational education. A high education level was defined as high professional education or university education. In line with published alcohol guidelines, low-risk alcohol consumption was defined as no more than 7 drinks/week for women and no more than 14 drinks/ week for men (16). Smoking was categorized as never, former, and current smoking. BMI was defined as a person's weight (kilograms) divided by square of height (meters). Levels of fasting glucose, lipids, HbA 1c, and serum creatinine were measured in venous plasma. Urinary albumin and creatinine were measured in a sample of urine collected in the morning. All measurements took place in certified on-site laboratories (12). The metabolic laboratory variables were measured using the same methodology at all but one center. This was also the case for HbA 1c, which were measured with high-performance liquid chromatography at all centers but one, which used affinity chromatography. Information on prior CVD was collected via questionnaires, as previously described (12). Information regarding nephropathy was derived from the albumin-to-creatinine ratio determined in the morning urine sample. Microalbuminuria is defined as an albumin-tocreatinine ratio of 30-300 mg albumin/g creatinine (17). Neuropathy was measured using a Horwell neurothesiometer (Scientific Laboratory Supplies, Nottingham, U.K.), a device that assesses the vibration perception threshold (VPT) at the distal phalanx of the hallux. The neurothesiometer was used at six centers (not at the VU and Leiden University Medical Centers). Vibration thresholds were tested eight times for each participant. Mean vibration threshold then was calculated as the mean of the six highest vibration thresholds. A VPT of 25 V or higher indicated neuropathy (18). --- Statistical Analysis Descriptive statistics were used to assess the characteristics of participants from seven ethnic groups, and these data were expressed as the mean 6 SD, median (range), or number (percentage). We examined the association between ethnicity and diabetes distress with logistic regression, with the Caucasian group as the reference. The regression model was cumulatively adjusted for age, sex, education level, smoking, alcohol use, BMI, cholesterol profile, HbA 1c, diabetes medication, lipid medication, antihypertensive medication, prior CVD, nephropathy, and neuropathy. Although systolic and diastolic blood pressures were recorded, we did not include these measurements as covariates in the analyses because neither systolic nor diastolic blood pressure was significantly different among the ethnic groups (data not shown). Blood pressure therefore could not confound the association between ethnicity and diabetes distress. SPSS Statistics version 23.0 (IBM Corp, Armonk, NY) was used to carry out statistical analysis. The significance level was a = 0.05. Two-tailed P values #0.05 were considered significant. --- RESULTS --- Patient Characteristics Table 1 describes the characteristics of participants in all seven ethnic groups, of whom 3,684 were Caucasian, 83 were Asian, 51 were Moroccan, 92 were African, 134 were Latin American, 46 were Turkish, and 101 were Hindustani-Surinamese. The Caucasians and Asians had the highest mean age of all ethnic groups. Among the Caucasians, Asians, Latin Americans, and Hindustani-Surinamese, most participants had an intermediate level of education, whereas most Moroccan, African, and Turkish participants had a low level of education. The mean BMI of Asian, Moroccan, and Hindustani-Surinamese participants was lower than that of Caucasian participants, whereas the mean BMI was higher in African, Latin American, and Turkish participants. Both the lipid profile and mean HbA 1c levels were more favorable in Caucasians than in the ethnic minority groups. Caucasians and Asians mostly used oral diabetes drugs. It is notable that Turks had the lowest use of oral diabetes medications and the highest use of insulin. Lipid-modifying medication use was lower among Moroccans, Africans, and Latin Americans than among Caucasians, whereas it was higher in the other ethnic groups. Among Caucasians, 9.0% had microalbuminuria and 2.1% had macroalbuminuria, whereas these percentages were higher in other ethnic groups. The highest percentage of albuminuria was found among Turks. The largest proportion of participants with neuropathy was found among Moroccans, and the largest proportion of participants with prior CVD was found among Asians. --- Ethnicity and Diabetes Distress Compared with the prevalence of diabetes distress among Caucasians (5.8%), this prevalence was higher in all other ethnic groups, ranging from 9.6% among Asians to 31.7% among Hindustani-Surinamese. Table 2 depicts the odds of having diabetes distress, both unadjusted and after adjustment for nine possible confounders (age, sex, education, smoking, alcohol use, BMI, HbA 1c, CVD risk factors, and medication use), with Caucasians as the reference group. In unadjusted analyses, compared with Caucasians, participants with a Moroccan, African, Latin American, Turkish, or Hindustani-Surinamese ethnicity had a 4-to 7.5-fold greater odds of reporting diabetes distress, whereas the risk was not significantly different for Asians. After adjustment for age, sex, and level of education (model 2), the odds ratios decreased slightly, mainly because of age. Additional adjustments for the lifestyle factors smoking, alcohol use, and BMI (model 3) resulted in slightly altered odds ratios, caused mainly by differences in alcohol use and BMI. After adjustment for cholesterol profile and HbA 1c (model 4), the odds ratios decreased slightly and were no longer statistically significant for the Moroccan group. The odds of having diabetes distress did not change after additional adjustment for diabetes medication, lipid medication, and antihypertensive medication (model 5). Figure 1 shows the odds of having diabetes distress for all ethnic minorities after full adjustment, with Caucasian participants as the reference group. While we also further adjusted the model for primary and secondary/tertiary care, this did not affect these results (60% of participants were in primary care and 40% were in secondary/tertiary care). Additional adjustment for prior CVD, neuropathy, or nephropathy in subgroup analysis did not change the association between ethnicity and diabetes distress (Table 3). --- CONCLUSIONS The main finding of this cohort study was that an important association exists between ethnicity and diabetes distress among multiple minority groups with type 2 diabetes living in the Netherlands. Diabetes distress was 2.7-to 4.4-fold higher among African, Latin American, Turkish, and Hindustani-Surinamese minority groups than among Caucasians. This association was found to be independent of age, sex, education level, lifestyle factors, lipid profile, HbA 1c, CVD risk factors, medication use, or diabetes complications. Diabetes distress was also higher among patients of Asian and Moroccan descent than among Caucasians, although this association was not statistically significant. These findings may have important implications for the delivery of care to individuals with type 2 diabetes from ethnic minorities. Our current observations are in agreement with results of the second Diabetes Attitudes, Wishes and Needs (DAWN2) study, which reported higher diabetes distress among African American, Hispanic, and Chinese American minority groups living in the U.S. (9). In addition, in a recent article, Schmidt et al. (19) showed that diabetes distress was more prevalent among ethnic minorities, in particular Moroccans, than among native Dutch persons, whereas the associations for Turkish and Surinamese were explained by differences in socioeconomic status and HbA 1c. However, these studies included both individuals with type 1 and individuals with type 2 diabetes, whereas we included only individuals with type 2 diabetes. We investigated whether clinical variables such as glycemic control, level of education, and lifestyle factors confounded the association between ethnicity and diabetes distress. While some of the adjustments did indeed attenuate the observed associations, differences in diabetes distress remained statistically significant. Furthermore, in our study the association of ethnicity with diabetes distress was independent of complication status and care setting, both of which have been found to associate with diabetes distress (10) and are known to differ among ethnic minority groups (20). Our results support the notion that being a member of an ethnic minority in a Western societyd independent of other factorsdis associated with an increased risk of diabetes distress relative to being from the ethnic majority. Several possible explanations exist for the difference in diabetes distress between ethnic minorities and Caucasians observed in our study. Unmeasured factors related to ethnic origin (e.g., migration-related stress, language problems, health illiteracy, comprehension issues, and cultural differences) could contribute to diabetes distress (21)(22)(23). Also, religion and religious coping strategies might affect diabetes distress and helpseeking behaviors. For example, Moroccan and Turkish persons are generally Muslim and perhaps rely more on religion to help them cope with the burden of diabetes, rather than seeking social support or professional help (24). Although they report high levels of distress on the PAID questionnaire, showing any sign of distress to family and friends may be considered a weakness and therefore is experienced as shameful; this leads to more use of "private" coping and the avoidance of disclosing personal distress (25). Furthermore, previous findings suggest that acculturation (integration, assimilation, separation, and marginalization) is associated with depressive symptoms in Turkish minorities in the Netherlands (26). Previous studies have shown that attempted control by family members or friends is related to more psychological symptomatology in adults with diabetes but may result in improved diabetes self-efficacy (27). Diabetes self-efficacy is defined as a person's beliefs about their ability to adhere to self-care behaviors (28) and is known to be related to both psychological distress and ethnicity (28,29). Snoek et al. (8) found that relatively high levels of diabetes distress were associated with less diabetes self-management. In addition, differences in cultural perceptions of disease could be involved. People from ethnic minorities are known to differ in their perception of emotional or physical experiences as problems (30). Finally, ethnic minority groups have more concerns about medication use (31) and may receive less adequate treatment for comorbid depression (32) than do Caucasians. We strongly recommend, based on our results, that awareness of the high levels of diabetes distress in ethnic minorities with type 2 diabetesdand of the consequences for diabetes selfmanagementdbe increased among health care professionals. Several studies have shown that poorly regulated diabetes will not be resolved unless diabetes distress is reduced (30,33). Offering psychological support that aims to reduce distress should therefore be considered. Patients with diabetes distress are highly responsive to intervention (34). One strategy to reduce diabetes distress is to offer self-management support (35), which is relatively inexpensive and easily available for large patient populations (36). As an alternative, such interventions could focus on emotional themes underlying diabetes distress. It is known that if distress increases over time as a result of unsuccessful diabetes management, the person's attention narrows and they become less able to apply "behavioral solutions" (37). This increased diabetes distress thus leads to even poorer coping. As a consequence, the patient is faced with unrealistic expectations and goals and is unable to gain new knowledge and skills, and any attempts to discuss inaccurate personal beliefs and perceptions become self-defeating (38). Reduced diabetes distress is associated with improved well-being and may also result in better diabetes-related health outcomes (39). In support of this notion, the American Diabetes Association's Standards of Medical Care in Diabetesd2017 recommends routine screening of patients with type 2 diabetes for diabetes-related distress. The new guidelines suggest that providers screen all of their patients with diabetes using standardized tests for these conditions (40). The strengths of this study include the large study population, the wide variety of measures included (covering medical, sociodemographic, and psychological parameters), and the sevengroup stratification. This allowed for a detailed analysis of ethnic differences in diabetes distress among a cohort of people with type 2 diabetes from various care settings in the Netherlands. Furthermore, we included people with type 2 diabetes from various regions in the Netherlands, adding to the study's external validity. The study also has some limitations. First, ethnicity was estimated from participants' country of birth and the judgement of study nurses. A better estimation of ethnicity could be achieved by adding the country of birth of both parents. Second, a relatively high number of patients had missing data on complication status and were therefore excluded from those specific analyses. Nevertheless, these adjustments had hardly any effect on the observed associations. Third, we included only individuals who were able to read and write Dutch. This meant that we included only assimilated groups of migrants, which may have led to an underestimation of the observed associations. Fourth, we cannot rule out the possibility of participation bias, because people with a language barrier could not adequately fill in the questionnaires. However, eliminating participation bias by providing questionnaires in multiple languages would have increased the number of participants in each ethnic group, which would have increased further the strength of the association. Finally, sample sizes of the various ethnic minority groups were relatively small because of the populationbased design of this study. However, this is, to our knowledge, by far the largest study from the European continent. This study shows that non-Caucasian ethnicity is associated with significantly higher levels of diabetes-related emotional distress in individuals with type 2 diabetes in the Netherlands. This association occurs independent of demographic characteristics, lifestyle factors, glycemic control, medication use, cardiovascular risk factors, diabetes complications, and care setting. A better understanding of the underlying causes of diabetes distress and the specific care needs of distressed patients is called for in order to improve health outcomes.

Diabetes distress among patients from ethnic minorities is still poorly understood. We investigated the association between ethnicity and diabetes distress among ethnic minority groups of people with type 2 diabetes in the Netherlands, focusing on the possible effects of glycemic control, lifestyle factors, cardiovascular risk factors, and diabetes complications.Cross-sectional data from the Dutch Diabetes Pearl cohort included people with type 2 diabetes from primary, secondary, and tertiary diabetes care programs. We used the 20-item Problem Areas in Diabetes Survey (PAID) scale to assess diabetes distress; a score ‡40 is considered to represent high distress. Ethnicity was estimated on the basis of country of birth. Sociodemographic and lifestyle data were self-reported; cardiovascular and metabolic data were retrieved from medical charts. Logistic regression analysis determined the association between ethnicity and diabetes distress, with Caucasians as the reference group.Diabetes distress scores and ethnicity were available for 4,191 people with type 2 diabetes: 3,684 were Caucasian, 83 were Asian, 51 were Moroccan, 92 were African, 134 were Latin American, 46 were Turkish, and 101 were Hindustani-Surinamese. Overall, participants in minority groups had worse health outcomes than those of Caucasian descent, and diabetes distress was more prevalent (ranging from 9.6 to 31.7%, compared with 5.8% among Caucasians), even after adjusting for age, sex, education level, alcohol use, smoking, BMI, lipid profile, HbA 1c , medication use, and the presence of diabetes complications.Among people with type 2 diabetes in the Netherlands, ethnicity is independently associated with high diabetes distress. Further research is warranted to explain the higher prevalence of diabetes distress in minority groups and to develop effective interventions.

INTRODUCTION Internet use increased rapidly in early 2020 when the world faced the COVID-19 pandemic (Feldmann et al., 2020). In Indonesia, the whole population remains at home. Most students are engaged in remote or online learning from their homes via the Internet due to the implementation of social distancing measures to prevent the spread of COVID-19 (Kemdikbud RI, 2020). Generation Z (Gen Z) are people born between 1997 and 2012. In 2019, they were between the ages of 9 and 22 years. They already enjoy a lifestyle with digital communication devices like cell phones, smartphones, iPhones, and social media applications. Their lives are dominated by searching for online information (Dimock, 2019). Gen Z's communication style differs from the other generations (Raslie & Ting, 2021). The nature of Gen Z is unique. They have an innate comfort in enjoying the visual world. Since childhood, they have been accustomed to digital devices and enjoy various conveniences through technology. However, they also have high social and cultural awareness and the ability to innovate according to the challenges of their time (Muliadi, 2020;Wood, 2023). They spend longer on social media than Millennials, Gen X, and Baby Boomers (Pratiwi, 2023a). However, Gen Z also needs to be given special attention because they tend to have worse levels of mental health (Pratiwi, 2023b). There was a study involving 109 Gen Z respondents in Bandung about their experiences in the COVID-19 pandemic. The results of this study reveal that they have much time at home. They use their time to study remotely or online, clean the house, exercise, nap, and perform daily prayers. The rest of the time is used for playing games and interacting with social media. Their remaining time on social media is 3-6 hours a day (Hadita, 2022). The research conducted by the University of Indonesia revealed that the risk of internet addiction among adults in Indonesia increased during the COVID-19 pandemic (Siste et al., 2020). Gen Z can also experience internet addiction, smartphone addiction, and even gaming addiction due to excessive activity using all these devices, as is the case in South Korea. Even in South Korea, they built anti-internet addiction camps specifically to deal with young people who are addicted to the Internet, games, and smartphones (BBC, 2019;Lee et al., 2016). In early 2022, the COVID-19 pandemic subsided, and Indonesia is gradually recovering from the impact of this infectious disease. However, efforts to prevent the transmission of COVID-19 are still being carried out with a comprehensive national immunization program (Aditama, 2022). Gen Z wants to return to normal activities and study as before (Hadita, 2022). What will be the level of utilization of the Internet, smartphones, and social media in 2022, following two years of online studying with the Internet, once COVID-19 is no longer a factor? The study addresses many vital questions: What is the extent of internet usage among Generation Z? Which electronic gadgets do they use for online browsing? What is their level of knowledge regarding internet usage? Lastly, how do they engage with social media platforms? The internet penetration rate in Indonesia is increasing from 2022 to 2023, reaching 78.9% (Azizah, 2022). Cell phone ownership also followed this development (BPS, 2020). In DKI Jakarta, 82.63% of the population accesses the Internet, and 98.55% of users use smartphones (Kemkominfo RI, 2019). excessive internet use hurts Gen Z, especially those who are still teenagers. For this reason, this research aims to describe the behavior of using the Internet, smartphones, and social media among Generation Z in South Jakarta. The results of this research will help understand the behavior of Gen Z. They will be valuable input in efforts to prevent the negative impacts of using the Internet, smartphones, and social media. --- METHODS The design of the study was descriptive. A survey method was employed to collect data, with a sample size of 50 respondents. The respondents are individuals from the Gen Z cohort, specifically aged between 12 and 15 years, residing in South Jakarta. The respondents were selected using a purposive sampling method. Respondents were given the option to participate voluntarily. The variables focused on in this research were internet use, smartphones and social media, and information about how to surf the Internet properly. The survey was conducted in April 2022, when the COVID-19 pandemic began to subside. The research instrument is a question sheet in the form of open and closed questions consisting of 4 groups: Internet use, smartphones, social media, and information on Internet use. Respondents were asked to answer questions according to current conditions. The data analysis performed was a univariate analysis and was presented in the table. --- RESULTS The respondents for this research were 50 people from Gen Z who lived in South Jakarta. Respondents consisted of 27 (54%) girls and 23 (46%) boys aged 12-15 years (Figure 1). Most Gen Z have one smartphone and access the Internet for 1-8 hours a day using wifi; 40% use the Internet for 1-4 hours, and 44% use the Internet for 4-8 hours. Generation Z frequently used WhatsApp and Telegram as means of communication. YouTube, TikTok, and Instagram were their favored social media platforms (Table 1). --- RESEARCH Gen Z likes to be on social media daily. They often share stories and post photos and videos on social media. Social media was a means for them to find entertainment and learning materials. They rarely seek information through social media, and when they get information from social media, they seldom believe it (Table 2). The majority of Gen Z individuals acknowledge their proficient use of the Internet. They also get this information from parents and siblings (98%), teachers (98%), and peers (66%), with the percentage of teachers and parents or family members at home being more significant than that of peers (Table 3). --- DISCUSSIONS Gen Z's life is synonymous with the Internet, smartphones, and social media. The results of this research are facts that prove this. Gen Z spends 1-8 hours enjoying the Internet with their smartphone and interacting online with social media. Research data shows that 40% of Gen Z use 1-4 hours for social media, and 44% use 4-8 hours. They get easy internet access with wifi facilities. We can find almost the same conditions in other places (Maulidina & Ridho, 2020;Restu, 2023). Excessive internet use will risk internet addiction in users and decrease their quality of life (Jeong et al., 2020). This risk will be more significant for children and adolescents (Nurbaiti, 2022a;Kurniasanti et al., 2018). The brain in adolescence is in a rapid development process to reach maturation. Excessive internet activity can disrupt brain development (Nurbaiti & Djauhari, 2021). For this reason, efforts to prevent excessive internet use among Gen Z, many teenagers, need attention (Nurbaiti, 2022b). Smartphones are used by 94% of Gen Z in South Jakarta. Among them, 22% have two or more smartphones. Besides computers, laptops, or tablets, smartphones are commonly used in conjunction with the Internet (Maulidina & Ridho, 2020). Smartphone use among teenagers also triggers smartphone addiction. This device makes it easier for users to immerse themselves in the pleasure of touch (Need for Touch), primarily as more and more social media can be accessed via smartphone (UNAIR News, 2020) WhatsApp is loved by 98% of Gen Z. Their choices are YouTube (42%) and TikTok (28%) as the applications they like. WhatsApp is a unique feature when used in group communication. During the Covid-19 pandemic, WhatsApp groups became an alternative study space. In one group, they can share files or images in the WhatsApp group (Nurbaiti, 2022b;Putri et al., 2022). Gen Z uses social media to communicate and seek entertainment. They like to share stories, photos, and videos via social media. Some of them even use social media to facilitate learning activities. YouTube and TikTok are very popular with Gen Z. Surveys in America note that 1 in 5 teenagers often access these platforms (Restu, 2023). This research also shows that Gen Z in South Jakarta likes YouTube and TikTok. --- ORIGINAL RESEARCH TikTok is a popular choice for Gen Z. This social platform contains short, entertaining videos that accommodate their need for personal expression. Around 60% of TikTok users come from Gen Z (Muliadi, 2020). Gen Z can build positive things using TikTok as a medium for education and social activity (Firamadhina & Krisnani, 2021). Gen Z likes to communicate and seek entertainment through social media, so they are the main target for online advertising. However, Gen Z is not a group that is easily tempted by advertisements because they are aware of what they need (Acoba, 2018;Utamanyu & Darmastuti, 2022). Another fact about Gen Z is that some of them have a consumerist lifestyle. An example is the impulsive purchase of online products (Khairunnisa et al., 2023). Gen Z still needs direct social interaction, especially in the education process. Face-to-face communication and interaction can overcome shortcomings when studying online during the COVID-19 pandemic (Szymkowiak et al., 2021). This research shows that Gen Z correctly obtains information about using the Internet from teachers (98%), parents and family members (98%), and peers (66%). Education on internet use is an essential process for all of us. Many parties can take part in protecting Generation Z so that they can use the Internet wisely. Steps can be taken, including families making rules for internet use at home, teachers socializing good internet usage in the school environment, and the government issuing policies supporting a wise internet environment (Kurniasanti et al., 2019). Research by Zsila and Reyes (2023) has found that social media addiction can negatively impact the mental health of Generation Z individuals. Excessive social media use has been linked to depression, anxiety, loneliness, self-harm, and even suicidal thoughts. Research findings also suggest that passive use of social media, such as scrolling endlessly on TikTok or Instagram, may be associated with decreased well-being over time (Zsila & Reyes, 2023). Excessive social media use is also associated with anxiety, depression, low self-esteem, poor sleep, lack of motivation, and poor social skills (Pantic, 2014). --- CONCLUSIONS Gen Z's Internet, smartphone, and social media use in South Jakarta were similar to the characteristics of other Gen Z currently. Their lives go hand in hand with using the Internet, smartphones, and social media. They usually spend 1-8 hours on social media using smartphones and Internet. They often enjoy using YouTube, TikTok, Instagram, WhatsApp, and Telegram. They post photos, videos, and stories on social media, but they rarely trust the information there. Gen Z obtains information about internet use from the people around them. To prevent the risks of excessive internet use, smartphones, and social media, educational efforts for Gen Z need to continue.

The amount of internet usage has increased following the COVID-19 pandemic in 2020. In order to reduce the transmission of COVID-19, the government implemented a policy mandating remote learning. Generation Z finds learning online at home enjoyable, and there is a growing trend of using the Internet for educational purposes. This study examined the characteristics of Internet, smartphone, and social media usage among Generation Z during the COVID-19 pandemic, which is expected to diminish by early 2022. A descriptive study was conducted, and data was collected through a survey of 50 respondents belonging to the Gen Z demographic, specifically aged 12-15 years, in South Jakarta. The research variables encompassed internet usage, smartphone usage, social media usage, and information regarding internet usage. The data analysis performed univariate analysis. The research results showed that the characteristics of Gen Z in South Jakarta are spending between 1 and 8 hours on social media. Most of them owned and used smartphones. Popular platforms for Gen Z were YouTube, TikTok, WhatsApp, Instagram, and Telegram. They posted photos, videos, and stories on social media but rarely trusted the information there. WhatsApp groups were also used for the learning process. Gen Z got information about internet usage from the people around them. To prevent the risks of excessive use of the Internet, smartphones, and social media, education for Gen Z still needs to be a concern.

Background Following a robust rollout at the end of 2020, the introduction of the various SARS-CoV-2 vaccines was met with enthusiastic adoption in the U.S., and by mid-January, 2021, the basic reproduction number had fallen below 1.0 in over 40 states [1], indicating a slowing of transmission and raising the possibility of eventually attaining herd immunity [2]. Epidemiologists and public health scientists know what happened next: early successes in vaccinating the general population led to declines in COVID-19 incidence and fatalities [3], but, then, perhaps unexpectedly, things hit a wall. By April, 2021, the daily count of vaccine doses administered to the U.S. population, which to that point had been ascending quickly, essentially went over a cliff and, since July, 2021, has risen again only modestly [4]. This hesitancy to get vaccinated, fueled in part by misperceptions reinforced by media coverage, has led to multiple adverse social, political, and public health outcomes. Most notable has --- Open Access *Correspondence: [email protected] Baylor University, One Bear Place # 97236, Waco, TX 76798, USA been the presence of a large enough "space" of unvaccinated individuals to enable highly transmissible variants to adapt, evolve, emerge, and spread throughout the unvaccinated population and into a vaccinated population beginning to experience waning immunity. As of this writing, the U.S. has finally emerged from a healthcare crisis exceeding in magnitude the experience with the first wave in 2020 [5], but once-optimistic talk of herd immunity has mostly been relegated to the back burner [6]. Ongoing surveillance and predictive models suggest that another outbreak of cases may be on the horizon later in 2022 [7]. Throughout 2021 and 2022, the public health sector has been struggling to identify the large pockets of vaccine hesitancy and to answer a simple question: who are these individuals? Anecdotally, it is widely held that SARS-CoV-2 vaccine-hesitant individuals in the U.S. are mostly at the far rightward end of the political and religious spectrum. The provenance of the broader and longer-standing antivaccine movement, however, suggests something more complex [8]. The contemporary spread of the antivaccine ideology is owed in large part to Andrew Wakefield, a now delicensed British physician who famously authored a study suggesting that the MMR vaccine was responsible for autism, a study later exposed as fraudulent [9]. After an investigation, The Lancet retracted the paper, ten of the co-authors disavowed the paper, and Wakefield left the U.K. in disgrace, settling in Austin, Texas, but not before causing what has been called "the most damaging medical hoax of the last 100 years" [10]. Once in the U.S., his discredited findings fueled something of a social movement that soon gained traction on both coasts and in certain urban centers, via adoption by people espousing progressive attitudes toward politics, diet and health, animal rights, environmentalism, and spirituality and who sought nonmedical exemptions for their children's required vaccinations [11]. These views resemble those of the earliest antivaccinators nearly 200 years ago [12]. Within a few years of the autism controversy, due in part to the debunked study's endorsement by several highly visible televangelists, notably the principals of Kenneth Copeland Ministries, the antivaccine movement fanned outward to politically conservative evangelical, Pentecostal, and fundamentalist Christians throughout the U.S. [13], and more recently to Somali Muslim immigrants in Minnesota [14] and ultra-Orthodox Jews in New York [15]. While political and religious opposites, both of these population groups-secular progressives and religious conservatives-seem to be implicated in SARS-CoV-2 vaccine hesitancy, according to media reports. Yet despite these much commented upon observations, there has not yet been empirical confirmation through representative national probability-sample data. This will continue to be a significant issue for the U.S. White House COVID-19 Response Team and its technical advisors helping to craft the nation's strategy to reach the still substantial vaccine-hesitant segment of the population [16], both in the U.S. and globally [17]. Aside from vaccine hesitancy, a related challenge emerged months before a vaccine was made availableindeed almost as soon as the first COVID-19 cases appeared and well before the virus' genome had even been sequenced. That issue has been skepticism or disbelief that the coronavirus is as pathogenic or virulent as advertised, or is even real [18]. Public health professionals are likely familiar with the tales of alleged conspiracies [19]-some perhaps semi-plausible early on (e.g., SARS-CoV-2 is a lab-weaponized pathogen), others quite ridiculous (e.g., COVID-19 is not a viral disease but is due to radiation from 5G towers) [20]. Vaccine hesitancy has spawned its own bizarre conspiracies [21], such as Bill Gates installing microchips or magnetic nanobots in the syringes for purposes of government surveillance, and its own disinformation [22], such as the vaccine not really being a vaccine, or the vaccine causing AIDS. Over two years into the pandemic, at the time of this writing, skepticism and hesitancy are twin impediments that, together, have created enough suspicion to have partly derailed the vaccine rollout and to have contributed to depressed rates of immunization throughout the U.S. [23]. This in turn may present a substantial impediment to finally reducing the incidence of COVID-19 cases to a manageable level, even to moderate endemicity. Vaccine hesitancy has been shown to have historical, political, and sociocultural antecedents [24]. Clusters of unvaccinated individuals exist within every country, even those with high overall rates of immunization [25]. Early in the COVID-19 pandemic, national data indicated that 31.6% of the U.S. population was unsure about receiving a vaccine, and 10.8% stated that they would refuse [26]. These numbers suggest that months before a vaccine had even been developed, herd immunity may have been a pipe dream. Moreover, there are other downstream effects: SARS-CoV-2 vaccine hesitancy has already contributed to sudden and substantial declines in routine childhood vaccinations [27,28], and American and Canadian veterinarians now report an impact on pet owners' hesitancy to vaccinate their dogs and cats [29]. As was noted several years ago, "Determinants of vaccine hesitancy are complex and context-specific-varying across time, place and vaccines" [30]. In other words, the present situation may be sui generis and present a unique set of circumstances with unique antecedents, although, as with vaccine hesitancy in general, political, religious, and socioeconomic determinants are typically observed [31]. The present study seeks to provide some confirmation or clarification as to the identity of those U.S. adults who endorse COVID-19 skepticism and/or SARS-CoV-2 vaccine hesitancy. Based on prior reports of political correlates of misperceptions about COVID-19 [32,33], along with decades of exposés of the antivaccine movement [34], it is hypothesized that skepticism and hesitancy will be observed more among the far ends of the political and religious spectrum-the most politically and religiously conservative Americans, as well as the most politically progressive secularists-with the highest rate of compliance in between these two poles. It is also hypothesized that skepticism and hesitancy will be greatest among those of lower socioeconomic standing in terms of financial resources and education. To test these expectations, data will be analyzed from a newly released national probability survey of the adult population of the U.S. --- Methods This study analyzed data from the sixth round of the Values and Beliefs of the American Public Survey (VBAPS), a stratified random probability sample of 1248 U.S. adults ages 18 and older living in all 50 states and D.C. The survey was conducted by the Gallup Organization in accordance with all relevant guidelines and regulations. Data were collected from January 27 to March 21, 2021, using mail and web surveys (AAPOR1 response rate = 11.3%). Weights were included to adjust the sample to known demographic characteristics of the U.S. adult population by geographic Census region, age, gender, race/ethnicity, and education based on the 2020 Current Population Survey [35]. Many of the measures in the VBAPS are not available in other national surveys, so this is a new and unique source of information on determinants of COVID-19 beliefs and attitudes. Outcome measures for the present analyses were COVID-19 skepticism ("The dangers of the COVID-19 pandemic are exaggerated by mainstream media") and SARS-CoV-2 vaccine hesitancy ("A vaccine for COVID-19 should not be trusted"), with response categories for both ranging from 1 = strongly disagree to 5 = strongly agree. For purposes of this study, binary variables were created for strongly agree or agree (coded 1) compared with strongly disagree, disagree, or neither agree nor disagree (coded 0). Agreement was estimated across categories of several exposure variables, including both political and religious measures. Political variables used in the present analyses were political party identity (coded 1 = strong Democrat to 7 = strong Republican), political orientation (coded 1 = extremely liberal to 7 = extremely conservative), and Presidential voting preference (dummy variables for wanting Biden, Trump, or some other candidate to win). Religious variables assessed Bible beliefs ("The Bible means exactly what it says. It should be taken literally, word-for-word, on all subjects, " compared with three other categories of personal beliefs about the Bible) and belief in God (a series of dummy variables indicating "no doubts, " belief in a "higher power or cosmic force, " and a few response categories collapsed into agnostic and atheist). Percentages of respondents who agreed with the skepticism and hesitancy items were also estimated separately by categories of gender, marital status, race/ethnicity, age, urbanicity, region, education, annual family income, and whether the respondent had "been infected by COVID-19" or had "lost a close relative or friend to COVID-19. " All analyses were conducted using Stata 15. Prevalence rates were estimated separately for COVID-19 skepticism and SARS-CoV-2 vaccine hesitancy by categories of the political and religious variables and by sociodemographic categories, using complete data for each bivariate association. A series of multivariable models was then estimated using binary logistic regression, with multiple imputation for missing cases [36], a standard procedure for epidemiologic and population-health research [37]. Dependent variables were used in the imputation process, but their imputed values were deleted prior to estimating the models. Results are based on five imputed datasets, and were comparable when listwise deletion was employed and when additional imputed datasets were analyzed. The final imputed sample used for the logistic regression analyses contained 1222 respondents. Findings are reported as prevalence odds ratios with associated 95% confidence intervals. --- Results Table 1 shows that COVID-19 skepticism and SARS-CoV-2 vaccine hesitancy are strongly associated with Republican and conservative political preference, in an almost linear fashion (Figs. 1 and2), and similarly with conservative religious beliefs about the Bible and God. The prevalence of skepticism among Trump voters (69.3%) compared to Biden voters (12.0%) is especially pronounced. Sociodemographic determinants are less clear-cut: minimal differences across most categories except for substantially greater skepticism among males, married individuals, non-Black people, the rural population, and those without a graduate degree, and greater vaccine hesitancy among Black and Hispanic people, the rural population, Southerners, individuals with a high school education or less, and those in the lowest categories of family income. For education and income, there are gradients with hesitancy but mostly not with skepticism, except for less skepticism among those with college or a graduate degree. A history of COVID-19, in oneself or a family member or close friend, does not predispose for skepticism, but modestly so for hesitancy. In Table 2, for skepticism, odds ratios for the political and religious variables remained statistically significant (i.e., 95% confidence intervals did not include 1.0) even after adjusting for effects of all of the sociodemographic variables. For vaccine hesitancy, this was true for the political but not the religious indicators. For both outcome variables, significantly higher adjusted odds were observed for Republican party identity and conservative political orientation. In other words, the Did not lose relative or friend 32.9 (938) 9.0 (935) stronger one's affirmation of each of these two political constructs, the greater the odds of skepticism and hesitancy. In additional results (not reported in Table 2), stratifying by whether or not one had been infected with the SARS-CoV-2 virus or had lost a relative or close friend to COVID-19 did not substantively alter the results for skepticism for either political variable. For vaccine hesitancy, having been infected with SARS-CoV-2 modestly reduced the odds due to Republican (OR = 1.3, C.I. = 1.0-1.7) and conservative (OR = 1.3, C.I. = 1.0-1.8) preference; having lost someone close did likewise for Republican (OR = 1.1, C.I. = 0.9-1.4) and conservative (OR = 1.1, C.I. = 0.9-1.5) preference. --- Discussion As seen in these analyses, our hypotheses were half right and half wrong. In these data, COVID-19 skepticism and SARS-CoV-2 vaccine hesitancy do not appear to be phenomena of both poles of the political and religious spectrum, as anticipated, but largely products of identification with the political and religious right. This finding is consistent with results of a recent online study [38], with European data [39], and with a U.S. Census Bureau household survey [40]. The findings for U.S. Presidential voting preference were, frankly, stark, though perhaps not unexpected [41]. The results for education and income suggest that skepticism and hesitancy may not be entirely a matter of lack of knowledge or lack of resources, and thus the solution may not be primarily about more health education or better access to vaccines. The problem instead may be philosophical and ideological and perhaps this is why the unvaccinated have proven so intransigent [42]. Note also that, in absolute numbers, skepticism does not inherently translate into vaccine hesitancy, but prevalences of the latter are still suboptimal for ending the pandemic. Still, the two issues are not as linked as one might have expected. Up to now, exposing the myths inherent in skepticism and hesitancy has been ineffective in countering resistance to immunization. Noncompliance with primary-preventive measures remains "a significant impediment to suppression of SARS-CoV-2 spread" and thus requires more creative approaches [43]. For example, providing evidence of the dangers of communicable disease exposure to unvaccinated individuals, especially vulnerable loved ones such as children, was found to be a better strategy to combat antivaccination attitudes pre-COVID-19 [44]. People are jealous of their beliefs and ideologies, but, one hopes, are more jealous of the wellbeing of their family members. Regardless, as has been observed since early in the vaccine rollout, efforts to address the persistent lacuna of immunization have met with strident pushback, motivated in part by political and religious zealousness [45]. The observation that personal experience with COVID-19, in oneself or a loved one, did not mitigate the effect of politics on skepticism and only barely reduced its greater odds for hesitancy should raise alarms. The expectation that both COVID-19 skepticism and SARS-CoV-2 vaccine hesitancy will fade as more and more people, or those whom they know, fall victim to the disease may not be accurate. Nor are opinion leaders as significant here as might be hoped. One should recall that in August, 2021, speaking at rally of supporters, when former President Trump implored the crowed to get vaccinated he was met with "booing and jeering" [46]. For good reason, many epidemiologists and physicians are pessimistic about immunization coverage ever reaching a level that will end the pandemic through attaining herd immunity [47,48], although so far this sentiment is not unanimous among biomedical scientists [49]. --- Conclusion As noted, efforts to increase immunization in the U.S. through public education may be inadequate; resistance appears ideological, not primarily the result of lack of access to accurate information. Nor is it clear that additional federal expenditures to facilitate increased access to vaccines would be money well spent. While there is an observable prevalence gradient with income, even in the lowest income categories the rate of hesitancy is not much higher than that of the most politically conservative respondents or of Biblical literalists. Moreover, according to our findings, adjusting for the effects of income did not reduce the greater odds due to political preference. Lack of financial resources may not be the overriding barrier to vaccine access here as others have concluded [50], although it is surely a co-factor. Other solutions therefore may need to be considered, including broader government mandates, which at the time of this writing have been implemented in places and been met with widespread political resistance [51], in some instances violent [52]. The alternative is to stand by while new variants have the opportunity to emerge, adding to the increasing fatality count and continuing to overburden a medical care system that has already found itself at the breaking point multiple times during the pandemic. Unless and until the immunization rate increases very substantially-and at present that does not appear likely in the near term-the COVID-19 pandemic may continue to persist until enough people are exposed to and infected by SARS-CoV-2 that they either acquire lasting immunity or are culled from the population in numbers that expand the fatality count far past where it is at present. The ideological roots of the present crisis of skepticism and hesitancy appear to be a downstream legacy of decision-making early in the pandemic that was motivated by political as well as scientific considerations, and the results presented here suggest that the challenges being faced may not be close to resolving. --- Availability of data and materials The data source used for the current study is a national survey conducted by the Gallup Organization. More information on the survey, including contact information and accessibility can be found at https:// www. baylor. edu/ baylo rreli gions urvey/ (Dr. Paul Froese, director). --- Authors' contributions JL was the primary author of the paper. MB was the primary data analyst. Both authors took part in both aspects of the paper. The authors read and approved the final manuscript. --- Authors' information --- Declarations Ethics approval and consent to participate This is a secondary analysis of national survey data from an anonymized data source containing no personal identifiers. Access was provided by the Gallup Organization, the polling organization that conducted the survey, under contract with Baylor University. No experimental or other research protocol was involved, nor were human subjects recruited by the investigators, therefore the study was exempt from human subjects review. All methods were carried out in accordance with relevant guidelines and regulations. --- Consent for publication Not applicable. --- Competing interests None. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

The enduring presence of COVID-19 skepticism and SARS-CoV-2 vaccine hesitancy is an ongoing impediment to the global response effort to the current pandemic. This study seeks to identify determinants of skepticism and vaccine hesitancy in U.S. adults.: Data are from the Values and Beliefs of the American Public Survey, conducted in 2021 by the Gallup Organization in conjunction with Baylor University. The survey used stratified random probability sampling of the U.S. adult population (N = 1222). Outcome measures were respective single items assessing COVID-19 skepticism and SARS-CoV-2 vaccine hesitancy. Exposure variables included political, religious, and sociodemographic indicators, and moderators assessed personal history of COVID-19 and losing a relative or close friend to COVID-19. Results: Skepticism and vaccine hesitancy were strongly associated with conservative and Republican political preference and conservative religious beliefs, and less so with socioeconomic status. Personal experience with COVID-19 did not mitigate the effect of politics on skepticism and barely reduced the odds for hesitancy. Results confirm that attitudes toward COVID-19 are politically and religiously conditioned, and are especially a product of conservative political preference.Skepticism about COVID-19 and hesitancy regarding SARS-CoV-2 vaccination are highest among the political and religious right. Efforts to increase immunization through public education may be inadequate; resistance appears ideological. Other solutions may need to be considered, which risk widespread pushback both politically and religiously motivated.

Introduction Sexual education is a lifelong process of acquiring information on sex and forming attitudes, values and beliefs. It involves sexual development, sexual and reproductive health, interpersonal relationships, affection and intimacy 1. Abstinence-only sexual education teaches the adolescents to abstain from premarital sexual intercourse because of the advantages it offers. Such advantages include prevention of unintended pregnancies and prevention of contracting HIV/AIDS and other sexually transmitted infections. According to the Federal Law of the United States of America (USA), every school-aged child should not engage in sexual activity 2,3. The USA Federal law further advocates sexual activity within the confines of monogamous marital relationship to guard against adverse psychological and physical effects associated with premarital sexual activity. Bearing children outside marriage has serious effects on the child, the mother and the society as a whole [3][4][5][6]. The Federal Law of United States of America also posited that, abstaining from sexual activity outside marriage allows for maturity and understanding of self 3,7. Studies disclose that it is comprehensive sexual education and not abstinence-only that will delay first sexual activity 2,3,[8][9][10][11][12][13][14]. Abstinenceonly sexual education lacks strong evidence of effectiveness because of faulty designs 3,15,16a. Abstinence-only sexual education does not positively affect the sexual behaviour of adolescents, lacks the message of sexually transmitted infections to its recipients and the positive effect in a few cases does not last for a long time 16b,17. According to Bruckner and Bearman 18 and a study by the Alan Guttmacher Institute 19 those that embrace abstinence-only sexual education still have sex before they get married. There is no difference between adolescents that accept abstinence-only sexual education and those who do not in terms of number of sexual partners and ages of first sexual intercourse 17. Abstinence-only sexual education does not reduce the scourge of HIV/AIDS 16b. Bennett and Assefi saw the failure to provide adolescents with information about contraception as a serious weakness of abstinence only sexual-education 20. The proponents of abstinence-only sexual education frown at the role of comprehensive sexual education in emphasising so much on the reliability of contraceptives while de-emphasising their failure rates and the possibility of contracting new sexually transmitted diseases including HIV/AIDS 21. They further frowned at the double message of comprehensive sexual education such as encouraging the delay of first sexual intercourse and promoting the use of contraceptives 22. They are also blamed for stressing the possibility of contracting sexually transmitted infections to the extent of falsifying information to establish the negative aspect of comprehensive sexual education 1,23,24. Studies further reveal that sexual educators do not stress enough on sexual intercourse or bring in sensitive issues such as homosexuality and abortions. The adolescents posited that the basic message is that they should not have sex 19,[25][26][27]. The proponents of comprehensive sexual education attributed the ineffectiveness of condoms and contraceptives to poor-quality research 28. Abstinence-only sexual health education is also blamed for withholding information on the positive aspects of sexual relationships, while magnifying the emotional risks and pitfalls associated with premarital sexual activity 1,24. Modern lifestyle is characterised by a high rate of broken marriages, predisposing an individual to having many sexual partners 1. According to some studies the age of first marriage has risen to 30 years, with a fifth of such marriages ending in divorce within a period of five years 29. Conversely the age at first sexual intercourse has dropped to 16 years with very few people having their husbands as their first sexual partners 30,31. The data on adolescents' sexual behaviour in the developing world with a high prevalence of HIV/AIDS suggests unacceptability of abstinence-only sexual education. Some countries do not accept abstinence-only sexual education even in the face of the high prevalence of HIV/AIDS. Some countries expect sexual educators to encourage adolescents to delay their first sexual experience while also providing education on contraception and sexual health services 32. In some countries, the requirements for teaching any type of sexual education are clearly outlined for the sexual educators 33. Most studies reveal comprehensive sexual education as the preference of parents and adolescents [34][35][36]. Studies advocate comprehensive sexual education for unmarried sexually active adolescents 3. Studies also recommend abstinence-only sexual education with information on contraception and risk-reduction behaviour for the few sexually inactive adolescents. These align with the position of Collins and Priya that parents and adolescents prefer comprehensive sexual education to abstinence-only sexual education 37. Most studies revealed that the opinion of adults differs on the type of sexual education to teach adolescents. Most of the adults feel 7 th to 9 th graders should be taught 'abstinence only' while some advocate the teaching of contraception use 38. Most studies also show that a greater percentage of adolescents prefer sexual orientation that will teach them about the use of contraception and sexually transmitted infections than abstinence-only sexual education 39. Other studies reported the positive outcome of abstinence-only sexual education ranging from reduced sexual activity, pregnancies, abortions to more successful deliveries 40. In line with this, John and Jemmott disclosed the success of abstinence-only sexual education in reducing sexual activity among youths 41. In support of the findings of this new study, Rector submitted that out of 15 scientific evaluations of abstinence-only sexual education, 11 of them demonstrated its effectiveness in reducing adolescent sexual activity 42. --- Amendments from Version 1 We appreciate immensely the efforts of the reviewers of our original manuscript. Their various suggestions have been incorporated accordingly to produce a second version of our work. All pointed out areas such as measures of efficacy of abstinence-only sexual education which were not addressed by the study are put up as suggestions for a follow up study. Conclusion and other areas pointed out have been amended accordingly to rule out any form of confusion. Comparison by religion is ruled out since Christianity is the dominant religion. Some aspects of the work without any negative effect on the entire work are removed to shorten the manuscript to avoid a possible loss of interest by some readers. Some text descriptions are removed where tables are self explanatory. Efforts are made to clarify that this work only examines the perspectives on abstinence-only sexual education which was found by previous studies as an effective for promotion of sexual health. --- See referee reports --- REVISED A survey of the National Campaign to Prevent Teen Pregnancy in 2001 showed that 93% of abstinence sexual education came from the society 43. They concluded that abstinence-only sexual education is the only 100% effective method to prevent teenage pregnancy and sexually transmitted diseases 4. They further reminded that condoms cannot provide 100% protection against unplanned pregnancy and sexually transmitted infections and also that premarital sex can lead to life threatening health problems such as abortion and its associated complications 5. They feel burdened that sexual, contraceptive and HIV information can provoke early sexual initiation among the adolescents 5. Abstinence-only sexual education has positively produced a corresponding decrease in teenage pregnancy 4. Studies have demonstrated that religion acts as a deterrent to early sexual activity 5. In line with this, many adolescents submitted that morals, values and religious beliefs significantly influence the decision to have sex or not. In deciding whether or not to have sex, the Organization of Concerned Women for America also decried the outcome of sex without love or responsibility supported by public policies. This results in the breakdown of nuclear families, increases crime, poverty, teen births and AIDS which in turn negatively affects the health of the general public. This only shows lack of values 5. --- Statement of the problem Adolescents are the future and so they require proper guidance that will propel them into responsible productive adults useful to themselves and their nations. Adolescence is a remarkable period characterised by the quest for experimentations with drugs, alcohol and sexual activity saddled with numerous life threatening adverse effects. The Nigerian Association for the Promotion of Adolescent Health and Development, (NAPAHD) found that, a hospital based research study revealed that 80% of patients with abortion complications in hospitals are adolescents. Studies in Nigeria have also shown that most female adolescents by the age of 15 have already had their first sexual intercourse 6,45. The same applies to male adolescents. In Nigeria, complicated abortion, sexually transmitted infections and HIV/AIDS, sexual coercion, unplanned and unwanted sexual activity and unwanted pregnancies and babies, drop outs from schools and homelessness abound and are very common features with the adolescents 6. Most Nigerian adolescents do not receive correct sexual information while some are ignorant. Hold back their potentials and also affect the nation negatively 6. The intense outcome associated with adolescent sexual activity necessitates the search for a positive way out and thus, the main objective of this study. --- Purpose of the study According to Focus on the Family group, sex should be avoided the same way as the use of guns, tobacco, alcohol and drink-driving 46. They sternly condemn the advocacy for the use of condoms against unwanted pregnancies and sexually transmitted diseases in favour of abstinence-only sexual education 47. Previous studies document the advantages of abstinence-only sexual education in reducing adolescent sexual activities and the associated health problems. This study investigated the perspective of secondary school adolescents in Nigeria on abstinence-only sexual education as an effective tool for promoting adolescent sexual health. The findings will help in planning informed corresponding intervention programmes. --- Material and methods An analytic descriptive survey design was used for the study. The research population comprised of all public secondary schools in three southern geopolitical zones of the Niger Delta Region of Nigeria. The States were Rivers, Akwa Ibom and Cross River. A multistage sampling technique was used for selecting 2020 senior secondary school (SS1-SS3) male and female students from the three states. Proportionate sampling technique was used in selecting the number of participating schools from the metropolis of each state. Five schools were selected from each of Cross River and Akwa Ibom States. Ten schools were selected from Rivers State. Respondents from each school were also selected proportionately. A total of 702 respondents were drawn from Akwa Ibom State, 510 from Cross River State and 808 from Rivers State. Participants were within the age range of 10-19 years. A questionnaire entitled Questionnaire on Nigerian Secondary School Adolescents Perspective on Abstinence-Only Sexual Education (QNSSAPAOSE) was used in eliciting information from respondents. The test/re-test reliability method was used to establish the internal consistency of the instrument. The instrument had a reliability coefficient of 0.75 established with Pearson Product Moment Correlation Coefficient (r). The instrument was divided into sections A and B. Section 'A' sought information on respondents' demographic characteristics. Section 'B' sought information on respondents' perspectives on Abstinence-Only sexual education. Questions were closed and open ended. The YES/NO questions attracted two points for positive responses and one point for negative responses. Questionnaires were administered with the aid of ten trained research assistants. Questionnaires were administered and collected on the spot to enhance a high return rate. Completion of the questionnaire was voluntary. Out of a total of 2020 questionnaires sent out, 2013 were returned and 13 questionnaires were not usable yielding a usable number of 2000 questionnaires. The return rate, therefore, was 99.1% (2000/2020). Hypotheses were formulated and tested. Frequency counts, percentage, Standard Deviation and Pearson Product Moment Correlation were used in analysing data. Approval of individual school management was obtained prior to execution of this study in their schools. Participants' consent was also obtained. Participation was voluntary and anonymity was also assured and maintained. --- Results All of the participants in this study were adolescents between the ages of 10 and 19 years (Table 1). Three research questions and six hypotheses were formulated in this study. Each one of them is addressed as follows: What are the perspectives of adolescents on the advantages of abstinence-only sexual education? Seven positive statements representing the advantages of abstinenceonly sexual education were made (Table 2). The general perspective of Nigerian secondary adolescents on the advantages of abstinence-only sexual education was negative. A greater number of the respondents did not agree with the statements highlighting the advantages of abstinence-only sexual education. However, taking into consideration the number of respondents from each age group that agreed with the statements highlighting the advantages of abstinence-only sexual education, it was found that the youngest age group (11-13 years) of adolescents ranked first. This implies that this group had the highest number of positive respondents in line with their total number when compared with the young (17-19 years) and younger (14-16 years) groups. Age group 11-13 years therefore demonstrated the likelihood of accepting abstinence-only sexual education. What are the perspectives of adolescents on the disadvantages of abstinence-only sexual education? The same number of statements was also made on the disadvantages of abstinence-only sexual education (Table 3). Generally, more respondents agreed with five while disagreeing with the last two of the seven statements highlighting disadvantages of abstinenceonly sexual education. Of the total number of respondents that disagreed with the statements of disadvantages, the young adolescents (17-19 years) ranked first followed by the younger ones (14-16 years). More of the young adolescents out of their total number disagreed with the statements of disadvantages when compared with the responses of the other two groups. This might be a factor of a better understanding than the younger age groups. Moreover, the problem might not be with the type of sexual education but mostly the willingness to practice. What are the perspectives of adolescents on the acceptance of abstinence-only sexual education? The general perspective of the respondents to five statements representing the acceptance of abstinence-only sexual education was negative. More respondents were against abstinence-only sexual education demonstrating a negative perspective on the acceptance of abstinence-only sexual education (Table 4). For instance 1686 out of the total respondents of 2000 objected to wanting a strong abstinenceonly message. The youngest adolescent group (11-13 years) had the highest number of respondents out of their total number of those in favour of abstinence-only sexual education. For instance, more of them when compared with other age groups wanted sex to be saved until marriage. More of them (11-13 years) also wanted a strong abstinence message and education. This indicates that the youngest group of adolescents were more likely to accept abstinence-only sexual education than other age groups. Is there any significant relationship between age and perspective on abstinence-only sexual education? A significant relationship existed between age and the students' perspective on abstinence-only sexual education. (r = 0.123**, N = 2000, P <unk> 0.01) (Table 7). Young age specifically had an influence on the respondents' perspective on abstinence-only sexual education in this study. Null hypothesis is rejected. Is there any significant relationship between religion and perspective on abstinence-only sexual education? Respondents belonged to different religious organisations but only 31 out of the total respondents of 2000 were pagans (Table 6). A greater proportion were Christians which numbered up to 1877 --- Gender and perspective on abstinence-only sexual education? A greater number of male and female adolescents demonstrated a negative perspective to the advantages of abstinence-only sexual education. This is deduced from their responses to the statements reflecting the advantages of abstinence-only sexual education. Out of the total number according to gender that responded in favour of the advantages of abstinence-only sexual education, more females were in favour of three statements while more males were in favour of four statements. More male and female respondents disagreed with five of the statements reflecting the disadvantages of abstinence-only sexual education and agreed with two of the statements. Out of the total number of those that did not agree with the disadvantages of abstinenceonly sexual education, more males responded to four of the statements while more females responded to three of the statements. An outright negative perspective on the acceptance of abstinenceonly sexual education was demonstrated by both male and female adolescents. A greater number of male and female respondents reacted negatively to statements which were in favour of abstinenceonly sexual education. More females responded in favour of acceptance of abstinenceonly sexual education in two statements out of five. More males responded to three of the statements. Specifically, more males than females advocated for sex to be saved until marriage and also wanted a strong abstinence message and education. This study found that boys are more likely to accept abstinence-only sexual Null hypothesis 1: There is no significant relationship between age and their perspective on abstinence-only sexual education. Null hypothesis 2: There is no significant relationship between gender and their perspective on abstinence-only sexuality education. education than females. This finding might be connected with the larger number of males among age 11-13 years group of adolescents with a likelihood to accepting abstinence-only sexual education than other older adolescents (Table 16). There was no significant relationship between gender and perspective of the adolescents on abstinence-only sexual education. (r = 0.051, N = 2000, P <unk> 0.05) (Table 11). Gender had no influence on perspective of abstinence-only sexual education in the study. Null hypothesis is therefore retained. Was there any joint effect of independent variables (age, gender, religion, ethnicity and parent's occupation) on perspective of abstinence-only sexual education? A significant joint effect existed between the independent variables (age, gender, religion, ethnicity and parent's occupation) and perspective on abstinence-only sexual education (F (5, 1994) = 13.085; R = 0.178, R 2 = 0.032, Adj. R 2 = 0.029; P <unk> 0.05) (Table 13). About 3% of the variation was jointly accounted for by the independent variables. The null hypothesis is therefore rejected. Would there be any relative effect of independent variables (age, gender, religion, ethnicity and parent's occupation) on perspective of abstinence-only sexuality education? There is a relative contribution of each of the independent variables on the dependent: age (<unk> = 0.115, P <unk> 0.05), gender (<unk> = 0.042, P > 0.05), religion (<unk> = -0.117, P <unk> 0.05), ethnicity (<unk> = -0.016, P > 0.05) and Parent's occupation (<unk> = 0.021, P > 0.05) (Table 14). Hence, while age and religion made a significant contribution, gender, ethnicity and parent's occupation were not significant. Would there be any significant relationship between perspective of abstinence-only sexual education and age, gender, religion, ethnicity and parent's occupation? A positive significant relationship existed between perspective on abstinence-only sexual education and age, a negative significant relationship between perspective of abstinence-only sexual education and religion but no relationship between perspective of abstinenceonly sexual education and gender, ethnicity and parents' occupation (Table 15). --- Discussion Young age is an important factor in the success of abstinence-only sexual education. According to Massey, three significant periods exist where values are learnt. They are the imprint period, modelling period and socialisation period 48. During the imprint and modelling periods, children learn through instructions and modelling. During these periods, the behaviour of children is formed from instructions given to them and examples before them. Therefore, if abstinence-only sexual education is taught between the imprinting to modelling period, it might produce positive results in adolescent lives. This could be an explanation for the larger number of respondents that opposed abstinence-only sexual education. The major finding of this study shows that Nigerian secondary school adolescents generally have a negative perspective towards abstinence-only sexual education. Out of a total population of 2000 respondents, abstinence-only sexual education was accepted by only 314 respondents and rejected by 1686. Those that had the tendency to accept abstinence-only sexual education were within the age bracket of 11-13 years and are the youngest group of adolescents. The highest number of those that advocated for sex to be saved until marriage and also wanted abstinence-only sexual message and education belonged to the youngest group of adolescents. A greater number of respondents that were not in favour of abstinence-only sexual education belong to the 14-19 years age bracket and are the older adolescents. These are the ones within the age of socialisation already. With their exposure to different views about premarital sex, it is likely that most of them are already sexually active and have already taken a position for premarital sex. The message of abstinence-only sexual education at this stage might not be accepted with ease. Null hypothesis 3: There is no significant relationship between religion and their perspective on abstinence-only sexual education. Null hypothesis 4: There will be no joint effect of independent variables (age, gender, religion, ethnicity and parent's occupation) on perspective of abstinence-only sexuality education. Religion had a negative significant influence on the adolescents' perspective on abstinence-only sexual education with the highest number of respondents being christians. Previous studies revealed that almost no religion supported premarital sexual activities. The studies of Concerned Women for America, demonstrated that religion acts as a deterrent to early sexual activity 5. Many adolescents in previous studies posited that morals, values and religious beliefs significantly influence the decision of whether to have sex [3][4][5]. Probably, those that accepted abstinence-only sexual education would have been influenced by their religious beliefs. This also implies that teaching of religious values could be a useful tool for inculcating the values of sexual intercourse within the confines of marriage. Fortunately 93.9% (N = 1877) of the total respondents were Christians 4.3% (N = 86) were Muslims and 1.6% (N = 31) belonged to traditional religion. Only 0.3% (N = 6) were pagans. Christianity should therefore teach abstinence-only sexual education period. A greater number of male and female Nigerian secondary school adolescents generally have a negative perspective towards abstinenceonly sexual education. Out of the number that favoured abstinence-only sexual education, more females than males were in favour of some statements period, in other statements more males than females were in agreement. Surprisingly, more males than females advocated for sex to be saved until marriage and also wanted abstinenceonly sexual message and education. One would have expected a more positive response from the females than the males because a greater proportion of the respondents in this study were females 64.9% (N = 1298) of the total respondents. This finding is also sad because it is the females that suffer most from the adverse effects of premarital sexual activities. They are the ones that drop out from schools in the advent of pregnancy. They suffer the adverse effects of early pregnancy and child birth or abortion so one would have expected more females than males to advocate for sex to be saved until marriage. At the other hand more males than females advocating for sex to be saved until marriage and also wanting strong abstinence message and education might not be unconnected with the high number of males among the 11-13 years adolescents (Table 16). This study has revealed age 11-13 years as the group with a more likelihood of accepting abstinence-only sexual education than the other two groups of adolescents. Ethnicity did not make any significant contribution to perspective of adolescents on abstinence-only sexual education. This shows that adolescents are all the same anywhere and everywhere. This study also revealed that about 80% of the total respondents could not define sexual education. Probably most of the responses would have been informed by ignorance. --- Conclusion This study therefore concludes that young adolescents within the age bracket of 11-13 years demonstrated the likelihood of being more receptive to abstinence-only sexual education. This finding suggests the need to teach abstinence-only sexual education at an early age by parents who are the first contacts, in religious organizations by sunday school teachers and then by teachers in schools where abstinence-only sexual education is preferred. This will in turn form a part of the values and belief system of the child which might not be easily compromised. --- Translations to health education practice Knowing the perspective of adolescents on abstinence-only sexual education is the key to knowing the right intervention programme to design and the approach to adopt for the implementation of such programme. This study reveals that abstinence-only sexual education will not work for majority of the respondents because it is not acceptable to them. For instance, only 314 out of 2000 respondents agreed with statements reflecting acceptance of abstinence-only sexual education while 1686 respondents disagreed. A very important finding of this study is that abstinence-only sexual education might only impact positively on those within the imprint and modelling periods of development. This includes those within the age range of 11-13 years. Those within the age bracket of 14-19 years seem to have formed their opinion already from different socialisation processes. This finding suggests the need for a second study only for those within the imprint and modelling period. A new study should address the efficacy of abstinence-only sexual education and the willingness by the young adolescents to practice messages of abstinence-only sexual education. The intervention programme based on the findings of this study recommends a two-dimensional approach. Comprehensive sexual education for the older age group (14-19 years) and abstinence only for the younger ones (11-13 years). The implications of this study are directed specifically to parents who are the primary caretakers of the children. They have the first contact with the children during the early years. The religious organisations and the school health educators are also very important role players. Findings from this study suggest the introduction of different programming strategies aimed at teaching abstinenceonly sexual education as a way of life within the early ages. This might help in safeguarding adolescents especially the females from Religious organisations should never relent in teaching the morals and values of abstinence-only sexual education to the youngest group of adolescents with the likelihood to accept. This should be commenced early enough right from their Sunday school classes with negative beliefs and design some remedial programmes for them to reduce the negative influence on other adolescents through the process of socialisation. It can be seen from this study that many problems faced by adolescents as a result of pre-marital sexual activity are avoidable. the numerous life threatening adverse effects associated with premarital sexual activities. Parents must have a sound knowledge of sexual education so that they can serve as effective teachers to the children. This is the reason it is very good also to empower today's adolescents with the knowledge of sexual education and the benefits as future parents. Preparing the adolescents of today to become knowledgeable parents of tomorrow might assist in creating a subsequent future of reduced cases of teenage pregnancies, deliveries, abortions and sexually transmitted infections. Whatever is wrong today can be corrected through adequately prepared adolescents who are the future of any nation. --- Author contributions Obonganyie P Inyang trained research assistants who helped in administering the questionnaire. She coordinated the administration, retrieval and sorting to separate useful copies of questionnaires from those not useful. The title of the work was conceived by Mfrekemfon P Inyang. She planned the research, reviewed the work, did the analysis and wrote the article. --- Competing interests No relevant competing interests were disclosed. --- Grant information The author(s) declared that no grants were involved in supporting this work. --- Open Peer Review Competing Interests: No competing interests were disclosed. I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above. --- Author Response 04 Jun 2013 Mfrekemfon Inyang, University of Port-Harcourt, Rivers State, Nigeria Much thanks to the referee for the comments. In this study the authors only looked at the Nigerian secondary school adolescents' perspective on abstinence-only sexual education. The purpose of this study is also clearly stated. The authors have not indicated anywhere in the study of evaluating the efficacy of abstinence-only sexual education, but only intend to do that in a follow up study. This study is not an intervention study so it could not have shown if abstinence-only sexual education could change attitude. It is always necessary to clarify needs before following up with appropriate intervention. The intent for a follow up intervention study has been indicated. The aspect of religion has been earlier addressed. The statement is a recommendation to educate the adolescents with the knowledge of what is right or wrong early in life. Numerous studies have proven the transforming power of information. It has also been established that most of the wrong-doings of adolescence are either due to wrong information or outright ignorance. --- Thanks. Dr. M. P. Inyang Competing Interests: There are no competing interests. --- Laxmi Baxi Department of Obstetrics and Gynecology, NYU Langone Medical Center, New York, NY, USA The title gives the impression that it refers to the entire population; however, it is true only for the youngest group -further clarification is needed. --- <unk> The majority of individuals in this study are Christian, so comparison by religion does not appear to be adequate. The authors need to emphasise this and may describe such findings in a single statement. --- <unk> The manuscript is too long, as is the introduction. It should be shortened to ensure the reader does not lose interest. --- <unk> Several text descriptions should be omitted as tables are self explanatory, e.g. table 1. <unk> Furthermore, it should be mentioned that the data only refers to a particular group of the population. --- <unk> Competing Interests: No competing interests were disclosed. --- I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable --- Warren Foster --- McMaster University, Hamilton, Canada The authors examined secondary school student perspectives on abstinence-only sexual education through the use of a questionnaire. The sample size is robust, the statistical methods are appropriate, and overall the paper is well written. Unfortunately, the title of the paper is misleading. Specifically, the authors only examined student perspectives and there were no measures of efficacy of abstinence-only sexual education. The data suggests only that students of the youngest age group are more likely to be receptive to abstinence-only sexual education but no evidence is presented on whether the students would or in fact do incorporate these messages into their lives. The authors also suggest that the data shows that there is a significant relationship between religion and student perspective; however, the numbers of students in other religions than Christianity is too small to make meaningful conclusions. Moreover, it is unclear if the students simply identify with a particular religion or actively participate in the religion. Finally, the conclusions reached and the implications for health education practice are overstated. The most that can be concluded from this study is that young adolescents are potentially more receptive to abstinence-only sexual education. It cannot be concluded that abstinence-only sexual education will be more or less effective than no education or alternative education messages in shaping student sexual practices. Competing Interests: No competing interests were disclosed. I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above. --- Comments on this article Version 1 Author Response 29 Apr 2013 Mfrekemfon Inyang, University of Port-Harcourt, Rivers State, Nigeria I appreciate the referee's observation. We did not address measures of efficacy of abstinence in our study. This is why this study suggested a second study for those within the age range of 11-13years who demonstrated potential receptivity to abstinence-sexual education. This can be found under translations to health education practice. Measures of efficacy of abstinence is going to be specifically added. --- Conclusion: The conclusion of the study says that based on the potential receptivity of the young ones to abstinence-only sexual education, then it can work with a timely and early introduction in the lives of young adolescents that will grow up with it. This requires teaching abstinence-only sexual education at an early age as what is right and obtainable by parents who are the first contacts. Children learn through instructions and modelling during the imprint and modelling periods which spans between 11-13years (Massey, 2011). This means working on the potential found in this study to develop along that line. It is not concluded that abstinence-only sexual education will be more or less effective than no education or alternative education messages in shaping student sexual practices. --- Religion: Respondents were asked for religions they were actively involved with. Data analysis showed Christianity as the dominating religion. This explains the emphasis put on Christianity to teach abstinence-sexual health education. However, this was not surprising since Christianity is the dominating religion in the study area. Competing Interests: There are no competing interests. The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias • You can publish traditional articles, null/negative results, case reports, data notes and more • The peer review process is transparent and collaborative • Your article is indexed in PubMed after passing peer review • Dedicated customer support at every stage • For pre-submission enquiries, contact [email protected]

The success of any type of sexual education programme depends on the knowledge and preparedness for practice by adolescents. A recent study has found that an 'abstinence-only' sexual education programme is effective in reducing sexual activity among adolescents. Knowledge of abstinence-only sexual education and preparedness for practice as an effective tool for promotion of sexual health among Nigerian secondary school adolescents was studied. An analytic descriptive survey design was used for the study. The research population comprised of all public secondary schools in three southern geopolitical zones of the Niger Delta Region of Nigeria. A multistage sampling technique was used to select 2020 senior secondary school (SS1-SS3) students as sample for the study. A partially self-designed and partially adapted questionnaire from an 'abstinence-only versus comprehensive sex education' debate, from debatepedia (http://wiki.idebate.org/), entitled 'Questionnaire on Nigerian Secondary School Adolescents' Perspective on Abstinence-Only Sexual Education (QNSSAPAOSE)' was used in eliciting information from respondents. Hypotheses were formulated and tested. Frequency counts, percentage and Pearson Product Moment Correlation were used in analysing data. A greater proportion of secondary school adolescents in this study lacked knowledge of sexual education. About 80% of the respondents could not define sexual education. The general perspective on abstinence-only sexual education was negative, as revealed by the larger number of respondents who demonstrated unwillingness to practice abstinenceonly sexual education. Specifically, of those who responded in favour of abstinence-only sexual education, the youngest group of adolescents (11-13 years) and the male respondents were more likely to accept this type of education than the other groups. Poor1 2 3 version 2 (revision)

Introduction International migration has been growing exponentially. According to the United Nations Global Migration Database [1], the population of international immigrants around the world increased from 153 million in 1990 to 280 million in 2020, suggesting that people are moving to and residing in countries not of their origin. Considering this trend migration is becoming a global issue of concern across the world. There are a few top countries attracting the majority of migrants for settlements, including the United States of America, Germany, Canada, Great Britain and Australia. Australia, while being known as a "country of migrants", was not the destination of non-European immigrants for a long time, due to its White Australian Policy (1901 to 1975). Based on this policy, only people from European countries, mainly from Great Britain, were offered migration to Australia [2]. However, following the change of Australian White Policy and the introduction of the Racial Discrimination Act (1975), the composition of Australian migrant population has changed drastically and immigrants from around the world started to be granted visas to settle in Australia [3]. Today, Australia as an open society with growing economic opportunity is an attractive destination for immigrants from different countries. That is why over 200,000 new migrants select Australia as their permanent destination every year. As a result, 30 percent of the country's 24 million people had been born abroad [4] and almost half of the population was either born in another country or have at least one parent born overseas [5]. This includes thousands of people with refugee backgrounds living in Australia, including people from Afghanistan. A series of tragic events in Afghanistan were the main cause for Afghan refugee to be settled in Australia. The invasion by the Soviet Union in 1979 with civil wars have placed Afghanistan among the top three refugeeproducing countries in the world [6]. Today, there are over 2.6 million Afghan refugees living in Iran and Pakistan, and many more thousands scattered in 70 different countries, including Australia [7]. According to the Australian Bureau of Statistics [8], there were 46,799 Afghanistan-born people living in Australia in 2016, the majority of them arrived in Australia as refugees [9]. The presence of Afghan people in Australia dates back to the 1880s when Afghan people and their camels were brought to Australia to explore the dry central regions and outback of Australia. It is estimated that around 3000 Afghans were involved in camel driving work for over sixty years [10]. They played an important role in Australian transport and made tremendous contributions to the development of Australia. The Existing Ghan Train between Adelaide and Darwin is one of the surviving examples of their contributions. Among recent Afghan arrivals in Australia as refugees, there are many successful stories in terms of adjustments and contributions to Australian society. The following are a few examples of these successful Afghan people in Australia: Dr. Homa Forotan, as a former Afghan refugee, fled Afghanistan with her parents to live as a refugee in Pakistan in the 1990s, and, after a few years, they decided to apply for Australian refugee permanent visas and eventually arrived in Brisbane, Australia, in 2005. Now, she is a cardiologist in Brisbane and believes that the support she received from Australian people and the government helped her to believe in herself [11]. Another example is Yalda Hakim, who arrived in Australia with her parents as Afghan refugees, where she took advantage of her new society to be a successful internationally-known journalist, working with the SBS and BBC. Hussain Sadiqi is also a Hazara-Australian who was the captain of the Afghanistan national team of martial arts. He became a Taliban target and had to flee Afghanistan; he arrived in Australia by a fishing boat with many other refugees in 1999, when he was 18 years old. He spent 6 months in a detention centre and was then granted a temporary protection visa. Today, he is a well-known martial artist actor in the Australian film industry, and he has won several international festival awards for movie fight scenes. However, in spite of these successful examples of integration of some Afghan refugees into Australian society, there are still some serious challenges for others, even after receiving citizenship. In this regard, some evidence shows that the white Australian Immigration Policy still operates in some form within Australian society [12][13][14], upon which, words like "race", "different" and "minority" in public discourse refer to the indigenous Australians, Muslims, Arabs, Asians, and Africans [13]. This means that different intersecting factors like race, religion, gender, and class specify one's status in the "hierarchy of citizenship" in society [15]. This suggests that certain members of society are considered as firstclass citizens, while other citizens are regarded as strangers, due to their different racial characteristics from the majority [15]. So, being second-class citizens will deprive members from an upward social mobility and will create serious challenges for them in terms of their integration process. These challenges have even led to poorer health condition among Afghan refugees compared with the general Australian population [16][17][18]. These issues are real, and this study documents the views of many people from Afghanistan who have gone through the long process, from being refugees to today being Australian citizens, who have been living in Perth for many years. Therefore, this study aims to understand what integration challenges still exist in Australian society for Afghan-Australians and how Australian governments should be aware of these issues and provide relevant support for the Afghan community to overcome these challenges and become valued additions to Australian society. --- Theoretical Framework Migration studies have a long history in sociology. It began with the study of immigration and its consequences in the United States by the Chicago School of Sociology in the second half of the 20th Century. Scholars, such as Robert E. Park and Ernest W. Burgess, developed an assimilationist theory and called it the Melting Pot Theory [19]. This was the dominant theory in migration and remained as such until the 1960s. However, sociologists seriously challenged the dominant assimilationist trajectory of the Chicago School of Sociology [20]. Therefore, as immigration has become a critical issue around the world since the 1980s, this field has witnessed huge developments in research, conceptual frameworks, and theoretical perspectives. This study uses Ager and Strang's [21] successful integration framework. This framework includes ten core domains of successful integration, which are divided into four areas: markers and means (employment, housing, education, and health), social connections (social bridges, social bonds, and social links), facilitators (language and cultural knowledge, safety and stability), and foundation (rights and citizenship). Figure 1 shows the connections within the core components of integration in Ager and Strang's conceptual framework (2008). around the world since the 1980s, this field has witnessed huge developments in research, conceptual frameworks, and theoretical perspectives. This study uses Ager and Strang's [21] successful integration framework. This framework includes ten core domains of successful integration, which are divided into four areas: markers and means (employment, housing, education, and health), social connections (social bridges, social bonds, and social links), facilitators (language and cultural knowledge, safety and stability), and foundation (rights and citizenship). Figure 1 shows the connections within the core components of integration in Ager and Strang's conceptual framework (2008). In this framework, "citizenship and rights", as a necessary foundation, affects the experiences of immigrants in the receiving society. In addition, the "social connection", or as Bourdieu (1986) calls it, the "social capital" of immigrants in the receiving society, occupies an important role in the process of integration. Scholars have divided "social connection" into three categories: "social bonds (with family and co-ethnic groups), social bridges (with other communities) and social links (with the host country institutions structures of the state)" [22]. Furthermore, the authors also consider language and cultural knowledge as a necessary element for successful integration. This could not be achieved without the existence of safety and security in a society. This framework is a very useful tool for the use of survey questions, as well as interviews. There are several recent studies that have applied this model for measuring integration among refugees around the world. Flug and Hussain [23] undertook Ager and Strang's framework to understand the integration experiences of refugees in the UK, and they found that language barriers, employment challenges, and experiencing racism are refugees' major problems. In another study that was in the UK and Japan in 2021 [24], Phillimore and colleagues applied this framework to compare refugees' integration in two research fields. The finding highlighted the importance of social connections in successful integration within both countries. Ziersch and colleagues [25], in another study in Australia, used this model to understand the relationship between integration and social determinants of health. They concluded that challenges of integration can negatively affect refugees' health and well-being. In this framework, "citizenship and rights", as a necessary foundation, affects the experiences of immigrants in the receiving society. In addition, the "social connection", or as Bourdieu (1986) calls it, the "social capital" of immigrants in the receiving society, occupies an important role in the process of integration. Scholars have divided "social connection" into three categories: "social bonds (with family and co-ethnic groups), social bridges (with other communities) and social links (with the host country institutions structures of the state)" [22]. Furthermore, the authors also consider language and cultural knowledge as a necessary element for successful integration. This could not be achieved without the existence of safety and security in a society. This framework is a very useful tool for the use of survey questions, as well as interviews. --- Method --- Study Design and Recruit of Participants There are several recent studies that have applied this model for measuring integration among refugees around the world. Flug and Hussain [23] undertook Ager and Strang's framework to understand the integration experiences of refugees in the UK, and they found that language barriers, employment challenges, and experiencing racism are refugees' major problems. In another study that was in the UK and Japan in 2021 [24], Phillimore and colleagues applied this framework to compare refugees' integration in two research fields. The finding highlighted the importance of social connections in successful integration within both countries. Ziersch and colleagues [25], in another study in Australia, used this model to understand the relationship between integration and social determinants of health. They concluded that challenges of integration can negatively affect refugees' health and well-being. --- Method --- Study Design and Recruit of Participants In this study, our research design included a mixed methodology to study the Afghan community in Perth. For collecting data, we used a questionnaire, individual interviews, and focus groups. We developed close contact with the Afghan community in Perth by attending various religious and cultural events. Data collection took place in the Perth metropolitan area, Western Australia, between September 2020 and May 2021. The population sample in the survey phase included 102 participants of both sexes, who acquired refugee status first and were then granted Australian citizenships after several years. Within the qualitative phase, 13 participants of both genders were individually interviewed and also 13 individuals participated in two gender-based focus groups. The majority of the interviews and focus group discussions were conducted in the Persian (Dari) language, but English was also used in a few interviews. In studying the Afghan community, the first author of this study, started to contact Afghan community leaders, introducing the research project and asking them to introduce him to research participants. Therefore, the researcher started to attend Afghan religious, social and cultural events all across Perth, to recruit participants for the research survey, as well as to invite them to participate in interviews and focus groups. Convenience and snowball sampling were conducted in the quantitative and qualitative data collection phases, respectively. Semi-structured individual interviews were conducted within safe and secure venues, including two campuses of Joondalup and Mt Lawley of Edith Cowan University in Perth. Other interviews and focus groups took place at participants' preferred locations across Perth. Two focus group sessions, with 8 and 5 members from the male and female groups, respectively, were also conducted in Kings Park and Riverton Library in Perth. At the start of the interview and focus group, a consent form was discussed with participants, and they gave their consent, either in writing or orally. This research was conducted according to the National Statement on Ethical Conduct in Human Research, as well as the Edith Cowan University's Conduct of Human Research Ethics Policy (approval number 2020-01617). --- Study Measures In conducting the survey, a short one-page questionnaire in both Persian/Dari and English languages was developed and distributed among respondents by the first author of this study, who is fluent in both Persian and English. The first part of the questionnaire assessed demographic characteristics, including age, sex, education level, employment status, length of stay in Australia, ethnicity, and religion. In addition, the second part of the questionnaire was designed based on Ager and Strang's integration framework, according to which all ten domains of integration were evaluated via ten questions. Subsequently, semi-structured individual interviews and focus group discussions were conducted based on the conceptual framework. In this regard, the researcher introduced himself and explained the aims of the interview and then tried to ask general questions about different domains of integration to encourage participants to share their experiences of integration into Australian society in the format of Ager and Strang's integration framework. In the focus groups, a similar structure was repeated and questions raised by the first author and participants contributed through discussion. All interviews and focus groups were transcribed and read several times to reach data immersion. The first author was responsible for coding, which was subsequently reviewed by other authors. After discovering no new data and achieving data saturation, the process of collecting qualitative data was stopped. Then, the transcribed interviews were analysed using thematic analysis. --- The Participants' Profile There were 102 respondents in the quantitative phase of the study, 54 percent were men and 46 percent were women; almost half of them were born in a country other than Afghanistan, mainly Iran and Pakistan (44% together). Regarding religion and ethnic affiliation, the majority of them are Hazara (65.4%) and Muslim-Shia (80.8%). The average number years of living in Australia for survey respondents, interview participants, and focus group members of this study were 8.5, 7.2, and 9.8 years, respectively. The participants had been living in Australia for an average of 8.5 years. Table 1 shows the demographic information of the participants. The survey demographic information showed that 21.2% of participants were illiterate in their native Dari language, two-third of them were women. Additionally, only 27.3% of those who had a "4-years degree", as well as none of those who had master's degrees, were women. Moreover, the majority of respondents were from the Hazara ethnic group, which is in line with the proportion of total Hazara people in Australia. Data also show that about 51 percent of respondents spoke both English and Persian/Dari during the day, whereas 33.3 percent usually spoke only Persian/Dari. Nevertheless, Persian/Dari was more convenient for 79 percent of them. Within the qualitative phase, 13 participants were recruited for individual interviews. The lengths of the interviews were between 45 and 90 min long. Eight interviews were conducted with men and five with women, one of which was conducted via telephone. Their age range was between 19 and 62, most of them were in their 30s, the majority were first generation immigrants and identified as Muslim. In addition to individual interviews, there were two focus group sessions, one for men with 8 participants and one for women with 5 participants. The average length of the focus group sessions was approximately 2 h. --- Quantitative Findings The descriptive statistics of the main variables in the survey showed that the majority of respondents, 86.9 percent, "definitely" or "most probably" wanted to live in Australia for the rest of their lives, and only 1.6 percent wanted to move to Afghanistan if it becomes a peaceful country. This is not surprising, because the situation in Afghanistan has been unstable over the past four decades; particularly, at the time of data collection, the U.S. declared that it was planning to withdraw its troops from Afghanistan. Therefore, returning home to the country was not an option for participants at the time of responding to this survey. According to the findings, most participants, 56.4 percent, considered both Australia and Afghanistan as their homelands, and 20 percent considered only Afghanistan and 21.8 percent chose Australia as their homeland. Our findings also showed that the respondents were not interested in following Afghanistan's news and only 40.4 percent of Afghans mentioned that they "sometimes" follow the news from Afghanistan media. As many Afghan refugees still suffer from the war-trauma-related issues, it can be a form of "self-care" for participants to hear less bad news from their home country. Table 2 demonstrates the main findings regarding the different domains of integration among the research respondents. Based on the theoretical framework, all the research participants possessed the foundation of integration, because they were all Australian citizens. After several years of living in Australia, they also did not seem to have major issues in some areas, such as housing, education, health, and safety. However, the findings indicated that they had challenges in some integration domains, such as employment and social connections, as well as language knowledge and proficiency. According to the findings, 54.8% of respondents were working, two-third were men. In addition, the majority of unemployed respondents, 83.6%, as well as all those who worked as unpaid volunteers and houseworkers were women. Findings on employment also indicated that the level of education among more than 80% percent of unemployed Afghans was "high school" or lower. In addition to this, the mean of respondents' satisfaction with their current employment was 42.5 out of 100, which is lower than the average. In addition to employment, social connections are another domain of integration in which respondents had the lowest scores compared to other domains. Afghans' social connections are divided into two categories: connections within their community and connections outside their community. Although their connections outside their community were weak, particularly among those who declared having English language barriers, surprisingly, even their social connections within the Afghan community were also weak. Finding shows that the weakest connections with Afghan community were among those who declared that they have "no religion", as well as those who were born and lived in Iran. The majority of Afghan community members believe in Islam, more than 90 percent of the respondents, and their connections usually occur during Afghan religious events and gatherings, where non-religious Afghan members do not participate and consequently miss the opportunity to strengthen their social connections. Ethnic identity is also another important factor in creating inner-group social connections within the Afghan community, but these ethnic divisions do not exist for the Hazara people who came from Iran. In the area of language proficiency, the majority of respondents who had the lowest level of English proficiency were Afghan women, as well as those with lower education levels. Based on traditional Afghan gender roles, women are usually responsible for managing domestic chores, while men work outside. Due to this fact, women's opportu-nities for improving their English language are limited. This is also true of participants with lower education levels who cannot improve their English proficiency by developing their education. --- Qualitative Findings Undertaking a bottom-up approach in thematic analysis, the researchers started with transcriptions, going through the data and noting similarities in the interviews. In this regard, dozens of simple codes were created and, in following step, they were grouped together to find patterns in the data. To do this, researchers drew a map of codes to understand the relationship between them in order to create themes. Finally, three main themes emerged from the qualitative data, indicating the main challenges for former Afghan refugees in Australia: employment barriers, socio-cultural barriers, and challenges within the Afghan community in Perth. These challenges can be categorised under integration markers and means, as well as social connection (within and outside the Afghan community). Figure 2 shows the themes. managing domestic chores, while men work outside. Due to this fact, women's opportunities for improving their English language are limited. This is also true of participants with lower education levels who cannot improve their English proficiency by developing their education. --- Qualitative Findings Undertaking a bottom-up approach in thematic analysis, the researchers started with transcriptions, going through the data and noting similarities in the interviews. In this regard, dozens of simple codes were created and, in following step, they were grouped together to find patterns in the data. To do this, researchers drew a map of codes to understand the relationship between them in order to create themes. Finally, three main themes emerged from the qualitative data, indicating the main challenges for former Afghan refugees in Australia: employment barriers, socio-cultural barriers, and challenges within the Afghan community in Perth. These challenges can be categorised under integration markers and means, as well as social connection (within and outside the Afghan community). Figure 2 shows the themes. --- Employment Barriers For many participants, finding a job that is related to their skills and qualifications was very difficult. A part of this difficulty relates to problems with recognition of overseas qualifications in Australia. Sadiq is a participant who came to Australia as a refugee in 2015 and was awarded an Australian citizenship recently. He holds a bachelor's degree in chemistry from Kabul University and worked as a high school teacher in Afghanistan, but Australia does not recognize his qualification. In this case, he has to enroll in university courses again, or change his field of study. --- "They don't recognise my qualifications here. So, I have to study 8 years if I want to be a teacher [in Australia], but I don't have time. I have to work and make money to support my family in Afghanistan. That's why I'm working in construction field now. " In this regard, several studies have shown that a significant relationship exist between employment-related difficulties and recognition of qualifications in Australia [26,27]. As a result of this barrier, the process of employment is a highly challenging factor --- Employment Barriers --- Socio-Cultural Barriers --- Challenges --- Within Afghan Community --- Employment Barriers For many participants, finding a job that is related to their skills and qualifications was very difficult. A part of this difficulty relates to problems with recognition of overseas qualifications in Australia. Sadiq is a participant who came to Australia as a refugee in 2015 and was awarded an Australian citizenship recently. He holds a bachelor's degree in chemistry from Kabul University and worked as a high school teacher in Afghanistan, but Australia does not recognize his qualification. In this case, he has to enroll in university courses again, or change his field of study. "They don't recognise my qualifications here. So, I have to study 8 years if I want to be a teacher [in Australia], but I don't have time. I have to work and make money to support my family in Afghanistan. That's why I'm working in construction field now." In this regard, several studies have shown that a significant relationship exist between employment-related difficulties and recognition of qualifications in Australia [26,27]. As a result of this barrier, the process of employment is a highly challenging factor for many Afghan immigrants, and not all of them are able to either negotiate their situation or afford to continue tertiary studies in Australia. Reza is a refugee; when he arrived in Australia as a refugee in 2015, he had a Bachelor's degree in sociology from Afghanistan, along with a few years of work experience with an international organization in Afghanistan. In spite of his university degree and work experience, as well as his good command in English proficiency, he was not able to find a job during his first year living in Australia. He discussed two main reasons for his failure and said: "they didn't recognise my qualifications and also all jobs require local experience". Even after graduating from an Australian university, he ended up working in a field that was not his first choice.... so, I had to go to university again but how? I needed the IELTS certificate, as well as paying the tuition fees... Finally, I graduated from a university, but I still wasn't able to find a job in my favourite field [International Security Studies], so, I had to shift to the field of Social Work and finally could find a job after a couple of years. Discrimination in the process of the finding a job, as well as in the workplace was another challenge that many Afghan participants face in Australia, regardless of their citizenship status. According to Dion [28], immigrants may experience discrimination because of the place they came from or they may be seen by others through their skin colour, language, or religion. Zari, as a Hazara woman, is an example. She is in her 20s and mentions that she has been facing many barriers in finding a job, because of her hijab. "I haven't been able to find a job mainly because of my Hijab. Even some employers have said this to me directly. My uncle is an owner of a business in Perth, but even he doesn't hire me for my Hijab... That's why I have to look for a job only in Afghan community." However, discrimination in the workplace is not merely limited to the hijab. Sanam, another female participant, who does not wear Hijab but still faces the same barriers. She introduces herself as a modern lady who respects the modern values of Australian society and has limited connections with the Afghan community, mainly "because of their traditional mindset". Nevertheless, she has several stories about discrimination in career promotions. Here is one of them: "I had an Australian employer that was telling my colleague that if she [the participant] was paler, I would use her in the other section [which was more important]. I heard it myself... You are a foreigner in the workplace, no matter how much you try to be similar to them." Nevertheless, some participants view this discrimination in a wider perspective. A participant who had been living in Australia for almost 30 years believes that discrimination in the workplace is not limited to only to people with an Afghan background, but it is extended to all non-white people in Australia in general. Naser shared: "When you look at the Australian government, from parliament to governmental organizations, the absolute majority of employees are those with White background. So, non-white people can only pursue their employment in non-governmental sections. That's why you see lots of businesses in shopping centres with Asian or African owners. In fact, the multicultural Australia is only represented in shopping centres not in government or public organizations." Participants also discussed more issues and challenges within the employment area, such as language barriers, lack of job security, and exploitation in the labour market. While participants discussed challenges, they also discussed some of their plans and strategies that they will employ to secure their employment in the future. Some of the changes include changing their names to Anglo-Australian names. They were also very serious to further enhance their English proficiency as a key factor in securing relevant employment and building a proper career. --- Socio-Cultural Barriers Former Afghan refugees have been facing several challenges in the context of socioeconomic aspects of their lives in Australia, one of which is the problem of revealing their identity as Muslim-Afghans in Australia. Due to negative feedback that most of participants received from Australian society about Afghanistan, mainly after 11 September 2001, they were uncomfortable with expressing their Afghan identity while communicating with people outside of the Afghan community. Ali and Saeed told their stories about hiding their Afghan background since they had received discriminatory reactions. They share: "At first, sometimes people would ask me: where are you from? And I would say: I am from Afghanistan. Then they would say: Oh, Taliban. Or they'd say do you know Osama Bin laden? So, I realised that I don't have to tell them the truth. Since then, whenever somebody asks me where are you from? I say Tajikistan or Uzbekistan. (Ali, 39 years old)" "My younger brother is a student at [... ] University. One day I was working with him in a construction project and the man that we were working for asked my brother what do you study? My brother replied: piloting. Then the man said: "oh. Okay, you plan for hijacking". It really made me sad and for a couple of days and decided not to tell people my nationality anymore. (Saeed, 41 years old)" As a result, after the tragic events of 11 September, Muslim immigrants have been targets of discrimination in Western countries and this has led them to hide their identities. This particularly makes them feel that they are seen as "others" in these societies. In addition to Muslim-Afghan identity, some participants were not comfortable with their former refugee identity. This is mainly because Afghan refugees in Australia are known as "boat people", which refers to the way in which they came to Australia. However, there is a generational difference between participants in accepting their refugee background. Younger participants tend not to talk much about their refugee status in the past, while older participants easily talk about being refugees sometimes in the past. These older participants also complained about the way that Australian people and media view refugees. During las decade, Australian media negatively focused on Afghan refugees who arrived in Australia by boats and called them "boat people". Sara is in her 60s and has been living in Australia for over 30 years. She had a pharmacy in Afghanistan in the 1970s, but had to flee Afghanistan as a result of the Soviet Union invasion in the 1980s. She shared: "It's always been so difficult for me to explain people that we were fleeing from violence and war. We didn't want to leave our country voluntary, but we had to do that. People here don't know anything about war, violence... they just blame refugees for coming to Australia... if they know a little bit more about refugees, their attitude will be changed." Another social barrier that was mentioned by many participants is that they are not able to broaden their social network to include other Australian people. Several studies have also shown that there are challenges for young immigrants in developing their social network within the Australian multicultural community [29,30]. This is even true about participants that have been trying to be assimilate into their new society. Sanam and Zari shared: "No one can understand from my appearance that I am Afghan or Asian. No matter how you dress like them or how good your language is, people believe that you have come from a foreign country... Finding an Australian friend has been so difficult for me. (Sanam, 24 years old)" "I can't find any non-Afghan friend here, whenever I try to communicate with a guy, after a while, either I find them different or they do. That's why I've stopped looking for a friend outside the Afghan community." (Zari, 28 years old) Some participants also revealed some challenges that they faced with their children who are the second-generation Afghans. First generation Afghan refugees would like to maintain their culture, religion, and language, while their children are reluctant to accept these values. The gap between the two generations has become bigger, in the case of parents who have not spent enough time helping their children to learn about their culture and language. "My son is completely different from me. I speak Dari with him but he replies in English. But for me, the most important thing is religion. I am not so strict about language... I tell him if you speak two languages, that would be an advantage, but if you don't believe in Islam, that's a shame." (Naser, 51 years old) "I have several Afghan friends who are really hardworking... they haven't taken the time to help their children learn to speak Farsi. Now, their children speak English with them and they can't understand them. I think it is a really sad story, because your kid is the most valuable thing that you have in the world. (Saeed, 41 years old)" --- Challenges within Afghan Community In addition to facing the above barriers, former Afghan refugees face some challenges within their own community in Perth. One of these challenges to is related to the fact that Afghanistan is a multiethnic society and peoples have been involved in ethnic divisions for a long time. Unsurprisingly, these divisions have affected the Afghan diaspora in Perth as well. Thus, there is no single Afghan community in Perth, rather there are several. Each ethnic group has its own members in terms of organising social or cultural events. Mehdi, who has been living in Perth since 1992, has witnessed the changes that happened in the Afghan community over time. "In those first years the number of Afghans in Perth was low; ethnicity was not so important and there was a single community. But after arriving more Afghan refugees in the following years, and especially after establishment of Taliban and deteriorating the ethnic divisions in Afghanistan, the situation in Afghan community in Perth changed. So, every ethnic group started to build a community for itself. Today, there are several of them which are independently active." These divisions are even strengthened by religious differences that exist within Afghan ethic groups. Afghan Hazaras are mostly Muslim-Shia, while other groups like Pashtuns and Tajiks are Muslim-Sunni. Thus, this has created some challenges for those former Afghan refugees who care about their religious belief. Naser shared: "We [Shia] used to have a problem when someone from our community died, because we didn't have a place to wash a corpse before burying and we had to ask [Sunnis] to use their place. But they sometimes refused, because they don't respect the way that we wash a corpse, and would say it is not recognised in Islam." Another challenge that several participants mentioned is that most Afghan community events and gatherings are religion-based, which is due to the mindset of Afghan community leaders. This factor may prevent people with different belief systems from being attracted to that community. Saeed is one of these people, who shared: "Almost all events are about religious things. I tried to add some other programs, like Persian Poetry reading program for some passionate people that I knew, but those narrow-minded people didn't allow us to continue our program. They would suggest run a Quran class instead.... I even tried to run some events independently, for example I managed to celebrate the Women's day last year, but those leaders have a huge influence among Afghans, and they'd advised people not to join the program and finally we run the program with only a few women." Many Afghan refugees, before arriving in Australia, lived in neighbouring countries of Pakistan and Iran, for many years. Some of them even were born in these countries and have never been in Afghanistan. This has added more complexity to the Afghan community in Perth. Thus, in addition to ethnic/religious divisions, there is another classification division based on the country that Afghan refugees have lived in before coming to Australia. Hamed is a 34-year-old Afghan-Australian who was born in Iran. Checking his Facebook profile before conducting an interview, the researcher realised that he is a big fan of football, and surprisingly there was the flag of Iran in his cover photo. He has never been in Afghanistan and is not interested in Afghanistan issues. Instead, he tries to talk more about Iran than Afghanistan during his interview. He shared: "I am not much in contact with Afghan community here, we are not similar, to be honest. I rather have a few Iranian friends and that's enough... Those Afghans who've come from Pakistan also have their own connections with Pakistani people, and it is true about those who have come from Afghanistan directly." Moreover, there have been some changes within Afghan families after migrating to Australia that created some new challenges, specifically for Afghan women. These challenges mainly are caused by the culture of masculinity, as well as by the changing of the gender roles in the family. In Afghanistan, a man/husband is recognised as a bread winner and decision maker within the family and all family members usually follow his orders. In addition, there is a traditional definition of gender roles in which a man is responsible for work outside, while woman is expected to manage domestic chores. However, after migrating to a country like Australia, which does not recognise these definitions, the dominant position of father, as well as other male gender roles within the Afghan family has been gradually changing. In this regard, there are still issues in relation to power distributions within the family. For example, husband is expected to make decisions on almost every aspect of family affairs and activities, including controlling their income and their relationships outside of family and even father can make decision on to whom his daughter should marry. Since these are in contrast with the values of a modern society like Australia, it may lead to tension within the family. Zari is an Afghan-Australian who has

This paper explores, analyses, and documents the experiences of Afghan-Australians who arrived in Australia as refugees and were granted citizenship after living in Australia for several years. This research adopted a mixed method of qualitative and quantitative approaches and surveyed 102 people, interviewed 13 participants, and conducted two focus-groups within its research design. Analysis of data indicates that former Afghan refugees gradually settled down and integrated within Australian society. They value safety and security, open democracy and orderly society of Australia, as well as accessing to education and healthcare services and opportunity for social mobility. However, since the integration is a long process, they are also facing some challenges in this area. Findings of this study show that Afghan-Australians require more support from Australian governments to overcome some of these challenges particularly securing employment within their area of interests and professional occupations that they have qualifications and experiences from Afghanistan. They are also experiencing broader challenges in the area of socio-cultural issues within Australian society. Since the Afghan community is an emerging community in Western Australia, they require more support from local government to enhance their ethnic cohesion and solidarity.

, there have been some changes within Afghan families after migrating to Australia that created some new challenges, specifically for Afghan women. These challenges mainly are caused by the culture of masculinity, as well as by the changing of the gender roles in the family. In Afghanistan, a man/husband is recognised as a bread winner and decision maker within the family and all family members usually follow his orders. In addition, there is a traditional definition of gender roles in which a man is responsible for work outside, while woman is expected to manage domestic chores. However, after migrating to a country like Australia, which does not recognise these definitions, the dominant position of father, as well as other male gender roles within the Afghan family has been gradually changing. In this regard, there are still issues in relation to power distributions within the family. For example, husband is expected to make decisions on almost every aspect of family affairs and activities, including controlling their income and their relationships outside of family and even father can make decision on to whom his daughter should marry. Since these are in contrast with the values of a modern society like Australia, it may lead to tension within the family. Zari is an Afghan-Australian who has been living in Australia for almost 8 years. She shared: "A few years ago, I started to work at [... ], but all my income was controlled by my husband. I even didn't know that how much is my income and how is being spent in our life. But after a while I started to complaining and he was responding by limiting my freedom. Since then, I couldn't go to my friends' gatherings." In addition to financial issues, there are other challenges that Afghan women are facing and dealing with. Female participants shared stories about their experiences and other Afghan women that they know who have had similar experiences. Zari shared: "A friend of mine who is a young lady, has a strict Afghan father who imposes lots of restrictions on her relationship with others, particularly with boys. So, what she does is that she goes out with her boyfriend from 11 pm to 6 am, when her father is asleep....my other friend has been forced to marry with a relative who lives in Afghanistan; they've never met each other." --- Discussion Adjusting to a new society is not an easy journey for everyone, particularly for refugees who may have not been ready for this change. However, the host society can assist refugees to have a successful integration into their new society. The participants of this research, who are former refugees, have already been provided with support by the Australian government, which means that they have the same rights as other Australians. That is why the majority of the participants consider Australia as their homeland and about 90 percent of them would like to live in Australia for their rest of life. However, this study showed that while former Afghan refugees enjoy some aspects of Australian society, at the same time they face some serious challenges in their new society. Challenges include employment, socio-cultural atmosphere of society, as well as socio-religious barriers within the Afghan community in Perth, Australia. This is evidence showing that, while they are legally Australian citizens, they are not fully accepted as members of society. This is in line with Lazarus's research that "some citizens more equal than others" [15]. Nevertheless, according to the quantitative findings, only 1.6 percent of Afghan participants want to return to Afghanistan, which means that, in spite of these challenges, they still prefer to live in Australia. Now that Afghanistan has been taken over by Taliban as of 15 August, 2021, the situation of Afghan refugees around the world has changed drastically; according to the UNHCR, over 550,000 Afghans have been forced to flee their homes and seek refuge in other countries. Therefore, returning to Afghanistan is not an option for Afghans living in Australia at the present time. The experience of four decades of war in the participants' home country is still alive in Afghans' memories. All these have generated poor health conditions, as well as psychological issues for them, including concerns about safety and post-traumatic stress disorder [31,32]. This was even obvious in interviews with participants, where most of them reported losing someone from their family, or friends in the war in Afghanistan. That is why, according to our quantitative findings, safety and stability were perceived to be the highest score among the domains of integration. This is in line with research that was conducted by Abedin [33] in which safety was one of the main reasons affecting Afghan women's sense of belonging in Swedish society. Nevertheless, considering the traumas they have experienced, as well as possible health problems, many Afghan-Australians need to be supported in terms of mental health and psychological services, especially by services that are culturally sensitive. However, at the top of the list in our research, employment stands as the biggest challenge for Afghan-Australians. As one study notes, "employment is the most common means through which people can express their creativity and find ways to fully participate in society" [34] (p. 61). However, as some studies have shown, there exists a large inequality between the earning and income of migrants as compared to the rest of the population in Australia [35,36]. According to the 2016 Census, the average weekly income for Afghanistan-born people is $371, while the median weekly income for Australian-born residents is $688, which represents a significant difference [8]. The census also reported a rate of unemployment among those who were born in Afghanistan of 17.8 percent, which is far from the 6.9 percent unemployment rate for all Australians. Several studies have discussed the challenges that immigrants have been facing in Australia, indicating the prevalence of overt and covert discrimination in the Australian labor market (e.g., [18][19][20][21]31]). It seems that COVID-19 has worsened the situation for migrants in Australia, which is beyond the scope of this paper. In addition, socio-cultural challenges reported by participants can impact their integration within Australian society. A big part of these challenges is related to the discrimination that Afghan-Australians perceive from society, in which they are often identified like "terrorist militants", and not as victims of terrorism. Nevertheless, as is discussed in this paper, these barriers can be, on the one hand, structural barriers like racism, discrimination, and lack of social capital in the host society or, on the other hand, ethno-religious factors, such as parental disapproval of participation in particular events or because of different definitions of gender and social roles [22]. As a result, while there are structural barriers within Australian society that narrows migrant social networks, in many cases an immigrant's ethno-religious beliefs also do not allow them to easily communicate with other people outside their cultural boundaries. An interesting finding in this paper is related to the challenges that former Afghan refugees face within their Afghan community in Perth, Australia. According to Daley [37] refugees' community support can play a vital role in the development of social bonds, bridges, and links that are key indicators of integration in the host society. However, ethnic and religious tensions in the Afghan community deprived members of having a unified community in Perth. This is not limited to the Afghan community in Perth, however, as Abraham and Busbridge [38], in their research, reported the existence of similar tensions among the Afghan community in Melbourne as well. This lack of cohesion in the Afghan community not only limits social interactions for community members with the wider community in Australia, but also prevents them from having a unified voice to represent community needs to the Australian government in order to receive more support. --- Conclusions During the recent migration of Afghan people to Australia, despite the fact that migration has been a complex and difficult change for them, in this journey, Afghan people have been successful in establishing thriving communities across Australia. This research discussed distinctive characteristics of the Afghan community in Perth, including their contributions and connections to Australia, as well major challenges. Afghan refugees have found Australia to be their new home and the overwhelming majority of them consider Australia as their permanent country of residence and only 1.6 percent of the participants indicated that they may return to Afghanistan if it is safe in their replies. However, in spite of the Australian government's support in granting many Afghan refugees citizenship, they still cannot enjoy the full advantages of citizenship rights in practice. The Afghan community in Perth, with the strong commitment to make Australia as their permanent home, strongly needs and deserves to receive the Australian government's support. In addition, the inclusion of Afghan people within the broader Australian society facilitates the integration process and the media can play a vital role in developing the local community's awareness about refugees, their backgrounds, as well as their potential contributions to Australian society. This social inclusion is particularly crucial for further enhancing the quality of the Australian multicultural society. To do this, conducting more research on some aspects of Afghans' lives in Australia, particularly in the context of employment, as well as their relationship within the Afghan community, would be helpful. However, in this difficult time, where the Taliban have returned to power in Afghanistan, and crisis after crisis has been happening in Afghanistan, the Afghan community in Australia also has been experiencing a very difficult time along with high anxiety and so many facing unwelcome news; it is time that the Australian government provide urgent counselling services with cultural sensitivity to those who desperately need them. These services should be part of an "ethnic particular service", and not a mainstream service. Afghans are highly concerned about the future of their country, the lives of their family and relatives in Afghanistan, as well as uncertainty of many Afghan refugees with temporary protection visas in Australia. While at the present time, the Australian government has allocated 3000 out of its 13,750 annual humanitarian visas to those fleeing Afghanistan, the government can do more. Therefore, in line with over 300 Australian organizations, businesses, and community groups that signed an online letter, demanding from Australian government to assist the Afghan community, recommendations are listed as follow: 1. Assist Afghan people who are at grave risk in Afghanistan, particularly those who have worked with the Australian government by facilitating evacuation. --- 2. Dedicating more humanitarian visas to Afghan refugees to protect lives at risk. These can be announced as special refugee visas. --- 3. As millions of people in Afghanistan have recently become displaced, many are at risk of hunger and lack of shelter. Thus, the Australian government can increase its aid to the region, for example, by supporting organisations that assist people in Afghanistan. --- 4. Extend permanent protection to over 4000 Afghan refugees on temporary protection visas in Australia, as it is not safe for them to return to Afghanistan. 5. Assist Afghan-Australians with urgent family reunion visas, particularly for those who are at greater risk, including ethnic minorities and relatives of those who have worked with the Australian government in Afghanistan. 6. Taking more Afghan refugees, the Australian government can take advantage of this opportunity to provide workforces to Australian regional areas that are in desperate need of workers, mainly as a result of closed borders from the COVID-19 pandemic. --- Data Availability Statement: Not applicable. --- Conflicts of Interest: The authors declare no conflict of interest.

This paper explores, analyses, and documents the experiences of Afghan-Australians who arrived in Australia as refugees and were granted citizenship after living in Australia for several years. This research adopted a mixed method of qualitative and quantitative approaches and surveyed 102 people, interviewed 13 participants, and conducted two focus-groups within its research design. Analysis of data indicates that former Afghan refugees gradually settled down and integrated within Australian society. They value safety and security, open democracy and orderly society of Australia, as well as accessing to education and healthcare services and opportunity for social mobility. However, since the integration is a long process, they are also facing some challenges in this area. Findings of this study show that Afghan-Australians require more support from Australian governments to overcome some of these challenges particularly securing employment within their area of interests and professional occupations that they have qualifications and experiences from Afghanistan. They are also experiencing broader challenges in the area of socio-cultural issues within Australian society. Since the Afghan community is an emerging community in Western Australia, they require more support from local government to enhance their ethnic cohesion and solidarity.

Introduction Despite advances in the field of infertility medicine and its availability, disparities are said to exist in the utilization of infertility services globally ( [1]; Ombelet [2]; Shapiro et al. [3]; de Mouzon et al. [4]; [5]). In the USA, racial/ethnic disparities are reported to constitute a major influence on both the utilization and outcome of infertility services (Quinn and Fujimoto [6]; Dieke et al. [7]; [8]). This is regardless of the success story of the first US conceived infant born in 1981, which was achieved through assisted reproductive technology (Centers for Disease Control and Prevention [9,10]). An improved form of fertility treatments in which male and female gametes or embryos are processed in the laboratory (commonly an in vitro fertilization [IVF] and related procedures) to achieve procreation. The use of ART and the number of fertility clinics providing ART services continue to surge across the country; sadly, racial disparity continues to militate against equal access and delivery (Adamson et al. (2006, as cited in Reefhuis et al. ( 2008)) [11]. According to the CDC [9,10], access to these infertility services is an "emerging public health priority," for which, in 1995, "CDC began collecting data on ART procedures performed in fertility clinics in the United States as mandated by the Fertility Clinic Success Rate and Certification Act of 1992 (FCSRCA) (Public Law 102-493 [ October 24, 1992])." These reports are obtained from 52 reporting areas (the 50 states, Puerto Rico, and the District of Columbia) [9,10]. Reflecting of total information on reports from "present multiple measures for successful ART intervention, (including the percentage of ART procedures and transfers that resulted in pregnancies), live-birth deliveries, singleton live-birth deliveries, and multiple live-birth deliveries." Although ART helps millions of American achieve pregnancy, the reported trend shows states and racial disparities with the latest in 2016, revealing a total of 197,706 ART procedures (range: 162 in Wyoming to 24,030 in California); an intent to transfer at least one embryo was performed in 463 US fertility clinics and reported to CDC [9,10]. These procedures resulted in 65,964 live birth deliveries (range: 57 in Puerto Rico to 8638 in California) and 76,892 infants born (range: 74 in Alaska to 9885 in California). Nationally, the number of ART procedures performed per 1 million women of reproductive age (15-44 years), a proxy measure of the ART use rate, was 3075. ART use rates exceeded the national rate in 14 reporting states/areas (Connecticut, Delaware, the District of Columbia, Hawaii, Illinois, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Utah, and Virginia). ART use exceeded 1.5 times the national rate in nine states, including three (Illinois, Massachusetts, and New Jersey) that also had comprehensive mandated health insurance coverage for ART procedures (i.e., coverage for at least four oocyte retrievals)," [9,10]. Several types of research have affirmed this position ([12] cited by [3]) with individuals of racial/ethnic minority groups, generally demonstrating poorer ART utilization and health outcomes than non-Hispanic white patients. Humphries et al. [13] in a systemic review concluded that current evidence points to race and ethnicity as a strong predictor of poorer clinical pregnancy rates and/or outcomes after ART treatment. This was from a review of twenty-four studies, where five (5) US registry-based studies showed that black, Hispanic, and Asian women had a lower ART procedure success when compared to non-Hispanic white racial women. These differences also persist even when adjusting for disease (infertility) severity, age, insurance coverage, marital status, income, and educational status (Nelson (2002; as cited in [3])). Race/ethnicity, therefore, seems to be an important predictor of ART utilization and outcome. However, which race-associated factors are responsible for these differences in ART utilization as well as outcomes has remained controversial in many studies. Some authors have suggested genetic predisposition difference in the distribution of FMR1 gene mutations [3] and socioeconomic and cultural differences (Seifer et al., (2010) (Wellons, et al. (2002), Seifer et al. [14], and Feinberg et al. [15]). Prevention of reproductive health disparities affecting utilization of ART requires "monitoring race-specific infertility prevalence, treatment patterns, and related risk factors to identify, guide, implement and monitor effective public health action strategies to safeguard reproductive health" [9,10] (World Health Organization [16]). Given the differential utilization or outcome of a specific service in healthcare delivery or practice like ART, the service suggests a barrier to infertility care which in turn populates reproductive health disparities among underserved populations [17]. Then, there is the need to establish/ elucidate this association and inform literature, thus improving practice and utilization in the USA as well as globally. This study seeks to address this need by analyzing a secondary dataset using a cross-sectional design. The objective is to study the association of maternal race on the use of ART (IVF) in the USA in a nationally representative sample, as reported in the 2017 CDC Natality Public Use file, with a view to describe this association. --- Methods --- Data Secondary datasets were obtained from the Centers for Disease Control and Prevention 2017 Natality Public Use file. A crosssectional observation of the racial distribution of ART use among a total US population of 325,719,178 of all races and ethnic origin was made for the reported year. A total of 67,554 responded to questionnaire question on ART, and 3,864,754 birth record counts were reported in the review by the 50 states, including the district of Columbia and US Territories. A total of 48 states and the District of Columbia (excluding South Carolina and Tennessee) reported information on the type of infertility treatment used, representing 96.4 percent of 2017 births. This cross-sectional study acquired sociodemographic information and clinical/obstetrics and gynecological history maternal race (explanatory variables) as well as infertility treatment utilization ART (outcome variable) from women living within this reporting States and US territories. --- Participants A secondary dataset of 67,554 respondents (who had responded to the questionnaire question section on ART use) of the women (3,864,754 reported birth records) who fell within the reporting states and US territories constituted participants for the study. Women who received natal care outside the reporting coverage were not reported in the file and thus were ineligible for the study. Male partners were also excluded to eliminate duplication of responses. --- Study variables The independent variable for the study was the race, and this was obtained from a recorded reported mother's race of a seven categories consisting of non-Hispanic White (only), non-Hispanic Black (only), non-Hispanic American Indians and Alaska Native (AIAN) only, non-Hispanic Asian (only), non-Hispanic Native Hawaiian or Other Pacific Islander (NHOPI) only, non-Hispanic (Mixed race), and Hispanic and also two ethnic-based subcategories of non-Hispanic (all races) and Hispanic statuses. The dependent variable was the use of ART entailing primarily in vitro fertilization (IVF); other methods include gamete intrafallopian transfer (GIFT), and zygote intrafallopian transfer (ZIFT). This outcome (binary) of interest was obtained from the self-reported use of ART as noted from the report, noted as the reporting flag. Several other variables were added during the analysis of Race and ART use to adjust for potential covariate and confounding effects. These cofounders include sociodemographic variables (mother's age (under 35 years and 35 to 54 years), previous cesarean (yes/no), fertility enhancing drug (yes/no), payment method (Medicaid, private insurance, self-pay, others), and educational levels) and obstetric and gynecological/clinical variables (gonorrhea infection (yes, no), syphilis infection (yes, no), chlamydia infection (yes, no), pre-pregnancy diabetes (yes/no), previous cesarean (yes/ no), and fertility enhancing drug (yes/no). These final confounding variables for the adjusted models were selected after performing a 2-stage process. First, a series of statistical tests were performed to assess frequency distribution and then checked for characteristics of a relationship with ART use. Confounders that were significantly (p <unk> 0.05) associated with either ART use were considered in the next stage, involving binary logistic regression analysis for association correlation. The list of potential confounders was further reduced using multivariate logistic regression modeling and a backward elimination stepwise regression approach, keeping only those variables that were significant at p <unk> 0.05. Giving the large sample size (67,554 ART use respondents of the total race population from all races N = 325,719,178, live birth record n = 3,864,754, and having 48 out of 50 reporting state), statistical analysis was powerful enough to evaluate the association between variables. --- Statistical analysis This was carried out using IBM SPSS Statistics for Windows, version 26 (IBM Corp., Armonk, N.Y., USA). Tests included descriptive statistics, Chi-square tests for categorical variables, and binary logistic regression analysis. Both unadjusted (OR) and adjusted (AOR) odds ratios with 95% confidence intervals were calculated to approximate associations with variables having a p value of less than 0.01. OR and AOR were extrapolated and checked for significance and variation between unadjusted and adjusted. --- Results --- Descriptive analysis From the dataset, 3,864,754 live births were reported out of a total US population of 325,719,178 of all races and origin. A total number of 42,846 women reported use of ART treatment out of 67,554 respondents (study sample). Table 1 demonstrates Gonorrheal, syphilis and chlamydia infections were seen to have a low prevalence of 0.3%, 0.1%, 1.8% respectively; Prepregnancy Diabetes (0.9 %). Bivariate analysis using the Chisquare test as displayed in Table 2 showed a statistically significant association between each of the explanatory variables and the outcome variable. Vis-a-vis maternal race, mother's age, marital status, gestation, payment method, educational status, fertility enhancing drug, previous cesarean section, and smoking status all have a p value of 0.01 each, respectively. Gonorrheal (0.43), syphilis (0.31), and chlamydia infections (0.93) were seen not to have significant p values; Also, was Pre-pregnancy Diabetes having a p value of 0.04. --- Regression analysis --- Unadjusted regression Table 3 presents the odds of having an ART treatment based on the independent variables and confounders examined (vis- The unadjusted odds ratios at 95 % CI for the association between maternal race and ART use was 87% higher among non-Hispanic Asian (only) when compared to the non-Hispanic white race (only). It was 141% for mothers of age category (35 to 54 years), 11% for the unmarried, 36% for those with a college degree, and 144 % for those who smoke. The unadjusted odds ratios at 95% CI for the association between maternal race and ART use was lower for non-Hispanic AIAN only (11% less likely), non-Hispanic NHOPI only (40% less likely), and Hispanic (11% less likely). However, non-Hispanic (mixed race) 1.00 (0.57-1.75) 95% CI and non-Hispanic Black only 1.05 (0.99-1.06) 95% CI had an approximately equal association (ART use) when compared to the non-Hispanic white (only). --- Adjusted regression When adjusted for the confounding variables, the OR for this relationship was of the higher side for many of the variables. Unmarried marital status shows a 0.07 (1.11-1.04) OR decrease, while all the following shows an increment OR at 95% CI: previous cesarean 0.11 (0.97-0.86) and payment method (self-pay 0.04 (0.73-0.69) and other 0.31 (0.96-0.65), the exception here being private insurance, having a 0.05 decrease (0.97-0.92) on adjustment). --- Discussion The disparity in access to healthcare continues to pose a challenge to equitable healthcare delivery. If this is seen reflected in ART treatment delivery, its dose calls for concern as it further deepened an already established inequality. Although a zero absence of barrier in a healthcare delivery setting may be practically non-realizable, the global concern is equal access to health for all, irrespective of sociodemographic background. According to [18]; as cited in [19]), the fertility rate in the USA has decreased to the lowest point, as at the first quarter of 2016 having 59.8 babies for every 1000 women of ages 15 to 44, a near half the rate of the baby boom peak in the 1950s. Hamilton et al. [20] in their second annual series vital statistics rapid release statistic carried out on US birth data of 2017 (based on a 99.93 % of birth record by maternal age and race/Hispanic origin) report a downward 2% ( 3,853,472 total births) from that of 2016, representing the lowest number in 30 years. This federal data statistics trend depicts an unmistakable sequela of women delaying motherhood mainly because of sociodemographic reasons (maternal race/ethnicity, age, educational status, marital status, income, etc.) and associated clinical/obstetric and gynecological implications. Bellieni [19] explains the paradox seen in western countries where fertility is socially discouraged by a mindset to exploit later in life, on expectations of gains of delayed parenthood, and the uninformed understanding of the associated risk of sterility provocation (infertility) as well as the greater gap between population. Our findings from pieces of literature and data analysis brought to the discussion, the association between our explanatory variables and ART use in the USA, as well as bringing to limelight the necessity to Xray the disproportionately high impact of racial/ethnic sociodemographic influence/implications on members of already underserved racial/ethnic minorities (e.g., Blacks, AIAN, Hispanic/Latino, NHOPI), underserved individuals of all racial/ethnic subgroup, as well as dwellers of rural America or communities. Right now, the future does not look promising for an improved equal maternal access to ART utilization and outcome. The prognosis as of now is even made worse, following the sudden new inclusion of COVID-19 (SARS-CoV-2) to the list, of it having a possible impact on fertility and assisted reproductive technologies [21]. All of this represent and constitute the palpable necessity, pressure, and demand for an available and accessible ART service and utilization for all women medically requiring it, regardless of maternal race or ethnicity. The data presented in this study suggest that maternal race is strongly associated with an increased likelihood of having an ART treatment. This was also obvious when controlled for age, marital status, education, payment method, smoking status, and previous cesarean. Non-Hispanic White women are relatively higher than those of non-Hispanic Black, non-Hispanic AIAN, non-Hispanic NHOPI, and Hispanic racial/ ethnic statuses in utilizing ART. This finding supports the result of other work. Humphries et al. [13] cited that race and ethnicity, especially Blacks, were strong predictors of poorer outcomes on ART treatment. A racial and ethnic difference in pregnancy rates following intrauterine insemination (ART) was also observed to be lower among American AIAN (prevalence PPR 0.60, 95% CI 0.43-0.83) and Black (PPR 0.53, 95% CI 0.44-0.63) compared to White women in a study by Craig et al. [22] and Janitz et al. [17]. Also, a reduced ART service utilization was also observed for Hispanic women compared to non-Hispanic women as observed by Janitz et al. [17] in a binary racial/ethnic perspective. Dimitriadis et al. [23] concluded that race affects timely access to infertility care, diagnosis, and treatment cycle, having some racial groups waiting significantly longer prior to seeking evaluation. That of [24]; as cited in [3]) stated that some disparities are attributable by some degree to differences in access to care, economic, educational, as well as cultural factors. However, Shapiro et al. [3] concluded on two major findings: that utilization of ART is inversely related to fertility rates in the general population and that the lower utilization of ART among Black and Hispanic women is mainly due to lower demand rather than barriers to access. Dieke et al. [7] in a running commentary put forward that ART utilization was higher for women in states with an insurance mandate for ART treatment. Also stating that, the Asian or PI had the highest utilization in 2014, while the AIAN women having the lowest utilization rate. Nonetheless, it was concluded that though insurance mandates may increase access to infertility treatments, they are not sufficient to eliminate these disparities. Our findings suggest a possible likelihood of confounders accounting for or constitute a compounded barrier to ART u t i l i z a t i o n, m o s t l y t h r o u g h a c o m b i n a t i o n o f sociodemographic status and obstetrics and gynecological/ clinical conditions (explanatory variables) affecting access or indication for infertility treatment. Sociodemographic barriers likely heighten this phenomenon among those minority women who desire fertility services but are unable to utilize them due to lack of access, affordability, or social acceptance in their communities. Also, obstetrics and gynecological/ clinical conditions like maternal medical conditions (e.g., diabetes mellitus, hypertension), addictions (e.g., drugs/ smoking status), and/or previous surgery (e.g., cesarean section) may account for indications or contraindications for ART utilization. Likewise, one who is infected with an infectious disease like COVID 19, Zikka virus, Human immunodeficiency virus (HIV), Syphilis, Gonorrhea, Hepatitis, Chlamydia, etc., may present as a barrier. [25]; as cited in [3]) put forward that a decline in fertility rates in the general population was observed after the year 2007, the period following the last economic recession. The declines in fertility rates were pronounced among underserved minorities, which has previously been attributed to the disproportionate effects of the recession on incomes of these minorities. On the other hand, Martin et al. (2015; as cited in [19]) added that fertility rates were seen to increase among women in their 30s and 40s, over the past three decades, as a result of more women delaying pregnancy to older reproductive ages. This is aside the increase in the number of same-sex couples and single individuals seeking parenthood, contributing to increased use of ART treatment generally and in particular among older women. According to Bellieni [19], scientific works of literature continue to be crystal clear on recommendations for a maternal less risky age range of childbearing to be between 20 and 30 years, regardless of the perceived benefit of delayed motherhood. However, the US data on delayed motherhood reflected adherence militating factors against this recommendation which summarily present as a strong and constant socioeconomic pressure on women. Multiple prior studies reported that success rates of ART varied among different racial/ethnic groups [26], as cited in [3]). Thus, to knowledge, this study is the first to look at the association between a maternal race with a view to providing useful dimensions to counseling and ART treatment for an individual patient. This may help mitigate the various barriers that may be unevenly confounding to maternal race. The data used, therefore, add further credence to the previous hypothesis that the race/ethnicity of the patient may be a truly independent predictor of use and success with infertility treatment. We had also felt a sense of relief that in our little contribution to existing works of literature. We had, while finding an association (between maternal race as a predictor variable and ART use as outcome variable), primarily only looked at only one gender (female) and only one factor (female factor), but in reality, this has holistically touched on the primordial medically defined infertility treatment associated factors for any couple (above 50%). Considering that, the following as shown on rmanetwork.com represent contributory makeup factors in the assessment of a patient with an infertility diagnosismalefactors (35 %), female factors (35 % contributory), a combination of male and female causation factors (20 %), and unexplained factors (10%). This study benefited from a nationwide report by ART clinics to SART of racial data in a significant percentage (96.2%) of ART use. However, a respondent population of 67,554 out of 3,864,754 live births on ART use posed a limitation to external validity. This is in conformity to the work of Quinn and Fujimoto [6] that assert that only a minority of the US population has access to treatment. Lack of external validity was found also to be a limitation in some other ART disparity studies with some having a low report of 39.6% [3]. The result of this study highlights the importance of reporting race when inputting demographic data of patients. This usually reduces the burden of accounting for missing data values and is therefore likely to be a major source of bias in this study. Maternal age, payment method, and educational status observed as strong confounders observed in this study should be addressed in future studies. Again, as an addendum, and of interest, future studies may also see the need to factor in the impact of the ongoing COVID-19 pandemic on reproductive health/ART utilization in America and globally. Missing data values, in the study, were accounted for and clean up using standard allocated numbers. Likewise, confounding variables not meeting significance were eliminated additional analysis in the study. Thus, fertility treatment used, gonorrhea, syphilis, chlamydia, and pre-pregnancy diabetes covariates were not included in the regression analysis. --- Conclusion The study finds the utilization of ART in the USA to be associated with maternal race, suggesting that the odds are lower for racial/ethnic minorities to ART utilization. Future research in this area may, therefore, help to provide improved ART treatment access, irrespective of maternal race. --- Code availability CDC Code book User guide to the 2017 Natality Public use file was accessed from https://ftp.cdc.gov/pub/Health_ Statistics/NCHS/Dataset_Documentation/DVS/natality/UserGuide2017. pdf. --- Authors' contributions Both authors have contributed equally to the work. Data availability Data was access from the CDC free access file in SPSS format. Available from https://www.cdc.gov/nchs/data_access/ vitalstatsonline.htm and https://ftp.cdc.gov/pub/Health_Statistics/NCHS/ Dataset_Documentation/DVS/natality/UserGuide2017.pdf. --- Declarations Ethics approval Not applicable. --- Consent to participate Not applicable. Consent for publication Both authors consent to publication. --- Conflict of interest The authors declare no competing interests. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Despite advances in the field of infertility medicine and its availability, disparities affect the accessibility status worldwide. Racial disparities could potentially affect the utilization of assisted reproductive technology (ART). We aimed at studying the association between maternal race and the use of ART treatment in the USA. We analyzed a secondary dataset (2017 Natality) obtained from the Centers for Disease Control and Prevention. This cross-sectional study acquired information on maternal race as well as ART utilization from women living within the reporting States and US territories. We analyzed the data using descriptive, bivariate, and regression analysis. A total of 3,864,754 live births out of 325,719,178 US races and origin populations were reported for the 2017 review year. A total of 42,846 women who had a live birth reported utilization of ART out of 67,554 respondents. The Chi-square test showed a statistically significant association between maternal race and the use of ART treatment, p value 0.01. Unadjusted regression odds of the utilization of ART at 95 % confidence interval (CI) was 87% higher among non-Hispanic Asian women as compared to the non-Hispanic White. We also found higher odds for maternal age 35-54 years 2.41 (95% CI 2.34-2.49), maternal education (above college degree) 1.36 (95% CI 1.31-1.42), and non-smoking status 2.44 (95% CI 2.02-2.94). Compared to the non-Hispanic white race, the adjusted regression odds were lower for all other racial/ ethnic minorities except for the non-Hispanic Asian 63% (95% CI 1.09-2.44) and non-Hispanic mixed race 59% (95% CI 0.81-3.10) subgroups. The study finds the utilization of ART in the USA to be associated with maternal race.Association . Maternal race . Assisted reproductive technology . USA This article is part of the Topical Collection on Medicine

Background Growing patterns of chronic disease have emerged in all populations that have adopted modern Western sedentary lifestyles and there is evidence that sufficient levels of physical activity can help prevent, manage and treat these diseases [1]. There is however, little understanding of how physical activity can be successfully reintegrated into daily lives in a rapidly changing environment characterised by labour saving technology, low physical energy requirements and the ready availability of energy dense convenience foods. Historically, health initiatives in Western developed countries have generally attempted to address the negative impacts of sedentarism by targeting individual behaviours through general practice interventions, social marketing campaigns and public health policy. Western biomedical models have generally conceptualised physical activity as a separate health related exercise requiring a minimum of thirty minutes of moderate activity to be inserted into an individual's routine on most days of the week. As the complexity of the determinants of physical activity emerge and there is acknowledgement that the driving forces lie mostly beyond the control of the individual and outside the health sector; social, environmental and economic factors are receiving more attention [2]. The creation of 'active living' communities through multi-level, environmental interventions and policy has more recently been recognised as a promising strategy for urban and suburban settings [3]. To be relevant, meaningful and more likely to be effective in specific populations these strategies need to consider and understand social influences and the different ways various cultures conceptualise activities and health [3,4]. Nationally, Indigenous Australians suffer a disproportionate burden of chronic disease compared to non-Indigenous Australians. Rates of circulatory disease are ten times those of the non-Indigenous population [5]. Obesity and physical inactivity are among the leading contributors to this burden [6]. It is estimated that in remote Northern Territory (NT) communities 77% of the 17 year difference in life expectancy between Indigenous and non-Indigenous populations is due to preventable chronic disease [7]. As little as fifty years ago, these remote Indigenous NT societies were subsistence-based and lived in an environment that required maximum energy expenditure for minimal energy density intake and had to conserve physical energy to survive [8]. Although there is no accurate information about current physical activity levels in remote Indigenous populations, the recent and sudden exposure to a sedentary lifestyle and the accompanying elevated burden of diabetes and obesity, suggest that being physically active is possibly even more important in the prevention of chronic disease in this population than it is amongst others [9]. However, the drastically lower levels of health and well-being in the Indigenous population are also associated with a range of social factors stemming from colonisation such as: racism, dispossession, social isolation, poverty, inadequate housing, unemployment and lack of appropriate services, as well as higher rates of smoking, alcohol abuse, poor diet and inactivity [7,9]. Similar health and social issues are experienced by colonised Indigenous people worldwide [10]. There are only a few examples of documented physical activity related initiatives in remote Indigenous communities and little evaluation of their outcomes to learn from [2,9]. The success of Indigenous health promotion strategies however, is known to be influenced by culturally specific perceptions of health and the social norms arising from them [11][12][13]. The available literature highlights the importance of making links between Indigenous culture and physical activity for initiatives to be effective in Indigenous settings and that these links involve "innovative ways of combining approaches to improve cultural pride, cultural identity and self-esteem with physical activity promotion" [9], p.40]. The literature also suggests that the ineffectiveness of interventions to modify high risk behaviours in Indigenous populations is likely to be due to a lack of understanding of the wider sets of social meanings attached to those behaviours and of the everyday lives of Indigenous Australians [14,15]. Cardiovascular health and lifestyle programs have also been shown to be more successful and sustainable when they involve Indigenous community members, local knowledge and leadership [16][17][18][19]. Overall, the need to work in partnership and to establish cultural links with physical activity has been shown to be important in Indigenous contexts but there is an absence of how these links and partnerships are made from an Indigenous perspective. Collaborating to identify the Indigenous meanings and lived experience of physical activity in a remote context has been acknowledged as a priority in health promotion. This research aimed to provide some perspectives on physical activity as seen by Indigenous people living remotely in the NT. --- Methods --- Study design This qualitative study drew upon phenomenological ethnography and was guided by participatory action research and community development principles. Community-based co-researchers were employed and worked together with the first author (a Darwin based migrant Australian of British descent), in the design, facilitation and implementation of the field work. --- Study population and study sites There are approximately 82 remote communities in the NT. These are spread out over an area of approximately 1,400,000 square kilometres. Recently these communities were grouped into 8 regional shires. An email seeking expressions of interest about participating in the research project was initially sent to all NT regional shire managers, all community clinics and as many community council offices and organisations as possible. Representatives from approximately 7 communities responded and were followed up with phone calls or visits to ascertain the level of interest. During a period of closer consultation with two of the remote community areas the people expressed concern about chronic disease and physical inactivity in their community and saw the research project as an opportunity to improve the situation. An agreement to participate in the project was established and support letters were signed by representatives of the community councils, organisations and traditional owners. Situated nearly 2000 km apart, one study site is in the tropical Top End and the other in the arid Central Desert region of southern NT. The Top End research setting has an estimated Indigenous population of 1300 and is approximately 400 km from the nearest town. The Central Desert setting has an estimated population of around 800 people and is approximately 800 km from the nearest town. Both communities are isolated and only accessible by light plane for months at a time every wet season. English is a third or fourth language to both groups. Five Bininj (the people of the top end community) were employed as community-based co-researchers in the Top End community. Attempts to recruit Yapa (the people of the desert community) co-researchers in the Central Desert community were unsuccessful. This was mainly due to intra-community conflict and unrest and the subsequent temporary departure of many people from the community preceding and during the field work period. Yapa and non-Indigenous service providers assisted with recruitment and facilitation of interviews in this community. Like most remote NT communities these settlements have a unique history, with a population made up of traditional owners and combinations of various clan and language groups who are not all originally from that region. Extensive and complex Indigenous legal systems, ancestral connections, kinship structures and skin groups (moieties) define identities, relationships, roles, social standing and power [20]. Various traditional and contemporary laws, customs, beliefs and spiritual traditions together with those brought in by Europeans and technology are woven into the fabric of everyday life in the settlements. Typical to most community settlements, those in this study cover a small geographical area and are surrounded by vast expanses of undeveloped vegetation and wildlife. They feature: very basic and often inadequate infrastructure and services, including a council office, school, clinic, an aged care facility and shop; mostly dirt and some sealed roads; overcrowded houses; running water; power (often supplied by diesel generators or solar systems); some shade structures; a small number of cars; a basic outdoor basketball court; football ground; playground and recently in a few communities -a pool. --- Data collection and analysis Purposive and opportunistic methods were used to recruit participants over a five month period with the aim of interviewing at least 30 community members or until data saturation. We sought a balance of genders and a range of ages over 16 in our study group. Available male and female community members and co-researchers over the age of 16 were invited in person by the researchers or service providers to be interviewed. Semi-structured, in-depth interviews were held in informal locations chosen or deemed suitable by co-researchers, participants or service providers. Trialling unstructured interviews enabled broad guiding topics to be formulated which were then used to semi-structure the discussions. Interviews were recorded with informed consent using digital voice recorders. Those held in local language were interpreted by bilingual co-researchers, although most were held in English. Five of the participants contributed commissioned paintings representing perspectives and meanings of physical activity. Observations of community infrastructure, physical activity and social and cultural events were recorded in a journal by the first author. Interviews were transcribed externally and reviewed for accuracy. Thematic analysis was used to organise and interpret the material. Persistent themes recurring across the interviews and paintings were identified, grouped and contextualised in relation to each other and to current social, cultural, Indigenous and public health literature. The interpretation and findings were compiled into a booklet and poster using quotes and images of the themes. These visual mediums were used by the co-researchers and available participants to read over, reflect upon and communicate with the first researcher about the results. In this way the thematic interpretation of the data was discussed and then confirmed as representative of the views of those involved in the study. --- Ethical approval The study received ethics clearance through the Human Research Ethics Committees of both Northern and Central Australia. Ethical issues related to this project involved inadvertent researcher dominance, inappropriateness or disrespect as a result of lack of awareness of and insensitivity to, cultural differences. The possibility of this occurring was limited as much as possible by adhering to the Cultural Respect Framework and the principles for ethical conduct in Aboriginal and Torres Strait Islander Research as outlined by the National Health and Medical Research Council (NHMRC) -reciprocity, respect, equality, responsibility, survival and protection and spirit and integrity [21]. The project was also informed by the NHMRC Road Map priorities, principles and processes, and tried to ensure: a holistic view of health; community involvement in the development and conduct of the research; enhancing skill development around research methods and health promotion by employing community-based research assistants; and producing practical outcomes relevant to Indigenous people and their service providers [22]. The researcher sought guidance from community-based Indigenous organisations, community members, communitybased research assistants and traditional owners regarding matters of cultural protocol and sensitivity, and attempted to maintain continued awareness of issues around; communication style differences; language barriers; informed consent and appropriate feedback. --- Results All participants from the Top End community (number =17, male = 9, female =8) were aged between 26 and 65 years. Due to time and distance limitations, community issues and staff recruitment difficulties in the Central Desert, only females (number =6) already involved in some form of organised physical activity were recruited. This included two younger participants aged between 16 and 25 years. The total number of respondents was 23. Findings from the study were fed back to community members, participants and stakeholders. In this paper the findings are contextualised by some observations from the journal of the first author and organised around three key themes: i) the economic meanings of physical activity; ii) the social organisation of physical activity; and iii) physical activity on traditional country. Participants requested to be referred to as Bininj or Yapa and individuals have been given pseudonyms. --- Setting the scene: some observations Even with the presence of cars and the heat in tropical and desert communities, people of all ages often go about their business on foot in small groups, mainly in cooler parts of the day, at a slow pace, walking on the roads and well worn dirt paths. Despite the service providers who participated in this project reporting a preference by community members for travel by car even over short distances, the lack of supply and maintenance of cars and licenses, and the cost of fuel, means that many people also walk when getting around in the community. At various times throughout the year residents go "out bush", into the nearby and vast areas outside the community to collect resources, hunt food, participate in ceremony, swim, or visit traditional country, other communities and distant towns. Team sports like football, basketball and softball are popular among the young and small league competitions are found in many remote communities. However, it is hard to find examples of ongoing programs capable of increasing activity levels of the majority of adults who do not play sport and the many at risk of developing or already suffering from chronic disease. The visible physical activity promotion that does occur in these communities happens indirectly through school sports and annual community festival activities; and directly through visiting health professionals, and where available, communitybased public health staff [23,24]. These are often short term, one-off or irregular programs driven by people from outside the community or by those who only stay a short time and thereby are often based on non-Indigenous cultural perceptions of physical activity and health [23]. --- Work and 'walkabout': the economic meanings of physical activity During interviews it was apparent that engaging in physical activity just to improve health did not factor highly if at all, in the day-to-day lives of the participants. Rather for most Bininj and Yapa participants there were strong associations between physical activity, the land and work. Some Bininj men even defined the meaning of 'physical activity' as most closely related to that of the Kunwinjku word 'durrkmirri'; meaning, 'to work'. When describing physical activities participants referred mainly in their stories to those with specific purpose related to the consumption and distribution of natural and cultural resources including the acquisition of food, collecting wood for fires, fibres and dyes for weaving and painting, the building of shelters and the passing on of knowledge through walking, ceremony and dancing. Walking was mainly described by Bininj and Yapa women as an essential activity with many purposes. Cynthia: Walking and doing something at the same time. Like looking and digging. Us women we love to hunt. (Yapa woman, 46+) Groups of significant others, usually family, were always involved in the activities. This relationship between physical activity, work and family can be seen in the comments of Yapa woman, Lorene: Lorene: Well, when I go, we go as a family and I take the kids and we go out for a walk, digging for witchetty grubs & picking up wood... walking around the bush... mainly on weekends... Like whatever (bush foods) you get, you have got someone there making fire or digging the hole ready, getting fire wood & getting other little pieces & getting leaves for the (food) to be cut on. (Yapa woman, [25][26][27][28][29][30][31][32][33][34][35][36] Even with the influence of the paid working week and shops these traditional food-related work practices continued in the lives of the participants. Ruby: At the shop they always sell buffalo meat, that's all, and bullock meat. But when we go out during weekends we can get magpie geese, turtle, file snake, or pig, fresh ones, or flying fox, and (the) other thing is, we can get mud mussel from the creek and just boil it!... Out in the bush we can get yabbies, prawn... red claw... Sometimes we go for a swim, drag the net, get everything -all sorts of fish! (Bininj woman, [26][27][28][29][30][31][32][33][34][35] In the increasingly sedentary setting of the remote community, the responses of Bininj and Yapa men and women demonstrated that, where possible, they were attempting to maintain subsistence-based, physical work activities on weekends. Campbell: Well, we do fishing (and) hunting, but that's only for one day, maybe two days. Just on weekends; Saturday and Sunday." (Bininj man, 36-45). Nancy: Probably just on Saturdays. Take (all the) ladies, go for a walk, Look for witchetty grubs. (Yapa woman, 36-45) Consistent with the literature, participants described how wealth exists not only in the natural resources themselves but in people's connection to those resources which are demonstrated through their presence, interaction and the passing down of knowledge [25,26]. The physical process of tending to the renewal and reproduction of these resources, the manner of collection or hunting of food, building of shelters from local materials, preparing bark for painting or pandanus for weaving or dancing in ceremony, involves physical activity but is also valuable work performed for the benefit of the group rather than for the individual. A painting created by a female participant from the Central Desert with the intention of expressing and promoting physical activity shows food sources being collected by four women and brought together to eat and share in a space in the centre of the painting. Footprints depict how walking has been utilised not only to collect the food from each place but also plays a part in sharing what has been collected. Another interpretation of healthy physical activity portrayed in a painting by one of the Top End male participants called 'Bininj erecting a bark shelter', is an example of an Indigenous activity which is likely to go unrecognised as physically active work. As is shown in the painting, creating dwellings according to cultural and environmental necessity is for this participant a legitimate, health promoting activity. Examples of some of these kinds of work becoming recognised and a form of paid employment are found in some communities in limited supply in the form of land management and youth diversion programs. The management work involves native vegetation regeneration, weed, feral animal, water and fire control and the youth activities involve hunting, camping and walking on traditional country with elders. Eugene, a male participant working as a ranger to manage natural resources in the Top End community, described how this kind of work allowed him to meet his traditional physical activity and dietary needs, and the needs of the environment, in a culturally appropriate way. Eugene: Well I walk a lot. You know how you see on T.V, you know, people riding and walking and exercising? Well... you know, we (are) working (as rangers) and exercising at the same time. I mean when you go out there (in the bush), you know, we have kangaroo or goose for (a) feed... we'll (remove) weeds (at the same time), we walk like five kms or twenty kms, go (a) long way... and we carry that (heavy) spray pack. Well that's mainly what we do see? Constantly walking and that's our exercise for the whole day (Bininj man, 36-45). Similarly, Dan described how living and working on one's own country contributes to Bininj economic and health needs: Dan: Like in another man(s) country (you) might be poor but when you go back homeyou'll be rich! You've got own land see? "I want (those who are homeless in town) to come back home and live a better life so we can see (them) and make a longer life. (Bininj man, 26-35). To participants, 'the bush' was not considered a remote, inhospitable place, it was home; a place where these physically active forms of mostly unpaid work and environmental management were essential and provided a reliable source of food, security, social cohesion and pride. --- Activity, time and place: the social organisation of physical activities Participant responses highlighted different cultural preferences for the way physical activities are constructed and organised, the spaces they take place in and the times they occur. Many said they and other adults did not and would not utilise physical activity programs or spaces, such as swimming pools, youth centres, basketball courts or walking groups. The comments of the Bininj women reflected a willingness and preference to 'do things (their) own way, in the bush' rather than the 'Balanda (European) way'. Swimming in a billabong was seen to be more culturally appropriate than swimming in a public pool. Bininj and Yapa women's comments exemplified how for them to be active or to be seen to be active, in public pools or jogging, was viewed as culturally unacceptable. For example: European forms of physical activity, facilities and resources were also often seen as being suitable for children only. Would --- Do you think people like doing Balanda exercise? Eugene: (They) wouldn't (like it). I think... the kids would be interested, but with adults, I doubt it. Like, you could try anyway. Yeah. You could ride a bike around here but... no, only kids ride bikes around here. (Bininj Man, 36-45). 'Fitness' from a sociological perspective has been defined as "a relationship between a person's psychomotor capacities and... the socio-material organisation of space, time and activity. In this context, fitness is a secure and comfortable relationship with a socially organised and materially constructed environment" [27]. The material construction of the community setting was described by most participants negatively, as uncomfortable and not lending itself to 'fitness' in this sense. People described themselves as feeling more comfortable and able to be active in their own cultural spaces, in private and in accordance with their own purposes and seasonal time structures. Physical activities carried out in 'the bush' were described as different to those carried out in the community according to the nature of the environment and the comfort of the individual's relationship with it. Campbell: (Sam, a Balanda man) likes to run, he likes the exercise. Yeah. Funny thing (though), out in the bush it's not worth it. Because it is hot, you get dehydrated quick(ly) & you feel dizzy, (and) all that. (Sam) was saying, 'Stay healthy, stay fit' (in the community) but out there (in the bush) he was crying! And my cousin, he's short, fat & unfit, not really fat (but) solid... (and) he was fine walking (in the bush). (Bininj man, 36-45) Campbell also described an experience and observation of the differences between his comfortable and secure way of walking quickly in the bush compared to the'slow and steady' way of walking of someone not familiar with their surrounds. Campbell: (When I was walking in the bush) I was just walking and climbing and walking... walking after dark. And she (Balanda) was walking slowly, steady, in her step over the rocks. And she ask(ed) me "how (did) you manage to walk like that the whole (time)?" (I said), "I grew up with it. My legs (have) got all the balance. I can run, I also run fast on the rocks. I don't fall down. If I'm walking out (there and) there's no sugar bags (native bees hive), I know which grass I can eat that (has) got sugar in it (to keep me going). (Bininj man, 36-45) Appropriate times to be physically active were seen mainly as part of an existing relationship between place, time, people and activities as determined by the seasons, the sun, weather patterns, environmental conditions and the availability of food sources. Comments about the role of seasonal timing and the availability of certain foods in the organisation of physical activities support previous observations that in Indigenous Australian cultures human activities are traditionally seen to be primarily designated by season and location (Broome in [20]. Additionally, the availability of a vehicle and fuel to reach those locations was also considered necessary and appropriate. To illustrate the importance of the relationship between activity, place and time, Bininj women and men described when they go looking for yam or goose in this way: Notably, it was only those already diagnosed with chronic disease or obesity ('Balanda sicknesses') that considered participating in, or were considered likely to participate in, 'Balanda' or 'Kardiya' (Central Desert word for non-Indigenous) exercise programs like a walking group: Why do you want to go walking in a walking group? Nancy: I want to lose weight. Also I am a diabetic too. (Yapa woman, [36][37][38][39][40][41][42][43][44][45] Mimi: My mum and my other aunty they (were) talking about walking around because they are really big you know? Because they are diabetic as well, they have got big stomachs. (Yapa young woman, [16][17][18][19][20][21][22][23][24][25]. How --- Being 'on country': physical activity on traditional country When it comes to physical activity and health we can see from the interpretation of the perceptions, experiences and meanings expressed by participants in these themes why interaction with the natural environment is important. Natural and cultural resource management and associated activities that occur 'on country' were identified as the preferred health promoting physical activities. This last theme reveals how being on country is an important part of re-establishing physical activity as part of everyday life in a remote community and presents organised outdoor programs as a strategy to support this connection. Both male and female, Bininj and Yapa participants often talked about traditional country or 'the bush' as an active healthy place and the community as an inactive, unhealthy place. Marcia: When he came in... he said something about going bush and I said, "oh, that's good, I'm glad you're going out bush, get away... And I said, "I don't want to go visiting you on 'the machine' (dialysis). You want to be on that machine or what?... and he cracked up laughing and said, "no, that's why I'm going bush -I just want to get away from the temptations...(Bininj woman, 46+) Activities on country were also viewed as an educational opportunity that could facilitate knowledge transfer about how to find bush foods and survive in 'the bush'. Noel: That's teaching them -walking the kids across the land. (That) teaches them what's good for them... and that's about physical exercise that we('re) talking about... I've experienced living off the land; out in Central Arnhem Land. That's it! It was good. People used to walk miles and collect their food...and that's healthy living and they would live in that (healthy way)... Yeah, I mean that (bush walking) is addressing those bush tucker (issues), and teaching them how to find that bush tucker... Those (are the) sort of experiences that they (health promoters) should be looking at because that's part of it, it will (make) that healthy mind (know how) to find a healthy'manme' (local word for food). Proper one, right one! (Bininj man, 36-45) Both male and female, Bininj and Yapa recounted taking part in annual organised bush walks on country that allowed them to fulfil their cultural obligations to their ancestors and their country. Ruby described the necessity of these connections: Ruby: That's why we have to take them (kids, bush walking) so the old people can show them the country and the names of the place. They need to talk about and learn about the name of the places and who belongs to that place. That's why myself and (my husband) always join in for the bush walks so when we go there they always tell us, "well this is your great, great, great, grandmother country, or grandfather. Like when we stayed there, (at the) first camp, we told (them) "this place here you call your grandfather country..." (Bininj woman, [26][27][28][29][30][31][32][33][34][35] One of the paintings prepared by a male participant shows traditional dancing on country as a form of healthy physical activity. In this image, two Bininj men are participating in a dance as part of a traditional ceremony. The figure on the left is an old man holding clap sticks and covered in ceremonial body painting designs. The figure on the right is a young boy learning from his elder, copying and learning the story of his movements, so that in the future he can pass them on to the next generation. The artist explained that his ancestors were taught to dance by mimi spirits and that dancing was still an important and enjoyable part of his culture today. Walking and being on country in this way is described as a necessity undertaken for a particular meaningful social purpose. What is the best way to spread the word about why working your body... doing exercise (and) being fit is good? Campbell: I think it's best for us (to) like take people out for a bush walk. It's the best (way) to tell them and also to make them understand what life risks are. I think that bush walking is it! It's great! (It is) talking (about) everything about life... (Bininj man, 36-45). Participants looked to their cultural traditions and activities associated with maintaining their heritage as a way forward in overcoming social, cultural, economic and health disadvantages. A desire and willingness to conduct and lead these kinds of activities was expressed by a number of male and female, Bininj and Yapa. Dan: Like for me I am thinking I don't want all (these community problems) to happen in the future, I reckon for me, (I want) to make (a) program, an education programwe want to do camping... Get axe, firewood, billy canlike that. We are going to stay there (out bush). That's how I'm looking at it. Make music... bush tucker... you learn what's healthy in the bush. Go hunting or fishing... make spear... That's how I'm looking at it and that's how I want to be too. Get them to tell their story maybe... write songs... travel around to other communities... learn other people's culture... (Bininj man, [26][27][28][29][30][31][32][33][34][35] Naomi: I'd like to take people out bush and show them our way. Show them (Balanda) all that bush tucker, you know? We talk about that all the time. (Bininj woman, 36-45) --- Discussion The meaning of work, space and place in relation to physical activity, as expressed by the participants in this study, suggests that supporting various forms of natural and cultural resource management may be an appropriate physical activity health promotion strategy. It is important that these land management and related employment opportunities involve access to and physical engagement with 'the bush' and traditional country. Meanings and definitions of work have been contested between Indigenous and settler culture since contact. Involved in this contest are diverging perspectives of what constitutes work, leisure, time and payment and how they link together in our relationship to natural resources and the economic environment [23]. Even though Indigenous Australians have sophisticated systems of natural and cultural resource management and work-related activities which continue still and develop where possible in the lives of participants today, Non-Indigenous Australians in general have tended to only recognise ways of working, forms of economics and resource management similar to their own as legitimate [20]. Subsistence activities were often perceived by early settlers as lazy and Indigenous people were reported as being "strongly adverse to labour" (Organ in [20], p.10]). Europeans called these kinds of activity 'walkabout'; a term still used in the present day and still often associated with laziness and unreliability rather than work [20], p.10]. So despite a long history of a strong working culture unemployment and poverty remain as main determinants of poor Indigenous health in Australia today [8,9]. Even though the inherently physical subsistence activities such as hunting magpie goose or digging for honey ant have not been recognised by non-Indigenous people in general as economically legitimate forms of work, they continue today and have not been totally replaced by Bininj or Yapa with other forms of food production which are likely not to require as much physical activity. The different concepts of work and wealth as explained by the participants are embedded in necessary physical engagement with a particular environment and help explain why on traditional homelands or outstations, where infrastructure and services are few, those who are participating in natural and cultural resource management activities as part of daily life are often the healthiest [14]. Knowing more about how, why, where and when people walk is also an important factor in understanding how to promote physical activity in remote communities. It has been noted by movement specialists that in a cultural context communities consider "uniformity of movement behaviour" as essential for protecting the "stability of the community spirit" (Laban in [28], p. 165]). The notion that physical activity is not a culturally or socially neutral activity is all the more pertinent when viewed from Bourdieu's concept of habitus [29]. Habitus is the psychological and physical embodiment of the social and cultural structures in which we live; it is the way the social and cultural world is written into our perceptions, our bodies and actions. The movements required by Western-style physical activity routines imposed on Indigenous people in remote communities thus speaks of a disparity between the non-Indigenous and Indigenous habitus. This has been described by Indigenous academic Bronwyn Fredericks when she stated that Western-style exercise environments "do not necessarily fit within the contexts of the materiality of our bodies" [30]. European style physical activity programs rely upon the Gregorian calendar and the 24 hour clock. The seasonal concepts of time in the tropics and the desert as manifested by different, specific and subtle changes in the environment challenge the unquestioned assumption of the European calendar and clock time as the norm (Adam in [20]). As Glenda Go_djalk, a Yolngu health worker from Marthakal homelands on Elcho Island explains: You can't tell the water: "Hey, Stop, stop, stop! Low tide, low tide now! High tide, high tide now! This way! Now that way! It has its own timing for going in, and going out, when it will become rough, and when it will become calm. Yes, this is the same for our bodies [31]. Such cultural differences between Indigenous and non-Indigenous ways of constructing and organising activities, places and times, need to be better understood to promote active living successfully in remote communities. Physical activity programs led by Indigenous people that incorporate the use of customary spaces, seasonal and biological timings are more likely to be meaningful in remote contexts [32,33]. The findings suggest that being and walking on country and regular participation in associated activities are central to the cultivation of Indigenous health in remote areas. These kinds of physical activity allow for the maintenance of social and cultural traditions that are seen to provide a link between health, the environment and economics in this context. Youth diversion and outstation programs in the NT involve facilitating trips to homelands and on to country for people to participate in a range of such activities. As well as increased physical activity, outdoor programs like these have been shown internationally to improve various dimensions of health including wellbeing, self-efficacy, confidence, independence, peer relations and mental health [26,34,35]. Overall, the participants involved in this study and the themes that they brought up suggest that in a remote Indigenous context being physically active and healthy is about interacting productively and meaningfully with the environment. It is about performing the cultural activities including, but not limited to, resource management and sharing stories about country which provide opportunities for the inter-generational transfer of knowledge and the maintenance of cultural heritage. Internationally, Indigenous people repeatedly associate their extreme illhealth and poverty with their experiences of dispossession and associate solutions with reclaiming access to land [10]. This study shows that in remote Indigenous communities in the NT people refer to being active on traditional country as a solution to many problems deeply embedded in a history of colonisation. These findings support the importance of employment in caring for country and land management programs which enable people to fulfil social and cultural obligations, educate young people, escape the stresses of settlements and

Background: The burden of chronic disease in Indigenous Australia is more than double that of non-Indigenous populations and even higher in remote Northern Territory (NT) communities. Sufficient levels of physical activity are known to reduce the risk of chronic disease and improve the health of those already suffering from chronic disease. It has been identified that effective promotion of physical activity in Indigenous settings requires the diverse cultural perspectives and participation of Indigenous people. However, Indigenous concepts of physical activity are not represented in the public health literature and examples of Indigenous involvement in physical activity promotion are scarce. This study aimed to explore and describe local perspectives, experiences and meanings of physical activity in two remote NT Indigenous communities. Methods: Qualitative research methods guided by ethnographic and participatory action research principles were used. Semi-structured interviews conducted with 23 purposively selected community members were the main source of data, augmented by five commissioned paintings by community-based artists and observations recorded in a journal by the first author. Results: The findings reveal that in this cultural context the meaning of physical activity is embedded in socially significant and economically necessary physical engagement with the environment. Participants described physical activities associated with Indigenous natural and cultural resource management, customary spaces, seasonal timing and traditional education as creating and protecting health. These activities were viewed not only as culturally appropriate physical activities that contribute to health but as legitimate, physically active forms of social organisation, education and employment that help to build and maintain relationships, wealth, resources and the environment. Conclusion: This different construction of physical activity in remote Indigenous communities highlights the importance of involving Indigenous people in the development and implementation of physical activity promotion. Physical activities associated with traditional Indigenous cultural practices and being active 'on country' need to be viewed as legitimate health promotion activities. Exploring further ways to enable Indigenous people in remote NT to be involved in creating viable active livelihoods on 'traditional country' needs to be considered as imperative to health improvement.

involve facilitating trips to homelands and on to country for people to participate in a range of such activities. As well as increased physical activity, outdoor programs like these have been shown internationally to improve various dimensions of health including wellbeing, self-efficacy, confidence, independence, peer relations and mental health [26,34,35]. Overall, the participants involved in this study and the themes that they brought up suggest that in a remote Indigenous context being physically active and healthy is about interacting productively and meaningfully with the environment. It is about performing the cultural activities including, but not limited to, resource management and sharing stories about country which provide opportunities for the inter-generational transfer of knowledge and the maintenance of cultural heritage. Internationally, Indigenous people repeatedly associate their extreme illhealth and poverty with their experiences of dispossession and associate solutions with reclaiming access to land [10]. This study shows that in remote Indigenous communities in the NT people refer to being active on traditional country as a solution to many problems deeply embedded in a history of colonisation. These findings support the importance of employment in caring for country and land management programs which enable people to fulfil social and cultural obligations, educate young people, escape the stresses of settlements and earn an income [25] thereby improving the communities health, economics and environment [13,15]. Indigenous cultures have considerable knowledge of human-environment relations and there is growing international evidence to support the promotion of Indigenous cultural and ecological practices to improve health [26,[36][37][38]. Similarly, environmental health and development specialists are recommending that better management of land degradation, biodiversity loss, soil erosion, food insecurity and water quality decline, as well as reducing 'unsustainable consumption', "will have long-term and sustained beneficial effects on human health" as a whole [39]. Indigenous subsistence economies and their achievements in sustainable development are beginning to be acknowledged: Indigenous peoples' economies now represent the greatest continuity with pre-industrial modes of production and traditional livelihoods in the contemporary world. These economies, representing sustained interaction and adaptation with particular locations and ecosystems, are among the longest-standing and most proven examples of "sustainable development" in the twenty-first century [10]. Recognising the social and cultural dimensions of humanenvironment relations plays an essential part in the sustainability of human health [38], yet this link between natural and cultural resource management, economics and physical activity in an Indigenous context has only begun to be examined. This connection, as demonstrated through this study, is important in creating what is viewed as 'a healthy society', as outlined in the Ottawa Charter of Health Promotion: the way society organises work should help create a healthy society and... the sustainable use of the natural environment and resources must be addressed in any health promotion strategy [40]. Public health strategists can recognise Indigenous perspectives and capacity and facilitate this connection also by working together with outdoor education and youth diversion programs. Particularly relevant are environmental and socio-ecological outdoor education models that prioritise the importance of human-environment relations [34,35]. These models take a strength-based approach, support activities that are coordinated with ecological rhythms and promote the active use of existing spaces in which people already feel comfortable and secure [32]. They create opportunities for increased physical activity, connection with nature, culture, social networking and Indigenous and non-Indigenous skills transfer [29,30]. For example, an account of an ethnoecological education camp in Indigenous Canada describes how the Indigenous and non-Indigenous organisers aligned the program and themselves with the physical, emotional, mental and spiritual aspects of the Dene mountain people of Canada's north, as well as modern science and technology [26]. In this way "the camp brings Indigenous instructors to the forefront as mentors who have something not only useful, but indeed necessary, to teach us" [26]. The participants in this study expressed the need for programs like these to improve physical activity and health. --- Limitations There are many challenges involved in the process of engaging meaningfully and effectively with people living in remote Australian communities for research purposes due to a difficult history, language and cultural differences, and varying cultural constructions of power, time and priorities. This study was conducted over one year, was only able to explore the views of a limited number of community residents and a majority of those were female in the desert community. The inability to recruit community-based researchers and more participants in the desert community was attributed to a period of intra-community conflict that had preceded the field work and resulted in a significant proportion of the community moving away temporarily. This included 3 people who had initially expressed a strong interest in the research positions moving away from the community during the field work period. This meant that we were not able to document male perspectives or those of less active females in the desert community. However, despite the resulting differences in recruitment methods, participant numbers and the gender imbalance between the groups, many views and experiences described by participants in relation to physical activity were similar within and across the two communities. The men and women, the active and the inactive all described a distinct preference for traditional physical activities in 'the bush'. The strengths of this study were: the involvement of local co-researchers where possible; the conduct of some interviews in first language; the ability of the co-researchers to translate the local language into English where needed; and the rich contribution made by the paintings of physical activity. Considering the paucity of literature available, as well the specific challenges of working with remote Indigenous communities, long-term research and interdisciplinary research teams working in partnership with communities are needed to develop effective physical activity health promotion models, guide implementation strategies and achieve substantial benefits. --- Conclusion These findings help to explain how from an Indigenous perspective the meaning of physical activity is connected to culture through work, diet, relationships, place, time and country. This helps to establish some of the sociocultural links between physical activity, health, the environment and economics in remote northern Indigenous Australia. Currently situated in a period of unprecedented global economic, environmental and lifestyle disease crises, public health advocates need to recognise Indigenous models of health promotion, resource management and associated contemporary and customary activities as sustainable, healthy systems of active livelihoods [25,41]. This paper demonstrates the necessity of collaborating with Indigenous community members when developing and implementing physical activity programs, as well as the legitimacy of enabling people living remotely to be active on traditional country as a broad inter-sectoral health strategy. --- Competing interests The authors declare that they have no competing interests. --- Authors' contributions All of the authors contributed to the conception and design of the project. JB provided supervision on all aspects of the study. ST and the communitybased co-researchers were responsible for data collection and the interpretation of the findings. ST analysed the data and prepared the manuscript; the remaining authors made revisions. All authors read and approved the final manuscript.

Background: The burden of chronic disease in Indigenous Australia is more than double that of non-Indigenous populations and even higher in remote Northern Territory (NT) communities. Sufficient levels of physical activity are known to reduce the risk of chronic disease and improve the health of those already suffering from chronic disease. It has been identified that effective promotion of physical activity in Indigenous settings requires the diverse cultural perspectives and participation of Indigenous people. However, Indigenous concepts of physical activity are not represented in the public health literature and examples of Indigenous involvement in physical activity promotion are scarce. This study aimed to explore and describe local perspectives, experiences and meanings of physical activity in two remote NT Indigenous communities. Methods: Qualitative research methods guided by ethnographic and participatory action research principles were used. Semi-structured interviews conducted with 23 purposively selected community members were the main source of data, augmented by five commissioned paintings by community-based artists and observations recorded in a journal by the first author. Results: The findings reveal that in this cultural context the meaning of physical activity is embedded in socially significant and economically necessary physical engagement with the environment. Participants described physical activities associated with Indigenous natural and cultural resource management, customary spaces, seasonal timing and traditional education as creating and protecting health. These activities were viewed not only as culturally appropriate physical activities that contribute to health but as legitimate, physically active forms of social organisation, education and employment that help to build and maintain relationships, wealth, resources and the environment. Conclusion: This different construction of physical activity in remote Indigenous communities highlights the importance of involving Indigenous people in the development and implementation of physical activity promotion. Physical activities associated with traditional Indigenous cultural practices and being active 'on country' need to be viewed as legitimate health promotion activities. Exploring further ways to enable Indigenous people in remote NT to be involved in creating viable active livelihoods on 'traditional country' needs to be considered as imperative to health improvement.

Introduction At present the World Wide Web (WWW) is undergoing a landmark revolution from the traditional Web 1.0 to Web 2.0 characterized by social collaborative technologies, such as social networking sites (SNSs), blogs, Wiki, and folksonomy [1]. As a fast growing business, many SNSs of different scopes and purposes have emerged in the Internet, many of which, such as MySpace [2,3], Facebook [4][5][6][7], and Orkut [2,8], are among the most popular sites on the Web according to Alexa.com. Users of these sites, by establishing friendship relations with other users, can form online social networks (OSNs), which provide an online private space for individuals and tools for interacting with other people over the Internet. Both the popularity of these sites and availability of network data sets offer a unique opportunity to study the dynamics of OSNs at scale. It is believed that having a proper understanding of how OSNs evolve can provide insights into the network structure, allow predictions of future growth, and enable exploration of human behaviors on networks [9][10][11][12][13]. Recently, the structure and evolution of OSNs have been extensively investigated by scholars of diverse disciplines. Golder et al. studied the structural properties of Facebook and found that the tail of its degree distribution is a power law which is different from the traditional exponential distribution of reallife social networks [4]. However, a mean of 179.53 friends per user for Facebook [4] or a mean of 137.1 friends per user for MySpace [2] is close to Dunbar's number of 150, which is a limit on the number of manageable relations by human based on their neocortex size [14]. Holme et al. studied the structural evolution of Pussokram and found that its degree correlation coefficient is always negative over time, i.e. disassortative mixing [15], which is in stark contrast to the significant assortative mixing for real-world social networks [16]. Viswanath et al. studied the structural evolution of the activity network of Facebook and found that the average degree, clustering coefficient, and average path length are all relatively stable over time [6]. Hu & Wang studied the evolution of Wealink [17,18] and found that many network properties show obvious non-monotone feature, including a sigmoid growth of network scale which was also observed by Chun et al. in Cyworld [19], and a transition from degree assortativity characteristic of real social networks to degree disassortativity characteristic of many OSNs which was also observed by Szell & Thurner in Pardus [20]. Despite the advancement, we find that to date most research on OSNs has focused on either the structural properties of a certain snapshot of networks or the multi-snapshots of evolving networks rather than detailed microscopic growth dynamics. For the research framework of network evolution from a macroscopic viewpoint it is usually hard to reveal underlying mechanisms and growth processes governing the large-scale features of the observed network structure. In this paper, to gain better insight into the growth of networks, based on empirical data, we study the detailed process of people making friends in an OSN from a microscopic point of view. Instead of investigating the global network structure or structural metric evolution, we focus directly on the microscopic user behaviors per se, i.e., we study the properties of a sequence of the arrivals of each edge or the formations of each friend relation. --- Data set In this paper, we will focus on Wealink, a large SNS in China whose users are mostly professionals, typically businessmen and office clerks [17,18]. Each registered user has a profile, including his/her list of friends. If we view the users as nodes V and friend relations as edges E, an undirected friendship network G(V, E) can be constructed from Wealink. For privacy reasons, the data, logged from 0:00:00 h on 11 May 2005 (the inception day for the Internet community) to 15:23:42 h on 22 August 2007, include only each user's ID and list of friends, and the time of sending link invitations and accepting requests for each friend relation. The finial data format, as shown in Fig. 1, is a time-ordered list of triples <unk>From, T o, When>. For instance, <unk>U 1, U 2, T 1 > indicates that, at time T 1, user U 1 sends a link request to user U 2, i.e., sends a friendship invitation to U 2, while <unk>U 2, U 1, T 6 > indicates that, at T 6, U 2 accepts U 1's request and they become friends, i.e., a new edge connecting U 2 and U 1 appears in the OSN. Thus only when the sent invitations are accepted will the friend relations or network links be established. The online community is a dynamically evolving one with new users joining the community and new connections established between users. --- Reciprocal behavior Like most OSNs, in Wealink, a user invites another user to be his/her friend; if the invited user accepts the invitation, a friend relation is established between them and a new edge connecting them appears (see Fig. 1). Thus the friendship is constructed by bilateral agreement. The degree of a user, i.e., the number of friends, will appear on his/her profile, which can be browsed by all the other users. During our data collection period, 273 209 sent link requests have been accepted and only 186 ones have not yet been accepted. Thus, in the following analysis, we will focus on the 273 209 sent link requests and their corresponding accepted ones with full temporal information. We first scrutinize the reciprocation of users, i.e., the sending of a link request from one user to another (as happens at T 1 in Fig. 1) causes following acceptance of the request (T 6 ). Fig. 2(a) shows the complementary cumulative distribution of the intervals between sending and accepting link requests in Wealink. It is clear that users often quickly responded to link requests and reciprocated them. The interval distribution decays approximately exponentially. The least squares fitting gives P c (t) <unk> e -0.011t with R 2 = 0.958. In fact, as shown in Fig. 2(b), 67.04% of all reciprocal behavior occurred within one day (24 hours) after the initial link requests and 84.25% of sent link requests were accepted within one month (30 days). Wealink informs users by email of new incoming link requests. It is quite possible that many users reciprocated requests as a matter of courtesy and respect. Recently inspired by the pioneering works of Barabási et al., there has been increasing interest for physicists and computer scientists in the research of human dynamics [21,22], which focuses on the time interval distribution between two consecutive actions performed by individuals. The examples of such temporal statistics include the inter-event time distribution between two consecutive emails sent out by users, two consecutive visits to a web portal by users, and two consecutive library loans made by individuals. Empirical studies have shown that many distributions of inter-event time follow a power law. However, the exponential reciprocation interval distribution is in distinct contrast to the power law distribution of waiting time in emails (i.e., the time taken by users to reply to received emails). The importance of different emails is different. A reasonable hypothesis is that there can be correlation between the importance of emails and reputation/status of senders or "social closeness" to senders. Thus users can reply to received emails based on some perceived priority, and the timing of the replies will be heavy tailed. In contrast, there is no obvious priority for the reciprocal behavior of users in OSNs; thus an exponential distribution will well characterize the reciprocation interval distribution. An interesting question is whether the users tended to reciprocate incoming link requests quickly regardless of how many friends the inviters or accepters had. To answer the question, we study the correlation between reciprocation time and the degrees of inviters/accepters at the time of sending link requests. Fig. 3(a) shows the density plot based on hexagonal binning for the relation between degrees of inviters k and reciprocation time t, where the cases with small k and t dominate. The Pearson correlation coefficient between k and t is -0.02, indicating slightly negative correlation. Fig. 3(b) shows the relation between k and mean reciprocation time t with logarithmic binning and error bars. t exhibits mild descending trend as k increases. Fig. 4 shows the relation between degrees of accepters k and reciprocation time t, which is similar to that shown in Fig. 3. The Pearson correlation coefficient between k and t is -0.05, also indicating mildly negative correlation. --- Types of users and edges The total number of users during the data collection period is N=223 482. Obviously the users can be divided into three classes: active users who sent link requests but have never received ones, passive users who received requests but have never sent ones, and mixed users who both sent and received requests. As shown in Tab. 1, we find that most users belong to the former two classes. For the very popular SNSs, such as Facebook and MySpace, due to the high activity and degree values of users, most users could be mixed ones. However, Wealink is a very professional OSN with a mean degree of only 2.53. The activity of most users is low; after joining in the OSN they either send link requests to a few old users (acquaintance in real life very likely) or receive link invitations from several old users. Among the mixed users, there exists obvious positive correlation between the numbers of times of sending invitations and accepting invitations, and the Pearson correlation coefficient is 0.48. As shown in Fig. 5, we find that the more link requests a user sends, the more requests he/she will receive, and vice versa. The finial network density is only 1.09 <unk> 10 -5, and what results in the sparseness? As shown in Tab. 2, the E=273 209 edges can be divided into four classes, and A-B type expresses that initially A users sent link requests to B users. "Old" means that the users have been in the network; they joined in the network some time ago and they either have sent at least one link request to other users or have received at least one link request from other users. "New" means that the users have joined in the network; however, they neither have sent link requests to other users nor have they received link requests from other users. It is shown that in Wealink most links are established by old users sending requests to new users (more than 30%) and new users sending requests to old users (approximately 50%). The number of edges of Old-Old type is relatively small, leading to the sparseness of the network. --- <unk>From, To, When> <unk>U 1, U 2, T 1 > Invite <unk>U 1, U 3, T 2 > Invite <unk>U 3, U 1, T 3 > Accept <unk>U 4, U 2, T 4 > Invite <unk>U 2, U 4, T 5 > Accept <unk>U 2, U 1, T 6 > Accept...... Network evolution Network evolution T 3 U 1 U 3 U 1 U 3 U 4 U 2 U 1 U 3 U 4 U 2 T 5 T 6 <unk>U 5, U 2, T 7 > Invite <unk>U 2, U 5, T 8 > Accept U 1 U 3 U 4 U 2 U 5 T 8 --- Temporal characteristics of linking We study the time interval distribution between two link events. As shown in Fig. 6, the distributions of intervals between consecutive sending link requests (i.e., between T 1 and T 2, T 2 and T 4, and so on in Fig. 1), accepting requests (i.e., between T 3 and T 5, T 5 and T 6, and so on in Fig. 1) and any two events (i.e. between T i and T i+1 (i <unk> 1) in Fig. 1) all follow a power law with a universal exponent 1.89, which diverges from the exponential distribution predicted by a traditional Poisson process and indicates bursts of rapidly occurring events separated by long periods of inactivity. Several peaks appear for intervals of an integral day in the tails of the distributions, indicating the daily periodicity corresponding to human life habits. --- Preferential selection Preferential selection means that, for a time-ordered list of individual appearance, the more likely an individual appeared before, the more possibly the individual will occur once again. We separate the preferential selection into two aspects: preferential sending and preferential reception. Preferential sending describes the mechanism by which users send new link requests with probability proportional to some power of the numbers of their sent link invitations before, and preferential reception describes the mechanism by which users receive new link requests with probability proportional to some power of the numbers of their received link invitations before. Fig. 7 presents the schematic illustration of sending and reception sequences of OSNs. The former is a time-ordered list of users sending link invitations, and the latter is a time-ordered list of users receiving link invitations. In both sequences, the more frequently a user appeared before, the more likely the user will occur once again. Let k i be the number of sent or received link invitations for user i. The probability that user i with frequency k i is chosen to send or receive a link request once again can be expressed as (k i ) = k <unk> i j k <unk> j.(1) Thus we can compute the probability (k) that an old user of frequency k is chosen, and it is normalized by the number of users of frequency k that exist just before this step [23,24]: (k) = t [e t = v <unk> k v (t -1) = k] t |<unk>u : k u (t -1) = k<unk>| <unk> k <unk>,(2) where e t = v <unk> k v (t -1) = k represents that at time t the old user whose frequency is k at time t -1 is chosen. We use [•] to denote a predicate (which takes a value of 1 if the expression is true, else 0). Generally, (k) has significant fluctuations, particularly for large k. To reduce the noise level, instead of (k), we study the cumulative function: <unk>(k) = k 0 (k)dk <unk> k <unk>+1.(3) Fig. 8 shows how the frequency k of users is related to the preference metric <unk>. <unk> <unk> 1 for both preferential sending and preferential reception, indicating linear preference. It is natural that, in the sending or reception sequence, the number of distinct users N increases with sequence length T. Fig. 9 shows the growth pattern of N with T for Wealink. N <unk> T, indicating that the appearance probability of new users is a constant, <unk> = N/T. According to the Simon model [25], based on linear preferential selection and constant appearance probability of new users, the complementary cumulative distributions of the numbers of sent invitations and received invitations for the users of Wealink follow a power law P c (n) <unk> n -( 1 1-<unk> ). Based on empirical data, for the inviters, we obtain <unk> = 0.53 and P c (n) <unk> n -2. 13, and for the receivers, <unk> = 0.35 and P c (n) <unk> n -1.54. Fig. 10 shows the distribution functions of the frequencies of inviters and receivers, and the tails of both distributions show power law behavior. The power law exponents achieve proper agreement with the predicted values of the Simon model, 1/(1<unk>). --- Preferential linking The degree distribution of Wealink shows power law features [17]. This kind of distribution can be produced, as indicated by the Barabási-Albert (BA) model [26], through linear preferential attachment, where new users tend to attach to already popular old users. In Wealink, as shown in Fig. 1, only when the sent link invitations are accepted can the inviters and receivers become friends and new edges appear in the social network. When new users establish friend relationship with old users, or new edges are established between old users, the old users with large degrees could be preferentially selected. --- <unk>From, To, When> <unk>U 1, U 2, T 1 > Invite <unk>U 1, U 3, T 2 > Invite <unk>U 4, U 2, T 4 > Invite...... Sending sequence: U 1, U 1, U 4,... Reception sequence: U 2, U 3, U 2,... <unk>From, To, When> <unk>U 1, U 2, T 1 > Invite <unk>U 1, U 3, T 2 > Invite <unk>U 3, U 1, T 3 > Accept <unk>U 4, U 2, T 4 > Invite <unk>U 2, U 4, T 5 > Accept <unk>U 2, U 1, T 6 > Accept...... <unk>U 5, U 2, T 7 > Invite <unk>U 2, U 5, T 8 > Accept <unk>U 5, U 2, T 7 > Invite U 5 U 2,, To test the preference feature for different types of link establishment, we separate the preferential linking into three aspects: preferential acceptance, creation, and attachment. Preferential acceptance implies that, the larger an old user's degree is, the more likely he/she will accept link invitations from other old users. Preferential creation implies that, the larger an old user's degree is, the more likely his/her link invitations will be accepted by the other old users. And preferential attachment implies that new users tend to attach to already popular old users with large degrees. For instance, in Fig. 1, at time T 6, a new edge appears between two old users U 1 and U 2. Old user U 2 who accepts a link invitation can be chosen by preferential acceptance, and old user U 1 who sends a link invitation can be chosen by preferential creation. At time T 8, a new edge appears between old user U 2 and new user U 5, and old user U 2 can be chosen by preferential attachment. Fig. 11 shows the relation between the degree k of users and the preference metric <unk>. We find that <unk> <unk> 1 for preferential acceptance, creation, and attachment, indicating linear preference. The property of linear preference for the network can be generalized to general OSNs. Mislove et al. studied the evolution of Flickr; they defined preferential creation as a mechanism by which users create new links in proportion to their outdegree, and preferential reception as a mechanism where users receive new links in proportion to their indegree [27]. They found that linear preference holds for both cases, i.e. users tend to create and receive links in proportion to their outdegree and indegree, respectively. Leskovec et al. studied the evolution of Flickr, del.icio.us, Yahoo!Answers, and LinkedIn, and examined whether the new users will preferentially link to the old users with large degrees [24]. They found that Flickr and del.icio.us show linear preference, (k) <unk> k, and Yahoo!Answers shows slightly sublinear preference, (k) <unk> k 0.9. LinkedIn has a different pattern: for low degrees k, (k) <unk> k 0.6, and thus the preference is not obvious; however, for large degrees, (k) <unk> k 1.2, indicating superlinear preference, i.e., the edges to higher degree users are more sticky and high-degree users get super-preferential treatment. Even though there are minor differences in the exponents <unk> for different networks, we can say that <unk> <unk> 1, implying that linear preference may be universal for OSNs. According to this linear preference, we put forward a realistic network model. Starting with a small network with m 0 nodes, at every time step, there are two alternatives. A. Growth and preferential attachment. With probability p, we add a new node with m 1 (<unk> m 0 ) edges that will be connected to the nodes already present in the network based on the preferential attachment rule of the BA model, i.e., the probability <unk> that a new node will be connected to old node i with degree k i is (k i ) = k i j k j. B. Preferential creation and acceptance. With probability q = 1 -p, we add m 2 (m 1 + m 2 <unk> m 0 ) new edges connecting the old nodes. The two endpoints of the edges are chosen according to linear preference (k i ) = k i j k j. After t time steps, the model leads to a network with average number of nodes N = m 0 +pt. For sparse real-world networks, p > q. When p = 1, the model is reduced to the traditional BA model. The model considers the introduction of new nodes and new edges, which can be established either between new nodes and old nodes or between old nodes. Most importantly, the model integrates linear preference for acceptance, creation and attachment found in the evolution process of real networks, and thus captures realistic features of network growth. The model has an analytic solution. Its stationary average degree distribution for large k is [28] P(k) <unk> k - 3pm 1 +4qm 2 pm 1 +2qm 2,(4) showing a scale-free feature. According to Tab. 2, we obtain p = p Old-New + p New-Old = 0.7941 and q = 0.1939. The links created between two new users are few and thus can be negligible. In addition, m 1 = m 2 = 1 for real growth of the network. Based on the parameters and Eq. ( 4), we obtain P(k) <unk> k -2.67. Fig. 12(a) shows the numerical result which is obtained by averaging over 10 independent realizations with p = 0.7941 and N=223 482. Its degree exponent 2.62 agrees well with the predicted value of 2.67. Fig. 12(a) also presents the complementary cumulative degree distribution of Wealink, and we find that the predicted value of the degree exponent 2.67 of the model achieves proper agreement with the real value 2.91. The difference between real and theoretical values may arise from the fact that p and q are time-variant variables and not constants. Fig. 12(b) shows the evolution of p and q and demonstrates the fact. --- Summary and discussion To conclude, we have unveiled the detailed growth of an OSN from a microscopic perspective. Our study shows that the distribution of intervals between sending and accepting link requests decays approximatively exponentially, which is in obvious contrast to the power law distribution of waiting time in emails, and there exists a slightly negative correlation between reciprocation time and degrees of inviters/accepters. The distributions of intervals of user behaviors, such as sending or accepting link requests, follow a power law with a universal exponent, indicating the bursty nature of the user act. We finally study the preference phenomena of the OSN and find that for preferential selection linear preference holds for preferential sending and reception, and for preferential linking linear preference also holds for preferential acceptance, creation and attachment. We propose a network model which captures real features of network growth and can reproduce the degree distribution of the OSN. It is noteworthy that, although there is a close relation between the microscopic growth of networks and global network structure or structural metric evolution, it is still quite hard to bridge the gap between macro and micro perspectives of OSNs. For instance, preferential linking may possibly supply some information on the degree distribution of networks; however, it may not tell us much about the other properties of networks, such as clustering or community structure. Thus to gain an indepth comprehension of OSNs, other microscopic behaviors of users, such as homophily, need to be studied in detail; a complementary research framework integrating macro and micro perspectives will also be indispensable.

We study the detailed growth of a social networking site with full temporal information by examining the creation process of each friendship relation that can collectively lead to the macroscopic properties of the network. We first study the reciprocal behavior of users, and find that link requests are quickly responded to and that the distribution of reciprocation intervals decays in an exponential form. The degrees of inviters/accepters are slightly negatively correlative with reciprocation time. In addition, the temporal feature of the online community shows that the distributions of intervals of user behaviors, such as sending or accepting link requests, follow a power law with a universal exponent, and peaks emerge for intervals of an integral day. We finally study the preferential selection and linking phenomena of the social networking site and find that, for the former, a linear preference holds for preferential sending and reception, and for the latter, a linear preference also holds for preferential acceptance, creation, and attachment. Based on the linearly preferential linking, we put forward an analyzable network model which can reproduce the degree distribution of the network. The research framework presented in the paper could provide a potential insight into how the micro-motives of users lead to the global structure of online social networks.

Background Despite recent signs of a decline in the prevalence of legal and illegal drugs, the use of alcohol and cannabis remains widespread among adolescents and young adults in contemporary European society [1,2]. Many people initiate alcohol and drug use during their years as a teenager [3,4]. According to a recent research report, 75.4% of the Flemish scholars (12-18 years) have drunk alcohol at least once during their lifetime [5]. A vast majority of these respondents (63.4%) even used alcohol during the past 12 months, of which 22% drank alcohol on a regular basis, that is, more than once a week. The same research reported a last year prevalence rate of cannabis among Flemish students of 11.7% of which 2.7% used it on a weekly basis. Approximately one out of five students had smoked cannabis once during their lifetime [5]. This substance use is however not without harm. Many studies acknowledge the acute and longer range health implications of this behavior, both on a personal and societal level [6,7]. Conceivable shortterm consequences of adolescent's substance (mis)use are an increased risk of accidental injury and death, of engaging in criminal and delinquent behavior, violence victimization, engaging in unsafe sexual practices, educational failure and depression and suicidal ideation. Beyond these immediate "threats", early alcohol and cannabis consumption is frequently associated with a heightened chance of developing substance use and dependence disorders, major depressive symptoms and other undesirable health outcomes in later adulthood. Notwithstanding these health implications, for most teenagers it holds that adolescence is a phase of experimentation in the first place. A striking observation is that age specific rates of alcohol and drug use, and related to this conformity to peer pressure and fear of peer rejection, peak in adolescence and drop sharply when entering adulthood [8][9][10]. Nevertheless, given the possible health consequences, the (ab)use of alcohol and cannabis remains a serious public health concern and considerable and continuing efforts need to be conducted to develop effective interventions in this domain. Therefore, insight in risk factors that constitute teenage substance use is imperative. Several studies have shown that in adulthood socioeconomic differences in alcohol and drug use are relatively consistent [11][12][13]. People from higher socio-economic strata tend to drink more often but in smaller quantities, while their counterparts in lower socio-economic strata tend to consume alcohol less often but in larger quantities and in a more problematic way. While the existence of class-based differences in alcohol and cannabis use among adults is well established, socio-economic differences in adolescents' substance use are far less investigated, and consistency in present evidence is lacking. According to some research reports, the association between socio-economic status (SES) and substance use in adolescence is similar to patterns found in other life stages, where a lower SES is associated with increased incidence rates of alcohol and cannabis consumption [13][14][15][16]. On the other hand, several studies couldn't corroborate these findings, reporting an insignificant, diminished or reversed relationship between a teenagers socio-economic position and his/her alcohol and cannabis use [17][18][19][20]. As a possible explanation for these inconsistencies, West and others [21,22] referred to the occurrence of "a process of equalization", in which a transition is taking place from health inequality in childhood to relative equality during adolescence. According to these authors, this "process of equalization" is rooted in the defining characteristics of adolescence as a transitional period (i.e. the growing importance of peers, school environment and youth culture), by which the influence of familial background gets sharply curtailed, resulting in a homogenizing effect [21,22]. However, it can be questioned whether this equalization process is a real-existing phenomenon or rather should be interpreted as an artifact, reflecting the way in which an adolescents' socio-economic position is traditionally measured, that is, using information about the parents' socio-economic status (i.e. parental educational level, parental occupational level and family income) [23][24][25]. As youngsters strive to obtain more autonomy from their parents and develop their own identity, their social position gets increasingly determined by their own choices and life course plans [24,[26][27][28]. Consequently, the use of merely parental SES markers as indicators of social status during adolescence may not be sufficient [27,29,30]. Following Bourdieu and Passeron [31] and others [24,26] we state that it may be more appropriate to use information about adolescents' educational level, as an indicator of their current individual social position, since this determines to a large extent ones future social class group. According to Bourdieu and Passeron [31] the educational system plays a fundamental role in the reproduction of social inequalities. This reproduction works through a combination of selection and socialization processes. Selection refers to the differential validation of cultural capital in schools: students that possess the "right" (i.e. dominant middle class) cultural capital have greater chances for academic success, while other students flounder at lower levels of education. That way, scholars from lower social classes tend to concentrate in the lower status education types [32][33][34][35]. However, schools also socialize students into particular cultures: higher status education types socialize students towards the dominant middle class cultures, while lower status education types socialize towards lower class cultures [31,34,35]. Hence, the existing social order is maintained and social inequalities are even reinforced. Considering people's health behavior, the awareness of such a process of reproduction calls the question whether social gradients in alcohol and cannabis use merely are the result of differences in students' familial background or whether schools reinforce these inequalities. In Flanders, the educational level of a student can easily be assessed by the type of education the student follows. The Flemish secondary school system is highly tracked and mainly consists of three different types of education that can be ranked in difficulty level from vocational, technical, to general secondary education. General education is a type of education that provides students with a firm theoretic knowledge foundation for going into higher education. Technical education is both practice-and theory-orientated, so that students can either enter the job market directly or continue their studies in higher education. Vocational education is a mere practical type of education and prepares students to enter the job market directly. Unlike the existing literature on socio-economic differences, this research was conducted within a multicultural environment, i.e. Belgium's capital region. The particular nature and assembly of this urban region shapes a unique context for this study, since it additionally allows us to examine whether the impact of an adolescents' socio-economic position, as determined by both familial background and education type, on his/her consumption pattern, interacts with ones' cultural origin. Brussels is characterized by a large degree of ethnic diversity, clearly reflected in the composition of the students' population in Dutch-speaking secondary education. Hence, migrant students make up about 60% of the secondary school student body. Furthermore, previously conducted analyses [see [36]] illustrated that 75% of this migrant group are Muslims. Since Muslims are overrepresented within the migrant population and due to the fact that Islamic cultures religiously and often legally prohibit the use of alcohol and cannabis [37,38], one can expect that SES differences in substance use are much more clear for native than for migrant students. As such, their religion may act as a buffer for the effect of education type on risk behaviors such as drinking alcohol or using cannabis. In summary, the present study contributes to the literature by elucidating the role of education type in students' alcohol and cannabis use, controlling for parental SES markers (i.e. parents educational level and work status). The aim of the study is twofold. First, in line with Bourdieu and Passeron [31], we expect that differences in students' consumption pattern not merely reflect the different backgrounds of these students, but that the school environment contributes something unique. Second, given the strong representation of migrants from Islamic countries in Brussels' schools, and given the enforcing rule of abstinence that is dictated in Islamic culture, we expect education type to have a an effect on native students' substance use in the first place. --- Methods --- Data The data used in this study were derived from the 'JOPmonitor Brussels'. This is a self-report school survey administered by the Youth Research Platform in 2009-2010. The data consists of a sample (N = 2,513) of 12-20 year old students in all grades (i.e. 1st to 6th year) of Flemish secondary education in Brussels. All Dutchspeaking secondary schools (N = 42) in the Brussels region were repeatedly invited to participate: 32 schools (76.2%) agreed to participate, a number that is quite high for this kind of surveys. The reason why the remaining ten schools did not participate is due to the fact that Flemish schools are commonly swamped with survey requests from researchers, generally resulting in a 'first come, first served' outcome. In each of the participating schools, classes were selected randomly based on study year and type of education. This in order to achieve a balanced representation of the Brussels student population. After being informed about the purpose and the voluntary nature of their participation, 88,6% of the students in these 32 schools (N = 2,513) actually filled in the questionnaire. The remaining 11.4% could not participate due to absence caused by illness or class excursions. The questionnaire was administrated in the presence of a researcher and a teacher during regular class periods. The data collection was approved by the Ethics Committee of the Faculty of Psychology and Educational Sciences of Ghent University. Since the subdivision between the previously described education types only exists from the 3rd year onwards, we selected respondents in the 3rd to 6th year only, resulting in a subsample of 1,488 students (i.e. 61.8% of the total sample). Despite the high representativeness of the data on both the school and class level [39] this study suffered, like most public survey data, of a substantial percentage of missing values. The variables responsible for most missing observations were the outcome variables (all around 27%), parental work status (22.7%), and parents' educational level (23.6%). The'mi' package in R developed by Gelman et al. [40] for multiple imputation was used to deal with the missings in the model. --- Measures In this paper we measured last month incidence rates of (1) beer/wine, (2) spirits and (3) cannabis by using a dichotomized scale (0) 'abstainers and moderate users' versus (1) 'frequent users'. 'Frequent users' are those adolescents who used respectively beers/wines, spirits or cannabis more than three times in the last month. Familial socio-economic background was operationalized by using information about the educational and occupational level of students' parents. Regarding the parents' occupational status we distinguished between three categories: 1) families where both parents are unemployed; 2) families where only one parent is employed and 3) families where both parents are employed. Likewise, the parents' educational level consists of three categories: 1) 'low-educated families' where none of the parents obtained an university degree or equivalent, 2)'middle-educated families' where only one parent has a university degree or equivalent, and 3)'high-educated families' where both parents have a university degree or equivalent. Education type in secondary school consists of three main categories: general education, technical education, and vocational education. Control variables included gender (male vs. female), age and cultural background (native vs. migrant students). 'Native students' are those who have the Belgian nationality, speak French or Dutch with at least one of their parents and have at least one parent of Belgian origin. --- Results --- Education type and student characteristics The results in Table 1 illustrate that the inflow of students in the different education types is strongly determined by their socio-economic background. Considering the two extremes within education type, it becomes clear that 51.7% of the students in general education grow up in a two income-family. In vocational education this is the case for only 32.1% of the students. In contrast, we counted approximately twice as much unemployed households in vocational education (24.4%) as compared to general education (12.9%). That students' education type can't be isolated from their familial background, becomes even more apparent if we regard differences in parents' educational level. The proportion of students living in 'high-educated families' is significantly higher in general education (32.4%) than in technical (17.1%) and vocational education (13.3%). For migrant students, we observed a similar trend: vocational education included the highest percentage of migrant students (77.4%) followed by technical (57.4%) and general education (52.2%). Gender differences within various education types were not significant. --- Education type and substance abuse In general, a fifth of all respondents drunk beer or wine on more than three occasions during the 30 days prior to the survey. This is roughly twice the number of students who frequently (i.e. more than 3 times) used spirits (10.4%) and thrice the number of frequent cannabis users (6.7%). Regarding the association between substance use and education type, a significant association was found for spirits and cannabis. Hence, the proportion of students who used these substances was significantly higher in vocational education, respectively 15.1% and 11.4%, than in technical education (respectively 11.3% and 7.2%). General education counts the lowest percentage of frequent strong spirit (7.9%) and cannabis (4.2%) users. These findings are in line with the raised expectation that frequent substance use is more concentrated among students in lower educational tracks. --- Multilevel analysis Because cluster sampling methods were used to collect the data and since the dependent variables are dichotomous, multilevel logistic regression techniques were used for the multivariate analysis, with the schools as the higher level units of analysis. The models were fitted in each multiple imputed dataset. The obtained coefficients and standard errors were then pooled according to Rubin's rules [41]. Finally, regular Wald tests are used to evaluate the null-hypotheses that the respective coefficients are equal to zero in the population. In Table 2, results are shown for the analyses on the full sample of 3 rd to 6th year students. Concerning the sociodemographic factors, gender, age and cultural background, were significantly associated with frequent alcohol and cannabis use. As expected, boys and older students and native students use alcohol and cannabis more frequently. As we hypothesized, our analyses display a clear effect of education type on students' consumption pattern, and these effects prove robust for differences in socio-economic background. Remember however that further analyses should indicate whether these effects are equally strong in both samples of native and migrant teens. Controlling for all other variables in the model, vocational students were significantly more likely than their counterparts in general education to use spirits or cannabis on a regular basis (OR = 3.22 and OR = 4.04 respectively). In addition, vocational students were also inclined to drink beers and wines more frequently, although the level of statistical significance was rather low here (OR = 1.75). With general education as a reference group, no substantial differences were observed for students in technical education. With regard to students' familial social position, significant effects were found only for parents' educational level. Frequent alcohol users are more often found among children of high than low educated parents. Such a finding was however not found for cannabis use. The parents' work status has no effect on adolescents' alcohol and drug use. Table 3 and Table 4 repeats the same analyses for native and migrant group students separately. This allows us to examine whether the effects of education type differ according to one's ethnic or cultural background. In accordance with Table 2, both native and migrant students' family background proved to be of limited importance. While effects of parents' work status were absent, significant effects were again found for parents' educational level. Regardless of a students' ethnic/ cultural background, growing up in a 'high educated family' increased the risk of frequent alcohol use. The findings in Table 3 further show that, for native Belgians, education type clearly structures their alcohol (spirits) and cannabis use (OR = 4.63 and OR = 6.44 respectively). These differences are again most pronounced between students in general education and students in vocational education. However, also students in technical education drink significantly more spirits than their counterparts in the highest track. No differences between educational tracks were observed for drinking beers and wines. The results in Table 4 also show that, for students from another ethnic/cultural background, no significant differences were found between educational tracks. The only exception is for cannabis use, where we found that migrant students in vocational education use cannabis more often than their counterparts in general education. --- Discussion Starting from the debate on whether a process of equalization in young people's health is taking place, we posited that in adolescence a shift is taking place away from one's familial background to a social position that is increasingly determined by students' own educational career. In the present study we focused on whether alcohol and drug use differs between students from different educational tracks, independent of their familial socio-economic background. Our results clearly confirmed this hypothesis. Students in lower educational tracks use alcohol and cannabis more often than students in upper educational tracks, although this finding held only for native students. We also illustrated that the effects of education type are not reducible to differences in students' familial background characteristics. These findings are in line with the literature on the role of the educational system in the reproduction of inequalities (Bourdieu and Passeron, 1977). Schools socialize students into particular youth cultures. Although some of these socialization is intended and part of the school culture, much of it is an unintended consequence of the interactions among students and between students and staff. A question that arises immediately when interpreting these findings is which processes within these schools can explain the link between tracking and substance use. Here, research has shown that the specific subcultures that arise in lower educational strata are characterized by a culture of demotion, feelings of held back with bleak future perspectives, feelings of futility, frustration and strain, or eroded self-esteem [42][43][44][45]. Furthermore, teachers in lower tracks have much less expectations from their students (i.e. Pygmalion effect) and also leadership styles clearly differ between education types [46]. All these experiences may push these students towards delinquency or substance use, not only to achieve alternative sources of status in the peer group, but also to handle the strain caused by their negative experiences in these lower educational tracks. The multicultural character of our study sample additionally allowed us to explore whether the effects of education type differed according to students' ethnic or cultural background. As we expected, effects of education type were observed in the sample of native students, while these effects were absent or negligible in the sample of migrant students. We explained this finding by referring to the buffering effect of religion in the relationship between education type and substance use. Students from Islamic background make out the majority in the group of students from another ethnic background, and these students not only profess their religion in a much more conscious way, they are also the product of families and cultures that put more emphasis on religious values such as for instance abstinence than their native Belgian counterparts [47]. A somewhat remarkable finding of our study was the positive association between parents' educational level and alcohol use. Students from families where parents are highly educated used more alcohol in the last month, independent of the students' educational level. Such a finding was not found for cannabis use. A possible explanation may be that students from high SES backgrounds simply have a more attractive financial background, allowing these youngsters to buy alcoholrelated products [20,[48][49][50]. For cannabis on the other hand, the financial strength of students' background may be of minor importance. For instance, cannabis is often consumed and shared in larger groups of teens, and especially the amount of cannabis needed to get high is much cheaper in price than the amount of alcohol needed to get drunk. Furthermore, cannabis is legally prohibited, and is dependent on having the 'right' connections or resources to buy these products, and these connections may be much less dependent on the familial SES background of these students. Other explanations are however possible and further research is necessarily to explain these results. This study suffered from a few limitations. A first limitation is that this study is based on self-reports, making it plausible that the these self-reports are biased by social desirability pressures. A second limitation relates to the recall period in the study (i.e. incidence rates of alcohol and cannabis in the last month). Since students may experience difficulties in remembering exactly how many times they used these substances during the last 30 days, it may be more accurate to focus on shorter recall periods. Also, the analyses presented here are based on outcomes that measure whether the adolescents consumed alcohol or drugs more than three times in the last month, and thus problematic drinking behavior is not really measured. Thus, it may be that the differences between the educational tracks, but also between the migrant and native groups, become even more pronounced when using measures of binge drinking, alcohol intoxication, etc. Because of the restrictions of our alcohol outcomes, our study did not let us to conclude that the so typical'reversed pattern' -moderate use of alcohol in higher socio-economic groups, less frequent but more problematic alcohol use in lower socio-economic groups-is also observed in adolescence. However, the literature indicates that once teenagers grow older and enter adulthood, such a pattern may start to unfold [11][12][13]. --- Conclusion Consistent with equalization theory, this study found little effects of parental SES markers on students' substance use, and where such effects existed they marked social gradients in a reversed pattern. Following West [21,22], adolescence is characterized by a "process of equalization", but only for what concerns their familial socio-economic background. Our study showed, however, that when using education type as a proxy of one's individual socio-economic status, clear differences persist among native Belgians in different educational tracks. Thus, structural causes of inequalities are still prevalent, but they work via the schools' students are ascribed to. For future research this means that it may no longer be sufficient to determine adolescents' social position merely on parental SES indicators. Rather, it is required to chart students' own social position as well, for which education type can be used as a proxy. Otherwise, researchers are blind for the diversity and SES based social stratification that still persists within Belgian, but also other European, schools. --- Author details Authors' contributions HB conducted all statistical analyses, wrote, and coordinated the paper. DC co-wrote the manuscript, RB provided statistical support, and NC was coresponsible for the data collection and supervision of the Brussels youth monitor and provided theoretical additions and other comments to the manuscript. All authors commented on drafts, have read the revised manuscript, and have approved the final version. --- Competing interests The authors declare that they have no competing interests. --- Richter M, Vereecken

Background: Research regarding socio-economic differences in alcohol and drug use in adolescence yields mixed results. This study hypothesizes that (1) when using education type as a proxy of one's social status, clear differences will exist between students from different types of education, regardless of students' familial socioeconomic background; (2) and that the effects of education type differ according to their cultural background. Methods: Data from the Brussels youth monitor were used, a school survey administered among 1,488 adolescents from the 3rd to 6th year of Flemish secondary education. Data were analyzed using multilevel logistic regression models. Results: Controlling for their familial background, the results show that native students in lower educational tracks use alcohol and cannabis more often than students in upper educational tracks. Such a relationship was not found for students from another ethnic background.Results from this study indicate that research into health risks should take into account both adolescents' familial background and individual social position as different components of youngsters' socioeconomic background.

Bisexual men are at increased risk for depression, anxiety, and problematic substance use compared to both heterosexual and gay men (McCabe, Hughes, Bostwick, West, & Boyd, 2009;Ross et al., 2018), and they are also at increased risk for HIV compared to heterosexual men (Caceres et al., 2018;Friedman, Wei, et al., 2014). As such, there is a critical need to develop interventions to improve their health and wellbeing. Many interventions are labeled as being for "gay and bisexual men," but they are typically designed for gay men and few bisexual men are included in efficacy trials (Feinstein, Dyar, & Pachankis, 2017). In recent years, HIV prevention programs have been developed for behaviorally bisexual men (i.e., men who have sex with men and women; Harawa et al., 2013), but these programs are not tailored to the unique needs of self-identified bisexual men. This parallels the general trend in research on bisexual men, most of which has focused on their sexual behavior more so than their identity (Bostwick & Dodge, 2018). However, bisexual identity is highly stigmatized, especially among men (Dodge et al., 2016;Friedman, Dodge, et al., 2014;Herek, 2002), and bisexual individuals face discrimination from both heterosexual and gay/lesbian individuals (Feinstein & Dyar, 2017). These stigmarelated experiences not only contribute to negative health outcomes (Brewster, Moradi, Deblaere, & Velez, 2013;MacLeod, Bauer, Robinson, MacKay, & Ross, 2015), they also present barriers to accessing healthcare (e.g., bisexual individuals report negative experiences with providers; Dodge, Jeffries, & Sandfort, 2008;Dodge et al., 2012;Eady, Dobinson, & Ross, 2011;Page, 2004). Developing tailored interventions for self-identified bisexual men has the potential to improve their health while affirming their identities and preparing them to cope with stigma and discrimination. In order to inform the development of interventions for self-identified bisexual men, it is important to understand their preferences for different intervention components. This is consistent with the Institute of Medicine's (2001) recommendation for healthcare to be able to respond to individual preferences, which first requires assessing them. Previous research has found that most behaviorally bisexual men prefer individual interventions due to privacy concerns, but some prefer group interventions to meet other bisexual men (Dodge et al., 2012). To our knowledge, no previous studies have examined intervention preferences among self-identified bisexual men, and there is reason to believe that their preferences may be different than those of behaviorally bisexual men. First, in samples of youth and adults, large proportions of self-identified bisexual men do not report bisexual behavior (e.g., Everett, 2013;Mustanski, Birkett, Greene, Rosario, Bostwick, & Everett, 2014). As such, findings from previous studies that have focused on behaviorally bisexual men may not generalize to self-identified bisexual men who have not engaged in bisexual behavior. Second, it has been suggested that sexual identity is the dimension of sexual orientation that is most relevant to experiences of disadvantage and discrimination (Vizard, 2014), and that the often public nature of sexual identity (relative to the often private nature of sexual behavior) may contribute to being targeted for victimization (Everett, 2013). In fact, accumulating evidence suggests that being more open about one's bisexual identity is associated with increases in depression and substance use (Feinstein, Dyar, Li, Whitton, Newcomb, & Mustanski, in press), highlighting the relevance of bisexual identity to health. Finally, there is some evidence that self-identified bisexual men differ from behaviorally bisexual men who do not self-identify as bisexual in important ways. Specifically, compared to behaviorally bisexual men who do not self-identify as bisexual, self-identified bisexual men are less likely to be married, less likely to have children, more likely to have ever had anal sex, and more likely to describe their sexual behavior as "predominately homosexual" (Lever, Kanouse, Rogers, Carson, & Hertz, 1992). These findings suggest that self-identified bisexual men may be embedded in different social communities than behaviorally bisexual men who do not self-identify as bisexual, which may affect their access to and engagement with healthcare as well as their needs and preferences for interventions. Interventions may need to address different topics and be delivered in different formats to meet the unique needs of self-identified bisexual men. For example, given the invisibility of bisexuality (Bostwick & Dodge, 2018), men who identify as bisexual may be particularly likely to prefer interventions that connect them to other bisexual men in the interest of building community. In sum, the lack of research on intervention preferences among self-identified bisexual men represents a critical barrier to developing tailored interventions to improve their health and wellbeing. To address this gap, we examined preferences for different intervention components in a subset of cisgender men who identified as bisexual who participated in a larger longitudinal cohort study. We focused exclusively on cisgender men because the broader cohort only included individuals who were assigned male at birth and, as such, there were no transgender men in the cohort. In addition to examining intervention preferences, we also examined their associations with stigma-related experiences, outness, and recent sexual behavior. --- Methods --- Participants and Procedures We used data from a subset of 128 cisgender men who identified as bisexual who participated in RADAR-a longitudinal cohort study examining influences on HIV and substance use among young men who have sex with men. In order to enroll in the broader cohort, participants were required to be 16-29 years old, assigned male at birth, Englishspeaking, and to report a sexual minority identity (e.g., gay, bisexual) or sex with a man in the past year. The subset of participants included in the current analyses were required to identify as bisexual and male. Given that the broader cohort only included individuals who were assigned male at birth, there were no transgender men in the cohort. Cohort members completed a psychosocial survey, a network interview, and biomedical specimen collection every six months and were compensated $50. Data collection began in February 2015, but the intervention preference measure was added after data collection had begun; it was administered to all participants who identified as bisexual and male who completed a visit between December 2016 and June 2017. The current analyses used data from the study visit during which each participant completed the intervention preference measure. All procedures were approved by the Institutional Review Board. The analytic sample ranged in age from 18 to 32 years (M = 22.21, SD = 2.87) and was racially/ethnically diverse (36.7% Black or African American, 30.5% Hispanic or Latino, 19.5% White, 9.4% multi-racial, 3.1% Asian, 0.8% other). Most participants (84.4%) were HIV-negative, whereas 15.6% were HIV-positive. --- Measures Intervention preferences.-Participants were asked four questions: (1) "Which of the following topics are important to address to improve the health and wellbeing of bisexual men?" Response options included: mental health; drug/alcohol use; HIV/STI; coping with discrimination/victimization; dating/relationships; other (please specify). Participants could select multiple response options. ( 2) "What type of program would you be most comfortable with?" Response options included: a program for bisexual men only; a program for gay and bisexual men together; I don't have a preference. (3) "Would you want the program to be one-on-one (a program facilitator meeting with just you) or in a group?" Response options included: one-on-one; group; I don't have a preference. ( 4) "Would you want the program to be online or in-person?" Response options included: online; in-person; I don't have a preference. Anti-bisexual discrimination.-The Anti-Bisexual Experiences Scale (Brewster & Moradi, 2010) is a 17-item measure that is administered twice (separately for discrimination from heterosexual and gay/lesbian individuals). It measures three types of anti-bisexual discrimination, including sexual orientation instability assumptions (e.g., "People have acted as if bisexuality is 'just a phase' I am going through"), sexual irresponsibility assumptions (e.g., "People have treated me as if I am obsessed with sex because I am bisexual"), and hostility (e.g., "People have not wanted to be my friend because I identify as bisexual"). Items were rated on a scale of 1 (never) to 6 (almost all of the time) and means were calculated for discrimination from heterosexual (<unk> =.93) and gay/lesbian individuals (<unk> =. 92). Internalized binegativity.-The 5-item internalized binegativity subscale from the Bisexual Identity Inventory (Paul, Smith, Mohr, & Ross, 2014) measures a negative affective response to being bisexual (e.g., It's unfair that I am attracted to people of more than one gender). Items were rated on a scale of 1 (strongly disagree) to 7 (strongly agree) and their mean was calculated (<unk> =.73). Outness.-Participants were asked, "How out are you to people around you?" Response options included: 0 = not out to anyone; 1 = only out to a few select people; 2 = out to most people; and 3 = out to everyone. Recent sexual behavior.-The HIV-Risk Assessment for Sexual Partnerships (H-RASP; Swann, Newcomb, & Mustanski, 2018) is a self-administered, computerized interview designed to assess sexual behavior and associated situational/contextual variables at the level of the sexual partnership. In the current study, participants reported on up to four sexual partnerships during the preceding six months. For analyses, a dichotomous variable was created to reflect whether or not participants reported vaginal or anal sex with any female partners in the preceding six months. --- Results First, we examined the proportion of participants who endorsed each intervention preference. Then, we used chi-square tests, independent-samples t-tests, and one-way analysis of variance tests to examine the associations between intervention preferences and stigma-related experiences (discrimination and internalized binegativity), outness, and reports of recent female sexual partners. --- Topics. The largest proportions of participants endorsed mental health (84.4%) and HIV/STI (77.3%) as important to address, but the majority also endorsed dating/relationships (70.3%), drug/alcohol use (62.5%), and coping with discrimination/victimization (63.3%). Six participants (4.7%) identified additional topics (e.g., self-love and understanding, bisexual erasure). Topic endorsement was not associated with discrimination, internalized binegativity, outness, or reports of recent female sexual partners. --- Target population. Results are presented in Table 1. Approximately one-quarter of participants (27.3%) preferred an intervention for gay and bisexual men together, 18.8% preferred an intervention for bisexual men only, and 53.9% did not have a preference. Participants who preferred an intervention for bisexual men only reported more discrimination from gay/lesbian individuals (compared to those who preferred an intervention for gay and bisexual men together) and more discrimination from gay/lesbian and heterosexual individuals (compared to those who did not have a preference). Participants who preferred an intervention for bisexual men only were also more likely to report recent female sexual partners compared to those who preferred an intervention for gay and bisexual men together. Similarly, participants who did not have a preference were also more likely to report recent female sexual partners compared to those who preferred an intervention for gay and bisexual men together. Target population preference was not associated with internalized binegativity or outness. --- Format. Results are presented in Table 2. Approximately one-third of participants (36.7%) preferred a group intervention, 21.1% preferred an individual intervention, and 42.2% did not have a preference. Format preference was not associated with discrimination, internalized binegativity, outness, or reports of recent female sexual partners. --- Delivery. Results are presented in Table 3. Approximately half of participants (55.5%) preferred an inperson intervention, 10.2% preferred an online intervention, and 34.4% did not have a preference. Participants who preferred an in-person intervention and participants who preferred an online intervention both reported more discrimination from heterosexual individuals compared to those who did not have a preference. Delivery preference was not associated with discrimination from gay/lesbian individuals, internalized binegativity, outness, or reports of recent sexual partners. --- Discussion The cisgender men who identified as bisexual in our study endorsed a range of topics as important to address in programs aimed at improving their health. Although mental health was most commonly endorsed, all of the topics were endorsed by the majority of participants. These findings highlight the need for interventions for self-identified bisexual men to address multiple health issues (including, but not limited to, HIV/STI) and to attend to psychosocial issues (e.g., dating/relationships, discrimination/victimization). The importance of attending to dating/relationships is consistent with efforts to integrate HIV prevention and relationship education for young male couples (Newcomb et al., 2017). Although interventions have been developed to address the impact of discrimination on sexual minorities' health (Chaudoir, Wang, & Pachankis, 2017), none have been developed to address the unique challenges facing self-identified bisexual men. Preferences for other intervention components varied. A slightly larger proportion of participants preferred an intervention for gay and bisexual together compared to an intervention for bisexual men only; these men may believe the health needs of gay and bisexual men are more similar than different. However, those who reported more discrimination, especially from gay/lesbian individuals, were more likely to prefer an intervention for bisexual men only. These men may recognize the unique challenges facing bisexual individuals (Feinstein & Dyar, 2017) and may not feel comfortable discussing their concerns with gay men. Participants who reported recent female sexual partners were also more likely to prefer an intervention for bisexual men only. Given that bisexual men are often blamed for transmitting HIV to women (Friedman et al., 2017;Jeffries, 2014), those who have had recent female sexual partners may be particularly aware of the unique challenges facing bisexual men and, as such, interested in discussing their concerns with other bisexual men who may have had similar experiences. In contrast to research on behaviorally bisexual men (Dodge et al., 2012), a larger proportion of participants preferred a group intervention compared to an individual intervention. Meeting other bisexual men and being part of a community that shares the same identity may be more important to selfidentified bisexual men than to behaviorally bisexual men (who may not identify as bisexual). Self-identified bisexual men may also feel more comfortable participating in a group intervention with other bisexual men. Finally, participants reported a preference for an in-person intervention compared to an online intervention, which may also reflect the importance of being part of a community that shares the same identity. Participants who preferred an in-person intervention and participants who preferred an online intervention both reported more discrimination from heterosexual individuals compared to those who did not have a preference. We did not assess whether discrimination occurred in-person or online, but it is possible that those who experienced in-person discrimination preferred an online intervention, whereas those who experienced online discrimination preferred an in-person intervention. Of note, a sizeable proportion of participants did not report intervention preferences (34.4% to 53.9%). Given the lack of interventions for self-identified bisexual men, they may simply be interested in an intervention tailored to their unique needs regardless of the format. The current findings should be considered in light of several limitations. First, given that all of our participants were cisgender men living in Chicago, it will be important to examine intervention preferences among transgender men and in different geographic regions. Transgender people are particularly likely to use labels that reflect attractions to more than one gender/sex (e.g., bisexual, pansexual; James, Herman, Rankin, Keisling, Mottet, & Anafi, 2016), and transgender people who identify as bisexual or pansexual tend to face greater hardships than those who identify as gay, lesbian, or heterosexual (Movement Advancement Project, 2017). As such, transgender people who identify as bisexual or pansexual may have unique needs and preferences for interventions. Further, there is some evidence that the experiences of people who identify as bisexual differ from those of people who identify as pansexual, queer, or fluid. For example, Mitchell, Davis, and Galupo (2015) found that people who identified as bisexual reported experiencing more interpersonal hostility from gay/lesbian people and feeling less connected to the LGBT community than did people who identified as pansexual, queer, or fluid. For that reason, it will be important to examine intervention preferences among people who use labels other than bisexual to describe their attractions to more than one gender/sex. Of note, cisgender men are more likely to use the label bisexual to describe their attractions to more than one gender/sex than they are to use other labels (e.g., pansexual, queer; Katz-Wise, Mereish, & Woulfe, 2017), and cisgender men are less likely than cisgender women and gender minorities to identify as pansexual or queer (Morandini, Blaszczynski, & Dar-Nimrod, 2017). Therefore, it may be particularly important to consider the range of labels used to describe attractions to more than one gender/sex in samples of cisgender women and gender minorities. Additionally, preferences for different intervention components may depend on the focus of the intervention (e.g., mental health versus HIV/STI). Finally, we did not ask participants why they preferred different intervention components, which future qualitative research could address. Limitations aside, the current study is the first to examine self-identified bisexual men's intervention preferences and findings will help inform the development of tailored interventions to address their unique needs. In sum, interventions for self-identified bisexual men would benefit from addressing a range of health and psychosocial issues. While in-person and group interventions may appeal to larger proportions of self-identified bisexual men, the appeal of an intervention for gay and bisexual men together compared to an intervention for bisexual men only may depend on individual and social/contextual factors. Our findings also highlight the importance of attending to the needs of self-identified bisexual men as their intervention preferences differed from those of behaviorally bisexual men in prior research. As noted, there are no evidence-based interventions tailored to the unique needs of self-identified bisexual men. The lack of funding available for research focused on bisexual men's health presents a major barrier to developing such interventions. From 1989 through 2011, less than 1% of all studies funded by the National Institutes of Health (NIH) focused on LGBT health and, of those, only 14.8% involved bisexual men (Coulter, Kenst, Bowen, & Scout, 2014). Additionally, while foundation funding for research on LGBT populations reached a record high in 2016, less than 1% specifically focused on bisexual individuals (Funders for LGBTQ Issues, 2016). The recently established NIH Sexual and Gender Minority (SGM) Research Office has and will continue to play a critical role in advancing research on SGM health, but there is still a need for increased attention to the health disparities affecting bisexual men. Policy makers can also help to reduce these disparities by enacting laws and allocating funding dedicated to this population. Identity and minority stress as a function of preferred intervention format. Table 3. Identity and minority stress as a function of preferred intervention modality.

Self-identified bisexual men are at increased risk for negative health outcomes, but there are no interventions tailored to their unique needs. In order to develop interventions for this population, it is first necessary to understand their preferences. As part of a larger study, 128 cisgender men who identified as bisexual reported on their preferences for different intervention components. Large proportions of participants prioritized addressing both health (e.g., mental health, HIV/STI) and psychosocial experiences (e.g., dating/relationships, discrimination/victimization). A slightly larger proportion of participants preferred an intervention for gay and bisexual men together compared to an intervention for bisexual men only. However, those who reported more discrimination and recent female sexual partners were more likely to prefer an intervention for bisexual men only. Larger proportions of participants preferred a group intervention compared to an individual intervention and an in-person intervention compared to an online intervention. These findings highlight the importance of addressing both health and psychosocial experiences in tailored interventions for self-identified bisexual men. Further, while in-person and group interventions may appeal to larger proportions of self-identified bisexual men, the appeal of an intervention for gay and bisexual men together compared to an intervention for bisexual men only may depend on individual and social/contextual factors.

INTRODUCTION Learning is wrong one part especially in life human. Start from man born until end his life no time free from activity this. Sort of mandated in the 1945 Constitution paragraph 4 year if for educate life nation already arranged in law controlling base various regarding about learning as effort for upgrade intelligence Indonesian people. Regarding the demand Public Indonesia for support government, with method demand knowledge on institution learning official. Learning is effort aware to use grow ability source energy man go through the educational process with give encouragement as well as means study teach at school. Quality learning will affect quality something country in various field, by because that needed good quality from source energy man so quality good learning can materialized as well as able advance whole aspect forprogress country. Perceived learning process by everyone doesn't only occur within the school area, family area as well as inhabitant also bring influence in the learning process a. Rules that report the learning process no only tied at school is law No. 13 of 2015 which reads "street learning consists on learning official, non-formal, and informal as possible alternate Fulfill and enrich." Effort government for support success education official with stage replacement curriculum learning so could explore era growth. Curriculum School Below/Madrasah Ibtidaiyah already implemented since year the 2013/2014 teachings were spoken. Effort for Upgrade ability source power man worth owned height for progress nation and country is through education. Education given formally at school or environment society. According to Sadirman(2018) " education is change Act in demand child educate going to maturity". According to Dimyati and Mujiono (2013)" education is participant educate must capable study and grow to be able Becomes quality future generations." (Darsani, 2012) Participant moderate student in the educational process certain will experience a number of consequence that is from facet knowledge nor his skills. Education that very urgent in the life, by because that there is a number of component such as. curriculum, teachers, students, facilities and infrastructure, parents and environment very family role urgent in the educational process. Frequent bottlenecks occur in the educational process is low achievement study student. Because there is a number of factors that influence it. Naturally for reach achievement satisfying learning participant educate must do a number of principle, wrong sat that is must have motivation and spirit for learn. According to Sadirman (2018) motivation is changes that occur in the self somebody for reach something purpose. According to Dimyati and Mujiono ( 2018) "motivation that is must develop spirit study and try Becomes more good in the learning process so after done education can get good job (Sardiman, 2018). Next influencing factors achievement study student is parents work, because parents job that very role urgent in the learning process students, especially in need tree student like eat, dress and health and also there is need means and necessary infrastructure fulfilled, like tables, chairs, bags, tools write, book and others. All facility the will fulfilled if originating students from capable family and have many money, however if originating students from families whose parents work low or no capable that will impact bad to achievement study, because student that feel inferior and embarrassed for gather with his friends (Setiawan, 2017). This is what matters to achievement study students, so there is a students who are looking living for help his parents look for money, though actually age they not yet proper for work, however circumstances like this precisely will Becomes something lesson and spirit for a student that Becomes more good in study and later will become a successful person. On the contrary originating students from capable family usually will always pampered in matter whatever, the make student not enough focus in study so that will caused bad in achievement learn it (Rini, 2017). On basically, growth and development on child is something to be expected by everyone, good boy and also useful for everyone along no interested for do possible actions harm self alone and others. Hope this will far more easy materialized when parents must give attention to his son every day, though many must work done but educate child must permanent worked on. Something bunch social consisting from parents and child called with working family look for living for need family and her children namely father. As head obligatory father's family work, however sometimes Mother also work help economy family. Parents job that is something obligation because income utilized for need means and infrastructure study child at school nor at home, even in millennials naturally need study child many copy at the present time example like cell phones, laptops and so on. so from that material from parents very urgent for help activity study child. Slameto(2019) states there is two influential factor to achievement study student that is factor external consisting from factor family and school. Whereas internal factors that consist from health physical, psychological, and exhausted. Economic parents sufficient Definitely will prepare good facilities for his son, however on the other hand the parents are the economy low, not yet of course can give good facilities for his son (Slameto, 2019). Type work done by parents is something thing that can influential to economy family. According to Slameto in 2019 separately theory can explained economy family is something very situation tightly relation with study child. Obtained opportunity for parents through profession could Fulfill available facilities for child, and if have complete facilities for support smoothness study as well as could affect success study child. There is difference between parents who have steady job he capable guard security his family more calm, rather than working parents as odd jobs or unemployed. According to Ngalim Purwanto in 2018, ability economy in family in a manner live and no live could influence profession and education or state, and consider results education as well as accomplished work. Parents job also tightly relation with fulfillment need his son in matter means and infrastructure for support learn it. So that child Becomes spirit study and value obtained increase (Ngalim, 2018). Parents job depend on Skills and opportunity their education get. something different parental occupations naturally could influential against success study child. For example like mood, pattern foster, activity and interaction family also can cooperate with type profession certain. Example type influencing work attitudes, parenting processes, such as: parents who continue work impact bad that happened that is no attention again to his son especially on learn. Children whose parents work in the private sector have motivation discipline and different learning from family whose parents is a teacher or employee country. Working parents sector private no have attention to development academic child. While the parents are working as gurud could embed discipline in learn. could differentiate Among time play and time study to child, as well could motivating child for achievers in all matter and as well as other factors such as environment House could influence success study child. This is what it becomes influence achievement study student. Often found a students whose parents are economics low, inclined more lazy study because less means and available infrastructure at home. Child also no have enough time for study because must help his parents work look for money, from sisni can be seen that economy family that very influential to results study student (Slameto, 2019). Based on research at Adhyaksa Senior High School, Jambi City wrong one students whose parents work as labour, with income no stay and no can Fulfill need means and infrastructure his son, so very influential to motivation and results learn it. Parents income that very urgent for need study students, for Fulfill the required facilities for learn. By because that facility means and infrastructure is role main for reach achievement good study and maximum. --- METHOD Our group uses method quantitative (Fajri, 2018) get data through interview live source person namely the BK teacher. We analyze with total population 6 locales with total whole student as many as 192 students. as for the samples we examined 2 local with total student as many as 64 students. Method study quantitative obtain the data with focus on number or numeric in a research, besides that data is obtained from survey results of respondents through questions asked through google forms to student/at school (Saebeni, 2018). Formula problem in study that is is parents job influential to mark academic high school students. As for technique used namely Purposive Sampling. Purposive sampling is technique data retrieval random. Source of data used in study this are primary data and secondary data. Primary data is data obtained live from results interview about influence parents job to results study students of Adhyaksa 1 Senior High School, Jambi City. While secondary data is data in the form of existing documents at high school. Location the school we made object study is school private level medium above SMA Adhyaksa I City Jambi Jl. general. UripSumoharjo, No.33, Sungai Putri Village, Jambi City, Telanaipura, Code Heading 36122. Here's what happens object in study as many as 64 students SMA Adhyaksa 1 Jambi City in particular for class XII IPA and IPS (Sugiono, 2017) --- RESULT AND DISCUSSION From the observation process that has been carried out, there is actually data that clearly shows a relationship that can be considered a correlation between academic achievement and parents' work. The data from the research results can be seen in table table 2. From the table on obtained results that is parent students his work as Laborer as many as 17 students with percentage mark academic tall by 35% and low by 65%. As for parent students his work as civil servants as many as 15 students with percentage mark academic tall by 60% and low by 40%. Whereas parent students his profession as Retired as many as 13 students with percentage mark academic tall by 62% and low by 38%. And students are parents his work as Trader as many as 19 students percentage mark academic tall by 47% and low by 53%. Based on yield data research obtained, students whose parents work as laborer as many as 17 students, from 17 students the there is percentage mark academic tall by 35% and low by 65%. From percentage the We could knowing that students whose parents work as laborer tend worth academic low, however no close possibility for students whose parents work as laborer for worth academic tall as we are know, that no all laborer educated high, (Sunain, 2017) so that he not capable accompany his son study with fine, but no all laborer economic, laborer alone divided Becomes two class that is laborer daily permanent and laborer daily off (Krismiati, 2017). Laborer daily permanent tend economic good the article he work every day with fixed salary on something company or factory, salary the tree he accept also permanent every the month and categorize enough (Lavasani et al., 2011). Whereas laborer daily free tend economic not enough fine, the article he work in a manner not stay so that the income he accept also no erratic. Wages the tree he accept usually only enough for Fulfill need everyday, so he not enough capable for facilitate his son with means and infrastructure support good study. Students whose parents work as Laborer most mark academically earned tend low because low parent education student that, no all laborer earn tall so that could influential to the learning process student the. For example laborer daily let go which is not every day get work, by because that income from laborer the no erratic so that he difficult for complete means and infrastructure study his son. Whereas laborer permanent like laborer Factory the income he get tend permanent and enough for complete means and infrastructure study his son. Students whose parents work as most civil servants mark academically earned tend Tall because, because of parents the educated worthy so that he capable educate his son with ok. Besides That's an income civil servant permanent and enough for complete means and infrastructure study his son so that his son spirit for study and obtain high value. However no all parents could notice his son with good because busy outside house so that his son not enough get attention from parents so child the no have spirit in study so that child the obtain mark academic low (Krismiati, 2017). Students whose parents profession as retired most mark academically earned tend tall, even though his parents profession as retired will but their parents education worthy and capable for educate son, the income that he accept permanent and enough for complete means and infrastructure study his son. She also capable give decent life for his son so that child the have spirit in study and have mark academically inclined high. Students whose parents work as trader most mark academically earned that is stable Among tall and low, because income trader tend fixed, with sufficient income he capable give means and infrastructure good study for his son, so child the obtain mark high academics, for example 5 foot traders whose income only enough Fulfill need his life everyday, he no capable for give good facilities for his son so that child the mark academic tend low (Suyono, 2017). From table 2 above could we know that students whose parents work as laborer as many as 17 students and percentage mark academic student from the highest by 35% and low by 65%. From the data above could concluded level education parental end very influential urgent for student that the more tall parent education the more he potentially capable accompany his son in matter study so that child the not only get education from school or from teachers but their parents also capable accompany child the for deepen learning at school. Parental attention to his son also influential urgent to children, because parents tend to many choose time with his son very potentially that he capable understand character from his son, so he could direct his son becomes more ok. Parents income also very influential to children, parents who earn enough or more from enough could complete means and infrastructure study child the (Acar et al., 2015;Friesem, 2016). So that child the spirit in study so that he potentially get mark academic high (Hasgimianti, 2017). Besides parents occupation, parents education also influential to achievement study child, if parent education low of course the job you can get also what yes, wrong only one labour. Working parents as laborer of course could Fulfill need his son but depends with income that can. Because no all laborer earn stay. So from that achievement study child tend more low, because not enough his need the means provided by his parents. Slameto (2019) suggests that factor parents job of course very influence learning. If a child life in less family able, need basically no fulfilled, which affects his health. Consequence other is child always covered sorrow, make they no want to gather with friend her friend because of inferiority, which will hinder learning them. Even children must work look for living for helping his parents, even kids who don't have time for study also will hinder the learning process his. Besides poor health, condition bad family could cause deficiency means and infrastructure study for child (Slameto, 2019). Sometimes children have a lazy nature, if this trait if allowed will study diligently behind this friends. This lazy nature could be came from his parents. Parents do not pay attention his son. Let him go without ever worrying about him because I have a job then parents who have many children of course feel disturbed, so that attiontion to one of their children becomes less. This, the child is monitored all the time accompanied by their parents feel motivated to make their studies more active. From the results sample obtained wrong one student Adhyaksa initials IB from class XII IPS 3, ia wrong one student with circumstances economy not enough can, father work as entrepreneur not settle down, meanwhile her mother no work, will but his family, because condition family he often late in payment school tuition fees, school already take Policy with reduce half from tuition fee each per month. because condition family he also often insecure at times gather with friends, because that he often no enter school or skipping from school (Ridwan, 2019). Achievement learn very bad, the value obtained also very low because seldom enter school and seldom work assignments, the teachers too confused for give value, moment meeting increase class his name always enter into the student problematic, he also always get school grade point average ranged from 75-80. As Adhyaksa High School Counseling teacher, influence economy family to achievement study student very influential because of the average economy family weak students have a sense of inferiority moment gather with his friends, because of this sense of inferiority choose for no enter school and gather with lucky friends same with him. Can be known that child whose parents work as civil servants will get attention and motivation distant study more good because of his parents have many time for give spirit and home learning. in comparison with child whose parents non-PNS work. However there is a number of children whose parents are civil servants not enough get attention and motivation study, possibly his parents too busy with his job so that needy child spirit and motivation from parents his ignored so course, that very impact laborer to achievement study child. No all child whose parents work as civil servants, smart and always get good performance, that is all depends with his parents respectively each. No can be compared between working parents as civil servants and non-PNS. Because parents job no ensure success child. Students whose parents work as civil servants as much as many as 15 students with mark academic tall as much as 60% and low as much as 40%. Majority mark academic his students tend more high, because his parents salary as civil servants enough for Fulfill need study his son at school nor at home as well as with parent education acquired knowledge capable teach to his son. Already becomes obligation for parents for Fulfill need study child and pay attention to the learning process child at home and arrange time study child with ok. So that child get good achievement (Acar et al., 2015;Wells et al., 2019). But there is a number of students whose parents work as a value civil servant academic student tend low because parents don't attention to his son so that study child decreased, if child too ignored because of his parents busy work, it is will impact bad to results study child. parental role in the learning process very important. So as parents must can share time Among profession and attention to his son. If attention of his parents good certain results study child also ok (Hasgimianti, 2017). Although parents work as trader should child more spirit in demand knowledge no rather lazy laziness, because of parents work for child his school, parents work with full spirit however child study at school just play play just. Should as child understand with parents condition and can proud of it, with study hard and get good achievement that already Becomes something pride for parents. Parents already feel succeed in support his son school in form material nor physical. But inside study this could seen tend mark academic child more low, because lack of motivation and spirit study child. And likely child also less attention with his parents, because of his parents busy work. parents only think about profession his so could give means and proper infrastructure for son, no think about impact the bad. Parents are certainly very corcerned about the education of their children. Many parents want their children's education highest. No wonder parent's are looking educational intitutions that are considered good for their children. Parent's can forget that the institution that makes their child human in the family. Students whose parents profession as retired as many as 13 students with percentage mark academic tall by 62% and low by 38%, value academic student tend tall because income retired sufficient for fulfill means and infrastructure his son. even his parents profession as retired will but their parents education worthy and capable for educate son, the income that he accept permanent and enough for complete means and infrastructure study his son. She also capable give decent life for his son so that child the have spirit in study and have mark academically inclined high (Sunain, 2017). Children get high value no only his parents capable fulfill need means and infrastructure her, but parents also capable give attention to son, no too indulge child, and behave naturally just as parents. Of course parents who have retired have what an experience normal in matter educate son, because years year educate child his students at school. Certainly capable educate child his alone at home with well, so child get good achievement. Students whose parents work as trader as many as 19 students with mark academic tall by 47% and low by 53%, working parents as traderhave income earned that is stable because income trader tend still, if income earned could sufficient her children with give means and infrastructure complete study so child the will spirit in study so that value obtained that is tall so parents Becomes proud with results the achievements enough ok. While the parents are working as trader vegetables around with no income permanent only enough for Fulfill need his life everyday, then child the not enough get facility study so that study child decrease the achievements not enough well, as well parental attention that is not notice development study child could trigger child no learn. Although parents work as trader should child more spirit in demand knowledge no rather lazy laziness, because of parents work for child his school, parents work with full spirit however child study at school just play play just. Should as child understand with parents condition and can proud of it, with study hard and get good achievement that already becomes something pride for parents. Parents already feel succeed in support his son school in form material nor physical. But in study this could seen tend mark academic child more low, because lack of motivation and spirit study child. And likely child also less attention with his parents, because of his parents busy work. parents only think about profession his so could give means and proper infrastructure for son, no think about impact the bad. According to Syaifullah (1981) stated something circumstances parents social could influence attitude child to education in life human. For example namely economic status that can determine parental abilities in matter provide facility study for learning media at school. Besides that's a great teacher role urgent in a learning process so that could Upgrade achievement study student. Then the teacher must imagine in the learning process like make method delivery material interesting media and pleasant so that student could understand with, so student no experience fed up or bored on moment learning going on (Suyono, 2017). Trough the educational process undertaken highly educated parents receive knowledge, skills and emotional abilities that can help solve various problem faced by children, both related to meeting children and learning at school. Parent's with high sosia economic status do not experience much difficulty in buying books. Lessons pencils, rules necessary for their children's learning. Children's with appropriate socioeconomic backgrounds have more opportunities to be creative and have their needs met. Meanwhile, children whose parents come from economically disadvantaged backgrounds are unable to cover the child's education costs. This certainly affects the motivation children. The consequence of course, is that the child does not completed his studies solving actions carried out by using time and energy to participate in learning activities. From in this reality, teachers and parents play an important role and are responsible for the future, so that these students can complete all subjects and pass national exams with better results. The role of parents in a family is quite complex including guiding, nurturing, controlling and educating children. Achieving this task is not easy. Throughout the history of human civilization, the role of parents in shaping and coloring discipline and morals is enormous. The birth of a nation that is creative, innovative, superior, has high morals, is discipline dand has an open mind cannot escape the touch and polish of parents. In other words, the family environment is very important in still and socialize moral values and discipline in children. Real participation in simple things in the small family community and the active role of parents made the most fundamental contribution to the enlightenment of civilization. A beautiful family to create a good society. Academic discipline can be developed and encouraged through training and education or instill habits that must be started in the home environment, since childhood, and continue to grow. With discipline training, students should not be overly burdened with the existing rules at school. However, some students have not implemented the learning routine. Learning discipline is an effortto control one self and the mental attitude of individuals or society by developing obedience and respect for rules and regulations based on impulses and perceptions that arise from with in their hearts. Learn with discipline Is one of the things that guide students in learning at school. Learn discipline meaning often go to school. Student discipline can be started from regular habits includes students' ability to make good use of their time, a sense of responsibility for their teacher's homework, a sense of belonging and responsibility for class arrangements and preparing lesson schedules. With self-awareness carrying out study routines, hopefully all activities will run smoothly done every day at school can give good results according to pedagogical goals but also goals national Education (Fernet et al., 2020;Scherer & Siddiq, 2019). Learningis a process of gradual and continuous change in all aspects of behavior throughout life. Besides that, students must provided with intelligence emotional and social. Results study very tightly relation with performance student. Understanding performance study is results from individual and group could resolved and created. Achievement is ability real special achieved optimally as results study and also actiona necessary prevention done so far where student has control given material by the teacher then generated evaluation from results embodied learning in form certificate. There is a number of understanding profession according to experts namely : according to Ismantoro Dwi Yuwono (2011), work is purpose used for obtain income with ability physical, fine temporary nor stay. Whereas according to De George, work is something skills that can do activity tree for produce something living for sufficient life. From several expert opinion could concluded that profession is with do something activities that can obtain income for need his life. Parents really influential in hold role urgent in an educational process her children they have not quite enough answer to education child exists encouragement channeled motivation from parents to child really urgent will push attitude from child and change behavior in not quite enough answer discipline success a child in achievement learn influenced from parental way direct method study his son including concerns high emotional. Children have knowledge fast responsive in all potency education you have attitude a child good or bad also including to in parental way direct wrong and true something act level parental behavior education more tall possible more believe will ability they in teach and help child they in study conversely, for educated parents low most from they realize meaning not enough urgent his something education to his son so parents not enough attention to education his son. So from survey results can is known that circumstances impact parents job very influential to achievement study student for example from facet parents job income as well as education last parents. With exists facility adequate study like means and infrastructure study that is books, bags, shoes and other. Could motivating child in study so that child becomes spirit. In environment family is reflection from what have taught to his son from since early. this show that something parental motivation very role urgent in give encouragement as well as give opportunity for child for achievers more well, so you can please parents (Rini, 2017). --- CONCLUSION We as composer article could conclude that student At the Adhyaksa High School, Jambi City, data were obtained from students whose parents work as laborer as many as 17 students with percentage mark academic high 35% and low 65% due no all laborer educated tall so that he not enough capable accompany his son study with well, students whose parents work as civil servants as many as 15 students with percentage mark academic tall as much as 60% and low as much as 40% due he capable give means and sufficient infrastructure for his son so that child the have spirit in learn. Students whose parents profession retired as many as 13 students with percentage mark academic tall by 62% and low by 38% due though received salary only wages tree will but he capable give means and infrastructure his son. Students whose parents work as trader as many as 19 students with percentage mark academic high by 47% and low by 53% due income trader no stay and usually income the only enough for Fulfill life daily so that no capable for facilitate his son. Besides parental income, parental attention to readiness study child is wrong one factor urgent because child also need attention from parents to children the can grow spirit in study so that child the get good achievement.Then suggestions for researcher next expected capable do research to be come, able compare Among parents job to results study student. It's good for study next do with analyze other factors that can give quality study student have mark high quality.

This study analyzes student learning outcomes in terms of the parents of SMA Adhyaksa 1 Jambi City. This type of research is quantitative. The subjects of this study were class XII students of SMA Adhyaksa 1 Jambi City. As for the population, there are 192 students in class XII, with six classes; there are three classes for science and three for social sciences. In the science class, there are 29 students and 67 girls; Meanwhile, there are 50 students and 46 girls in the social studies class. The sample that filled out the Google form was 64 students from 2 science and social studies classes, the technique used was purposive sampling. The data collection method uses the Google form to compare students' academic grades with their parents work. Parents' occupations were used as the object of study, namely Laborers, Civil Servants, Pensioners and Traders; From the research results, it was found that there were 17 students whose parents worked as labourers with a high academic score percentage of 35% and a low academic score of 65%. Whereas for students whose parents are civil servants, there are 15 students with a high percentage of 60% and 40% with low academic scores. Meanwhile, students whose parents are retired have a high academic rating of 62% and a low academic rating of 38%. And students whose parents work as traders, 19 students with a high percentage of 47% and 53% with low academic grades.

Background Fear of vaccines dates as far back as vaccines themselves as evident by Edmund Massey's [29] Sermon titled "A sermon against the dangerous and sinful practice of inoculation." This appears to be the first objection to any forms of inoculation to prevent disease with Massey stating "Let us not sinfully endeavour to alter the Course of Nature" [29]. Next in notable vaccine objections was when the smallpox vaccine was introduced, "many skeptics [...] found it counterintuitive that deliberately infecting a person with a disease" [38]. This is when the world began to see a group of people who not only refused vaccination but made an effort to inform others of the "dangers" through propaganda. However this propaganda largely consisted of arguments of infringement of rights and anti-socialism [15]. Since the infamous 1998 paper by Andrew Wakefield, this was later retracted because it incorrectly related the MMR vaccine to autism, a group of people known as vaccine deniers or more commonly known as antivaxxers have been exponentially growing. A vaccine denier or anti-vaxxer will be defined in this study as someone who believes vaccines do not work, are not safe or refuse vaccines for themselves and their children if applicable. Claims about vaccine safety, efficacy, and adverse effects have been evolving and have now spread to almost every vaccine available. Surveys from the American Academy of Pediatrics found that the rate of parents who refused one or more recommended vaccines increased from 9.1% in 2006 to a staggering 16.7% in 2013 [30]. The problem being faced today is the wealth of information that is not only accessible but easily spread across social media platforms regardless of veracity. It is clear that the internet is now patching a significant time in health literacy and decision making. A survey by Fox [18] found "(72%) [Of] adult internet users say they have searched online for information about a range of health issues[...] (26%) adult internet users say they have read or watched someone else's health experience about health or medical issues in the past 12 months. And 16% of adult internet users in the U.S. have gone online in the past 12 months to find others who share the same health concerns. There are an estimated 58 milion followers on anti-vaccination pages across socialmedia ]1]. This study uses Social media propagation, to reach the study population. The purpose of this study is to evaluate the current knowledge and beliefs about vaccines in people who use social media and the differences between scores and demographics. Past research has found that the strongest influence for positive vaccine views is having factual knowledge [11]. This study hopes to determine how much factual information is known and what differentiates social media users who have adequate and inadequate knowledge. This cross sectional study hypothesizes that there are correlations between each of the individual demographic questions and the respondent's knowledge of vaccines. Knowing what differentiates people who have adequate and inadequate knowledge can be beneficial for determining how to reach people in future vaccine campaigns. This information could also be used in an attempt to better educate or correct misinformation by means of social media in those who classify under this study as having insufficient knowledge. Additionally this research will investigate the individual's beliefs about vaccinations in people who use social media, whether positive or negative. Uncovering individual characteristics in the respondents with their different beliefs can be helpful for deciding how to correctly reach each person group in the future in regards to vaccinations. We can use this information for educational purposes through social media to help those who were found to have negative or lower beliefs toward vaccinations. Knowledge about vaccines, both true and false, can be easily accessible but also easily confused. The internet has become a huge influence on vaccine knowledge [26] and the emergence of social media has created a vast community that allows multi-person discussion to happen instantaneously and with little supervision [9]. "Anti-vaccination activists use [social media] to disseminate messages, facts and beliefs that oppose some or all recommended vaccinations" [6]. Brunson [8] found that the most significant factor influencing parents online is the percent of parents within a parents online network that are nonconforming. Because of these factors refusal to vaccinate is becoming a concern of public safety. Regardless of vaccination rates, legitimate information about vaccine safety is "not reaching parents in an effective or convincing manner" [19]. This study seeks to assess whether factual knowledge is reaching social media users. This is because it has been found that the strongest predictor of positive attitudes towards vaccines is better knowledge [11]. In a study of Serbian University students it was found that 47.9% of participants thought that giving multiple vaccines at one time overloads the immune system [11]. Further research will determine if views such as the one above are perceived across all social media users and not just students. Investigations of the dissemination of vaccine knowledge across social media have been substantially less studied until recent years, especially with the development of the COVID-19 vaccine. In a study of vaccine attitudes in Twitter users Mitra et al. found that antivaccination tweets had a wider reach (seen by more people) than pro-vaccine tweets [31]. In a study of HPV vaccine tweets it was found that there was "an association between prior exposure to negative tweets about HPV vaccines and the subsequent posting of negative tweets about HPV vaccines" which allowed for the sharing of negative opinions to more susceptible people [14]. From the information aforementioned it can be predicted that the majority of information about vaccines on social media has a negative connotation. What is still unclear is the ability of the general population to distinguish between fact and fiction, and whether this information has an influence on their knowledge and eventually decisions. To determine what questions would be used and what is to be expected by demographic results this study looked at Duggan and Brenner's [13] journal article on social media users. This study found several useful social media user characteristics. Firstly, women are statistically significantly more likely to be social media users compared to men. This study also shows that with age, there is a statistically significant progressive decline in social media use. In terms of education they found that social media use was approximately equal across all levels of education. Equity across races was also seen. This demographic data is important for determining what questions should be asked, but also what can be expected in the results. It is important to determine what factors influence people to not vaccinate. Yaqub et al. found that "'distrust of doctors', 'distrust of government sources', and 'distrust of pharmaceutical companies' as reason for hesitancy" [40]. In that same study they found that very few people reported that they did not have access to adequate information. While this is important, it does not elaborate on the quality of information these people are using. A cross sectional study in Australia in 2012 found that although 92% of 452 parents reported that their children were adequately vaccinated, 52% reported concerns including but not limited to vaccine safety and source of vaccine knowledge [10]. However due to discrepancies in defining anti-vaxxers or people who are vaccine hesitant, this study will focus on what knowledge people have and not ask their vaccine practices. Rather, it will look into how far the anti-vax messages have reached and the characteristics of those participants that have or have not been swayed by these messages. Since the collection of the data related to this research the world has faced the COVID-19 pandemic. Social media and vaccine hesitancy has become a huge topic of discussion and research relating to combating COVID-19. Several studies worldwide have indicated that older individuals, females, those with higher incomes and those with higher education levels were more likely to accept a vaccine [34]. Research in this area has exploded due to fast paced development and deployment of the COVID 19 vaccine. Globally there has been varying intention to get the COVID-19 vaccine ranging from 41 and 89% [17]. A survey conducted in the United States of America showed only 57.6% of respondents intended to be vaccinated but also that "(62%) [of respondents] believed that sociopolitical factors and pressures may lead to a rushed approval for the COVID vaccine without the assurances of safety and efficacy" [27]. --- Methods --- Collection This research was conducted by a cross sectional multiple choice study created via Survey Monkey. Survey Monkey subscription was provided by Saint James School of Medicine and was chosen based on the platforms ability to export data to IBM's SPSS. Data of both qualitative (demographic) and quantitative (Belief and Knowledge Scores as described below) nature were collected. Survey was designed to be completed in 5-10 min. It was designed with no open ended questions, no intended question bias, and no implied judgement. Survey questions were validated by pilot participation and follow up to determine if there was any ambiguity that needed to be addressed. Demographic questions were designed to be all inclusive. Belief and Knowledge questions were based off of key arguments of vaccine deniers as determined by popular social media posts. Research of key argument involved exploring multiple social media platforms and investigating posts/comments regarding vaccine denial. --- Inclusion/exclusion Subjects were included based on completion of all parts of the survey. All demographics (country, race, gender, socioeconomic status, preferred social media platform) were included except for those under 18 years of age. Those participants under 18 were excluded because in most countries the age at which an individual can consent for medical treatment such as vaccination is 18 year old [39]. Therefore, vaccine decision making if deferred to parents and guardians over 18 years of age. Those participants who no do not consent to their information being used for research purposes (first question of the survey) were excluded from the study. Survey's that contained any missing data/questions were excluding from analysis as scoring could not be completed. The survey was designed to not allow advancement onto next question without answering the current question. If participants clicked out of browser before completing final question their survey was invalidated and subsequently not extrapolated for analysis. Survey was available for completion from August 15, 2018 till November 1, 2018. --- Sampling Snowball sampling of social media users was used. Snowball sampling was used to help perpetuate the survey through social media, where social media is the quality of referral. The study population was aimed at being as demographically diverse as possible among people who use social media. Snowball sampling was chosen in particular for its ability to perpetuate hard to reach communities, such as those who identify as vaccine deniers [23]. Additionally, using Facebook with snowball sampling is effective at diversifying the geographical scope and increasing the overall response rate [3]. --- Recruitment Subjects were recruited through the three largest social media platforms (Facebook, Twitter, Instagram) via a shareable web link and asked to consent before completion of the research survey. Initial survey link was posted publicly on social media platforms outlined above with information on the survey and encouragement to further share the survey once completed. A web survey was chosen due to its ease, speed, cost, and ability to obtain a geographically diverse population [20]. --- Questions All questions in the survey were only available in English language and can be found in the supplementary material, document 1 titled Survey. The first half of the survey consisted of demographics and questions pertaining to use of social media and its relation to vaccine information. The latter half of the survey had six questions relating to vaccine knowledge and six questions relating to vaccine myths. The method of scoring was designed by the authors to create a numerical scale for comparative analytics. Lack of knowledge and belief in myths is not a negative feature but more so an area of improvement and discussion. As such, the design only uses positive numerals for scoring each question. This scoring system was created by the authors of this study specifically for this research. Question content was selected by authors through observation of social media posts pertaining to vaccines (both pro vaccine and anti-vaccine content) for common misconceptions and rebuttals. The six vaccine knowledge questions were scored on a two point scale. Questions were scored by awarding two points for the answer of belief in the vaccine statement, one point for uncertainty, and zero points for the answer of disbelief in the statement. All questions were then totaled for a score on a 12 point scale. Higher values (9)(10)(11)(12) suggesting adequate vaccine knowledge, middle range (5)(6)(7)(8) suggesting some vaccine knowledge but with uncertainty and lower values (0-4) suggesting inadequate vaccine knowledge. This score could then be appropriately analyzed. The six vaccine belief questions were scored on a two point scale. Two points were given for the answer choice "disbelief in the vaccine statement", one point for uncertainty, and zero points for the answer of belief in the statement. All questions were then totaled for a score on a 12-point scale. Similar to knowledge values, higher values were indicative of disbelief in common myths, whereas lower values indicated a belief in common myths. This score could then be appropriately analyzed. --- Analysis All data analysis was conducted using IBM's SPSS. Significance testing was performed using the Welch test. This test was chosen based on the negatively skewed data distribution with non-homogeneity of variances and sample sizes [16]. The Welch test has historically been shown to better control Type 1 error for these parameters compared to other tests [35]. Post hoc analysis was completed with Games Howell due to its robustness and utility in non-normal distributions [21]. A standard P value of 0.05 was used for statistical significance but reported up to <unk> 0.001 which is the limit on statistical software. Raw and descriptive data is available from openICPSR.org project ID openicpsr-120,505 [5]. --- Results Of the 2517 respondents, 2417 were included in the analysis based on the inclusion/exclusion criteria. The age of participants showed 446 (18.5%) people aged 18-24, 715 (29.6%) people aged 25-34, 591 (24.5%) people aged 35-44, 394 (16.3%) people aged 45-54, 189 (7.8%) people aged 55-64 and 82 (3.4%) people over the age of 65. Females accounted for 80.1% of the respondents (n = 1937 people) and males accounted for 18.8% (n = 454 people). Respondents were predominantly North American with 70.3% (n = 1700) from the USA and 12.9% (n = 312) from other North American Countries. The remaining were divided into 7.4% (n = 180) Australia/Oceanic, 7.3% (n = 176) European, 0.9% (n = 22) Asian, 0.6% (n = 15) African, 0.5% (n = 11) South American, and 1 respondent from Antarctica. The education levels showed that 0.1% (n = 2) had no formal schooling, 0.4% (n = 10) completed elementary school (grade level 1-8), 21.2% (n = 509) completed high school (grade level 9-12/13), 22.3% (n = 540) completed an Associates (2 year) degree, 34.5% (n = 833) completed a Bachelor (4 year) degree, 13.9% (n = 336) completed a Master's degree, and 7.7% (n = 187) completed a Professional degree (PhD, MD, DC, DO, etc.). Individuals identified themselves as Lower class socioeconomic status (SES) comprised 9.8% (n = 238) of the population, 82.2% (n = 1987) as Middle class SES, and 7.9% (n = 192) as Upper class SES. Facebook is the most commonly used social media type in the population with 69.8% (n = 1688), followed by Twitter 15.6% (n = 378), Instagram 12.9% (n = 311), then other forms of social media 1.7% (n = 40). Other forms of social media were identified as Snapchat, Tumblr, Reddit, Pinterest, or using all platforms equally. Most people 47.9% (n = 1158) claimed to only spend 0-2 h on social media daily, followed by 40.9% (n = 989) using 3-4 h, 8.4% (n = 204) using 5-6 h, 1.5% (n = 36) using 7-8 h, and 1.2% (n = 30) using social media for over 9 h. Most respondents 92.7% (n = 2240) have seen posts on social media about vaccines and only 7.3% (n = 177) have not. These posts influence 5.4% (n = 130) of users to think vaccines are worse than previously thought, 13.6% (n = 328) to think vaccines are better than previous thought, 76.4% (n = 1846) claim to not have been influenced by the posts, and 4.7% (n = 113) had not seen any posts. Lastly, people claimed to trust doctors 89.4% (n = 2160) the most with their immunization related information/decisions. The remaining people trust the internet 4.1% (n = 100), family 2.0% (n = 48), peers and friends 2.3% (n = 55), social media 0.2% (n = 5) and the government 2.0% (n = 49) with their information and decisions. --- Knowledge Table 1, found in the supplementary document 2 labeled "Tables", depicts the frequency of knowledge scores in the sample population. As described in the methods, knowledge scores are based on a scale from 0 to 12 derived from 6 questions with answers ranked from 0 to 2 points. Scores toward 0 represent negatively skewed knowledge, or lack of correct information. Scores toward 12 represent positively skewed knowledge, or adequate vaccine knowledge. Scores of 6 represent uncertainness. Analysis of all demographic questions against the respondent's knowledge score was completed by Welch and then further analyzed by Games Howell. Explanations of why these tests were chosen can be found in Methods. When age was compared with knowledge scores a Welch statistical value of 0.763 and the significance of 0.576 (p > 0.05). Post hoc was not necessary. Gender analysis showed a Welch statistic of 1.627 with a significance value of 0.204 (p > 0.05). Post hoc analysis was not examined because there was no significance. Geographical Welch testing showed a statistic value of 11.552 with a significance of <unk> 0.001(p <unk> 0.05). Since this value is statically significant post hoc analysis was examined. North Americans (USA) has significantly lower knowledge scores compared to Europe (mean difference -0.78309, significance <unk> 0.001), and Australia/Oceania (means difference -0.84316, significance <unk> 0.001). North Americans (Other) also showed significantly lower knowledge scores compared to Europe (means difference -0.76122, significance 0.001) and Australia/Oceania (means difference -0.84316, significance 0.001). Values from Asia, Africa, and South American should be looked at with caution because of low responses. Antarctica was excluded from these calculations because there was only one respondent. Analysis of respondents highest level of education completed showed a Welch statistic of 13.030 and significance of 0.001 (p <unk> 0.05). Post hoc showed that those who completed a Professional degree had significantly higher scores than Bachelor's degree (means difference 0.55353, significance of 0.007), Associates degree (means difference 1.21578, significance of <unk> 0.001), and high school (means difference 1.11273, significance <unk> 0.001). Those with Masters Degrees were significantly higher scoring than Associate degrees (means difference of 0.80000, significance <unk> 0.001), and high school (means difference of 0.69695, significance of 0.001). Bachelor's degree holders had significantly higher scores compared to Associates degree (means difference of 0.66224, significance of <unk> 0.001) and High school (means difference of 0.55920, significance of 0.003). Those values from who have no formal school or only completion of elementary school should be looked at with caution due to low frequencies. Socioeconomic class compared to knowledge scores yielded a Welch statistic of 0.266 and a significance of 0.767 (p > 0.05). No further analysis was needed. The type of social media used compared to knowledge score showed a Welch statistic of 7.175 and significance of <unk> 0.001(p <unk> 0.05). Games Howell determined that Twitter users had significantly higher scores than Facebook (means difference 0.43812, significance of 0.001) and Instagram (means difference 0.69491, significance of 0.001). Hours spent on social media showed a Welch statistic of 2.531 and significance of 0.044 (p <unk> 0.05). Post hoc testing showed significantly lower values in those who use social media for 3-4 h compared to 0-2 h (means difference 0.33869, significance of 0.018). No other means from this analysis were significant. Whether or not a respondent had seen anything on social media about vaccines was not analyzed because there are only 2 categories and therefore the question is noncompliant with the Welch analysis. The influence of vaccine posts on social media had a Welch statistic of 145.202 with a significance of <unk> 0.001 (p <unk> 0.05). Post hoc testing revealed that those who now perceived their opinion of vaccine of being worse than previously thought had significantly lower scores compared to those who now think vaccines are better (means difference -6.36712, significance of <unk> 0.001), no influence/ change in opinion (means difference -5.83564, significance of <unk> 0.001) and those who had not seen anything (means difference -4.70483, significance of <unk> 0.001). Those who think vaccines are better after seeing social media posts had significantly higher scores compared to worsened opinions (as mentioned before), those who were not influenced (means difference 0.53148, significance <unk> 0.001) and those who have not seen anything (means difference 1.66229, significance <unk> 0.001). In addition, those who have not been influenced by posts had significantly higher scores than those who have not seen any posts (means difference 1.13081, significance <unk> 0.001). Lastly, those trusted for immunization related information and decisions was analyzed and found a Welch statistic of 83.032 with significance of <unk> 0.001 (p <unk> 0.05). Post hoc analysis showed those who trusted Doctors the most have significantly higher scores than those who trusted the internet (means difference of 5.32139, significance of <unk> 0.001), family (means difference 5.94306, significance <unk> 0.001), and peers (means difference 6.31957, significance of <unk> 0.001). Those who trusted the government the most also had significantly higher scores than internet (means difference 5.13429, significance <unk> 0.001), family (means difference 5.75595, significance of <unk> 0.001) and peers (means difference 6.13247, significance of <unk> 0.001). Trusting of social media should be looked at with caution due to low frequencies. --- Beliefs Depiction of the frequency of belief scores in the sample population can be found in Table 2, found in the supplementary document 2 labeled "Tables". As noted in the methods, the remaining 6 questions were scored on a two point scale resulting in a belief score from 0 to 12. Scores toward 0 represent negatively skewed beliefs or belief in common myths. Scores toward 12 represent positively skewed beliefs or disbelief in common myths. Scores of 6 represent uncertainness. The analysis of the demographic questions against the individual's belief score was completed by Welch and then further analyzed by Games Howell. Explanations of why these tests were chosen can be found in methods. When age was compared with belief score a Welch statistical value of 2.923 and significance of 0.013 (p <unk> 0.05). Post Hoc revealed 65-year-olds and older had significantly lower scores than 10-24-year-olds (mean difference -1.37750, significance 0.014) and 24-34 year olds (mean difference -1.19606, significance 0.047). Gender analysis showed a Welch statistic of 0.320 with a significance value of 0.728 (p > 0.05). Post hoc analysis was not examined because there was no significance. Geographical Welch testing showed a statistic value of 29.212 with a significance of <unk> 0.001(p <unk> 0.05). Due to this value being statically significant, post hoc analysis was examined. North Americans (USA) had significantly lower belief scores compared to Europe (mean difference -1.47989, significance <unk> 0.001), and Australia/ Oceania (means difference -1.81575, significance <unk> 0.001). North Americans (Other) also showed significantly lower belief scores compared to Europe (means difference -1.29021, significance <unk> 0.001) and Australia/ Oceania (means difference -1.62607, significance <unk> 0.001). Values from Asia, Africa, and South American should be looked at cautiously because of the low response rate. Antarctica not included in these calculations because there was only one individual who responded. The analysis of individuals with the highest level of education completed showed a Welch statistic of 17.789 and significance of <unk> 0.001 (p <unk> 0.05). Post hoc showed that those who completed a professional degree had significantly higher scores than those with master's degree (mean difference 0.74516, significance of 0.009), bachelor's degree (means difference 1.10881, significance of <unk> 0.001), associate's degree (means difference 2.02797, significance of <unk> 0.001), and high school (means difference 1.97009, significance <unk> 0.001). Respondents with master's degrees were significantly higher scoring than those with associate degrees (means difference of 1.28280, significance <unk> 0.001), and high school (means difference of 1.22493, significance of <unk> 0.001). Those with bachelor's degrees had significantly higher scores compared to those with associate's degrees (means difference of 0.91916, significance of <unk> 0.001) and high school (means difference of 0.86128, significance of <unk> 0.001). The values from those individuals that had no formal schooling, or only completion of elementary school, should be looked at cautiously due to low frequencies. Socioeconomic class, compared to belief scores, resulting in a Welch statistic of 0.028 and a significance of 0.972 (p > 0.05). No further analysis was needed. The type of social media users compared to belief score showed a Welch statistic of 8.011 and a significance of <unk> 0.001(p <unk> 0.05). Games Howell determined that Twitter users had significantly higher scores than Facebook (means difference 0.55094, significance of 0.001) and Instagram (means difference 0.98733, significance of <unk> 0.001). Hours spent on social media showed a Welch statistic of 3.162 and a significance of 0.016 (p <unk> 0.05). Post hoc testing showed significantly lower values in those who used social media for 3-4 h compared to 0-2 h (means difference 0.39195, significance of 0.034). No other means from this analysis were significant. Exposure to posts on social media about vaccinations was not analyzed because there were only two categories and therefore, noncompliant with the Welch analysis. The influence of vaccine posts on social media had a Welch statistic of 312.900 with a significance of <unk> 0.001 (p <unk> 0.05). Post hoc testing revealed that those who now perceived their opinion of vaccines of being worse than previously thought, had significantly lower scores compared to those who now think vaccines are better (means difference -7.97280, significance of <unk> 0.001), No influence or change in opinion (means difference -7.27248, significance of <unk> 0.001), and those who had not seen anything (means difference -4.97992, significance of <unk> 0.001). Those who thought vaccines were better after seeing social media posts had significantly higher scores compared to worsened opinions, as mentioned before, and to those who were not influenced (means difference 0.70031, significance <unk> 0.001), and those who had not seen anything (means difference 2.99288, significance <unk> 0.001). Also, those who had not been influenced by social media posts had significantly higher scores than those who had not seen any posts at all (means difference 2.29256, significance <unk> 0.001). Finally, those trusted in immunization related information and decisions were analyzed and found a Welch statistic of 150.953 with a significance of <unk> 0.001 (p <unk> 0.05). Post hoc analysis showed those who trusted doctors the most had significantly higher scores than those who trusted social media (means difference of 6.50713, significance of <unk> 0.001), family (means difference 7.28796, significance <unk> 0.001), and peers (means difference 7.41258, significance of <unk> 0.001). Individuals who trusted the government the most also had significantly higher scores than social media (means difference 6.87265, significance <unk> 0.001), family (means difference 7.65349, significance of <unk> 0.001) and peers (means difference 7.77811, significance of <unk> 0.001). Trusting of social media should be looked at cautiously due to its low frequencies. --- Discussion The significance found in this study can help us understand who is being influenced by posts about vaccines on social media. It is important to note that the sample in this study was originally started in North America; hence the vast majority of respondents reported residing on this continent. Beyond that, significant values from the statistics should be examined to determine what it means for this research and for further implications. --- Mean scores Overall this study found that respondents were very knowledgeable with a mean knowledge score of 10.4. Very few people had negatively skewed knowledge 0-4 (138 people, 5.7% of the total study population). Further investigation into those people that scored lowest would be able to show greater detail into the minds of those people and where the lack of factual knowledge is coming from or what hurdle needs to be faced. Respondents had mainly positive beliefs about vaccines with a mean belief score of 9.68, with a standard deviation of 3.14. Few people had negatively skewed beliefs 0-4 (236 people, 9.8% of the total study population). Looking deeper into the individuals that had lower scores would be able to show greater explanation into why their beliefs about vaccines were negative, and how we might be able to change these beliefs. --- Age In a study in 2009, 75.64% of people aged 18-64 were internet users and 74% of users aged 18-24 were social media users (Chou et al). This is important because the high social media use group from that study is now 27-31 years old and could now be making decisions about vaccines for their children. When looking at ages, we saw that 65 years and older tend to have more negative beliefs than those individuals that are 18-24 and 25-34, with a significant difference of 0.013 (p <unk> 0.05). Meaning, older people tend to believe the things they see on social media more than someone who is younger. We saw that in the previous research, older people are less likely to use social media [4], but in this study, those who did use social media scored lower. There were no significant correlations regarding age for knowledge scores. --- Gender and socioeconomic status (SES) A comprehensive research paper by UNICEF in 2013 [36] titled "Tracking anti-vaccination sentiment in eastern European social media networks" found that females are more likely to discuss developmental disabilities, chemical, toxins and potential side effects whereas males are more likely to discuss conspiracy theories, religion and distrust of the government. Across all genders and all SES, we saw no significant differences in knowledge or beliefs. This is significant to note because as previous literature and this one found, women are the predominant social media users. Since gender and SES have no impact on vaccine knowledge or beliefs we look deeper into other variables. --- Geography For geography, analysis was according to continent lived on and subsequently found statistical significance. North Americans compared to all continents are significantly less knowledgeable about vaccines and had more negatively skewed beliefs. Although this research does not give a reason to why North Americans are less knowledgeable and believe more myths it gives us insight into areas that need to be studied. There was little data available on Asia, Africa, and South America and none on Antarctica making analysis unreliable and therefore all of which should be disregarded. It is evident that Asia is showing a lower mean knowledge than other areas but without significance due to low frequency. This is paralleled when looking at beliefs. This may be explained by previous research has found that "among people who use the internet, those in developing countries often turn out to be more likely than their counterparts in advanced economies to network via platforms like Facebook and Twitter" [33]. Further investigation should determine if Asian countries do in fact have lower knowledge of vaccines and why this is. Some things to determine include whether there is a quantifiable difference between continental education, social media, use, cultural perceptions of vaccination, etc. --- Education There is a steady incline in mean knowledge and mean belief score as education level increases. Due to low response frequencies of "no education" and "elementary education" results pertaining to these categories should be disregarded. There are significant differences between all other education levels in an increasing fashion for both knowledge and beliefs. This tells us that those with higher education are in some way or another seeking and finding valid information about vaccines. "Formal schooling adds significant value to innate ability in the form of higher-order cognitive skills crucial to decisions about health" [2]. This is important in terms of whom to focus on for future vaccine education and propaganda. --- Social media platform A comprehensive research paper titled "Tracking Anti Vaccination Sentiment in Eastern European Social Media Networks" by UNICEF in 2013 [36] found that vaccine influencers (people or pages that speak publicly about vaccines, both positively and negatively) are most prominent on Facebook and Twitter. This study found that Twitter users are significantly more knowledgeable about vaccines than Facebook or Instagram users and also had more statistically significant positive beliefs about vaccines. Somehow, information is reaching Twitter users but not reaching other forms of social media. Upon further investigation research has showed "information sought from Facebook may be obtained socially (i.e. by asking other users), whereas the information sought on Twitter might be more cognitively based, such as academic or political information that is best gained by reading source materials, for which links are often 'tweeted' " [22]. --- Time on social media While the majority of people claimed to only spend 0-2 h on social media daily, those who spent more time, namely 3-4 h, were significantly more knowledgeable about vaccines and had significantly more positive beliefs about vaccines. Although not explored in this study, other studies have found that upwards of 90% of young adults use social media and this number is increasing every year and the majority of this time is spend on smartphones [37]. Possible reasons for the increased scores may be that those who spend more time on social media could be spending that time reading more deeply into conversations or information. Although it does not appear to be currently studied, there could also be an association between access to a smartphone and the ability to fact check information on the spot. --- Vaccine related posts One of the most interesting takeaways from the data analysis is how vaccine posts have influenced opinions. Those who reported that after seeing vaccine posts they now think vaccines are worse have significantly lower knowledge and belief scores. The opposite is true for those who reported more positive opinions since seeing posts, their scores were significantly higher. The importance of this is that people are able to accurately selfreport how these vaccine posts affect them. Those people who see posts about vaccines that make them think vaccines are "bad" have less knowledge or their knowledge has been changed from correct to incorrect with these posts. A European study found that over 40% of respondents had some degree of "negative feelings about vaccine safety" [28]. These "fear" posts are typically full of myths, although not inherently known to be myths by the reader, about vaccines and as a result people became more likely to believe the myths. While it cannot be inferred why this is happening it is an

Background: Social media use has become a mainstay of communication and with that comes the exchange of factual and non-factual information. Social media has given many people the opportunity to speak their opinions without repercussions and create coalitionS of like-minded people. This also has led to the development of a community know as anti-vaxxers or vaccine deniers. This research explores the extent to which vaccine knowledge has reached on social media. Methods: This cross sectional research explored the relationship between the spread of information regarding vaccines in relation to social media use. A sample of 2515 people over the age of 18 around the world completed the survey via a link distributed on Twitter, Facebook and Instagram. A series of questions on vaccine knowledge and beliefs were compounded to create an individual's "knowledge score" and a "belief score". Knowledge scores were ranked from low knowledge to high knowledge with increasing scores. Belief scores were ranked from belief in myths to disbelief in myths with higher scores. This score was then analysed, using a Welch test and post hoc testing when applicable, across demographics and questions relating to social media use. Results: Significant relations were found in both the knowledge and belief categories, many of which were similar findings between the two. North Americans had significantly lower knowledge and belief scores compared to all other continents. While the majority of people primarily use Facebook, Twitter users were significantly more knowledgeable. It was also found that higher education was correlated with higher knowledge and belief scores. Conclusions: Overall, these correlations are important in determining ways to intervene into the anti-vax movement through the use of social media. Cross demographics were not analysed in this study but could be in future studies. To better understand the social media exposures related to vaccine information a follow up structured interview research study would be beneficial. Note that due to the cross sectional nature of this study, causal relationships could not be made.

spending that time reading more deeply into conversations or information. Although it does not appear to be currently studied, there could also be an association between access to a smartphone and the ability to fact check information on the spot. --- Vaccine related posts One of the most interesting takeaways from the data analysis is how vaccine posts have influenced opinions. Those who reported that after seeing vaccine posts they now think vaccines are worse have significantly lower knowledge and belief scores. The opposite is true for those who reported more positive opinions since seeing posts, their scores were significantly higher. The importance of this is that people are able to accurately selfreport how these vaccine posts affect them. Those people who see posts about vaccines that make them think vaccines are "bad" have less knowledge or their knowledge has been changed from correct to incorrect with these posts. A European study found that over 40% of respondents had some degree of "negative feelings about vaccine safety" [28]. These "fear" posts are typically full of myths, although not inherently known to be myths by the reader, about vaccines and as a result people became more likely to believe the myths. While it cannot be inferred why this is happening it is an interesting statistic. The presence of this significance lets health/vaccine promoters, such as government; know that the vaccine posts with valid knowledge and promotion are working. Those respondents that reported they were not influenced by social media posts, had lower belief scores compared to those who said they were positively influenced by social media posts. Those who reported they were not influenced had similar knowledge scores to those who were positively influenced. This may be due to the ability to weed through information to find facts versus fiction. The internet is now the easiest and most accessible way to get information. A google search from the United States of America for the term "vaccination" yielded 71% anti-vaccination pages and only 29% pro vaccination pages [25]. Many of these anti-vaccinations sites claimed that vaccines contain "poisons" and were damaging to the human body [25]. "In many cases, [those who perceive vaccines as harmful] may become the only individuals who voice their opinions [32]. This allows for a greater propagation of anti-vax information rather than pro-vaccination and factual knowledge. Googling information about the MMR vaccine and autism only 51% of the search results yielded correct information [41]. Not only is it important to understand the diversity of search results but also the quality of the websites appearing. Many of "sites masquerading as official scientific sites, some web users may not question the veracity of such material" [12]. Although this study does not specifically focus on the internet, social media allows for easy sharing of not only opinions but also links to some of these "pseudo-science" websites. --- Trusted source Lastly, the analysis of who people trust most with their vaccine related information and decisions. Literature review revealed a study that found that parents who used the internet to get their information about vaccines were more likely to think children do not benefit from vaccines [24]. It is comforting to know that the vast majority of respondent in this study do in fact rely on the information they receive from their doctor and government. While many anti-vax campaigns are grounded on the fact that authorities are misconstruing vaccine information, the idea of big pharma, it is evident that that ideal is not being taken up by social media users. Those who trust the government and doctors with their information were significantly more knowledgeable about vaccine and more positive beliefs facts than those who trust another group. While it is known that many anti-vaxxers rely on social media to disseminate their message, very few people trust social media as a source. This is very good news for doctors whose patients seek information from them and government run websites such as health departments and health organizations. --- Limitations Some limitations faced by this research include the cross sectional nature of this study. Cross sectional studies cannot infer causation, only correlations. It also limits this survey to only surface information and not important individual details. In addition, a limitation may be that demographics were not cross analyzed. This means that while it was found that professionals had the highest knowledge score this research does not delve into the gender, location, age, etc. of these people. This same information could be analyzed in this way for further investigation. It is also important to note that this information is limited by some frequencies being too low for confidence. These areas would ideally have to be isolated and looked at again with higher number of participants. These categories were mentioned in the discussion and results stating they should be disregarded. This study is also limited by the snowball sampling method. This limits the study by which participants decided to share, and where the majority of their friends are located, as seen by the North American predominance. While great caution was taken when constructing this survey to be nonjudgmental, some valuable information is lost in the fact that respondents were not asked their vaccine practices. As with all online surveys, no interviewer was present. Respondents were not able to clarify answers that they felt needed explanation. Without an interviewer there is no probing into deeper answers. Information this study found could be used in the future for interviewer based research to determine a more in depth understanding of those who lacked vaccine knowledge. Another limitation of being an online survey is that there is no accountability for answers. Respondents could have clicked through the survey to complete all the questions as quickly as possible without regard for what the answer choices were. The information portrayed in this research by the authors favors scientific fact as described by other journal articles and not personal opinions/bias. Every effort was made not to use incriminating or offensive language and statements in the creation of the survey. This core of this research was completed by the authors when they were medical students. Saint James School of Medicine ethics committee approved this research and all of its modalities. Survey platform was provided by Saint James School of Medicine. --- Conclusions As social media continues to grow exponentially, it can be expected that anti-vaxxers will further spread their messages across these platforms. While this research does not delve into the totality and extent of the antivax movement on social media it does provide some insightful information. As mentioned previously, it is possible for governments and doctors to use this information to intervene and correct false information about vaccines and their safety. Given that trust in doctors resulted in significantly higher knowledge and belief scores, it is imperative that physicians create a trusting environment and relationship with their patients and guardians of pediatric patients. This also tells us that the information patients are getting from their doctors and the conversations they are having are influencing vaccine decision making in a positive way. What was not studied, and would be beneficial to investigate going forward, is whether doctors have the ability to change a patient's perspective on vaccines once they have been exposed to negatively skewed vaccine information. It is also evident that those who trust the internet with their vaccine information are exposed to more misinformation. There may be an avenue for addressing misinformation on the internet through this trusted source and that could be via physician social media influencers. We have started to see the importance of this avenue with the prevalence of COVID-19. Social media has create an easily accessibly way to reach credible sources such as CDC and WHO, who made their presence known during the height of the pandemic through social media campaigns [7]. This appears to be especially important for direct intervention on Facebook, where the majority of social media use is, and unfortunately lower rates of knowledge and more negatively skewed beliefs. In North America, actions should be taken to combat the misinformation and myths that are reaching its population more than anywhere else in the world. Further research is recommended to understand why some countries, age groups, social network users, etc. are not getting adequate information on vaccines. Future research would benefit from using a structured interviewer based approach to allow for expansion and clarification of answers. Further evaluation of the data collected here could yield more in-depth understanding of demographics. As mentioned before this research did not cross analyze demographics. --- Availability of data and materials Raw and descriptive data is available from openICPSR.org project ID openicpsr-120505. --- Abbreviations Anti-Vax/ Anti-vaxxer: Anti-vaccinator / Vaccine denier; HPV vaccine: Human Papilloma Virus vaccine; MMR vaccine: Measles Mumps Rubella vaccine; UNICEF: United Nations International Children's Emergency Fund; SPSS: Statistical package for social sciences; SES: Socioeconomic status --- Supplementary Information The online version contains supplementary material available at https://doi. org/10.1186/s12889-021-12114-8. Additional file 1. Survey. Anti-Vax research survey. This file includes the questions and multiple choice options that were used to build the research's SurveyMonkey Survey. Additional file 2: Table 1. Frequency of Knowledge Scores 0-12 with percent of total sample population. Table 2. Frequency of Belief scores 0-12 with percent of total sample population. Authors' contributions SB; Lead author, designed analytic scoring/method and data analyst. RM; coauthor, co-conceived experimental idea, coordination of timeline, data collection. All authors have reviewed and approved the manuscript. --- Declarations --- Ethics approval and consent to participate This research was approved by the ethics committee of Saint James School of Medicine. This research included a consent page present at the beginning of the survey which included consent to participate. Participation was limited to those over 18 years of age. Failure to consent resulted in termination of the remainder of the survey. --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Social media use has become a mainstay of communication and with that comes the exchange of factual and non-factual information. Social media has given many people the opportunity to speak their opinions without repercussions and create coalitionS of like-minded people. This also has led to the development of a community know as anti-vaxxers or vaccine deniers. This research explores the extent to which vaccine knowledge has reached on social media. Methods: This cross sectional research explored the relationship between the spread of information regarding vaccines in relation to social media use. A sample of 2515 people over the age of 18 around the world completed the survey via a link distributed on Twitter, Facebook and Instagram. A series of questions on vaccine knowledge and beliefs were compounded to create an individual's "knowledge score" and a "belief score". Knowledge scores were ranked from low knowledge to high knowledge with increasing scores. Belief scores were ranked from belief in myths to disbelief in myths with higher scores. This score was then analysed, using a Welch test and post hoc testing when applicable, across demographics and questions relating to social media use. Results: Significant relations were found in both the knowledge and belief categories, many of which were similar findings between the two. North Americans had significantly lower knowledge and belief scores compared to all other continents. While the majority of people primarily use Facebook, Twitter users were significantly more knowledgeable. It was also found that higher education was correlated with higher knowledge and belief scores. Conclusions: Overall, these correlations are important in determining ways to intervene into the anti-vax movement through the use of social media. Cross demographics were not analysed in this study but could be in future studies. To better understand the social media exposures related to vaccine information a follow up structured interview research study would be beneficial. Note that due to the cross sectional nature of this study, causal relationships could not be made.

INTRODUCTION Despite the increased availability and use of modern communication technologies, contemporary society seems more polarized on key social issues than ever before, McCoy et al. (2018); McCoy and Somer (2019). The rise of social media as an important channel for entertainment, shopping, news and communications over recent years has led to the phenomenon of the influencer. These are groups or individuals with a strong persuasive power due to expertise, personality, fame or notoriety. Charismatic opinion leaders have the ability to influence the behavior or opinions of others, Tur et al. (2021). Influencers are now a common feature of social media platforms such as Facebook, Instagram and Twitter, as well as mainstream media, where they promote products, causes and opinions. Although the image of an influencer promoting an extreme position on an issue would be most familiar in the socio-political sphere, the promotion of fact-based information is often part of public discourse on important issues. In this case the influencers tend to represent a government or public institution rather than providing an individual point of view, and the messaging tends to be framed in a more neutral tone. Examples of this include the Centre for Disease Control and Prevention, CDC, on public health (including vaccination), or the US Environmental Protection Agency, EPA, on climate change. Despite their ubiquity, the role of influencers in affecting social attitudes and political opinions is not well understood, Riedl et al. (2021). Although it is likely that exposure to extremist messaging increases polarization, very little is known about the complex relationship between influencers, the community (to be influenced), the broader societal context (the type of society) in which the interactions between influencers and their followers occur or type of message (neutral or extremist) being transmitted. To examine the way influencers affect societal polarization, we create an Agent-Based Model of a society, into which we introduce an influencer. Varying the confidence threshold at which an agent would 978-1-6654-7661-4/22/$31.00 <unk>2022 Crown have a positive interaction with another agent (that is, one in which each agents belief moves slightly closer to the others, and a social connection is strengthened) enables relatively closed, moderately open and very tolerant societies to be simulated. Similarly, the confidence threshold for influencer-agent interactions is varied to simulate an influencer with narrow, moderate and wide reach. By increasing the probability that an influencer will interact with an agent, it is possible to amplify the influencer's effect. Since the message to be broadcast doubtlessly affects the interaction, we consider two types of influencers: one having a neutral message (that would be intended to promote consensus in the real world), and the other taking an extremist position on an issue. Under these varied parameter settings it is then possible to observe the effect of the influencer on the rate of polarization, interpreted as both the ideological extremitization of belief (as each agents belief becomes increasingly polarized), and psychological distancing (through the formation of polarized clusters). Results show that an extremist influencer will always increase the rate of polarization in a society by increasing the rate of both ideological separation and psychological distancing. Neutral influencers having a narrow reach will also increase the rate of polarization due to the repulsive effect of their interactions. However, when neutral influencers have a wider reach, they slow the rate of ideological extremitization and ideological separation in more open communities. These experimental results offer insights into the design of communication strategies to slow the rate of societal polarization on key issues. In the following section we present a brief review of research into the social processes leading to polarization, the use of Agent-Based Modeling as a means to understand the social networks that result from human interaction, and the psycho-social models of influence and bonding that underpin these processes. We then introduce the Agent-Based Model of a society that includes influencers. Section 4 presents the design of experiments and results. We conclude the paper with a summary of results and discussion of the implications of our findings. --- LITERATURE REVIEW --- Societal Polarization In political psychology, polarization tends to be regarded as either issue-driven by support for opposing policy positions, Duffy et al. (2019), or manifest as distrust and intense dislike of the outgroup (the opposing ideological camp), Abramowitz and McCoy (2019); Druckman and Levendusky (2019). Recent studies propose that polarization results from both the ingroup (one's own group) position becoming more uniform and extreme as members of the ingroup interact, amplifying the consensus (ideological extremitization), and increased distancing from the outgroup as interactions between the ingroup and outgroup lead to dissent and hostility (psychological distancing), Bliuc et al. (2021). Thus, polarization results from these two processes operating concurrently: ideological extremitization of belief and psychological distancing as the difference between ingroup and outgroup belief increases. Classic studies in social psychology support this view by showing that communication in closed groups (or echo-chambers in the modern context), can lead to group members positions becoming more extreme, Myers (1975); Myers and Lamm (1976); Cinelli et al. (2021). These findings help explain why in groups that polarize because of opposing views on policy issues, for example (climate change supporters vs climate deniers), interaction within the group can lead to more extreme group positions and therefore a greater ideological distance between ingroup and outgroup. Previous research on polarization has shown that selective exposure to communication from ingroup on social media is likely to produce greater polarization (by increasing confirmation bias) Ram<unk>rez-Due<unk>as and Vinuesa-Tejero (2021); Modgil et al. (2021). Moreover, we know that influencers on social media platforms such as Twitter help spread extreme content faster Asatani et al. (2021). However, less is known about the effect of different types of influencers on polarization, and in particular about the potential 'buffering' effects of neutral influencers compared to their extremist counterparts in various types of societies. Although empirical studies have analyzed the diffusion of ideas through social networks, for example, Bakshy et al. (2011), and the structural position of influencers within polarized clusters, for example, Soares et al. (2018), the role of influencers in the polarization of society at large remains unclear. --- Agent-Based Modeling as a Means to Understand Society While models of social influence or opinion dynamics can be used to observe that polarization of belief is an outcome of the interaction between individuals, these models do not clearly describe the effect of this on social structure (see for example, Flache et al. (2017), and Ishii and Kawahata (2018)). Agent-Based Modeling, Macal (2016), is one means by which the social networks formed by the repeated interaction of individuals can be observed. These models enable the interaction of autonomous entities (agents), each having certain characteristics, to be simulated according to specific rules. By simulating successive interactions between agents over the longer term, the complex net effect of these individual interactions on a population at large can be observed. This makes them an effective tool to connect the micro-level assumptions about individual agent behaviour to macro-level patterns in artificial societies, Smith and Conrey (2007). They have been used, for example, to study opinion dynamics and social processes that may lead to the clustering of extreme beliefs and polarization in societies, for example, Nowak et al. (1990); Deffuant et al. (2000); Flache and Macy (2011); Turner and Smaldino (2018); Rychwalska and Roszczy<unk>ska-Kurasi<unk>ska (2018). We now extend this research to study the effect of influencers on these processes. --- Psycho-Social Models of Influence and Bonding Social influence can occur when there are interactions between members of different social groups or members of the same social group holding differing opinions, and these interactions result in attitude or opinion change. The rules governing the interaction between agents in the Agent-Based Model that follows are derived from three types of social influence models: • Assimilative, Friedkin and Johnsen (2011); Groeber et al. (2014), based on classic social psychological theory that individuals who are connected by a shared social identity always influence each other towards reducing the differentiation between them (through a process of averaging); • Similarity biased, Sherif and Hovland (1961);Festinger (1962); Axelrod (1997), whereby only individuals who are sufficiently similar can influence each other; and • Repulsive, Festinger (1962); Flache and Macy (2011); Dykstra et al. (2015), where it is assumed individuals who are highly different still influence each other, but towards increasing those differences. Thus, clusters may form as maximally oppositional views through a process of bi-polarization Flache et al. (2017). Pairs of agents form a social bond with each other when they hold a similar level of belief, McPherson et al. (2001). These bonds strengthen after repeated interactions between pairs holding a similar level of belief, but also weaken over time if the agents either fail to interact, or interact but now hold sufficiently different levels of belief Skyrms and Pemantle (2009). Following these processes, a social network of agents is formed, which evolves over time reflecting the change in belief held by agents, and the history of their interactions. Influencers are treated as agents having special properties in this modeling framework. They broadcast a belief to others in the society through their interaction. However, their belief on an issue does not change as a result of interactions with others, and they do not form social bonds with other agents. In this regard, we treat them as zealots or ideologues holding either a neutral or extreme stance on an issue, interacting with others while not being a member of the groups with whom they interact. Of interest to our study is the rate at which polarization occurs in the agent population, reflected in its ultimate form as the formation of distinct groups and extremitization of belief, and the effect of influencers on this process. --- AN AGENT-BASED MODEL OF A SOCIETY WITH INFLUENCERS --- Modeling Assumptions and Notation In the model that follows, agents are represented by X i, where i <unk> <unk>1,..., n<unk>. Each agent has a level of belief on a single issue (for example, vaccination or climate change etc., which may be positive or negative), B(X i ), where <unk>-1 <unk> B(X i ) <unk> 1<unk>. The single influencer is indicated by Y. The influencer also holds (broadcasts) a belief on the same issue, B(Y ), where <unk>-1 <unk> B(Y ) <unk> 1<unk>. A key difference between the agent and the influencer is that the belief of each agent varies over successive iterations as the result of interactions between agents, and between agents and the influencer. The influencers level of belief, however, remains constant throughout each trial. For our experiments we tested two scenarios, the case of a neutral influencer, where B(Y ) = 0 and an extremist influencer, where B(Y ) = 1. The strength of the social connection between each pair of agents, X i and X j, is is indicated by N(X i, X j ) <unk> <unk>0, 1,..., <unk>. This is used to construct the social network of agents, and varies over sucessive iterations as a result of interactions between agents. The confidence threshold between agents, Flache et al. (2017), CT X, is the maximum difference in belief between a pair of agents, in which each has confidence in the other as a source of information, and influences the other positively. When two agents are within this range during an interaction, their beliefs converge and the social connection between the pair is increased. Conversely, outside this range, agents beliefs diverge and the network connection decreases. In the experiments that follow we assume that all agents have the same confidence threshold for an agent-agent interaction, which remains constant throughout each simulation trial. The confidence threshold for an interaction between an agent and the influencer, CT Y, may be different to that for agent-agent interactions, reflecting a different level of trust in the influencer held by agents. It is also the same for all agents, and remains constant throughout each trial. The behavior when an agent interacts with the influencer is similar to the interaction between two agents with the exception that the influencers belief is unchanged and there is no social bond with the influencer to make or break. The probability of an agent interacting with an influencer during each iteration is given by p Y. The degree by which belief changes during an interaction between agents is proportional to the difference in agent belief, scaled by a constant rate, r, and truncated at either terminal during these interactions to maintain belief on the interval [-1, 1]. --- Design of the Simulation Parameters that remain constant during each simulation trial are: the confidence threshold between agents, CT X, between an agent and the influencer, CT Y, the proportional change in the belief of an agent at each interaction, r, and the probability that an influencer will interact with an agent in any iteration p Y. The model is initialized by randomly assigning a level of belief to each agent chosen from the uniform distribution on the interval [-1, 1]. Thus B(X i ) <unk> U(-1, 1), for i <unk> <unk>1, 2, 3,..., n<unk>. Agents have no social connections at this stage, thus N(X i, X j ) = 0 for all i, j. At each iteration a pair of agents is randomly chosen for interaction. To simplify notation, it is assumed that that B(X i ) <unk> B(X j ) and <unk> a = B(X i ) -B(X j ), so that <unk> a > 0. When <unk> a <unk> CT X agents have an assimilative interaction, where each agent's belief moves incrementally toward the other as B(X i ) <unk> B(X i ) -r<unk> a and B(X j ) <unk> B(X j ) + r<unk> a, and the network connection between agents is strengthened as N(X i, X j ) <unk> N(X i, X j ) + 1. When <unk> a > CT X agents engage in a repulsive interaction whereby B(X i ) <unk> min(B(X i ) + r<unk> a, +1) and B(X j ) <unk> max(B(X j )r<unk> a, -1). The network connection between agents is weakened as N(X i, X j ) <unk> max(N(X i, X j ) -1, 0). A second interaction, between a randomly chosen agent and the influencer, then occurs, with probability p Y. The difference in belief between the influencer and agent is calculated as <unk> i = B(X i ) -B(Y ). When |<unk> i | <unk> CT Y an assimilative interaction occurs, and agent belief moves closer to that of the influencer as B(X i ) <unk> B(X i ) -r<unk> i. Otherwise a repulsive interaction occurs in which B(X i ) <unk> min(B(X i ) + r<unk> i, +1) if <unk> i <unk> 0 max(B(X i ) + r<unk> i, -1) if <unk> i <unk> 0. A summary of the social network of agents, and individual agent belief, is then reported periodically. The operation of the simulation is shown in Algorithm 1. The algorithm was implemented in the R programming language, using the igraph package for social network analysis and graph plotting. --- The Simulation Algorithm Algorithm 1 Agent-Based Model with Influencer B(X i ) <unk> U(-1, 1) for all i <unk> <unk>1, 2, 3,..., n<unk> N(X i, X j ) = 0 for all i, j set CT X, CT Y B(Y ), p Y, randomize for I = 1 to total iterations do # agent-agent interaction randomly choose X i and X j where i = j and B(X i ) <unk> B(X j ) <unk> a = B(X i ) -B(X j ) if <unk> a <unk> CT X then B(X i ) <unk> max(B(X i ) -r<unk> a, -1); B(X j ) <unk> min(B(X j ) + r<unk> a, +1) N(X i, X j ) <unk> N(X i, X j ) + 1 else B(X i ) <unk> min(B(X i ) + r<unk> a, +1); B(X j ) <unk> max(B(X j ) -r<unk> a, -1) N(X i, X j ) <unk> max(N(X i, X j ) -1, 0) end if # influencer-agent interaction if x <unk> U(0, 1) <unk> p Y then randomly choose X i <unk> i = B(X i ) -B(Y ) if |<unk> i | <unk> CT Y then B(X i ) <unk> B(X i ) -r<unk> i else if <unk> i <unk> 0 then B(X i ) <unk> min(B(X i ) + r<unk> i, +1) if <unk> i <unk> 0 then B(X i ) <unk> max(B(X i ) + r<unk> i, -1) end if end if # periodic reporting if I mod reporting interval = 0 then calculate and report statistics end for --- Output from the Agent-Based Model The output from the model at each iteration is each agent's level of belief, B(X i ) and the matrix containing the level of connection between each pair of agents, N(X i, X j ). From these data, the social network formed by the agents was constructed and the number of clusters counted. In the simulation trials that follow, this information was recorded at initialization and then after every 1000 periods. Figure 1 shows the evolution of the social network formed during one simulation trial. Agent belief is indicated as a gray-scale with -1 shown by white, 1 by black and mid-gray being neutral. Social connections between agents are represented by edges, and the social network plot shows the formation of clusters of agents. Beginning from a random, disconnected state at initialization (I = 1) the progressive polarization of agent belief and the gradual clustering of agents having a similar level of belief can be seen over successive iterations. By I = 40000 the social network is completely polarized. Figure 2 presents summary output from the same simulation over successive iterations. The left panel shows the trajectory of each agents belief over the duration of the simulation, where the complete polarization of agent belief can be observed at approximately 35,000 iterations. The center panel shows the number of clusters over the simulation as a natural logarithm, from which it can be observed that the rate of decrease in the number of clusters is approximately exponential. The right panel shows the standard deviation of agent belief over successive iterations, commencing from an initial value of approximately 0.58, being <unk> x when x <unk> U(-1, 1), to approximately 1 when x = -1 or 1 with equal probability. In the analysis that follows, the rate at which polarization occurs is determined by two measures obtained from these summary calculations: the number of iterations at which two distinct (polarized) clusters have formed, representing the psychological distance between groups; and the number of iterations by which at least approximately 80% of agents have polarized (that is, hold a belief of either +1 or -1), reflecting the ideological separation between groups Bliuc et al. (2021). --- DESIGN OF EXPERIMENTS AND RESULTS --- Inputs to the Model Parameter settings for the simulation trials were set as follows: The confidence threshold for agentagent interactions were chosen from CT X <unk> <unk>0.1, 0.5, 0.9<unk>, reflecting relatively narrow, medium, and wide differences in belief over which agents would have an assimilative interaction and form a social connection. At these values of CT X, polarization is the natural outcome of the interactions described by Algorithm 1, with -1 <unk> B(X i ) <unk> +1, whereas CT X > 1 would typically result in consensus. The confidence threshold for influencer-agent interactions were also chosen from the same range. Thus, CT Y <unk> <unk>0.1, 0.5, 0.9<unk>. Influencer belief was set as B(Y ) <unk> <unk>0, 1<unk>, representing the case of a neutral influencer, having belief 0 and an extremist influencer having belief 1. The probability of an influencer-agent interaction in any iteration was p Y <unk> <unk>0, 0.1, 0.2, 0.5, 1.0<unk>, giving the base case, where there was no influencer interaction (p Y = 0) as well as increased levels of influencer activity to the point where there were as many influencer-agent interactions as there were agent-agent interactions. The multiplier for the proportional change in agent belief during each agent-agent or influencer-agent interaction, r, was 0.01 throughout. A population of 200 agents was created. Trials were fully factorial, with 100 repetitions at each parameter combination. Each trial ran for 50,000 iterations, with summary statistics reported every 1,000 periods. --- Measures of Polarization Two measures are used to evaluate the relative rates of polarization across these varied parameter settings. These are: The first reporting period during which two completely polarized clusters emerged; and the first reporting period at which at least approximately 80% of agents had polarized. The censoring of trials meant that for a small proportion of cases (1.2%) the trial terminated before two clusters were reached (that is, held a belief level of either +1 or -1). In these cases the terminal value of 50,000 iterations was reported. The polarization of 80% of agents was estimated as the period at which the standard deviation of agent belief was 0.93, this being the standard deviation of a population of 200 agents in total, comprising 80 agents having belief +1, 80 having belief -1 and the remainder having distribution B(X i ) <unk> U(-1, 1). Again, the censoring of trials meant that for a small proportion of cases (1.3%) the trial terminated before a standard deviation of 0.93 was reached. Again, the terminal value was reported in these cases. --- Results Figure 3 shows the number of iterations by which polarization, in terms of psychological distancing (n = 2 clusters), occurs as a function of agent-agent confidence threshold, CT X, influencer-agent confidence threshold, CT Y and influencer belief, B(Y ). This figure and the next can be viewed as two main panels showing results for the neutral influencer (LHS) and the extremist influencer (RHS). Each panel in the plot shows the effect of increasing the probability of an influencer-agent interaction during each iteration. Ignoring the effect of influencers temporarily by considering the only cases where p Y = 0, the results overall show the rate of clustering increases (that is, the time to form two clusters decreases) as the confidence threshold for agent-agent interactions increases. There is a small, and insignificant increase from CT X = 0.1 ( x = 35, 293, s = 7, 986) to CT X = 0.5 ( x = 34, 196, s = 8, 167). However, there is a significant increase in the rate of polarization when CT X = 0.9 ( x = 23, 230, s = 4, 945), confirmed by a two-sided t-Test (here and following tests), p 0.001. This results from the increased likelihood of assimilative iterations (and making network connections) due to the increased confidence threshold, reflecting a more tolerant society. Subgroups of parameter combinations that exhibit similar properties have been identified for the following analysis and discussion. Groups A, B and C show the period at which polarized clusters have formed for a neutral influencer at each parameter combination. These groups show the effect on cluster formation is mixed. Group A shows that increased interaction with a neutral influencer having a narrow confidence threshold increases the rate of clustering regardless of the confidence threshold for agentagent interactions over the range tested. This is because the majority of influencer-agent interactions in this situation are repulsive, increasing the rate at which agent's belief polarize and hence increasing the likelihood of assimilative agent-agent interactions. Groups B and C show results when the confidence threshold for influencer-agent interactions is relatively wider. The results here are mixed. Group B shows that the neutral influencer only increases the rate of clustering by a small amount for the most part, and the increase is not statistically significant. It is only when CT X = 0.1 and CT Y = 0.5 that the increase in p Y = 0.5 ( x = 33, 850, s = 9, 240) to p Y = 1.0 ( x = 27, 910, s = 8, 289) results in a significant increase in the mean time to polarization p <unk> 0.001. Group C shows that when the confidence threshold is high for agent-agent interactions (CT X = 0.9), increasing influencer activity slows cluster formation when the confidence threshold for influencer-agent interactions is high. For example, when CT Y = 0.9, increasing from p Y = 0.5 ( x = 26, 470, s = 4, 644) to p Y = 1.0 ( x = 33, 140, s = 7, 489) results in a significant increase in the average time to polarization p <unk> 0.001. Group D shows the periods at which of clusters (n = 2) form for an extremist influencer where it is evident that increasing the rate of influencer interactions always increases the rate of clustering regardless of the confidence threshold for agent-agent or influencer-agent interactions. Due to the consistency of these results they are presented without tests of statistical significance. Iterations (x 1000) 0.1 0.5 0.9 Figure 4: The number of iterations by which agent belief has polarized, (that is, B(X i ) <unk> +1 or -1) for approximately 80% of agents, by agent-agent confidence threshold, influencer-agent confidence threshold, probability of influencer-agent interaction and whether the influencer is neutral (LHS) or extremist (RHS). Figure 4 shows the number of iterations by which approximately 80% of agents are polarized in terms ideological extremitization of belief (that is, the standard deviation of agent belief <unk> 0.93), as a function of agent-agent confidence threshold, CT X, influencer-agent confidence threshold, CT Y and influencer belief, B(Y ). Ignoring the effect of influencers, the results overall show the rate of ideological extremitization increases as the confidence threshold for agent-agent interactions increases. There is a significant decrease in the average number of iterations at which 80% polarization occurs from CT X = 0.1 ( x = 15, 371, s = 1, 642) to CT X = 0.5 ( x = 14, 653, s = 2, 392), p 0.001 and again from CT X = 0.5 to CT X = 0.9 ( x = 12, 450, s = 1, 498) to, p 0.001. Looking now at the effect of the influencer, Group A shows that increased activity by a neutral influencer (B(Y ) = 0) having a narrow confidence threshold for influencer-agent interactions (CT Y = 0.1), always increases the rate of agent polarization due to the influencer-agent interactions being predominantly repulsive. Groups B and C show that when there is a wider confidence threshold for influencer-agent interactions, the increased activity by the neutral influencer always reduces the rate of polarization. The censoring of trials at 50, 000 iterations is evident for these two groups, indicating the extreme retardation of belief extremitization when influencer activity is high, that is, when p Y = 1.0. Group D shows that in the case of an extremist influencer the rate of polarization always increases in line with an increased rate of influencer-agent interactions. Due to the consistency of results within each group, they are presented without tests of statistical significance. --- SUMMARY AND CONCLUSION Taking the results in Figures 3 and4 together, Group D shows that increased activity by extremist influencers always increases the rate of polarization by increasing both the rate of ideological separation (polarization of belief), and the rate of psychological distancing (the formation of polarized clusters) regardless of the confidence threshold for agent-agent or influencer-agent interactions. This result confirms the commonly held concern that messaging by extremists of any political persuasion or ideology will increase the likelihood of polarization within communities. The effect of a neutral influencer is less straightforward. Group A shows that a neutral influencer having a narrow confidence threshold for influencer-agent interactions will increase the rate of polarization, due to the largely repulsive influence of the interaction. This accords to real world experience when, in situations of ideological division (for example, on the issue of climate change) even when confronted by ostensibly neutral opinion leaders (such as scientists), people in opposing ideological camps may still maintain their belief regardless of the evidence in the influencers message. We think there are two possible reasons for this. Firstly, even a neutral message (that is, equidistant from either pole) may still be polarizing. For example, a balanced message by a scientist on climate change may still lead to confirmation bias in the pro-environmental camp (resulting in psychological distancing), and dis-confirmation bias in the deniers camp, with information more likely to be endorsed by the outgroup as unreliable (increasing ideological extremitization). Secondly, ineffective, non-charismatic influencers tend to be counter-productive as they increase the rate of polarization. Information from an unpopular (unconvincing) communicator is likely to be seen as an outsider, and hence unreliable, by both camps, leading to increased polarization in both. Groups B and C show that increased activity by a neutral influencer with a moderate to wide reach will always reduce the rate of ideological extremitization, that is the polarizing of agent belief. Although the effect of the neutral influencer on psychological distancing (clustering) is neutral when agents have a fairly narrow confidence threshold for agent-agent interactions, Group C shows that when the confidence threshold is wide, cluster formation is slowed. In either case, an important observation is that these clusters contain agents that are less polarized (that is, more moderate) in belief than would be the case if interaction with the influencer was reduced. In this regard, Group B can be thought of as being similar to a conservative community, where individuals form more moderate clusters, even though the rate of clustering is similar to society as a whole. Group C represents a very tolerant community, where a neutral influencer reduces the rate of clustering to a small degree, and has the potential to reduce the rate of ideological separation significantly. Again, even though clusters form in this community, they contain more moderate members than would be the case without the neutral influencer. Our findings suggest that rather than trying to stop or prevent polarization (which most research indicates is ineffective), evidence-based communication strategies could be used to slow the rate of ideological extremitization. For example, education campaigns may be employed to increase the publics tolerance to diverse opinions in combination with moderate messaging from charismatic opinion leaders. A limitation of the current work is that in the real world, individuals are confronted by multiple issues of varying significance simultaneously. In this situation, the tendency to polarize on a single issue may be overwhelmed by competing issues. Notwithstanding, this research offers an insight into the mechanism by which dissent may lead to polarization, and the effect of influencers in accelerating or mitigating that Betts and Bliuc process. Although different societal issues in varying social contexts will each require a unique strategy to address polarization, a greater understanding of the processes by which polarization occurs, and the role played by influencers, can only help improve public strategies to address this societal concern. This project has only employed a small range of parameter values to describe the artificial society. Future research includes testing the robustness of our current conclusions across a broader parameter space. --- AUTHOR BIOGRAPHIES JOHN M. BETTS is a Senior Lecturer and Director of Education in the Department of Data Science and Artificial Intelligence, Faculty of Information Technology at Monash University, Australia. His research applies computational modeling, optimization and simulation across diverse fields. Ongoing research with social and political scientists investigates the language use and social dynamics of online white supremacist communities to better understand the spread of influence and social cohesion in these groups. His e-mail address is [email protected]. His website is https://research.monash.edu/en/persons/john-betts. ANA-MARIA BLIUC is a social and political psychologist at the University of Dundee. She has a PhD in Psychology from the Australian National University. Her research examines how peoples social identities influence their behavior in a range of contexts including health, environmental (climate change), and socio-political (collective action and social change). Research into online communities (including far-right and white supremacist, and recovery from addiction) has examined how collective identities and behaviors are shaped through online interactions. Her email address is [email protected]. Her website is https://www.dundee.ac.uk/people/ana-maria-bliuc.

Societies can polarize when there is disagreement on important issues. The rise of social media in recent years has led to the phenomenon of influencers, who are now prominent in public debate, especially online. However, their effect on polarization is not well understood. To address this gap in knowledge, we create an Agent-Based Model of a society into which an influencer is introduced and their effect on polarization observed. Results show that an influencer holding extreme views will always increase the rate of polarization, with this effect increasing in line with their reach and activity level. The effect of a neutral influencer varies with the tolerance for opposing beliefs in the society: slowing the rate of polarization for relatively tolerant societies, but increasing the rate when societies are more conservative, or the influencer has narrow reach. Consequently, these results have implications for the design of influencer campaigns for social good.

UC is a social and political psychologist at the University of Dundee. She has a PhD in Psychology from the Australian National University. Her research examines how peoples social identities influence their behavior in a range of contexts including health, environmental (climate change), and socio-political (collective action and social change). Research into online communities (including far-right and white supremacist, and recovery from addiction) has examined how collective identities and behaviors are shaped through online interactions. Her email address is [email protected]. Her website is https://www.dundee.ac.uk/people/ana-maria-bliuc.

Societies can polarize when there is disagreement on important issues. The rise of social media in recent years has led to the phenomenon of influencers, who are now prominent in public debate, especially online. However, their effect on polarization is not well understood. To address this gap in knowledge, we create an Agent-Based Model of a society into which an influencer is introduced and their effect on polarization observed. Results show that an influencer holding extreme views will always increase the rate of polarization, with this effect increasing in line with their reach and activity level. The effect of a neutral influencer varies with the tolerance for opposing beliefs in the society: slowing the rate of polarization for relatively tolerant societies, but increasing the rate when societies are more conservative, or the influencer has narrow reach. Consequently, these results have implications for the design of influencer campaigns for social good.

INTRODUCTION The COVID-19 (coronavirus disease 2019) pandemic, together with the subsequent social distancing measures, caused drastic changes in the global economy as well as individual lifestyles, health and well-being. 1 These shifts have been accompanied by changes in family planning, including fertility plans. 2 Fertility intentions refer to the desire to have a --- WHAT IS ALREADY KNOWN ON THIS TOPIC <unk> Studies in individual countries have suggested that the COVID-19 pandemic has contributed to the change in people's fertility intentions. --- WHAT THIS STUDY ADDS <unk> Pandemic-induced isolation/quarantine and deteriorating mental health were associated with postponed fertility intentions, while living with a steady partner during social distancing measures was associated with accelerated women's fertility intentions. Original research child, which is influenced by an individual's current situation. 3 Fertility intentions have been recognised as a fairly reliable predictor of individual fertility-related behaviour. 4 Given that shifting fertility intentions manifest as changes in fertility patterns, 5 the changes in fertility intentions as a result of the pandemic would impact society at large through shifts in social and economic stability as well as population and migration policy. Thus far, studies have shown conflicting evidence regarding changes in fertility intentions related to COVID-19. Some studies in European countries (Italy, Poland, Germany, France, Spain), the UK and the United States (US) reported that 20% to 57.8% of respondents postponed fertility intentions, and 14.2% to 36.5% of respondents even abandoned their previous fertility plans because of the COVID-19 pandemic. [6][7][8][9] However, in the Italian and US studies, 10-20% of participants who did not previously intend to have children wanted a child sooner during the pandemic. 7 8 Previous studies have provided clear evidence regarding the impact of living with a partner on fertility intentions, demonstrating that living with a partner is a prerequisite for having children. 10 11 Economic recessions are often accompanied by a decline in fertility, as income reductions lead to a postponement of fertility intentions among women until their financial situation improves. 12 13 Additionally, women with better mental well-being are unlikely to postpone their desire to have children and are more likely to have children sooner. 14 Recent studies have further examined the impact of COVID-19-induced changes in family relationships, financial status and mental health on fertility intentions, 8 9 15-17 highlighting the associations between COVID-19-related experiences and fertility intention shifts. Moreover, the COVID-19 pandemic, with its infections and associated isolation measures, may also contribute to changes in women's fertility intentions. The International Sexual Health And REproductive Health (I-SHARE) consortium, comprising researchers from 30 countries worldwide, used a standardised survey instrument to investigate the effects of the first wave of COVID-19 on sexual and reproductive health (SRH). 18 This multi-country, cross-sectional study aimed to explore how fertility intentions have changed due to the COVID-19 pandemic, and identify factors associated with changes in fertility intentions among an international sample of women aged 18-49 years, with the hope of informing global health and demographic policy. --- METHODS --- Study design This cross-sectional study was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. 19 The I-SHARE consortium convened a group of SRH researchers to administer a common survey instrument in their respective countries as an online survey. 18 The online survey used social media, partner organisations, paid social media advertising, university websites, telephone interviews, and television or newspapers to recruit participants. 20 Individuals who agreed to answer the online questionnaire participated in the specific survey. A more detailed description of the study methods, including the study protocol, questionnaire development, participant recruitment and data collection, is available elsewhere. 18 In summary, a total of 30 countries participated in the survey, which was distributed through local, regional and national networks between 20 July 2020 and 15 February 2021. 20 The majority of countries employed convenience sampling, while two countries (Denmark, Czech Republic) used representative samples, and six countries (Sweden, Botswana, Uganda, Lebanon, Kenya and Argentina) utilised online panels. This study examines the associations between the COVID-19 pandemic and fertility intentions, and therefore focuses on the subsample of 18-49-year-old women. --- Participants Participants had to be aged 18 years or older, live in their respective participating countries, be able to read and understand questionnaires, have access to online surveys, and be willing to provide informed consent. Overall, 23 067 participants from 30 countries completed the survey. Based on the prespecified plan and considering the potential for greater heterogeneity, the I-SHARE group advised conducting multi-country studies with sample sizes of at least 200 participants (online supplemental table 1). 18 The eligibility criteria for participants included in this analysis were as follows: aged between 18 and 49 years, assigned female at birth, and answered the question regarding fertility intentions. The inclusion and exclusion processes are outlined in figure 1. Eventually, the --- Measures --- Dependent variable The dependent variable for this study was fertility intentions, and participants were asked specifically about the pandemic-related changes to their fertility intentions with the question "Have you recently changed your mind about having a child soon because of COVID-19?". The answer options were: (1) Yes, I have decided to postpone my decision to have a child in the near future (postponed); (2) Yes, I have decided I want to have a child sooner (accelerated); and (3) No, I have not changed my plans (not changed). --- Independent variables Key explanatory variables of interest included COVID-19 pandemic-induced exposures such as testing for COVID-19, isolation/quarantine, personal income loss, living together with a steady partner, and mental health during the pandemic. Testing for COVID-19, assessed by the question "Were you ever tested for COVID-19?" with the following answer options: 1=no; 2=yes, always negative; 3=yes, positive at least once. Isolation/quarantine was assessed by the question "Were you ever in (self-)isolation/quarantine because of symptoms or because you were in close contact with someone with COVID-19 or because you returned from a country that had a large number of cases?" with the following answer options: 1=yes; 2=no. Personal income loss was assessed by the question "Have you personally experienced a loss of income? " with answer options: 1=yes; 2=no. Living together with a steady partner was assessed by the question "During the COVID-19 social distancing measures is/was your steady partner living with you in the same place?" with the following answer options: 1=no, s/he stays elsewhere; 2=yes, the whole time/part of the time. The questionnaire covered four different mental health items: frequency of getting angry, frequency of feeling frustrated, frequency of feeling bored, and frequency of worrying about finances. Participants were asked about their mental health status during the COVID-19 social distancing measures compared with 3 months before the COVID-19 pandemic, assessed by the same question asked four times: "Does this happen more or less since the start of the COVID-19 social distancing measures?". Response options for the four questions were the same: 1=same; 2=more (a lot more/more); 3=less (less/a lot less). The response options were redivided into two categories: 1=same or less; 2=more. --- Confounding variables Age was measured as a continuous variable in the questionnaire and was divided into two categories in this study: 18-32 years old; 33-49 years old. Education was measured as: no formal education; some/completed primary education; some/completed secondary education; some college/university; completed college/ university. These were combined into three categories for regression analysis: no formal education-secondary education; some college/university; completed college/ university. Marital status was classified into four categories: single, no partner; not married, have a partner; legally married; widowed/divorced/other. Number of children was measured as a continuous variable in the questionnaire and was divided into three categories in this study: 0; 1; <unk>2. When asked about their religion, participants selected their religion from a list of country-specific options and then were classified into two groups: has no religious belief; has a religious belief. In terms of country-level variables, the participating countries were recoded as 21 : sub-Saharan Africa (SSA); Latin America and the Caribbean (LAC); North America (NA); East Asia and the Pacific (EAP); Europe and Central Asia (ECA). Following the classification of World Bank income levels, participating countries were recoded as 21 : high-income countries; uppermiddle-income countries; low-or low-middle-income countries. The minimum and maximum number of participants by country were reported. --- Statistical analysis Descriptive statistics were used to summarise the participating countries and participants' sociodemographic characteristics. Due to significant variation among the participating countries, the analysis was performed at the individual level. The Bonferroni correction was applied to account for multiple hypothesis testing. The comparison of sociodemographic characteristics between participants in this study and all reproductive-aged women from the I-SHARE survey is shown in online supplemental table 2. The distribution of fertility intentions across country regions, country income levels and participants' sociodemographic characteristics are also described (online supplemental table 3). In the multivariable analysis, multinomial probit regression was performed to examine the associations between fertility intentions and pandemic-induced exposures at the individual level. 22 For each independent variable we calculated the crude OR, the adjusted OR after controlling for all confounders, and the adjusted OR (AOR) after controlling for both confounders and independent variables (online supplemental table 4). Given the multi-country data used in this study, we adjusted the estimates to account for clustering of individuals within their respective countries. 23 The sensitivity analyses, which utilised countrycluster robust bootstrap estimation, are presented in online supplemental table 5, and countries or the country income level with more than 100 participants --- Original research are shown in online supplemental table 6. The associations between pandemic-induced exposures and fertility intentions for participants at different country income levels are shown in online supplemental table 7 and online supplemental table 8. All tests were twotailed, the significance level was set at 0.05, and the analyses were performed using Stata 16. --- RESULTS According to the flowchart of participants (figure 1), 10 672 women were enrolled in the study, with 10.2% having postponed fertility intention, 4.2% having accelerated fertility intention and 85.6% having not changed their fertility intention due to the COVID-19 pandemic. The participants in this study had a mean age of 29.9 years and a median age of 28 years (IQR: 24-35); 47.2% were not married but had a steady partner; most of the participants (60.1%) had completed their college or university; more than half (52.0%) had a religious belief; and the majority of the participants (72.9%) had no children (table 1). Table 2 shows the distribution of fertility intentions by pandemic-induced exposures. Participants with negative COVID-19 test results had a lower proportion of postponing fertility intention (9.4% vs 10.3%) --- Original research whereas those with positive test results had greater proportions of both postponing (13.7% vs 10.3%) and accelerating (5.8% vs 3.9%) fertility intention. A higher proportion of participants who had been isolated/quarantined reported postponed fertility intention (12.7% vs 9.3%) or accelerated fertility intention (5.0% vs 4.0%). Participants who experienced personal income loss were more likely to report both postponed (13.6% vs 8.4%) and accelerated fertility intentions (5.1% vs 3.8%). Participants who lived together with their steady partner were more likely to report both accelerated fertility intention (12.6% vs 9.2%) and postponed fertility intention (5.1% vs 3.2%). Compared with participants with equal or fewer mental health issues, those who had more frequency of getting angry (12.8% vs 8.2%), more frequency of feeling frustrated (12.1% vs 7.7%), more frequency of feeling bored (11.6% vs 8.6%) and more frequency of worrying about their financial situation (13.1% vs 7.1%) during the COVID-19 social distancing measures were more likely to postpone fertility intentions. In the multinomial probit regression analysis (table 3), after all variables are adjusted, the results show that participants who tested negative for COVID-19 were less likely to postpone fertility intentions (AOR=0.79; 95% CI 0.64 to 0.97) compared with those who did not test for COVID-19. Participants who had ever been isolated/quarantined were more likely to postpone fertility intentions (AOR=1.41; 95% CI 1.18 to 1.69) compared with those who had never been isolated or quarantined. When compared with those who did not experience a loss in personal income, the likelihood of postponing (AOR=1.29; 95% CI 1.12 to 1.49) or accelerating (AOR=1.22; 95% CI 1.06 to 1.40) fertility intentions increased among those who did. Participants who lived with their partner during the COVID-19 social distancing measures were more likely to accelerate (AOR=1.57; 95% CI 1.10 to 2.23) --- Original research their fertility intentions compared with those who did not. In terms of mental health, participants who reported increased frequency of getting angry (AOR=1.30; 95% CI 1.18 to 1.43), increased frequency of feeling frustrated (AOR=1.21; 95% CI 1.04 to 1.41) or increased frequency of worrying about their financial situation (AOR=1.43; 95% CI 1.16 to 1.75) were more likely to postpone their fertility intentions. The associations between pandemic-induced exposures and fertility intentions were robust in the sensitivity analyses (online supplemental tables 5 and 6). --- DISCUSSION The associations between the COVID-19 pandemic and reproductive-aged women's fertility intentions are investigated in this multi-country, cross-sectional study. Fewer than a third of those who changed their fertility intentions decided they wanted a child sooner, implying that many women may be postponing plans to expand their families, with long-term implications for the total fertility rate. This study expands the literature by using a multi-country approach to explore pandemic-related factors that contributed to the postponement or acceleration in women's fertility intentions. A relationship between COVID-19-related experiences and changes in fertility intentions is observed in this study. Women who received negative COVID-19 test results were less likely to postpone their fertility intentions. Similarly, women who were quarantined due to COVID-19 were more likely to postpone their fertility intentions. Underlying reasons explaining this trend may be the fear of the effect of COVID-19 infection on complications during pregnancy, as women may avoid conceiving in such circumstances. 2 In addition, women who lived with their partners during COVID-19 social distancing measures were more likely to accelerate their fertility intentions compared with women who did not change their fertility intentions. This may be due to the fact that the COVID-19 social distancing measures allowed women to spend more time with their partners, improving the quality of their relationships and encouraging them to expand their families. 15 24 The economic recession brought on by the COVID-19 pandemic was a crucial factor for fertility intentions among our respondents. Women in our study who lost personal income during the pandemic were more likely to postpone their desire to have children. This is consistent with the findings of Italian and Polish studies in which women who postponed or even abandoned their intention to have children cited reduced income as a major concern. 8 9 In contrast, a US study showed that women who were financially insecure reported newly considering pregnancy. 25 The current study also found that women who lost personal income were significantly more likely to want a child sooner. This may be explained by such women using this period to have children, as the barriers to childbearing diminished, and the available time for childbearing and childcaring increased. 26 Evidently a loss in personal income may influence fertility intentions bidirectionally. Another important topic was the mental health deterioration in association with pandemic-related social distancing measures. Our findings are in favour of the mental health burden produced by the COVID-19 pandemic having negative associations with fertility decisions. 27 28 Women who had a higher frequency of getting angry, frustrated, or worried about finances during the pandemic were more likely to postpone their desire to have children. This study supports the need to learn more about how the pandemic affects mental health and fertility decisions concomitantly. This study is not without shortcomings. Because the participants were mostly recruited through convenience sampling with no quota set, and those individuals who did not have access to or could not use electronic devices were also not represented, this study was deficient in representativeness. Additionally, the study's findings may be primarily driven by high-income countries, given that only a small number of participants were from low-or low-middleincome countries, and the majority were from highincome countries, resulting in limited generalisability. Furthermore, the number of participants varied widely between countries or regions, making it impractical to conduct inter-country or inter-regional comparisons for this study. --- CONCLUSIONS In this large, diverse, cross-sectional study of reproductive-aged women, few women wanted a child sooner following the outbreak of COVID-19 and there was substantial uncertainty about whether their intentions would change as the crisis continued. In the context of the pandemic, the COVID-19-related experiences, personal income loss, living together with a steady partner, and mental health among women might influence the fertility rate over the long term. Future research could further investigate specific factors related to the pandemic that influence the decision to postpone or accelerate fertility plans, as well as whether fertility intention patterns differ across countries and regional groups. --- Data are available upon reasonable request. The authors have affirmed their willingness to provide all anonymized participant data that form the foundation of the analysis, results, and conclusions presented in this paper. Data are available from Prof. Wei-Hong Zhang who can be contacted on Weihong. zhang@ ugent. be. The survey protocol and final survey instrument are available at https://sti.bmj.com/ content/97/2/88.longhttps://sti.bmj.com/content/97/2/88.long.

The COVID-19 pandemic, together with the subsequent social distancing measures, could lead to shifts in family and fertility planning. This study aimed to explore the associations between the COVID-19 pandemic and changes in fertility intentions among an international sample of reproductive-aged women. Methods A multi-country, cross-sectional study based on data from 10 672 women aged 18-49 years who participated in the International Sexual Health And REproductive Health (I-SHARE) study, which organised an international online survey between July 2020 and February 2021. Factors associated with changes in fertility intentions were explored using multinomial probit regression models. Cluster-robust standard errors were used to calculate model parameters. Results Of 10 672 included reproductive-aged women, 14.4% reported changing their fertility intentions due to the pandemic, with 10.2% postponement and 4.2% acceleration. Women who had ever been isolated/quarantined were more likely to postpone their fertility intentions (adjusted odds ratio (AOR)=1.41; 95% CI 1.18 to 1.69) compared with those who had not; women who lived with a steady partner were more likely to want children sooner (AOR=1.57; 95% CI 1.10 to 2.23) compared with those who did not; and those who reported a higher frequency of getting angry, feeling frustrated, or worrying about their finances were more likely to postpone their fertility intentions. The main findings were robust in the sensitivity analyses. Conclusions Most women who changed fertility intentions because of the pandemic have postponed intentions to expand their families. The pandemic-induced exposures were associated with these postponements.⇒ Changes in women's fertility intentions could cause uncertainty in fertility rates over the long term. Authorities should respond to the reproductive health needs of women during public health crises.

has long been considered to be influential in the initiation of drug use, with research demonstrating that identification with peer groups, cultures, or communities can serve as either a risk or a protective factor for deviant behaviors (Battistich & Hom, 1997;Duncan, Boisjoly, Kremer, Levy, & Eccles, 2005;Nasim, Belgrave, Jagers, Wilson, & Owens, 2007). However, the impact of belongingness in later stages of drug use and abuse is less understood. The correlates of long-term illicit drug use, including alienation from family and mainstream society, unemployment, and risky behaviors (e.g., drug dealing, property crime, and violence) are associated with a drug use subculture that runs contrary to many of the values of mainstream culture. As individuals gain skills (e.g., finding drugs, making money through illicit means) in this culture and social connection to its members a perceived belongingness to this culture may develop. Moreover, chronic drug use may create a context in which belonging to a subculture of drug use may supersede motivations or the perceived ability to belong to more mainstream, drug-free communities. To date, discussion of belongingness to drug subculture has been mainly anecdotal (Hughes, 2007;Bell, 2010;Singer, 2006;Mullen & Hammersley, 2006;McIntosh & McKeganey, 2001). Treatment providers consider it essential to encourage patients to break their affiliation with drug subculture and to develop new lifestyle patterns more conducive to abstinence. This may include distancing oneself from established social networks, refraining from illicit and/or maladaptive behaviors that generate excitement or a sense of worth, and participating in a more traditional society. However, patients often feel marginalized from traditional society due to drug use and/or delays in the attainment of traditional role responsibilities or achievements. In short, individuals in treatment are encouraged to forgo the sense of identity that the social connection, excitement, and sense of competence and mastery that belonging to a deviant subculture provides, without necessarily being able to enter successfully into the dominant culture. Although accounts of drug subculture are varied and complex, four clusters of cultural elements have been consistently identified across accounts of illicit drug subculture and communities: (1) alienation from mainstream society, (2) strong social ties with other drug users, (3) feelings of excitement/pleasure and (4) feelings of effectiveness/mastery deriving from illicit drug use and other associated risky behaviors. Taken together, these domains may comprise the core elements of an overarching construct of affiliation with drug culture. First, as individuals form stronger ties to drug subculture, they are increasingly isolated from mainstream culture roles and society. For example, only a minority of heroin addicts are employed prior to or upon admission to treatment programs (estimated employment rates of 11-35%; Platt, 1995;Substance Abuse and Mental Health Services Administration, Office of Applied Studies, 2008). Users also face significant stigma from the mainstream culture, including drug-free family members and care providers (Mullen & Hammersley, 2006;McLaughlin & Long, 1996). For example, one study of perceived discrimination in a sample of 1008 illicit drug users found that over 75% of the sample had been rejected by family members and 23% had been prevented from obtaining medical care due to drug use (Ahern, Stuber, & Galea, 2007). Furthermore, a strong association has been established between illicit drug use and homelessness (Fountain, Howes, Marsden, Taylor, & Strang, 2003;Neale, 2001), which results in further alienation and marginalization from the larger society. Second, strong social ties emerge among drug users. Drug use is a social activity; locating, purchasing, and consuming drugs is a bonding ritual for some (Neale, 2002). As use increases, one's social network shrinks and becomes populated mainly by other users (Van Der Poel & Van De Mheen, 2006). Qualitative accounts suggest that drug-using friends offer practical and emotional support; they may protect one another from overdose or share information about dealers (Neale, 2002). Although, as some suggest, these relationships may be founded mainly on mutual self-interest in obtaining drugs, several narratives suggest that acts of kindness and cohesiveness are not absent from the drug using community (Hanson, 1985;Neale, 2002;Singer, 2006). One study of social networks in individuals at-risk for HIV found that in a subsample of injection drug users, 64% of users reported overlap between their "drug networks" and "social support networks" (Suh, Mandell, Latkin, & Kim, 1997). Third, drug subculture offers shared value placed on the excitement and pleasure of drug use and the associated deviant behaviors. Among subcultures built around a specific drug, strong beliefs exist that one's specific drug of choice is the "best high" (Golub et al., 2005). For example, interviews with heroin users describe a sense of tedium and boredom with neighborhood life that drives users to relapse (Mullen & Hammersley, 2006). Rosenbaum (1981) uses the terms "the high" and "the life" to describe two reinforcing aspects of use. "The high" reflects the immediate pleasurable effects of substance use, while "the life" reflects the fact that for many, life as a regular user offers something to look forward to and brings more meaning, excitement, and opportunity than a "straight" lifestyle. Lastly, as individuals become entrenched in drug subculture, they may derive a sense of purpose and efficacy from success with the tasks associated with this culture (e.g., finding drugs, making money through illicit sources, avoiding being caught by authorities). One's self identity becomes tied up in drug use such that even one's positive sense of self derives from their drug habit or related illicit/maladaptive behaviors. In a series of qualitative interviews, heroin addicts describe how quitting resulted in a loss of focus and direction which was "akin to losing a job" (Mullen & Hammersley, 2006). For many, the "drug scene" may offer more opportunity for prestige and respect than the "straight" life (Neale, 2002;Hanson, 1985). Because many users have faced social exclusion and lack of opportunities prior to using drugs, delinquent behavior and drug use may provide a sense of mastery not afforded to them in the wider culture (Buchanan, 2004). Overall, the literature suggests that key elements of drug subculture include alienation from mainstream society, strong social ties with other drug users, and feelings of excitement and worth deriving from illicit drug use and associated behaviors (e.g., obtaining drugs, engaging in theft or drug dealing). The presence of these four themes across narrative accounts, qualitative interviews, and clinical experience with individuals struggling with addiction suggest that they are common elements of many drug users' experience. Importantly, belongingness to drug subculture is an individual difference variable that may impede engagement in and efficacy of treatment. For example, for opioid-dependent individuals, integration with a larger community (as measured by more positive perceptions of community and higher rate of employment) helps identify those who choose to seek treatment as compared to those who do not (Schwartz, Kelly, O'Grady, Mitchell, & Brown, 2011). In contrast, higher rates of drug use within one's social network is associated with reduced rates of treatment seeking and poorer treatment outcome in samples of injection drug users (Booth, Corsi, & Mikulich, 2003;Brewer, Catalano, Haggerty, Gainey, & Fleming, 1998;Gogineni, Stein, & Friedmann, 2000;Lloyd et al., 2008;Schroeder et al., 2001). Hence, drug use, treatment engagement, and treatment response are contingent on social group affiliation; increased exclusion from the dominant culture and stronger membership in drug subculture are associated with more negative outcomes. For individuals whose social network consists mostly of fellow drug users, who feel alienated from the larger community, and who derive strong feelings of mastery and excitement from their drug use, the perceived rewards of continued drug use may outweigh those of treatment engagement and sustained recovery. Thus measuring individual differences in belongingness to drug subculture may provide important clinical information in populations of illicit drug users. The purpose of this study was to describe the development and psychometric properties of a self-rated measure called the Belongingness to Drug Culture Questionnaire (BDCQ). The impetus to develop the BDCQ came from our clinical observation of the characteristics of a cohort of outpatients who continued to abuse drugs despite the combination of methadone maintenance treatment (MMT) and focused psychotherapy. These clinical observations were subsequently refined in relation to findings in the drug abuse literature (described above). Opioid replacement therapies are associated with positive outcomes from many patients; however, high rates of continued drug use, treatment drop-out, and subsequent relapse are also common (Kleber, 2008). Thus, the identification of barriers to treatment engagement and success is of particular importance in this population. Although belongingness to drug culture is likely elevated in non-treatment seeking illicit drug users, its measurement is especially relevant in treatment-seeking patients because it may elucidate characteristics that, when modified through treatment, may aid recovery. Accordingly, following early scale development, we recruited a sample of MMT patients as the initial validation sample for the BDCQ. In the current study, the internal consistency reliability and the convergent and incremental validity of the BDCQ were examined in a sample of individuals currently undergoing methadone maintenance treatment for opioid dependence. To remain consistent with our theoretical characterization of belongingness to drug culture, items were developed based on the four conceptual categories described above (alienation, social connection, excitement, and effectiveness). We hypothesized that both total scale scores and scores within these four conceptual domains would demonstrate adequate internal consistency. We hypothesized that BDCQ total scores would be positively associated with recent drug use and regular drug use in close interpersonal relationships, as well as negatively associated with desire to quit. Additionally, we hypothesized that BDCQ scores would be related to measures of perceived belonging to an undefined social group, impulsivity, drug use motives, and psychopathy. --- Method Participants The initial validation sample consisted of 96 opioid-dependent patients receiving methadone maintenance treatment (MMT) at one of two outpatient clinics in the Boston area. In order to meet eligibility criteria, participants were required to be at least 18 years of age, able to read and understand English, and currently receiving MMT. Details regarding the demographic characteristics of the study sample are presented in Table 1. The sample was 54% female and predominantly Caucasian (77%). Level of education varied, with approximately 16% with less than a high school degree, 54% with a high school degree (or equivalent), and 30% with more than a high school education. Participants had been receiving methadone and counseling treatment at their current clinic for an average of 33 months. Eighty-seven percent (n = 83) of participants reported that they had not worked in the past 30 days. --- Procedures Participants were recruited specifically for this study by posted signs in the methadone dosing area at the two participating methadone maintenance clinics. Following an explanation of study procedures, participants provided written informed consent and completed a packet of self-report questionnaires, and provided authorization for release of specific information from clinic files (see below). Participants received a $10 gift card for completing all study questionnaires. The Boston University Institutional Review Board approved all procedures. --- Measures Belongingness to Drug Culture-The Belongingness to Drug Culture Questionnaire (BDCQ; Appendix A) is a self-report questionnaire measuring individual affiliation with drug subculture. Twenty-eight items were developed from the four conceptual categories of drug subculture reviewed above: (1) excitement related to drug seeking and use, (2) sense of effectiveness in obtaining drugs, (3) social connection with other drug users, and (4) alienation from traditional society. To ensure that the questions reflected generally accepted definitions of these constructs, some items were derived from scales of acculturation, sensation-seeking, and social support, but modified for relevance to the context of drug use and subculture. To maximize the applicability of the BDCQ to heterogeneous groups, the readability of items was assessed during scale development and items were adjusted accordingly to best approximate American Medical Association recommendations for the reading level of patient materials (Ad Hoc Committee on Health Literacy, 1999). Likewise, formatting and linguistic elements that may negatively influence comprehension level of items (e.g., reverse-scored items, linguistic difficulty; McHugh, Rasmussen, & Otto, in press; Shumway, Sentell, Unick, & Bamberg, 2004) were not included in this measure. Responses are rated on a 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree). BDCQ total scores are calculated by summing all items. Impulsivity-The UPPS Scale (Whiteside & Lynam, 2001) is a 44-item self-report measure of four personality factors hypothesized to underlie impulsive behaviors: sensation seeking, (lack of) premeditation, (lack of) perseverance, and urgency. The measure has demonstrated strong internal consistency reliability for each of the four subscales and has been linked to self-report of risky behaviors (e.g., Whiteside & Lynam, 2003). In the current sample, the UPPS subscale scores demonstrated high internal consistency reliability (sensation seeking <unk>= 0.79; premeditation <unk>=.76; perseverance <unk>=.75; urgency <unk>=.82). Interpersonal belonging-The Interpersonal Needs Questionnaire (INQ; Van Orden, Witte, Gordon, Bender, & Joiner, 2008) is a 12-item self-report questionnaire designed to measure beliefs about the extent to which individuals feel connected to others. Using a 7point Likert scale, individuals rate the degree to which each item is true for them. The INQ contains 2 subscales: perceived belongingness and perceived burdensomeness. Higher scores on the INQ reflect greater levels of "thwarted" belongingness and burdensomeness, respectively. The INQ has been shown to have high internal consistency, and evidence for its construct validity is encouraging (Van Orden, et al., 2008;Van Orden, 2009). In the current sample, internal consistency was strong for both subscales (<unk>=.85 for belongingness and <unk>=.78 for burdensomeness). Motives for drug use-The Drug Use Motives Questionnaire (DUMQ; Cooper, Russell, Skinner, & Windle, 1992) is a 15-item self-report inventory designed to provide an assessment of coping, social and enhancement motives for alcohol use. A modified DUMQ was used to assess non-alcohol drug use motives (as utilized by Mueser, Nishith, Tracy, DeGirolamo, & Molinaro, 1995). All three subscales demonstrated strong internal consistency in the current sample (<unk>=.89 for coping, <unk>=.88 for social, <unk>=.83 for enhancement). Psychopathy-The Levenson Self-report Psychopathy Scale (LSRPS; Levenson, Kiehl, & Fitzpatrick, 1995) is a 26-item measure designed to measure psychopathic attitudes and beliefs based on Hare's Psychopathy Checklist (Hare, 1991). It has been strongly associated with more comprehensive self-reported and clinician-rated measures of antisocial tendencies in a number of populations ranging from college students to institutionalized criminal offenders, and has also been shown to predict antisocial behavior (Levenson, Kiehl, & Fitzpatrick, 1995;Seibert, Miller, Few, Zeichner, &Lynam, 2010;Brinkley, Schmitt, Smith, & Newman, 2001). In the current sample, internal consistency was adequate (<unk>=.75). We anticipated that psychopathy would be related to BDCQ scores because a number of traits characterizing psychopathy are also implicated in belongingness to drug culture, such as sensation-seeking, violation of social norms, and disregard for the law. --- Social network drug use: Participants were asked to list the initials of up to 3 people they feel closest to, and to report whether each person uses illicit drugs regularly. --- Desire to quit: Participants were asked to rate their desire to stop using drugs on a scale from 1-10, with 10 being the greatest desire, and 1 being the least. Days of drug use: Number of days of illicit drug use in past 30 days was also collected via self-report. --- Clinical record data: The following information was assessed from each participant's clinic record: 1) participants' oral toxicology test results for the past 3 months, and 2) date which participants initiated MMT. Oral swab drug tests are administered to patients on a random basis (15 or 24 times per year at each respective clinic). Due to limitations in the availability of these data, for 36 participants, drug test results were only available for one month prior to study participation. Given the presence of missing data and the infrequent nature of drug testing, the relationship of this data with BDCQ scores was considered a secondary analysis. --- Readability: The ability of patients to read and comprehend measure items is a critical element of measure validity. Readability formulas provide an estimate of the reading grade level required to understand a text. Three common readability formulas were used to examine the reading grade level of the BDCQ: the Flesch Reading Ease formula (Flesch, 1948), the SMOG Readability formula (McLaughlin, 1969), and the FORCAST formula (Kern, Sticht, Welty, & Hauke, 1976). The results of these three formulas were averaged to provide an estimate of reading grade level. This was done to minimize the limitations of any one formula. Instructions and measure items were evaluated separately. --- Statistical Analysis Data analyses were conducted using SPSS version 16.0. Item selection was a two-stage process. First, in order to confirm that the BDCQ items adequately represented the four conceptual domains of belongingness to drug culture described in the introduction, itemdomain correlations were conducted. All items with a corrected item-domain total correlation of <unk>.30 were removed in an iterative fashion starting with the poorest performing item, as per the recommendation by Nunnally & Bernstein (1994). Second, to ensure the unifactorial relevance of the scale and to maintain the intended use of the scale relative to providing a single index of strength of belongingness, we removed all items with a corrected item-total correlation of <unk>.30 in the same iterative manner. Inter-item consistency reliability (Cronbach's alpha) was calculated for total BDCQ scores and each conceptual domain based on the remaining items. Pearson correlations were conducted to evaluate the convergent and divergent validity of the BDCQ and other measures. Statistical significance was set at p<unk>.05 for all analyses. --- Results Descriptive statistics for all measures are presented in Table 2. Four items (see Appendix B) were removed from the BDCQ due to corrected item-domain total correlations of less than 0.3. Two additional items were removed following this due to corrected item-total correlations of less than 0.3. All remaining items had standard deviations ranging from 1.70 to 2.36. The final 22-item scale (see Appendix A) had high internal consistency reliability (<unk>= 0.88) and item-total correlations ranging from 0.30 to 0.66. Internal consistency for the four conceptual domains of the BDCQ ranged from <unk>= 0.58 (alienation) to <unk>= 0.78 (effectiveness; see Table 4 for all values). The readability analysis provided support for good readability relative to recommended standards, with instructions written at a reading grade level of 5 th grade and items at a level of 7 th grade. The inter-correlation matrix of total BDCQ scores and other study measures is presented in Table 5. BDCQ scores were significantly associated with self reported days of drug use in the past 30 days and a lower desire to quit. BDCQ scores were also associated with coping, enhancement, and social drug use motives. The BDCQ shared significant variance with both impulsivity (each of the four UPPS subscales) and the measure of psychopathy and was positively associated with greater levels of thwarted belongingness and burdensomeness as assessed by the INQ. Given that BDCQ scores were significantly associated with self-reported drug use, but not with occasional clinic-administered random drug test results, we also examined consistency between the two drug use variables within subjects. Forty percent (n = 38) of participants reported at least one day of illicit drug use in the past month, and forty percent (n = 38) tested positive for illicit drug use in the past month. The concordance rate between past-month drug use and toxicology results was 73% (see Table 3), a rate which is consistent with other studies (Darke, 1998). Of those who reported at least 1 day of illicit drug use in the past month, 10 (30%) had not received a positive drug test result in the previous month. Of individuals who reported no drug use over the past 30 days, 12 (25%) had tested positive for illicit drugs. --- Discussion The BDCQ appears to be an internally consistent measure that demonstrates favorable construct validity. In addition to high internal consistency for total scale scores, the items within the four conceptual domains also demonstrated generally adequate internal consistency. The items assessing "alienation" (<unk>=.58) had lower internal consistency than the other domains, which may reflect the heterogeneity of the items within this group (e.g. family rejection, perceived societal discrimination). Overall, these results warrant further exploration of the psychometric properties (e.g., latent factor structure) and clinical and research utility of the BDCQ. As expected, the questionnaire was significantly and positively correlated with impulsivity, psychopathy, thwarted belongingness (to a general, unspecified group of "others"), and drug use motives, constructs which are theoretically related to the conceptualization of belongingness to drug culture. These associations were of low to moderate strength, suggesting that the BDCQ is adequately distinct from these measures. The highest association (reflecting 36% shared variance) was with the measure of psychopathy, perhaps indicating the degree of alienation or non-acceptance of dominant cultural values in favor of an illicit subculture membership. Psychopathy is itself associated with substance use and linked to poor treatment outcome (Alterman, Rutherford, Cacciola, McKay, & Boardman, 1998). As a related concept to psychopathy, assessment of the perceived belongingness to a drug subculture attends to potentially modifiable contextual factors that maintain patients' drug use. It has been suggested that for drug-dependent individuals, many of the traits and behaviors assessed in antisociality and psychopathy measures might be better conceptualized as functional responses to and consequences of one's environment rather than true psychopathy (Darke, Kaye, & Finlay-Jones, 1998). The current research extends this perspective, underscoring the role of a drug user identity that may be a useful target for treatment. Individuals who scored higher on the BDCQ reported more frequent past-month drug use as well as lower desire to quit. Among the wide range of variables assessed that are thought to influence drug use, BDCQ scores were one of the few that were significantly associated with self-reported use. These results indicate that the BDCQ may be a useful clinical tool to identify those at risk for continued use during substance dependence treatment. Examination of BDCQ scores among non-treatment-seeking drug using individuals and prospective examination of belongingness as a barrier to treatment success will be important to understanding the value of this construct for understanding substance use disorders. Several of our hypotheses were not confirmed. Total BDCQ scores did not correlate significantly with drug test results. One possible explanation is that drug tests at the methadone clinics are administered approximately 1-2 times per month, which may not be not frequent enough to serve as an adequate measure of drug use. In the current sample, 30% of individuals who reported using within the past month received negative drug test results for that month, suggesting that use is likely higher than drug tests results indicate. In addition, we anticipated that belongingness would be associated with social and enhancement motives for drug use, but did not hypothesize an association between the BCDQ and coping motives. Further evaluation of why belongingness is associated with coping motives (e.g., via greater negative affect or depressive symptoms) will help to understand this association. The measure of peer use administered in this study asked participants to list their three closest relationships and whether or not each individual used drugs regularly. This peer use variable did not correlate significantly with BDCQ scores; however, it was significantly associated with self-reported drug use. This suggests that the BDCQ may not have captured important aspects of social network use that are tied to drug use. The social-related items such as "I belong when I am with other drug users" speak of other drug users in broad terms as a uniform population. These items may not acknowledge differences in attitudes toward drug users as a whole and peer users; participants could hold negative perceptions of drug users more generally that are inconsistent with their beliefs about their friends and family members who use illicit substances More detailed evaluation of drug use within social networks is an important area for future research. In addition, treatments such as 12-step programs or job promotion (contingency management) interventions for drug abuse (DeFulio & Silverman, 2011;Wong & Silverman, 2007) may directly modify a patient's social cohort, providing an additional context for examining the predictive significance of BDCQ in these programs as well as its modification by select treatments. This study has a number of limitations. First, all participants were methadone-maintained opioid-dependent individuals; thus the extent to which these results can be generalized to non-treatment seeking drug users or those receiving other types of treatment is unclear. Because all participants were engaged in treatment, we could not assess the degree to which the BDCQ could identify those most likely to drop out of or otherwise refrain from treatment. This is an important target for future study. Additionally, the relatively high education level as well as the high proportion of Caucasian and female participants may limit the generalizability of the current results. Future studies should include program-wide implementation of BCDQ in order to avoid potential selection bias based on these characteristics. Drug test results from the previous 3 months were available for only 60 participants, and were collected on a monthly schedule. Future examination of the BDCQ should examine its prospective prediction of drug use utilizing a higher density of objective drug measures. Additionally, it is important to note that our results also occur within the context of methadone clinic culture. While belonging to one culture does not preclude affiliation with another, it serves as a reminder that a number of cultures (whether ethnic, institutional, or other) may exert unique interactive effects on each individual. Finally, conclusions cannot be made regarding the factor structure of the BDCQ. Our results support the use of BDCQ total scale scores due to strong internal consistency; however, further research is needed to explore whether the conceptual domains drawn from in the development of the measure reflect unique latent factors under the higher-order factor of belongingness, which may have utility as measure subscales. The current study demonstrates that belongingness to drug culture may be importantly related to the course and treatment of substance dependence, and that the BDCQ may be a useful tool for clinical work and further study in this area. These results are in line with the growing literature arguing that treatment research must look beyond the physiological processes of addiction and overemphasis on the drug itself, and instead fully consider the context in which drug use occurs (Buchanan, 2004;Keane, 2007). Relatedly, research suggests that individual differences in "recovery capital," the sum of one's resources that can be used to recover from problematic substance use, are predictive of sustained recovery and life satisfaction in substance users (Cloud & Granfield, 2008;Laudet & White, 2008). Belongingness to drug culture may be one "ingredient" of negative recovery capital that may reduce motivation and engagement in treatment and ultimately, treatment success (Cloud & Granfield, 2008). Our initial findings encourage continued psychometric evaluation of the BDCQ as well as evaluation of the role of belongingness in the onset and maintenance of drug use disorders and its potential to be modified in treatment. --- The people I use drugs with care about me. I like when my drug use leads to scary situations. I am glad I do not have a normal life style. I would not fit well into a full time work life. Non drug-users do not understand me. Non drug-users are boring.

Illicit drug use frequently occurs in a context of a drug subculture characterized by social ties with other drug users, feelings of excitement and effectiveness deriving from illicit activities, and alienation from mainstream society. Identification with this subculture is recognized anecdotally as a barrier to recovery, but clear quantification of individual differences in perceived belongingness to the drug subculture has been absent from the literature. The purpose of this study was to describe the development and psychometric properties of a brief self-report measure designed to assess this construct, the Belongingness to Drug Culture Questionnaire (BDCQ). Ninety-six opioid-dependent, methadone-maintained participants completed the BDCQ, related self-report measures, and assessment of drug use patterns. The BDCQ demonstrated high internal consistency (α= 0.88) and was significantly associated with self-reported days of drug use in the past 30 days, desire to quit, impulsivity, psychopathy, and social, enhancement and coping drug use motives. These findings encourage continued psychometric evaluation of the BDCQ and study of the role of belongingness in the development and maintenance of substance use disorders.

Introduction Men who have sex with men (MSM) continue to bear the burden of HIV/AIDS in the United States, representing 57% of total HIV diagnoses in 2009 (1). Between 2006 and 2009, the number of men who were diagnosed with HIV transmitted through same-sex contact increased, while HIV diagnoses due to intravenous drug use and heterosexual contact decreased for both men and women (1). Recent studies of MSM indicate that high rates of new HIV infections may be driven by unprotected anal intercourse (UAI) with primary partners (2,3). For example, in a longitudinal study of young MSM in the Midwest, unprotected sex was nearly eight times more likely to occur in relationships described as "serious" compared to those that were "casual" (2). Moreover, using behavioral surveillance data from five U.S. cities, Sullivan and colleagues estimate that 68[95% CI 58-78]% of cases of HIV transmission among MSM originated from primary partners, attributed to more sex acts, more frequent receptive anal intercourse, and lower levels of condom use with those partners (3). Yet, throughout the epidemic, prevention efforts targeting gay men have primarily focused on the individual, thereby ignoring other important contextual factors such as relationship status and quality, despite previous research that has documented differences in sexual behavior between gay men in relationships and their single counterparts. Thus, investigating factors that hinder protective sexual behaviors for men in relationships may have important implications for future HIV preventive interventions for MSM. An increased likelihood of engaging in UAI with primary partners could be explained partially by relationship dynamics. While an association between positive relationship dynamics and sexual risk behavior may seem paradoxical, evidence suggests that factors such as closeness, investment in and dependence on the relationship, the desire for a stable and lasting relationship, and relationship satisfaction are associated with UAI (4,5). For example, as couples grow closer, taking risks such as engaging in UAI may be seen as part of building trusting, commitment and love (5). These findings highlight that in order to improve the effectiveness of HIV prevention efforts it is necessary to examine not only the individual-level influences on sexual behavior within relationships but also the relationship itself, since the relationship is the context within which UAI frequently occurs. Moreover, transmission of HIV through sexual contact is innately dyadic -that is, it requires the participation of two individuals in a sexual act. There is compelling evidence to indicate the importance of studying HIV risk among gay men at the couple level with data from both partners. However, few research studies employ this approach. Among gay couples, agreements about sex with outside partners are ubiquitous regardless of couple HIV status (6). Sexual agreements serve as a framework for the couples' decision to engage in or refrain from sexual behaviors that may place them at increased risk for HIV. They may include formal or informal agreements, as well as understandings or expectations, about sex with outside partners. Agreements may be closed (i.e., monogamous) or open, permitting a wide range of acceptable situations for sex with outside partners, including sex with a third person when both partners are present, separation of physical and emotional intimacy with outside partners, allowing "safe" sex with outside partners, or non-disclosure of outside sexual activity (6). Given the high rates of HIV infection among men in relationships, agreements about sex with outside partners are a likely prevention strategy for gay couples (3). Relationship dynamics that deeply influence the quality of the relationship may have a bearing on the agreements that couples make and thereby on couples' risk behavior for HIV. For example, couples with open agreements describe high levels of relationship satisfaction because a relationship that permitted outside sexual partners was less oppressive and hetero-normative (6). One study found that there is no difference in reported levels of relationship satisfaction between couples with open and closed agreements, though monogamous couples have higher levels of investment in their sexual agreements, trust, and commitment (7). Moreover, these same dynamics may influence whether these agreements are broken. Broken agreements may be common and can include a wide range of behaviors with varying implications for HIV risk. For example, for a couple with a closed agreement, kissing another man may constitute a break. A break for a couple with an open agreement could include disclosure of outside sexual behaviors to the primary partner when the agreement includes a "don't ask, don't tell" policy or having UAI when the agreement requires condom use for anal sex with outside partners. Higher levels of positive relationship dynamics, such as trust, communication and commitment, may decrease the likelihood of a broken agreement. For example, couples who do not communicate well may differently interpret the terms of an open agreement and thus be more likely to break the agreement. There is an emerging but limited body of literature examining the consequences of broken agreements (6,8). One study notes that in situations where agreements prohibit UAI with outside partners, broken agreements could increase HIV transmission risk for both partners (6). The present analysis examines relationship dynamics as predictors of broken agreements among concordant HIV-negative and HIV-discordant gay couples. We hypothesize that higher levels of positive relationship dynamics will be associated with a lower likelihood of breaks in agreements about sex with outside partners. --- Methods This analysis utilizes data from a longitudinal study of 566 gay couples recruited between 2005 and 2007 in the San Francisco Bay Area. Couples were eligible for the study if their relationship length was 3 months or longer and if both partners were at least 18 years old, fluent in English, had knowledge of their own and their partner's HIV status, identified as gay or bisexual, and were willing to participate. Members of eligible couples individually completed six audio computer-assisted self-interviews (ACASI) over a three-year period. Follow-up interviews were not conducted with couples who broke up between data collection points. The present analysis focuses on the subset of 263 concordant HIV-negative and HIV-discordant couples (n = 526 individuals) who reported having a sexual agreement and completed the baseline interview and the first follow-up interview one year later. The outcome of interest is a binary variable measured at the one-year follow-up interview, reflecting a break in the couple's agreement about sex with outside partners in the past year. All independent variables of interest and control variables are created using the baseline interview data. Primary independent variables of interest are previously validated scales for salient relationship dynamics. Cronbach's alpha statistics are calculated to assess reliability for the set of items included in the scales reflecting various dimensions of relationship dynamics for the study population. Relationship commitment measures the degree to which participants feel they will continue their relationship (8 items, <unk>= 0.91) (9). Social support measures the degree to which respondents and their partners provide each other various dimensions of social support, including attachment, social integration, reassurance of worth, reliable alliance, guidance, and opportunity for nurturance (24 items, <unk>=0.91) (10). Three dimensions of trust are examined (11). Predictability reflects the extent to which respondents feel they can predict their partner's behavior based on past experiences (5 items, <unk>=0.72). Dependability addresses the partner's trustworthiness (5 items, <unk>=0.66). Faith captures an aspect of trust not rooted in past experiences and reflects emotional security in the relationship and their partner (7 items, <unk>=0.84). Additionally, two subscales for communication patterns are examined: Mutual constructive communication around discussion and resolution of problems, expressing emotions, and understanding of views (7 items, <unk>=0.78); and mutual avoidance and withholding captures mutual avoidance of, withdrawing after and withholding after discussion (3 items, <unk>=0.63) (12). Individual scale items are reverse-coded as necessary so that higher scores on the scale indicate higher levels of the characteristic under consideration. Each measure is summed and entered into models as a continuous variable. Relationship dynamics are examined at the couple level in two ways. First, the average of both partners' scores on each factor is entered in models to assess differences between couples in the sample. Second, each partner's deviation from the couplelevel mean is also entered into models to examine differences within couples. Individual-level dichotomous control variables include race, income, employment status, and educational attainment. Dichotomous control variables at the couple level include whether the couple was interracial, self-reported couple HIV status, relationship length, cohabitation, and whether the agreement was open. Study procedures and measures are detailed in-depth elsewhere (7). All analyses are conducted using Stata 11.0 statistical analysis software (StataCorp, College Station, TX). F tests are used to assess statistically significant differences in the prevalence of and reasons for broken agreements by couple HIV status, with associated p values reported. Multivariate logistic regression models are employed to examine the association of relationship dynamics with broken agreements, controlling for individual-and couple-level characteristics. Due to the dyadic nature of the data, survey commands are employed, including adjustments for couple-level clustering and robust standard errors, for both the bivariate and multivariate analyses. --- Results Among the 526 men included in the analysis, 23% report breaking their agreement about outside sexual partners in the previous year. There is not a statistically significant difference in reports of broken agreements by couple HIV status, with 24% of members of concordant negative couples and 23% of members of discordant couples reporting a break (p=0.90). Among discordant couples, there is not a statistically significant difference in the proportion of reported breaks between the HIV-positive and -negative partners, though 27% of HIV-positive partners report breaks compared to 19% of HIV-negative partners (p=0.20). Participants describe a variety of reasons for breaking their agreement, with the most frequently reported causes being: feeling "horny" (96%); "the guy was really hot" (88%); being "weak" (85%); feeling deserving of "an exciting sex life" (78%); and lack of sexual satisfaction (74%). There are some statistically significant differences by couple HIV status with respect to reasons for breaking the agreement. Compared to members of concordant negative couples, members of discordant couples are more likely to report the break occurred because they: did not have to use condoms (41% vs. 20%, p<unk>0.05); did not have to engage in "safer" sex (47% vs. 15%, p<unk>0.001), or wanted to "bottom" (i.e., have receptive anal sex) (59% vs. 39%, p=.052). At the same time, they are less likely to attribute the break to not being able to control their urges (62%) compared to members of concordant negative couples (91%, p<unk>0.001). Among the men who report a broken agreement, 55% did not disclose the break to their primary partner. The most frequently reported reasons for not disclosing the break are: a desire to protect the relationship (89%); fear that their partner would not forgive them (73%); the belief that the break did not put their partner at risk of HIV (73%); and fear of depressing their partner (67%). In multivariate logistic regression analyses, the couple-mean score on each examined relationship dynamic is consistently and significantly associated with breaks in the expected direction (Table I). Members of couples with higher mean scores for commitment, mutual constructive communication, dependability, predictability, faith, and social support within the relationship have a lower odds of reporting a break in their agreement. For example, each 1-point increase in the couple-mean dependability score is associated with a 12% decrease in the odds of a broken agreement (AOR: 0.94, 95% CI: 0.90-0.99), meaning that couples with more dependability have lower odds of their members breaking the agreement relative to couples with less relationship satisfaction. Higher levels of mutual avoidance and withholding are associated with increased odds of a break (AOR 1.07, 95% CI 1.004-1.13), suggesting that couples with a more avoidant communication style have a higher odds of a couple member breaking agreement than do couples with a less avoidant communication style. Within couples, higher levels of predictability are negatively related to breaks (AOR 0.94, 95% CI 0.89-0.999), which means that couples with greater discrepancies in predictability within the couple had lower odds of a broken agreement by one of the men in the relationship. Also, each 1-point increase in the individual's deviation from the couple-mean for mutual avoidance and withholding is associated with an 8% increase in the odds of a broken agreement (AOR: 1.08, 95% CI: 1.01-1.16). This finding indicates that the higher the discrepancy in communication style between the partners, the greater the odds are that a partner breaks the agreement. --- Discussion To our knowledge, this is the first study examining the association of relationship dynamics with breaks in agreements about sex with outside partners among gay couples. We find that breaks in sexual agreements about sex with outside partners occurred in less than a quarter of the couples included in the analysis. Our hypothesis that positive relationship dynamics are protective against breaks in sexual agreements about sex with outside partners is partially supported. While the couple mean of each positive relationship dynamic is protective against broken agreements (i.e., the between-couple effect), each partner's deviation from their couple mean is generally not associated with breaks (i.e., the within-couple effect). For the sample as a whole, the most frequently reported reasons participants provide for breaking their agreements are situational. However, for members of discordant couples, breaks are also attributed to the desire to not use a condom or not practice safer sexual behaviors. More than half of participants who report broken agreements did not disclose the break to their primary partner. The differing causes of broken agreements between members of concordant negative and discordant couples have important implications for HIV prevention efforts. For the sample as a whole, many of the reported reasons for breaking agreements are situational and have a spontaneous quality, since they involve the participant breaking his agreement in the heat of the moment. This finding is worrisome, as it may indicate situations and emotional states that have the potential to sabotage even the best prevention efforts. More information about the context of the break, the agreement that was broken, and with whom (i.e., was the outside partner a regular partner or a partner of unknown HIV status), is desperately needed and may open potential inroads for HIV risk reduction. There are statistically significant differences in reasons for breaking agreements by couple HIV status that may reflect safer sex fatigue or habituation to risk among members of discordant couples. Similar to the above, more information about the context of the break -and, in the instance of discordant couples, the HIV status of the partner implicated in the break -may inform tailored prevention efforts for these men. While condom use and abstinence have been promoted as the most biologically effective methods of preventing HIV transmission for discordant couples, the development of new prevention approaches, such as targeted (in place of blanket) condom use promotion, for this population is crucial, particularly if condoms are not seen as a component of a satisfying sexual relationship. This analysis has a number of strengths, including the usage of longitudinal, dyadic data. In our regression models, all independent variables were measured at baseline, while the broken agreement outcome variable was measured at the one-year follow-up interview. While this analysis does not form the basis for a causal relationship, the availability of longitudinal data ensures that the predictor variables precede the outcome variable. Likewise, the usage of data from both members of each couple allows for a richer exploration of relationship dynamics. Both relationship dynamics and sexual risk behaviors involve inherently dyadic processes; in this case, between two men who are in a committed relationship together. The examination of relationship and sexual risk behavior variables at the couple level acknowledges the interpersonal and relational contexts that must be targeted in HIV preventive interventions for gay couples. Furthermore, these data were collected using ACASI, which may have increased participants' comfort with reporting broken agreements. Finally, our sample was diverse, representing men from various racial and ethnic groups, couples with both open and closed agreements, and concordant negative and discordant couples. The primary limitation of this study lies in the broad measurement of breaks in sexual agreements. Participants were only asked if they had broken their sexual agreement in the past year and not about the specific aspects of the agreement that were broken or the circumstances of the break itself. Thus, we are unable to separate the most risky breaks (e.g., UAI with an outside partner of unknown HIV status) from those that may be important to the relationship but have fewer implications for HIV risk in the near-term. While sexual agreements are extremely common among gay couples, little is known about the implications of breaks in sexual agreements for relationships, HIV risk, and overall mental health and well-being. The current study was not designed to explore broken agreements in-depth, though our future work with gay couples will attempt to address many of these measurement limitations in order to better understand the antecedents, contexts, and consequences of broken agreements. Further limitations of the study include the self-reported nature of the data, including particularly participant HIV status; possible underreporting of broken agreements due to social desirability bias, which may have been intensified because members of couples completed the ACASI surveys simultaneously; and the data are not population-based, which may limit generalizability. While it is not entirely surprising that members of couples with higher levels of positive relationship dynamics are less likely to report broken agreements, there are still important implications of this association for HIV prevention. Reducing the risk of HIV transmission may not serve as primary motivation for sexual agreements among gay couples. Rather, couples may establish agreements to express trust and love or to provide structure to the relationship with boundaries around sex with outside partners (6). Given that sexual agreements may be an important approach for HIV prevention, that gay couples may perceive agreements in the greater context of a healthy and satisfying relationship, and that the majority of couples in this analysis did not report broken agreements, promotion of positive relationship dynamics could also be an important aspect of prevention efforts targeting couples. Encouraging positive relationship dynamics, such as communication and trust, through skills-based interventions may allow couples to form clearer, equitable, and satisfying agreements; communicate dissatisfaction about agreements; minimize the occurrence of broken agreements; improve disclosure of breaks when they happen; and, ultimately, reduce HIV risk.

Agreements about allowing sex with outside partners are common among gay couples, and breaks in these agreements can be indicative of HIV risk. Using longitudinal survey data from both partners in 263 gay couples, we investigate whether relationship dynamics are associated with broken agreements. 23% of respondents report breaking their agreement. Partners who report higher levels of trust, communication, commitment, and social support are significantly less likely to report breaking their agreement. Promoting positive relationship dynamics as part of HIV prevention interventions for gay couples provides the opportunity to minimize the occurrence of broken agreements and, ultimately, reduce HIV risk.

Introduction Everyone values a good night's sleep. Professional nurses and social care staff, who have a strong sense of a duty to care, also need to acknowledge and respect that care home residents need a good night's sleep. Based on an extensive study of sleep undertaken in 10 care homes in South East England, involving 145 old people and 50 members of care staff, this article provides an insight into the importance of sleep in old age and considers the conflict between professional accountability and facilitating good sleep for older people living in care homes. The research methodology, which aimed to identify the determinants of poor sleep in care homes, is described and key findings are presented and discussed. The intention is to inform future developments in service delivery in care-giving environments and thereby to improve the quality of life experienced by older people dependent on care and support from professional caregivers. --- Older people's perspective We all know how difficult the day ahead can be when we have not slept well. Good night time sleep enhances cognitive and physical abilities, reduces the risk of falls, and also enhances quality of life (Livingston et al 1993;Leger 1994;Ersser et al 1999;Busto et al 2001;Martin 2002;Stepanski et al 2003). For older adults these are particularly important aspects of everyday life (Luff et al 2011;Venn and Arber;2011). However, studies undertaken in both North America and the United Kingdom indicate that the care home environment is not conducive to sleep (Alessi and Schnelle 2000;Kerr et al 2008). Sleep forms a significant part of the everyday life experienced by older people living in care homes and may dominate their lives. Their experience of sleep in a care home environment can be expected to differ from that of sleep 'at home' in a bed they may have slept in, with their partner for many years, surrounded by furniture, belongings and sounds they know well. Frailty or a medical condition is likely to have precipitated their move into residential care, in order for their activities of daily living (ADL) to be supported by care staff. These activities range from waking up in the morning and being prepared for the day, to settling down again at night with the reassurance that help is at hand if needed (Eyers 2007) --- Nursing perspective One of the main tasks for care home staff is to assist residents in their preparations for the day by supporting them in the vital tasks of getting up, washed and dressed in order to participate in daily life. While providing this support, care staff are also able to enquire how well the resident has slept and monitor their wellbeing, especially in conjunction with their physical and cognitive abilities on that day. In the evening, residents are supported in their preparations for the night by care staff who also facilitate their sleep hygiene. At night, care staff are on hand to safeguard residents, provide continence care and monitor their sleep. In the course of performing all these activities an ethical awareness of a duty to care is to be expected of both qualified nursing staff and care assistants (NMC 2008;Topss England 2004). Whilst all citizens have a duty of care, Registered Nurses (RN) and Social Care Workers are further bound by their professional code of conduct. Neglecting such a core value is seen as professional misconduct (Topss England 2004;NMC 2008). Care giving at night incorporates ensuring that the residents are safe and sleeping soundly. To successfully complete this task, night staff regularly monitor residents' wellbeing. Night time monitoring is seen to be reassuring to relatives, managers, care staff and regulatory authorities, but it is also important to consider whether it contributes to poor sleep of residents in care homes. Sleep is such a taken-for-granted yet important aspect of everyone's life. In recognition of this, the New Dynamics of Ageing Programme (http://www.newdynamics.group.shef.ac.uk/) funded interdisciplinary research into sleep in later life (SomnIA). One of the eight 'Work Packages' (WP) within the SomnIA (Sleep in Ageing, www.somnia.surrey.ac.uk) project aimed to identify the factors leading to care home residents' poor sleep at night (WP3). A further WP studied the sleep of community-dwelling older people and the findings enabled comparison between sleep at home and sleep in a care home (Meadows et al 2010;Venn et al 2010). This comparison highlighted the contrast between the two groups of older people and paralleled the findings of Alessi and Schnelle (2000) and Kerr et al. (2008) who documented the poor sleep amongst care home residents. The outcome of research undertaken in WP3 also informed two other 'WPs' within SomnIA that researched potential interventions to improve the sleep of older people living in the community and in care homes. In this article the focus is on the findings from the WP aiming to establish the reasons why older people living in care homes experience poor sleep. --- Research Methods Quantitative and qualitative data were collected from 145 male and female residents aged between 60-100 within 10 purposively selected care homes for older people in South East England. In order to ensure a high number of participants able to give informed consent, only care homes registered for 35+ residents were selected. The ownership of the homes covered nursing and non-nursing homes, local authorities, the voluntary sector, large corporate organisations and small businesses (Table 1). Geographically the homes were situated in suburban and rural areas and aimed to cover a spread of socio-economic groups. All the participating facilities met the required National Minimum Standards for Care Homes for Older People (DH 2003) and consequently were considered to fulfill the statutory criteria set by policy makers at the time of data collection. Homes registered to care for 'Elderly Mentally Impaired' residents were excluded from the study as this would have required a different methodology. Ethical approval for the study was given by the University of Surrey Ethics Committee and Local Authority Social Research Governance. The names of participants used in this paper are pseudonyms to preserve confidentiality. --- Table 1 about here Prior to conducting the research, information about the study was provided to the home management, all care home staff, residents and their relatives. On the day that the data collection commenced, a presentation was made to residents and staff in each home. Residents who had volunteered were cognitively able to participate in the study and each had a detailed conversation with the researchers before signing the consent form to participate in the study. Participation involved wearing an actiwatch (small activity and light exposure monitors) to record levels of movement continuously for 14 days and keeping a daily diary over a period of two weeks denoting their everyday activities. As the majority of the 145 participants were not able to personally maintain the diaries, researchers visited daily and chatted with the residents to elicit information about activities during the previous 24 hours. The diaries were maintained in a standardised questionnaire format which facilitated analysis using SPSS. The questions asked daily included how well the participant had slept, how often they had got up during the night and if they were disturbed during the night. An observational study was also conducted, covering dawn, dusk and during the night time in each of the participating care homes, totalling 250 hours of observations. Analysis of the observational data and field notes enabled the researchers to evaluate the validity of the daily information provided in the diaries. Each diary also recorded demographic information and the participants' dependency level based on the amount of time care staff stated was required to provide care each day (Eyers 2000;2007). As not all care staff were qualified nurses this was the most reliable, tested form of data collection. Information about individual participants' continence care plans was also elicited. Audio tape recorded semi-structured interviews were conducted with all 10 care home Managers and a further 50 members of day and night care staff. The structure of the interviews contributed towards identifying organisational factors influencing sleep and the perception that care staff had about residents' sleep. The interviews were fully transcribed and analysed using NVivo 8. To gain a deeper understanding of the work load night staff encountered, additional data was collected from the Managers to establish the care needs of all the residents in each home. --- Figure 1 about here This multi-method approach facilitated 'rigor without rigidity' (Fielding 2009: 428) for this study with vulnerable older people living in a complex community. Bringing together the qualitative and quantitative data facilitated data analysis that provided a deeper understanding of care home life and the interaction between older people and the staff employed to support them (Figure 1). The findings aim to inform policy and practice development as well as the other WPs within the SomnIA project. To integrate a practitioners' perspective into our analysis, two well attended 'Day Workshops' were held, which enabled discussion about the findings with care home managers and night staff. --- Quantitative Data Findings Diary data was collected from 145 participants who completed all or part of the two week daily diaries. There were 105 women and 40 men, covering an age range of 60 to 100 with the strongest representation of both genders in the age group 85-89 (Figure 2). --- Figure 2 about here Information about the participant's health care needs was provided by care staff and indicated that 27 needed under 90 minutes physical assistance per day whilst 79 received 90-180 minutes of physical care and support per day. Three to five hours of 'hands-on' care was provided to 30 residents, and 6 participants received over 5 hours of daily assistance to meet their health care needs. Data relating to dependency was missing from 3 participants. Background data collected about the care homes provided by care home matron/managers indicated that assisting and supporting residents to use the toilet was, alongside assistance in getting washed and dressed, one of the main forms of support provided to residents by care staff. From the observational data, it was clear that assistance with toileting and continence care is provided day and night. The data strongly indicated that such a sensitive, very personal bodily function that a healthy independent person undertakes in privacy, is a core element in the everyday (and night) life of vulnerable old people living in care homes. A key finding from the diary data was that the mean length of time care home residents spent in bed at night was 10 hours 50 minutes. However it would be erroneous to equate the hours in bed with long hours slept, as the majority of the participating residents reported spending a number of hours awake in bed each night (Luff et al 2011). --- Figure 3 about here The actigraphy (data from actiwatches) analysis established that many participating residents experienced fragmented sleep. Figure 3 shows an example of the actigraphic record for 48 hours of a typical care home resident with fragmented sleep. The spikes represent movement, and therefore when the resident is awake. Only where there is a flat line on the horizontal axis is the resident likely to be asleep. Figure 3 indicates that this residents' pattern of movement is little different during the day and the night-time on the Saturday, in particular there are peaks of movement at two hourly intervals during the night, probably reflecting routine checks by care staff. On the Sunday, there appear to be periods of sleep from 2.00am to 5.30am, and also in the afternoon from 1.30pm to 4.00pm. The observational data and interviews provide an insight into potential reasons behind residents' disturbed sleep. However, an indication comes from the diary data which showed that 65% of sleep disturbances were reported by residents to be related to toilet/continence care. This factor links with the qualitative findings; both the interviews and observations indicated that continence care was a dominant aspect of care home life. --- Qualitative Data Findings Based on grounded theory principles (Strauss and Corbin 1998), a key theme identified within the qualitative observational and interview data was the frequency of night-time monitoring by staff, which was driven by two requirements: Dead or alive? Reliably monitored by entering the bedroom to visually assess residents' wellbeing (are they still breathing?) Wet or dry? Reliably monitored through the physical evaluation of the continence pad or sheet. Not heeding these aspects of care giving could be interpreted to mean care staff are failing in their duty of care. Regular hourly checks were expected by managers and routinely performed by care staff, which was evident from the interview, observational and actigraphy data. Based on the interview data collected and discussions during the Day Workshops attended by Care Home Managers, it was clear that night staff generally do not intentionally wake residents at night. Interviewed care staff also tended to believe that 'their residents' sleep well. They made this evaluation on the basis of the following three key points: • 'their residents' did not take'sleeping tablets' • 'their residents' did not use the buzzer during the night • the sound of 'their residents'' snoring A care home manager said: 'They [residents] are checked hourly [during the night]... unless residents specifically request that we do not go into the room... they are checked hourly...'. Madge, a care assistant said: '... Checking on them... that's our work, we have to look after them'. However, to counterbalance the above two interview extracts thought needs to be given to the following statement made by a resident, Stanley: '... When the staff come into my room, I wake up'. Considering that a high proportion of the participants experienced fragmented sleep, it becomes clear that there is a potential area of conflict between care staff facilitating a good night's sleep and fulfilling a duty of care. Regular surveillance of night time sleep by care staff restricts residents' experience of restorative sleep, which could enhance their cognitive and physical abilities during the day. --- Conclusion This comparatively large scale study of care home life in England and the mixed method research approach aimed to ensure the collection of ethical, reliable and valid data in a complex environment involving vulnerable people. This study set out to identify determinants of poor sleep experienced by older people living in care homes. The findings contribute to understanding the reasons for poor sleep identified in other studies (Alessi and Schnelle 2000;Kerr et al 2008). On the basis of the evidence from this research it is important to address the issue of how care at night is provided so that it does not disrupt residents' sleep, while night staff still fulfill their duty of care. However, a limitation of this study is the lack of detailed enquiry into the actual understandings that care home staff have of a 'duty of care'. There is a need to consider what can be done to ensure the development of best practice that enables care home residents to sleep well while also reassuring care home staff, managers and relatives that, at night, all residents are safely asleep in dry beds. It is equally important to value the positive impact of good sleep on a resident's physical and cognitive abilities during the day. A good night's sleep for residents could improve both residents' and care staffs' day time experiences. Facilitating a good nights sleep furthermore exemplifies the effective implementation of person-centred care (Innes et al 2006). Further research is recommended to support and encourage the use of readily available night time sensors and to comprehensively review routine daytime and night-time care giving procedures to support the implementation of person centred care that enables restorative sleep. A key research goal should be to develop procedures by which care staff can fulfill their 'duty of care' for the well-being of residents at night, but without disturbing resident's sleep. Further research is called for to gain a deeper understanding of the links between policy and practice relevant to professional Codes of Practice established to provide night-time guidance within a wide spectrum of health and social care settings. Policy makers, care service regulators, managers, nurses and care assistants jointly have a duty of care to address this vital aspect of care home life. To succeed, this would require the regulators' support, and for the importance of sleep to be explicitly included in care staff training programs. It could well be a hard day's night to ensure care home residents can sleep like a log.

has been involved in the care home sector for 30 years. As a Registered General Nurse she initially worked in intensive care and then transferred to the care of older people in the community. For the last 12 years Ingrid has been undertaking research in care homes. Within the SomnIA (www.somnia.surrey.ac.uk) project her team undertook an in-depth data collection in 10 care homes in order to identify the determinants of poor sleep experienced by care home residents. Ingrid

INTRODUCTION The sex ratio at birth, or the secondary sex ratio (SSR), is defined as the ratio of male to female live births. Not only has the SSR served as a population health indicator (e.g., for fertility), but it has also played an important role as a social indicator pISSN 1975-8375 eISSN 2233-4521 (e.g., for son preference, gender equity, and couples' childbearing behavior) in many countries around the world, notably in Asia [1][2][3]. Korea is an Asian country that has undergone drastic changes in the SSR in the last few decades [4]. In Korea, the SSR began to increase abruptly in the mid-1980s, along with the use of ultrasound for prenatal sex determination [2,3,5]. The SSR then peaked in 1990, with a severely male-biased SSR of 1.165. Since the mid-1990s, the SSR has gradually decreased, reaching an average natural SSR of 1.050 in 2016. The stability and variability of the SSR have been proposed to be influenced by a variety of endogenous and exogenous factors. It has been suggested that several biological factors, such as parental hormone levels around the time of conception [6][7][8], follicular phase length [9,10], timing of conception during the menstrual cycle [7,9], and sperm characteristics [11], may play a role in the natural SSR. In addition, the SSR observed at the population level has been reported to vary according to socio-demographic factors such as parental age [12][13][14][15], birth order [12][13][14]16,17], race/ethnicity [12,14,18], and socioeconomic status (SES) [5,12,15,[19][20][21][22]. Despite the recent dramatic changes in the SSR observed in Korea, sparse previous research using nationally representative data has investigated the associations of socio-demographic factors with long-term temporal trends in the SSR. A study using national vital statistics data from 1981 to 2004 [5] assessed parental SES, as measured by education and occupation, in relation to the SSR, and suggested that maternal SES may be a more sensitive indicator of variations in the SSR than paternal SES. Of note, the aforementioned study reported that there had been changes in the association between parental SES and the SSR across three equidistant time periods (1981-1984, 1991-1994, and 2001-2004). Namely, before the mid-1980s, higher parental SES was associated with an increased SSR (i.e., an excess of male births). Since then, however, an inverse association between parental SES and the SSR was prominently noted in the early 1990s. With a more extensive evaluation of socio-demographic factors (e.g., parental age, education, occupation, and birth order) in relation to the SSR, the present study aimed to investigate the impact of socio-demographic factors on the decreasing trend in the SSR from 1997 to 2017 in Korea. --- METHODS --- Research Data Data from all live births registered with Statistics Korea from 1997 to 2017 were obtained from the Microdata Integrated Service (MDIS) [4]. With no missing values for infant sex, a total of 10 349 602 live births were included in the final dataset. Nonetheless, missing values for independent variables (birth order, n=44 --- Statistical Analysis All live births, the proportions of male and female births and the SSR (among all live births and according to birth order [first, second, and third and higher-order births]) during the study period were calculated. The linear-by-linear association test was used to test the significance of trends in the proportions of male and female births during the study period. To evaluate changes in the SSR and its associations with sociodemographic factors over time, the study period was divided into 7 phases: 1997-1999 (phase I), 2000-2002 (phase II), 2003-2005 (phase III), 2006-2008 (phase IV), 2009-2011 (phase V), 2012-2014 (phase VI), and 2015-2017 (phase VII). Logistic regression models were used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for a male birth, after adjusting for parental age (<unk>19, 20-24, 25-29, 30-34, 35-39, and <unk>40 years), education (primary, secondary, tertiary), occupation (non-manual, manual, and others [students, housewives, etc.]), and/or birth order (first, second, and third or higher). As birth order has been recognized as a factor responsible for imbalances in the SSR, a subgroup analysis was conducted according to birth order. A 2-sided significance level of 0.05 was used to assess statistical significance. Statistical analysis was performed using SPSS version 25 (IBM Corp., Armonk, NY, USA). --- Ethics Statement As all of the data used were de-identified, this study was classified as exempt research by the Institutional Review Board of Daegu Catholic University Medical Center. --- RESULTS Table 1 shows all live births, the proportions of male and female births, and the SSR from 1997 to 2017 in Korea. During the study period, the highest SSR (1.10) was observed in the late 1990s and the early 2000s. The SSR gradually decreased from 1.10 (in 1998-2000 and 2002) to 1.05 (in 2013-2016). When the SSR was calculated according to birth order, the decreasing trend in the SSR was not noticeable among first births. Among third and higher-order births, however, malebiased sex ratios were prominent in the late 1990s and the early 2000s (with the highest SSR being 1.46 in 1998), demonstrating a remarkable decrease in the SSR during the study period (Table 1). The number of all live births and the SSR according to sociodemographic factors in phases I-VII are presented in Table 2. The proportion of mothers aged <unk>35 years increased from 5.9% in phase I to 26.3% in phase VII; and that of fathers aged Socio-demographic Factors and the SSR <unk>35 years increased from 17.8% in phase I to 47.2% in phase VII. The distributions of parental education and occupation changed markedly during the same period. For instance, the proportion of college-educated mothers increased from 32.0% to 77.2%, and that of mothers with non-manual jobs increased from 10.1% to 32.4% (Table 2). Table 3 presents the ORs and 95% CIs for a male birth according to socio-demographic factors among all live births. Higher birth order was significantly associated with an excess of male births in phases I-VI (<unk>third vs. first, OR range, 1.03 to 1.35), but not in phase VII. Adolescent mothers had higher odds of a male birth than did mothers aged 20-24 years in phase I (<unk>19 vs. 20-24 years, OR, 1.04; 95% CI, 1.00 to 1.07; p<unk>0.05). However, older mothers had lower odds of a male birth than did mothers aged 20-24 years in phases II, III, and V (<unk>40 vs. 20-24 years, OR range, 0.92 to 0.95). Equivocal findings on the association of maternal SES with the SSR were noted. Namely, less-educated mothers had lower odds of a male birth than did collegeeducated mothers in phase II (primary vs. tertiary, OR, 0.94; 95% CI, 0.90 to 0.98). Meanwhile, compared with mothers with non-manual jobs, those with manual jobs (OR, 1.03; 95% CI, 1.01 to 1.05) and others (OR, 1.01; 95% CI, 1.00 to 1.02; p<unk>0.05) had higher odds of a male birth in phase III. The associations between socio-demographic factors and the SSR among first births, second births, and third and higher-order births are shown in Tables 4 and5, Supplemental Material 1. Comparable findings on the association between maternal age and the SSR were noted among first births (in phase I, <unk>19 vs. 20-24 years, OR, 1.04; 95% CI, 1.01 to 1.08; in phases II, III, and V, <unk>40 vs. 20-24 years, OR range, 0.87 to 0.93) relative to the corresponding findings among all live births. Paternal age was also found to be significantly associated with the SSR among first births, indicative of similar opposing directions toward infant sex at both age extremes (i.e., <unk>19 and <unk>40 years) (Table 4). However, when the analysis was restricted to third and higher-order births, the findings on the associations between parental age and the SSR were in contrast with the corresponding findings among all live births or first births, indicating that advanced parental age was associated with an excess of male births in phase I (Table 5). In addition, among third and higher-order births, fathers with low SES, as measured by education and occupation, had lower odds of a male birth than did those with high SES, particularly in some early phases (phases I-III) (Table 5). --- DISCUSSION The present study used data from the national birth registry to evaluate the effects of selected socio-demographic factors on the decreasing trend in the SSR from 1997 to 2017 in Korea. This study revealed that higher birth order was a strong predictor of an excess of male births in 1997-2014, but not in 2015-2017 (Table 3), possibly reflecting the reduction of an artificially-biased SSR among third and higher-order births over time [3]. In fact, the analysis according to birth order showed that changes in the SSR during the study period were not prominent among first births, but were noteworthy among third and higher-order births. Imbalances in the SSR observed in some Asian countries, such as China and India, are thought to be attributed to son preference and prenatal sex selection against females [2,3,5], which are social trends considered to be influenced by economics and cultural transmission [23,24]. As Korea has undergone drastic economic development and cultural changes during the last few decades, changes in sex selection practices, along with the implementation of policies against these practices, may have contributed to the rapid normalization of the artificially-biased SSR in Korea [24][25][26]. In addition, consistent with some-but not all-previous studies [12][13][14][15], the present study revealed that maternal age was significantly associated with differences in the SSR among all live births. Specifically, this study suggests that advanced maternal age (<unk>40 years) may be associated with a lower SSR, whereas young maternal age (<unk>19 years) may be associated with a higher SSR (Table 3). Similarly, in a study using data from the national birth registry from 1940 to 2002 in the United States, the 2 oldest maternal age groups (40-44 and <unk>45 years) had the lowest SSR (both 1.04) [14]. Another study using United States linked birth-infant death data from 1983 to 2001 found that mothers aged 15-19 years were more likely to give birth to sons than mothers aged 20-34 years, whereas mothers aged <unk>35 years were more likely to give birth to daughters Socio-demographic Factors and the SSR Table 5. Odds ratios (ORs) and 95% confidence intervals (CIs) for a male birth according to socio-demographic factors among third and higher-order births, phases I-VII Logistic regression models were used to estimate ORs and 95% CIs for a male birth, after adjusting for parental age, education, and occupation. *p<unk>0.05. [15]. Contrary to the present study, however, several studies, particularly those conducted in Western countries, have suggested that higher birth order may be associated with a decreased SSR, given that older age at childbearing tends to be associated with higher birth order [12,14]. For instance, the aforementioned study conducted in the United States showed that seventh and eighth and higher-order births had the lowest SSR (both 1.03) [14]. Of note, the present study revealed that the association of parental age with the SSR differed by birth order. This may have resulted from a mixture of the biological and artificial effects of this factor on the SSR (i.e., a mixture of naturally-varied and artificially-biased SSR), which may differ by birth order. The SSR among first births, which appeared to be within the natural range of variability, was found to be significantly associated with parental age at both extremes (<unk>19 and <unk>40 years), reflecting the potential biological impact of this factor on infant sex (Table 4). Meanwhile, the SSR for third and higher-order births, which appeared to be severely male-biased, particularly in the late 1990s (phase I), showed opposite results in terms of its association with parental age compared with the corresponding results among first births (Table 5). Among third and higher-order births in the late 1990s, as parental age increased, the odds of a male birth increased, implying possible birth cohort effects on the magnitude of son preference and couples' childbearing behavior [3,24,27]. Uncertain associations of parental SES with the SSR were noted among all live births and in the subgroup analysis according to birth order (Tables 345, and Supplemental Material 1). In comparison with a previous study using national vital statistics data from 1981 to 2004 [5], the present study did not clearly show that maternal SES was a more sensitive indicator of variation in the SSR than paternal SES during the study period. Instead, among third and higher-order births, low-SES fathers (SSR range, 1.21 to 1.32; data not shown) were found to have lower odds of a male birth than high-SES fathers (SSR range 1.37 to 1.49; data not shown), particularly in the late 1990s and early 2000s (phases I-III) (Table 5). This finding may be interpreted as reflecting the artificial effect of SES on the SSR, as high-SES fathers may have better access to sex selection techniques than low-SES fathers having a third and higher-order child. In terms of the biological effect of SES on the SSR, the Trivers-Willard (TW) hypothesis, a long-established hypothesis regarding the SSR, theorizes that parents in good condition tend to produce more sons, whereas parents in poor condition tend to produce more daughters from the perspective of evolutionary biology [28]. According to the TW hypothesis, high-SES parents are more likely to have sons, while low-SES parents are more likely to have daughters, allegedly due to higher male mortality during the gestational period [12,19,20,22]. However, the existing evidence on the TW hypothesis is controversial, with inconsistent findings on the association between parental SES and the SSR [12,15,[19][20][21][22]. For instance, in a study conducted in Sweden, more than 3 000 000 births between 1960 and 2007 were analyzed to examine various dimensions of parental SES (i.e., earnings, post-transfer income, wealth, parental wealth, education, and occupation) in relation to the SSR, and no substantive relationship was found between parental SES and the SSR [21]. In another study using Swedish national data for the years 1862 through 1991 [20], in contrast, economic contractions were found to be significantly associated with alterations in the SSR at the population level, implying a potential role of ambient social stressors such as economic stress in affecting the SSR [19,20,22,29]. Nonetheless, the biological mechanisms by which parental socio-demographic factors might influence the SSR are ob-scure. Another prevailing hypothesis regarding the SSR is the hormonal hypothesis proposed by James [6][7][8], who postulated that parental hormone levels around the time of conception are potential determinants of offspring sex. According to this hypothesis, high levels of estrogen (in the mother) and testosterone (in either parent) are associated with an excess of male births, whereas high levels of follicle-stimulating hormone and luteinizing hormone are associated with an excess of female births [6,7]. In light of age-related hormonal changes in humans, parental age may function as a biological determinant of the SSR. While the hormonal hypothesis applies to the determination of sex ratio at conception or the primary sex ratio, the economic stress hypothesis proposed by Catalano applies to subsequent adjustments to the primary sex ratio via sex-selective fetal loss [8,19,20]. Namely, since maternal economic stress around the time of conception causes increased testosterone secretion from the adrenal glands, it may increase the likelihood of conceiving a male according to the hormonal hypothesis. If maternal economic stress persists during the gestational period, however, it may increase the risk of spontaneous abortion, which disproportionately affects male embryos and fetuses according to the economic stress hypothesis. As such, prolonged economic stress experienced by mothers may compensate for the male-biased primary sex ratio, or could even result in an excess of female births [29]. The major limitations of the present study should be taken into account when interpreting the observed results. Although the present study utilized national birth registry data that covered nearly all births in the nation, the information regarding the parents, such as age and SES, may have been inaccurate. As for the information on infants, the SSR is not considered to be affected by recall bias. With sole dependence on information included in the national birth registry, various dimensions of parental SES (e.g., earnings, wealth, etc.) were not evaluated. For the same reason, further investigations of other potential determinants of temporal trend in the SSR (e.g., environmental factors such as exposure to endocrine disruptors [30]) were unfeasible. Normative changes related to the observed trend in the SSR were not quantitatively assessed, meaning that this study did not address how much normative changes have contributed to the decreasing trend in the SSR during the study period. Furthermore, although the study results were interpreted as reflecting a mixture of the biological and artificial effects of factors affecting the SSR, the SSR may not be an accurate indicator for evaluating changes in sex selec-Socio-demographic Factors and the SSR tion practices and policies targeting these practices within a population. As recently proposed, the sex selection propensity, which incorporates other relevant factors (e.g., family size and birth order) together with the SSR, may serve as a more accurate indicator of these changes [1]. In conclusion, the present study demonstrated that advanced maternal age and the reduction of the artificially-biased SSR among third and higher-order births may partially explain the decreasing trend in the SSR from 1997 to 2017 in Korea. The artificial effect of birth order on the SSR appeared to exceed the biological effect of birth order on the SSR from 1997 to 2014, but this was no longer the case starting in the mid-2010s (2015-2017; phase VII). The rapid normalization of the artificially-biased SSR in Korea in the last few decades is perceived to be a desirable phenomenon, with favorable social implications regarding gender equity and couples' childbearing behavior. Along with various socio-demographic factors, potential environmental factors affecting the SSR at the population level need more research attention to ensure the stability of the SSR within its natural range of variability. --- SUPPLEMENTAL MATERIALS Supplemental material is available at https://doi.org/10.3961/ jpmph.20.282. --- CONFLICT OF INTEREST The author has no conflicts of interest associated with the material presented in this paper. --- AUTHOR CONTRIBUTIONS All work was done by JB. --- ORCID Jisuk Bae https://orcid.org/0000-0002-5809-5394

This study investigated the effects of socio-demographic factors on the decreasing trend in the sex ratio at birth from 1997 to 2017 in Korea. Methods: Data from 10 349 602 live births registered with Statistics Korea from 1997 to 2017 were analyzed. The secondary sex ratio (SSR), defined as the ratio of male to female live births, during the study period

MAXIMIZING COLLECTIVE IMPACT ACROSS THE AGE-FRIENDLY ECOSYSTEM Chair: Holly Dabelko-Schoeny Co-Chair: Teri Kennedy Age-friendly initiatives have the potential to transform communities through inclusive policies and practices that meet the needs of people, families, and communities across ages and abilities. The ecosystem of age-friendly initiatives (communities, universities, public health systems, and health systems) has grown exponentially over the last decade. Despite shared values and principles, each age-friendly initiative largely operates independently with limited opportunities to share knowledge to maximize collective impact and social change. This theoretically-driven symposium provides an overview of each age-friendly initiative, challenges faced, and ideas for innovation within and across each initiative. Dabelko-Schoeny and Sheldon will utilize a macro application of the KAP (knowledge, attitude, practice) theory of behavior change to identify the vast opportunity for synergies and grand challenges across the age-friendly ecosystem. Greenfield will provide an overview of the development of the Age-Friendly Cities and Community Movement, and present ideas regarding challenges and opportunities for the future based on insights from an ongoing, community-engaged study in New Jersey. Montepare will describe the Age-Friendly University initiative and how age-inclusive practices in higher education can extend and enhance ecosystem efforts. Phillips and Wolfe will provide examples of Age-Friendly Public Health Systems including a description of a collective impact approach for state-wide age-friendly policy and practice adoption. Finally, Kennedy will offer approaches and models to advance the adoption and sustainability of age-friendly principles, programs, and policies by leveraging synergies and implementation science. This is an Age-Inclusivity in Higher Education Interest Group Sponsored Symposium. Frameworks to make communities, universities, public health systems, and health services more inclusive and equitable for older adults and people with disabilities have expanded over the last decade. Centering the voices of older adults, these "age-friendly" initiatives have sought social change through building knowledge, changing attitudes, and developing new practices and policies in a variety of settings. Despite the similarities in values and principles across these initiatives, these models of social change have largely operated independently from one another with unique administrative structures, networks, and resources. This presentation will utilize a macro application of the KAP (knowledge, attitude, practice) theory of behavior change to identify the vast opportunity for synergies across the age-friendly ecosystem, strengthening the potential for collective impact and lasting change. Often these initiatives have overlapping and intersecting goals and would benefit from intentional alignment across the initiatives. Modeled after the National Academy of Engineering Grand Challenges and the Grand Challenges for Social Work, we propose the Grand Challenges for the Age-Friendly Ecosystem, areas in need of action and support, for transformational change in our social, educational, service, and built environments to better meet the needs of our aging society. Abstract citation ID: igad104.0333 --- ADVANCING ADOPTION AND SUSTAINABILITY OF AGE-FRIENDLY INITIATIVES THROUGH SYNERGIES AND IMPLEMENTATION SCIENCE Teri Kennedy, University of Kansas Medical Center, Kansas City, Kansas, United States For communities interested in adopting age-friendly principles, the plethora of age-friendly initiatives (i.e., age-, disability-, and dementia-friendly cities and communities; age-friendly universities, health systems, and public health systems) can be overwhelming. For rural, frontier, underserved, and under-resourced communities, it can be challenging to consider tackling even one such initiative. A major barrier to implementation is the know-do gap between what is known and what gets accomplished in practice. This gap is widened as a result of overly complex models and practices that are difficult to disseminate and scale, difficult to reproduce, or do not apply across settings. The know-do gap can

is unknown whether these immune-related aging processes are associated with neurodegeneration, tauopathy, amyloid accumulation, and/or vascular issues in the brain. To address this gap, we utilized a set of recently assayed highly innovative markers of neurodegeneration, biomarkers of inflammation, and proportions of naïve lymphocytes (indicative of less immune aging), and cognitive dysfunction (errors on the Telephone Interview for Cognitive Status) in the Health and Retirement Study, a nationally representative sample of US adults over age 50 (N = 3985). Using structural equation modeling (SEM), we estimated latent factors representing neurodegeneration (neurofilament light chain (NfL), glial fibrillary acidic protein (GFAP), and phosphotau 181 (pTau181)), inflammation (c-reactive protein, IL10, IL1 receptor agonist, IL6, TNF receptor 1, albumin, and percentage of neutrophils), and immune aging (percentages of naïve CD8+ and CD4+ T cells). Results indicate inflammation is associated with neurodegenerative biomarkers (β = .31, p < .001) and with cognitive dysfunction (β = .08, p < .001) after controlling for age, race, gender, education, BMI, smoking, drinking, and immunosenescence. Neurodegenerative biomarkers mediated about half of the association between inflammation and cognitive dysfunction. It appears that greater systemic inflammation may be associated with brain damage and tauopathy. Interventions focused on reducing systemic inflammation may be successful in reducing neurodegeneration and cognitive dysfunction in older adults.

Introduction Australia is made up of distinct groups of Aboriginal and Torres Strait Islander peoples, each group has their own culture, language, beliefs and practices. Aboriginal and Torres Strait Islander peoples are the first peoples of Australia, and there is evidence of their existence in Australia many thousands of years prior to European colonisation. They are acknowledged and respected as the Traditional Custodians of the Land. The authors of this manuscript recognise the two distinctive First Nations populations of Australia; Aboriginal and Torres Strait Islander people as being the First People of Australia. This study has taken place across Australia on First Nations land, the term First Nations people has been used when referring to Aboriginal and Torres Strait Islander people of Australia in this manuscript. First Nations health workers are employed in tertiary health settings around Australia to improve cultural safety and enhance interactions between First Nations people and the Australian healthcare system [1]. Studies demonstrate improved health outcomes and communication when FNHWs are involved in care in tertiary healthcare settings [2,3]. Other studies also show that involvement of FNHWs in care has led to improved access to aftercare for children with chronic conditions [4,5]. Burns can be a devastating injury causing long term physical and emotional trauma [6,7]. Australia's First Nations children are admitted to hospitals as a result of a burn injury twice as often as other Australian children and have longer lengths of stay [8]. Serious burn injuries need specialised acute tertiary care and often require multidisciplinary aftercare for extended periods of time [6]. Given the reported benefits of their involvement in other care contexts [9], it is clear that FNHWs are important facilitators to the continuity of burn aftercare once a First Nations child has left a tertiary healthcare setting. Burn care is informed by Western biomedical concepts of health [10], despite First Nations family needs for best quality healthcare. Racism towards First Nations people in Australia has been a reality since colonisation and brings with it a power imbalance that is entrenched in our health systems and leads to poor health outcomes. Quality healthcare is such that it incorporates First Nations concepts of health and healing and delivery of care by FNHWs [11]. However, how FNHWs work in the healthcare system at the interface of Western biomedical care and First Nations concepts of health and healing in relation to childhood burns is unclear. To our knowledge, this is the first study to describe the role and function of FNHWs in multidisciplinary burn care for Australia's First Nations children. The desired, actual and perceived role of FNHWs is described by families and the FNHW themselves, along with data from the perspectives of multidisciplinary burn team members. --- Methodology and Methods --- The Coolamon Study This work comprises a substudy of the Coolamon Study, which examined a range of factors related to care and impact of burns in Aboriginal and Torres Strait Islander children who sustained a serious burn. As part of this study, Australia's First Nations children under the age of 16 years who had sustained a burn injury and present to a tertiary pediatrics burn unit were recruited from New South Wales, Queensland, South Australia and Northern Territory. Data collected from this study included sociodemographic information, out of pocket costs, functional outcome and measures of pain, itch and scarring. The study also explored barriers and facilitators of burn after care for the family. Health-related quality of life was measured using the PedsQL, and impact of injury using the family impact scale. Clinical data and treatment were also be recorded [12]. Within this larger study two sub-studies were conducted by JC and SF undertaking their PhDs [12]. The Coolamon Study [12] comprises four sub-studies. This paper reports on data from two sub-studies led by authors JC and SF to better understand the role and contribution to care by FNHWs in multidisciplinary burn care for First Nations children, data from the two sub-studies were combined. Data included incorporates the perspective of families on the perceived and desired contribution to burn aftercare by FNHWs from sub-study one. Data from stub-study two includes the descriptions of FNHWs' involvement in multidisciplinary burn care from the perspectives of the workers themselves and of multidisciplinary burns team. The combining of the two data sets is essential in facilitating a comprehensive understanding of the role of FNHWs from multiple perspectives and lived experiences regarding burn care for First Nations children and their families. Approval was granted by ethics committees in each state research was undertaken, as well as all the relevant Aboriginal health ethics departments. --- Analysis Approach --- First Sub-Study The first set of data was collected by the first author J.C., an Australian First Nations researcher, using indigenous research methodologies. The first author, J.C.'s worldview as a Murri woman impacted her standpoint in this sub-study [13,14]. J.C. sought to understand the barriers and/or facilitators in accessing burn aftercare using indigenous methods of yarning [15] and Dadirri [16]. A total of 59 individuals from 18 different families were asked to share their burn care journeys. Families were recruited purposively from the larger national study examining burn care in Australia's First Nations Children, the Coolamon Study [12]. Families were selected to ensure diversity of experience and access to burn aftercare. Families resided in communities across Australia and the Torres Strait Islands. These stories were audio recorded and transcribed verbatim. Families received a copy of their transcript and a follow up phone call was made to ensure the family's stories was accurately reflected. Data were analysed using a cyclical process which gave ownership of the story to the storyteller and empowered the voice of each family [14]. --- Second Sub-Study Interface research methodology [17] incorporating both Indigenous and Western biomedical knowledges guided the second sub-study for author SF, a non-Indigenous researcher. Sub-study two sought to explore how burns care is delivered, with a focus on the care of First Nations children and families. It also investigated factors informing burn care and explored how clinicians in burn teams use guidance documents and if such documents are appropriate for care of First Nations children. In the second sub-study, and using a semi-structure interview guide, author SF interviewed 76 healthcare professionals from six different multidisciplinary burn teams across five jurisdictions in Australia. Of the 76 participants interviewed, 11 were employed in specific First Nations health worker roles. These included one Aboriginal health practitioner and 10 Aboriginal or Indigenous liaison officers. Interviews were also audio recorded, transcribed verbatim and participants confirmed the transcripts were accurate and true. Data that discussed the specific role, either perceived or actual, of FNHWs have been included here. All other data was excluded for this analysis. Individually authors J.C. and S.F. thematically analysed their own data and then came together to consolidate their findings and discussed over all themes with the third author TM who has expertise in Indigenous methods and from there reached consensus on the key findings and research themes.., The coming together by the two researchers and their respective methodologies, provides an opportunity to demonstrate how research at the interface might deliver outcomes that marry Indigenous ways of doing and Western biomedical care. This is not unlike Ganma, described by the Yolgnu people as respectful two-way sharing of cultural knowledge and interaction between Aboriginal and non-Aboriginal people [14]. J.C., S.F. and T.M. met to discuss the data and existing themes, grouped the data (as per result below) and then synthesised all information specifically into three categories: (i) First Nations children's and families' perspectives (ii) First Nations health workers' perspectives and (iii) multidisciplinary burn team member's perspectives. --- Results Results from both sub-studies highlighted the role of FNHWs in delivery of burn care, from a patient and family perspective (study 1) and from a clinician perspective (study 2). The data have been arranged into three sections-the FNHWs' perspectives of care, the multidisciplinary burns team perspectives of the role, and the First Nations families' perspectives of the FNHWs' role. Quotes have been used throughout to illustrate the themes that were synthesised from the data and pseudonyms used to protect identities where necessary. --- First Nations Children's and Families' Perspectives Remote participants needing multidisciplinary teams for ongoing burn aftercare described the importance of the FNHWs in supporting children and families accessing burn aftercare. Families recounted how FNHWs provided access to essential tangible support and a culturally safe environment. Families also identified challenges related to the role of FNHWs and their availability to be involved in their child's care (Table 1). Several sub-themes were identified within these aspects of care. FNHWs were able to assist families with tangible support in transporting the child and family to appointments, with short-and long-term accommodation, food and taxi vouchers and filling in forms for the patient assistance transport scheme. "He needs to still see the physio. No, I don't have a car, we either catch the bus to the hospital or the health girls [FNHW] came and picked us up." "She [FNHW] took off my PATS [Patient Assistance Transport Scheme] form and faxed it off to me". "The [First Nations] health workers, they brought me up a few times for our appointments, I think a couple of times." "We didn't have anywhere to stay but she helped us with a place while he had his dressing done... every few days we'd go back to the hospital with taxi vouchers she gave us." --- Cultural Support Families voiced how dislocation from home and community was understood and valued by the FNHW. Due to FNHWs' advocacy between the multidisciplinary team, health services and the family, a connection was created between First Nations families and FNHWs. "I ended up giving her a gift after she [FNHW] done what she did, she supported me a lot and kept an eye on him and, so, I done her a nice painting she took home. She loved it." "Linda liked her [FNHW] she was happy when she saw her the next time we went for dressing 'cause we trusted her." "She [FNHW] understood where we come from and how different it is down here they [multidisciplinary team] don't understand." "We had to move from our community and no family is here but yeah, she [FNHW] helped me with like food vouchers, and sit down and have a cup of coffee and a yarn at the hospital house they sent us to so Maison could have dressings." --- Challenges and Fears Data showed that some children and families were frustrated that FNHWs were not available when they needed them. This was often because they were either busy with other patients or there wasn't a FNHW employed at the service. When FNHWs were available to provide support, family challenges and fears were alleviated. Although some families did not receive support from a FNHW it was express that they would have liked to have the support. "No, but I wish I did see one [FNHW] then she could have been with me when mum couldn't." (child) "I don't think I seen anyone else-any Aboriginal worker they said they had one but was sick." (mother) "Yes I did see an Aboriginal worker... Once for about 20 min. She was going to come back but she didn't come back. Obviously it was flat out." (mother) "It was so good having her [FNHW] there when they were talking them big words she would tell us what they meant." (mother) "The Aboriginal [liaison] worker helped get my family together down here... going to be here for months in Ronald Macdonald house... we were all living in separate places... so hard on Damien's father." (mother) --- FNHWs Perspective on Their Contribution to Burn Care FNHWs were essential to the cultural safety of the child and the family, often acting as advocates between family and medical staff. There was a clear message from FNHWs that the need for cultural awareness training for non-Indigenous health workers was imperative in providing cultural safety and support for First Nations families. Data showed the provision of two main areas of support by FNHWs to include tangible and cultural support. Data also highlighted the importance of working together and the challenges associated with the role (Table 2). Transport for the child and family was difficult from remote communities, and FN-HWs would organise the crucial transport for the child to receive ongoing burn aftercare, alleviating some of the stress family were often face with. "But if it's an outpatient we'll sort that out as well and help the family. We used to help them with taxi vouchers but now taxi voucher are getting really, really scarce and it's really frustrated because we get families who missed their flight, and I'm not going to tell a family catch a bus and go from here to there, it's not fair. So we have to fight for taxi vouchers." (Aboriginal Liaison Officer) "So at the moment my role is, I give health education to patients, I help them understand why they're in the hospital, I help with any issues around social stuff within reason because the ILOs [Indigenous Liaison Officers] are employed for that role but because sometimes they're short staffed... our roles overlap a little bit and that's the whole role of the AHP [Aboriginal Health Practitioner] is the primary health care, is the holistic approach to health so sorting out everything. I also advocate for patients who need spiritual healing or cleansing, unfortunately (this hospital) don't provide that service or support that service so I try my best to get that patient to communicate and navigate through the system with having that need addressed, the spiritual need addressed as well the Western medical need." (Aboriginal Health Practitioner) "... Aboriginal Liaison Officers who are based on the ground floor and they do outpatients but it's more escorting them to appointments, booking them back home and stuff... those guys [patient/family] if it was a burns clinic outpatient appointment they probably wouldn't receive really any support without this." (Aboriginal Liaison Officer) 3.2.2. Cultural Support Supporting First Nations families through cultural connection helped with effective communication between multidisciplinary team and families. This was expressed by FNHWs as an important and essential component of their role. "Cultural advice you know or cultural safety on the clients say if they, we do a research and find out the language, where they're from and what traditions they have in their community and if they, you know some of the patients don't want to be seen by females so we have to let them know. Just making sure that staff are safe too and the patient's safe." (Aboriginal Health Practitioner) "... where the patient really fully understands or the patient's parent or guardian fully understands what's going on with a particular patient and will make an informed decision around that patient's care. So there's a lot of cultural barriers around that and you know like for me I think our cultural awareness training really lets staff down because it only addresses the first part of the cultural continuum of going towards patient cultural safe, delivering a culturally safe service..." (Aboriginal Health Practitioner) "English is probably their third language it's still funny though, I don't know how to explain it but we can still connect." (Aboriginal Liaison Officer) --- Two Worlds Working Together FNHWs functioned as advocates for the child and family and as mediators when working with the multidisciplinary burns team. There was recognition that FNHWs and the multidisciplinary team need to work together for the benefit of the child and family in order to contribute to better healing outcomes. "I think if you understand what the organisation expects and you understand the upbringing that you were raised in and you balance them out you realise how you can actually go about it and bring both of the worlds together and that means the outcome for the families is, you're going to achieve something." (Aboriginal Liaison Officer) " I've just always been asked can you come and talk to them, or they're not going to turn up for their physio appointments or they're not getting out of bed for me can you go, them kind of things but I've never really been asked culture-wise stuff whether it's because they've been around long enough they've got that knowledge and experience, but yeah I've just sort of being asked, they're not getting up, they're swearing, behaviour, the parents need to turn up, that sort of stuff." (Aboriginal Liaison Officer) "Just letting the team know like if we get someone from remote we let the team know where this community is, how far the nearest hospital is because there's only usually clinics on the communities and so looking at things geographically and then like what their cultural background is because all Aboriginal cultures are different but same if you know what I mean. Yeah and just basically looking after them while they're here and making their stay comfortable and their journey here comfortable." (Aboriginal Liaison Officer) "We go there and just advocate on behalf of the Aboriginal patient and have feedback, it's in regards to the feedback, more of the patient." (Aboriginal Liaison Officer) "So when a family comes up for outpatients if they need a bit of extra support, they'll contact the social worker or myself." (Aboriginal Liaison Officer) --- Challenges at Work The lack of FNHWs employed was stated as an important contributing factor in not being able to deliver best practice burn care. FNHWs expressed their concerns that staff fatigue and subsequent assumptions resulted in multidisciplinary workers treating First Nations patients unjustly. "So fatigue management also comes into it because then the staff get tired and a patient comes in as a new admission so that new admission is not treated as a new admission because it's easier to go, actually this patient reminds me of the last patient that came here and they weren't very engaged so straight away, and it happens, that's a natural response when someone's fatigued." (Aboriginal Health Practitioner) "... we need more Aboriginal people in here especially AHPs [Aboriginal Health Practitioners]. I can just see a whole system of AHPs working across the hospital, less money because we'd be doing things right from the start, not putting down any other profession it's not about that it's about delivering a service. You know when we look at customer service if we look at big corporations that make lots of money, what do they do, they really focus on what their target group want and need. That's what it's all about." (Aboriginal Health Practitioner) "I'm one to a whole division so no-one thinks to refer sometimes, maybe they're afraid to refer because they might be seen as not being able or culturally competent to deal with that patient. And so there's one of me to the division so that's very tricky most of the time." (Aboriginal Health Practitioner) --- Multidisciplinary Burn Team Members' Perspectives Multidisciplinary burn team members have a varied understanding of the importance of the FNHW role. Perceived descriptions of the care provided by the FNHW role including meeting tangible and cultural support needs (Table 3). --- Tangible Support Multidisciplinary team members understood the role of FNHWs was to provide families with tangible support including provision of education and networking for aftercare appointments. They also perceive the role is to stop people from absconding against medical advice. --- "We don't have an Aboriginal Liaison Officer [ALO] at the moment because she's unwell but we are recruiting to a contract position and I would always take the ALO... with me to go on the first visit with a family, check the family, because I just think it's really useful and then we kind of divide the tasks so you know she might do some of the more liaison practical things and I will do the trauma stuff with the family and the assessment." (Social Worker) "I use them [Aboriginal Liaison Officer] all the time especially if patients abscond or I need to know more about family relationships like, do you know much about this mob or their family... And I also use them quite a bit when you're trying to arrange and negotiate appointments for outpatient clinics and where to from here." (Clinical Nurse Consultant) "... it is such a isolating environment down here, very different and I don't pretend to know how I can understand and address those things culturally and so I really look for their [FNHW] input so to try and get an understanding how the parents and the child's feeling and what their care needs are." (Medical Consultant, Surgeon) "We have one or two chronic rehabilitation patients post-burn who are Indigenous, they are very involved with the ALOs [Aboriginal Liaison Officers] and also very involved with social work in terms of helping to arrange transport particularly if they're from regional centres, to us and back from us." (Registrar) --- Cultural Support Multidisciplinary team members perceive the role of the FNHW to include the provision of cultural support including an appreciation of family circumstance and communication. Some team members stated the importance of having FNHW involvement to ensure the family and child from remote communities felt comfortable. "I've been in this hospital in different capacities over time from a very junior doctor to a registrar and now a consultant and I have noticed that there's been a difference in the input from our Aboriginal Liaison Officers and I'd like to see more input because when I was a junior doctor I remember them being there all the time on the ward with the families and you could really notice the comfort that families found from having cultural similarities with staff there." (Consultant) "We always make sure we get the Indigenous liaisons involved just simply because a lot of the time especially if they're coming from remote communities they may not have the family support and we know that family and community is a big thing for Indigenous people." (Occupational Therapist) --- Two Worlds Grappling Multidisciplinary burn team members understood the importance of working together and the need for FNHW involvement in burn care. However, there was evidence in the data that illustrated a lack of commitment towards involving FNHWs in burn team activities such as core multidisciplinary meetings. At the same time often criticising that FNHWs are unable to effectively relay important family information. --- "... I think both work, you can't have one [multidisciplinary team] without the other [FNHW] and I think probably there are ILOs [Indigenous Liaison Officers] being underutilised and under-resourced for the amount of demand that we have and certainly like the model of care in Queensland for our ILOs are more a supportive role, they don't take on a clinical caseload, they don't do hands on dressing changes or anything like that, they're more an emotional support and navigating the health system for the Aboriginal clients and Torres Strait Islander clients." (Occupational Therapist) "I think the ideal thing would be to get Aboriginal staff involved in their management, they sort of identify better with Aboriginal staff, the difficulty is finding those who are trained well enough but they do seem to respond better to those than us telling them something." (Consultant) "... sometimes we have to get the Aboriginal Liaison person in if we're having difficulty communicating, especially the kids that come down... sometimes it's really hard to engage with them, and you've got to get help in for that sort of thing. The indigenous kids we tend to get here coming from... tend to be different." (Surgeon) "I think their ability to communicate with the team has been lacking, maybe they're great at talking to the family but then feeding back and actually feeding our information back is probably, there's not a really strong link there. So from my perspective I think having stronger ALO [Aboriginal Liaison Officer] support, that liaison and educating me of what I need to do or how I can get through the other way, how can I get my message through via the ALO officer, probably involving them more would help in certain circumstances." (Physiotherapist) "I guess one of the obvious ones is asking for support from our ILO [Indigenous Liaison Officer] to facilitate meetings or education or discharge planning or resilience sort of stuff, to be having that supported conversation with me so that I'm being culturally appropriate and culturally aware." (Clinical Nurse Consultant) "We have Aboriginal Liaison Officers [ALOs] in the hospital and we're very quick to try and get them involved very early on but I think their capacity at times from my perspective has been, I think their ability to communicate with the team has been lacking, maybe they're great at talking to the family but then feeding back and actually feeding our information back is probably, there's not a really strong link there. So from my perspective I think having stronger ALO support, that liaison and educating me of what I need to do or how I can get through the other way, how can I get my message through via the ALO officer, probably involving them more would help in certain circumstances." (Physiotherapist) --- Challenges at Work Multidisciplinary team members recognised there are challenges between FNHW availability and being over worked due to inadequate resourcing of First Nations workers. It was also recognised that poor interaction with FNHWs by the multidisciplinary team members impacted outcomes for First Nations children. "When they're identified as Aboriginal or Torres Strait Islander we would always engage with the Aboriginal Health Officer... she's currently off at the moment and we don't have that service available to us." (Clinical Nurse Consultant) "... wonderful, she's worth her weight in gold, but we need ten of her." (Physiotherapist) "... occasionally we've got to get the Aboriginal and Torres Strait Islander Liaison Officer involved if it's a particularly sticky situation but most times we can deal with things in the usual manner." (Surgeon) "I understand it's about getting them home and back into their environment and the financial strain and things like that about getting back but I feel we're very mindful of those sorts of things but what we're pushing is long term scar, long term loss of movement of arms and hands, you know for that sort of outcome, but I do think probably the relationship between her [Aboriginal Liaison Officer] and us [burns team] probably isn't ideal which maybe then impacts onto those children." (Surgeon) --- Discussion Both studies specifically explored the role of FNHWs in delivery of burn care, both acute and in aftercare. Combining the data of First Nations author J.C.'s study informed by Indigenous methodology [18] with the data from non-Indigenous researcher SF who engaged interface research methodology [17] deliberate and purposeful. This gives a breadth of context (understanding the entire patient journey) and depth of understanding (how the role participates in the team and how the roles meets the needs of the families across the journey). The combination of data provides an opportunity to further explore the role of FNHWs in burn care for First Nations children First Nations ways of knowing, being and doing in the context of health and healing are not always present within the Western biomedical health systems that surrounds and informs initial burn care and burn aftercare. Multidisciplinary teams do recognise the importance of FNHWs, but systems are not set up to support resourcing because all categories said they're overworked. This further supports evidence about how the biomedical model excludes the lived experiences and knowledge of First Nations people [10]. This lack of inclusion is partly the result of a discrepancy between the perspectives of First Nations families, FNHW, and multidis-ciplinary burn team members regarding different ways of health and healing. The First Nations families experienced improved cultural safety and were able to access necessary tangible support through care provided by the FNHW. This supports other studies [9] that have highlighted the necessary role of FNHWs in better being able to understand the needs of First Nations families. Similarly, FNHW experience of being involved in burn care for First Nations families was that they understand their contribution to care and wanted to be involved in the multidisciplinary care yet were excluded. While First Nations families and health workers experienced and understood these needs, the health systems informing multidisciplinary burn care excluded First Nations ways of knowing being and doing. This was evidenced by the team's exclusion of FNHW from team meetings, and by the Western biomedical models lacks acknowledgement of the importance of the FNHW role and subsequent resourcing. Different ways of knowing, being and doing exist. Values are often placed on needs without considering these differences. This is evidence of colonisation through a disregard to different ways of knowing, being and doing. Care that is regardful of differences, improves dialogue and reduces power imbalances will lead to an experience of improved culturally safe care [19]. --- Two Worlds Collide There was an expectation verbalised by FNHWs that some health professionals only used the FNHWs role as trouble shooters for assumed difficult patients and did not value the professional role of a FNHWs. However, the diverse nature of services for First Nations children and families was expressed as essential by family members. FNHWs provided cultural and spiritual care, tangible support such as transport to aftercare appointments, accommodation and food vouchers. Families valued the FNHW for alleviating fears, providing cultural safety and advocacy due to a shared recognition of cultural knowledge and connection [20]. Members from the multidisciplinary team merely asked for assistance from FNHWs when there was a 'problem to fix' such as absconding. Ongoing aftercare appointments were facilitated by FNHWs only when all else failed due to team members inability to reach the family for aftercare appointments. From the data collected from members from the multidisciplinary team it was apparent that FNHWs are not valued for the quality of care that families had expressed [21]. --- Colonisation A power imbalance exists in favour of the Western biomedical model [19], especially as it relates to burn care for First Nations children in Australia [22]. Multidisciplinary burn team members showed power imbalances through their placement of value on the needs of First Nations families. Their values were assumptions, not grounded in evidence, and showcased the implicit bias within the system of multidisciplinary burn care. Furthermore, multidisciplinary burn team members sought input from FNHWs sometimes based on needs that were centred in a deficit mindset. For example, when they thought a First Nations family would not attend follow-up care due to dysfunction. The problematisation of First Nations people's needs is indicative of structural racism in Australia's health system [23]. --- Cultural Safety Cultural safety has been shown to improve the health outcomes of First Nations people in accessing mainstream healthcare [24,25]. Furthermore, engagement in critical reflexivity in healthcare has been shown to support the competence of non-Indigenous healthcare professionals. Support for cultural competence in policy documents contributes further to improved cultural safety. Good cultural competence leads to an experience of culturally safe care. First Nations families and health workers understood the FNHW role in enhancing an experience of cultural safety, whilst multidisciplinary burn team members grappled to acknowledge the role's contribution to achieving quality care. So, while multidisciplinary burn team members said they understood they role, the disjuncture between the two ways of knowing being and doing supports the idea that enhanced cultural training is imperative for multidisciplinary team members to move beyond the divide and contribute constructively to closing to gap. --- Strengths and Limitations Combining the two sets of data strengthens the reliability and validity of the issues surrounding FNHWs' crucial involvement in burn aftercare. Furthermore, this combining of data also resonates with the methodology used to inform sub-study two whereby Aboriginal ways of knowing, being and doing [26] are integrated with the Western biomedical standpoint [19]. It has also shown that by using Indigenous methods, which is instrumental in decolonising research approach that supports the empowerment of all Aboriginal and Torres Strait Islander peoples and communities in this study [13,27,28]. A limitation of sub-study two was not asking if any multidisciplinary team member identified as a First Nations person. Sub-study one only yarned with children and their family that were admitted to a tertiary burns unit and may have missed perspectives from families whose child with a burn was seen outside of the tertiary setting. --- Conclusions Delivery of culturally safe care is essential to ensure equitable health outcomes. This paper has highlighted the importance of inclusion of FNHWs in delivery of healthcare to Australia's First Nations children who need ongoing complex care. It is essential that FNHWs are active participants in the multidisciplinary care burn meetings and are encouraged and supported by the team members to engage in partnership of service delivery and ongoing aftercare for the child and family. This will require additional resourcing and additionally important changes to clinical hierarchies in the delivery of care, to ensure the important role of FNHWs is appropriately noted and rewarded. Informed Consent Statement: Informed consent was obtained from all participants involved in this study. --- Data Availability Statement: All data collected are stored in secure servers at The George Institute and to maintain Indigenous data sovereignty cannot be placed on an Open Data Platform. --- Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Burns affect Australia's First Nations children more than other Australian children, they also experience longer lengths of stay in tertiary burns units and face barriers in accessing burn aftercare treatment. Data sets from two studies were combined whereby 19 families, 11 First Nations Health Worker (FNHW) and 56 multidisciplinary burn team members from across Australia described the actual or perceived role of FNHW in multidisciplinary burn care. Data highlighted similarities between the actual role of FNHW as described by families and as described by FNHW such as enabling cultural safety and advocacy. In contrast, a disconnect between the actual experience of First Nations families and health workers and that as perceived by multidisciplinary burn team members was evident. More work is needed to understand the impact of this disconnect and how to address it.

Introduction A substantial number of studies have linked alcohol availability to a range of alcohol-related problems (Gruenewald, 2007;Livingston et al., 2007). In particular, there has been an almost uniform finding from cross-sectional studies that the density of alcohol outlets in an area is positively associated with rates of violence in that area. This finding has been replicated in a wide variety of U.S. settings (e.g. Scribner et al., 1999;Stevenson et al., 1999;Gorman et al., 2001;Lipton and Gruenewald, 2002;Nielsen et al., 2005;Reid et al., 2005;Gruenewald et al., 2006;Gyimah-Brempong and Racine, 2006) and in handful of Australian studies (Stevenson et al., 1999;Chikritzhs et al., 2007;Livingston, 2008a). An increasing number of longitudinal analyses have also found significant associations between alcohol outlet density and violence over time, providing stronger evidence that the relationship is causal (Norström, 2000;Gruenewald and Remer, 2006;Livingston, 2008b;Yu et al., 2008). These studies have focussed on assault rates, using either hospital or police data to estimate the rates of violence occurring in geographical areas. Thus, the violence generally examined in the current literature is largely public violence, often taking place in and around licensed premises. It is worth noting that Freisthler and colleagues have undertaken a series of studies that have demonstrated statistical associations between alcohol outlet density and child abuse and maltreatment, both cross-sectionally (Freisthler, 2004;Freisthler et al., 2004Freisthler et al.,, 2005Freisthler et al.,, 2007) ) and over time (Freisthler and Weiss, 2008). Studies of the relationship between alcohol outlet density and domestic violence are less common. Gorman et al. (1998) examined the association between outlet density and domestic violence (largely made up of intimate-partner violence, but also including violence against children), finding no relationship once socio-economic factors were controlled for. An Australian study examined police data using assaults on private premises as a measure for domestic violence, finding that total volume of beer sold in a region was related to rates of violence, but that geographic measures of outlet density (i.e. outlets per km 2 ) were not (Chikritzhs et al., 2007). In a recent paper, McKinney et al. (2009) used survey and ecological data in a multi-level study of selfreported intimate-partner violence. Their study found a significant association between alcohol outlet density and male-to-female partner violence, but no significant association for female-to-male violence. The study controlled for a substantial amount of individual-, couple-and zip code-level sociodemographic factors, and examined self-reported alcohol consumption along with outlet density, finding that the effect of outlet density on domestic violence was stronger for couples who reported bingedrinking behaviour. When the results were broken down by type of outlet, on-premise outlet density was significantly related to partner violence rates, while off-premise outlet density was not. Neither of the ecological studies (Gorman et al., 1998;Chikritzhs et al., 2007) controlled for the effects of spatial autocorrelation, which can result in biases in statistical testing (Gruenewald and Remer, 2006). Theoretically, a causal relationship between alcohol outlet density and domestic violence could be based on one of two mechanisms. Most obviously, it is plausible that higher levels of alcohol outlet density cause higher alcohol consumption, leading to increased rates of domestic violence. This is given some support by the clear links between alcohol consumption and domestic violence (e.g. Leonard, 2001;Foran and O'Leary, 2008) and by the studies finding positive associations between alcohol outlet density and consumption (Godfrey, 1988;Gruenewald et al., 1993), although this finding is not uniform (Pollack et al., 2005). Secondly, Freisthler (2004) has suggested that alcohol outlets alter residents routine activities, with increases in off-premise outlets resulting in a shift towards drinking at home (and thus higher rates of child physical abuse), and increases in on-premise outlets resulting in a shift towards drinking outside the home (with corresponding increases in child neglect). It is possible that similar changes to routine drinking activities directly bear on the likelihood of domestic violence incidents with, particular types of outlets encouraging different drinking patterns without necessarily altering overall consumption. For example, increases in pubs and hotels may lead to more frequent episodes of intoxication and thus increased domestic violence, while increases in packaged liquor outlets could shift drinking into the home, with potential increases in domestic problems due to increased exposure. Contrastingly, increases in restaurants and cafes may shift some residents to more moderate drinking patterns, thus reducing the likelihood of intoxication and domestic violence. It is worth noting that these two rationales are interconnected -routine drinking activities may change along with total consumption, with for example, increased consumption overall along with increases in drinking to intoxication. Based on these two theoretical perspectives, it is expected that overall densities of alcohol outlets will be positively associated with domestic violence rates, and that densities of pubs and packaged liquor stores will be particularly strongly associated with violence. It is possible that alcohol outlets and domestic violence are positively associated with no true causal link. For example, alcohol outlets may be markers of social disorganisation, or a lack of collective efficacy, which have been shown to have significant relationships with domestic violence rates (O'Campo et al., 1995;Miles-Doan and Kelly, 1997;Grisso et al., 1999;Cunradi et al., 2000;Bartolo, 2001;Browning, 2002;Pearlman et al., 2003;Cunradi, 2007). In this case, unless alcohol outlets represent specific causes of social disorganisation, any relationship between outlet density and domestic violence would only be detected when neighbourhood levels of social disorganisation were not controlled for. The results of Gorman et al. (1998) analysis from New Jersey, USA suggest this explanation, with the effects of outlet density disappearing once socio-demographic factors were controlled for. Contrastingly, McKinney et al. (2009) found a significant association between outlet density and partner violence even while controlling for a wide range of socio-demographic factors, consistent with a causal relation mediated by increased consumption. The present study tests the statistical association between alcohol outlet density and domestic violence in Melbourne, Australia. The study will provide further analysis of an under-researched question: is the availability of alcohol related to the incidence of domestic violence? The association between outlet density and domestic violence rates will be examined with and without socio-demographic control variables to examine whether any associations found with outlet density are mediated by socio-demographic factors. --- Materials and methods Aggregated data from administrative sources were used to examine the associations between alcohol outlet density and police-recorded domestic violence rates. Analyses were undertaken both with and without a range of neighbourhood characteristics controlled for to assess whether alcohol outlet density was associated with domestic violence over and above other measures of social deprivation. --- Geographical units The study uses postcode-level data for the greater Melbourne area, approximately 5,600 km 2 containing the city of Melbourne and its surrounding suburbs. Postcodes are the smallest unit for which reliable liquor licensing and police data are available, and are generally good approximations for local suburbs. In 2001 there were 222 postcodes in the greater Melbourne area, with approximately 3,350,000 residents. The analyses were limited to Melbourne to avoid the difficulties associated with analysing rural and urban postcodes in the same models. Initial analyses resulted in five postcodes being excluded from the study, leaving 217 postcodes for the final analyses. Two of the excluded postcodes represented particularly unusual land uses and low residential populations (i.e. an airport and a military base), while three others were innercity postcodes with very high outlet numbers and few residents. --- Measures Domestic violence incident data were provided by the Victorian Police Service from their Law Enforcement Assistance Program (LEAP) database. The data used are counts of "family incidents" recorded by the police for each postcode in the study area in the financial year 2001/2002. Family incidents include any incident reported to the police where the police determine that a crime has been committed and that there is an intimate relationship between the offender and the victim (e.g. married, divorced, child/parent, de-facto etc). The majority of these incidents involve incidents of domestic violence involving people in couple relationships (or whose couple relationship is now over), and around 80% of the incidents involved male offenders and female victims. This contrasts somewhat with survey derived estimates of domestic violence, which generally find roughly equal numbers of male and female victims (e.g. Graham et al., 2008). However, generally female victims experience more severe violence and are more likely to be injured (Graham et al., 2008). Thus, as offences involving police are likely to be at the more serious end of the spectrum, it is not surprising that the majority of offences dealt with in the current study involve female victims. The postcode recorded is for the locality in which the offence takes place, which is generally the victim's residence. It should be noted that using a policing driven measure of domestic violence creates the potential for biases in the analyses (e.g. policing effort may be more intense in some areas and thus more likely to record incidents in those areas). In addition, many incidents of domestic violence are likely to be excluded from police-based statistics with, for example, the Australian 2005 Personal Safety Survey (Australian Bureau of Statistics, 2006) finding that just 36% of female victims of physical assault reported the incident to the police. This represents a significant source of potential bias to the study, particularly if reporting rates vary along with the availability of alcohol. There is little published research on factors influence the reporting of domestic violence in Australia, although there is research from the US indicating, for example, that non-white victims are more likely to report domestic violence, so the risk of bias from this measure is not negligible. However, police data is often used in these kinds of analyses (e.g. Miles-Doan and Kelly, 1997; Bartolo, 2001;Britt et al., 2005;Gorman et al., 2005), and the use of a range of control variables related to police reporting rates (e.g. socioeconomic disadvantage) will ameliorate this bias somewhat. While previous studies have used counts or rates based on geographic measures, the fact that offences took place in residential settings meant that it was reasonable to calculate rates on a per 1,000 population basis. Alcohol outlet data were provided by the Licensing Branch of the Victorian Department of Consumer Affairs. The licensing data provided included full address details for each premise, including postcode. Two-hundred random records were checked to assess the accuracy of the postcode field, with accurate data noted in 98% of cases. Three categories of licences are examined in this study: packaged liquor, general and on-premise. Packaged liquor licences allow for the sale of alcohol for off-premise consumption only and are generally specialised alcohol stores or small grocery stores. General licences allow for the sale of alcohol for either on-or off-premise consumption and are largely pubs and taverns -places where drinking is the primary activity and where alcohol can be sold to be consumed elsewhere. On-premise licences allow only for on-premise sales and include cafes, restaurants, bars and nightclubs. On-premise licences include a broad range of premises, but are predominantly restaurants and small bars. In 2001, these licence categories made up 61% of all licences in Victoria, with the remainder made up of registered clubs, producer-distributors and limited licences. Correlations between licence densities of the three types were examined and, while on-premise and general licence densities were highly correlated (0.7), tests for multicollinearity in the final models did not find significant problems due to correlated dependent variables. Outlet densities were calculated for each of the three categories examined on a population basis as rates per 1,000 residents1. The remaining data for the study were sociodemographic measures sourced from the Australian 2001 Census of Population and Housing Basic Community Profiles (Australian Bureau of Statistics, 2002). The particular measures chosen were selected based on their significant associations with violence in previous studies of outlet density and on their significance in studies examining the ecological predictors of domestic violence. In particular, socio-economic status has repeatedly been linked with rates of domestic violence, as have measures of race and the proportion of residents who own their own homes (Miles-Doan and Kelly, 1997; Bartolo, 2001;McKinney et al., 2009). Thus, this study includes three measures of socio-economic disadvantage: a composite measure, the index of relative socio-economic disadvantage (IRSED) derived from 2001 census data, the unemployment rate for each postcode and the proportion of households in each postcode which were owner-occupied. The IRSED is a composite measure based on numerous variables including educational attainment, household income and single-parent families. This measure provides an overall estimate of the socio-economic disadvantage of an area, without introducing collinearity (correlation between two or more variables under analysis; Cohen, 2003), by including multiple measures of the same underlying construct, e.g. the proportion of people on low incomes and those with low educational attainment are highly correlated. Full details of the derivation of the IRSED have been published by the Australian Bureau of Statistics (2003). In addition, the proportion of the population in each postcode who spoke a language other than English was included as a measure of cultural diversity. Population instability has also been linked with social disorganisation, lack of collective efficacy and domestic violence (Morenoff et al., 2001;Stover, 2005), so a measure of population turnover (the proportion of the population in each postcode who had moved in the last five years) was also examined. Finally, a measure of population density (residents per km 2 ) was examined to account for the varying geographic sizes of the units of analysis. --- Analysis All analyses were undertaken using the "R" statistical software package (R Development Core Team 2006), with the "spdep" package (Bivand, 2006) used for all spatial analyses. The "spdep" package is an add-on to "R", providing a number of tools for spatial analysis, including calculation of spatial autocorrelation and spatially appropriate regression models. These tools allow for the appropriate analysis of data based on spatial units, where the underlying assumption of standard regression models that study units are independent is likely to be broken. The unit of analysis throughout the study was the postcode. The dependent variable for all analyses was the rate of police recorded family incidents per 1,000 residents. The independent variables were the three categories of alcohol outlet density, the IRSED index of disadvantage, the unemployment rate, the proportion of households owner-occupied, the proportion of residents who had moved in the previous 5 years and the population density (residents per km 2 ). Tests for multicollinearity were undertaken throughout the analysis. A regression model involving highly correlated independent variables (e.g. variables measuring poverty, unemployment and education levels) can produce misleading results, due to the overlaps in variance being explained by the correlated variables. Thus, models were checked to ensure that the independent variables included did not produce models overly affected by multicollinearity. An initial series of ordinary-least squares linear regression models were developed. The first model included only the alcohol outlet density variables as independent variables and the second model added all the socio-demographic variables. Diagnostic analyses undertaken at this point resulted in the exclusion of the unemployment rate and the proportion of households that were owner-occupied from the final model due to multicollinearity (variance inflation factors >5). This is not surprising, as both unemployment and housing tenure variables are included in the calculation of the IRSED index. The final model incorporating the alcohol outlet variables and the remaining socio-demographics was then produced. At this stage, an assessment of the level of spatial autocorrelation in the residuals of the final model was examined. Spatial autocorrelation was examined based on neighbouring postcodes. Thus, the connection matrix included 1s where postcodes were direct neighbours and 0s otherwise. The Moran's I coefficient of the residuals was significant (I = 0.12; P <unk>0.01) indicating possible bias due to spatial autocorrelation. An examination of the Lagrange Multiplier Test Statistics suggested that a spatial error model was appropriate to control for these possible biases, and a maximum likelihood estimation of a spatial error model was developed (Ward and Gleditsch, 2008). --- Results Descriptive statistics for the variables used in this study are provided in Table 1. The bi-variate correlations between each of the independent measures and the domestic violence rate are provided in the table, indicating a strong association between measures of socio-economic disadvantage and domestic violence, and a positive association between general licence density and domestic violence. The OLS regression models are presented in Table 2. The first model includes only the alcohol outlet density variables, and the second adds in the sociodemographic variables discussed previously. As mentioned in the previous section, the unemployment rate and proportion of households that owned their own home were excluded from the modelling as they were highly collinear with the IRSED index. The initial model finds a positive and significant association between general outlet density and domestic violence rates and a negative and significant association between on-premise outlet density and domestic violence. There is no significant association between packaged licence density and domestic violence rates. The model was not a good fit to the data, explaining only 9% of the variance in domestic violence rates. When the socio-demographic variables were added to the model, the direction and significance of the associations between the outlet density variables and domestic violence rates remained, although the magnitude of the effects for both general outlets and on-premise outlets were roughly halved. The IRSED was highly significant, suggesting that postcodes with lower socio-economic status had higher reported domestic violence rates. Similarly, the model suggests that postcodes with high population density have higher rates of domestic violence. The proportion of residents from non-English speaking households was negatively associated with domestic violence rates, and there was no significant relationship between population mobility and domestic violence. This expanded model provided a much better fit to the data, explaining around 55% of the variance in domestic violence rates. Tests of the Moran's I coefficient for each model indicated substantial levels of spatial autocorrelation. To ensure that the final results were not biased due to spatial dependence, spatial error models were developed. The results of these models are presented in Table 3. While the spatial parameter was significant in both models, the overall results are not markedly different from those derived using OLS techniques (Table 2). The fit of the full OLS and full spatial error models were compared using the Akaike information criterion (AIC) (Akaike, 1974). The AIC measures the goodness of fit of regression models across varying model-fitting methods (measures such as the R-squared are not appropriate to compare maximum likelihood estimated spatial models with OLS regression models). Lower values of the AIC represent models that use the fewest parameters possible to best fit the data. The OLS model had an AIC of 905.9, while the spatial error model had an AIC of 898.7, indicating a slightly better fit. Thus the spatial error model was considered the better fitting model and will be the focus of the discussion. In terms of the licence density variables, general licences were significantly positively associated with domestic violence, while on-premise licences were significantly negatively associated with it. An increase of general licence density of one outlet per 1,000 residents was associated with an increase of 1.35 reported incidents of domestic violence per 1,000 residents. Similarly, an increase of 1 on-premise outlet per 1,000 residents was associated with a decrease of 0.34 incidents of domestic violence per 1,000. --- Discussion This study has demonstrated a statistical association between the density of alcohol outlets and police-recorded domestic violence rates at the postcode level in Melbourne, Australia. The study finds a positive association between general (pub) licence density and domestic violence and a weaker, negative association between on-premise (restaurant, etc.) licence density and domestic violence. The study provides the first aggregate analyses of the relationship between alcohol outlet density and rates of domestic violence to adequately control for the spatial dependence of the units of analysis. In addition, it examines specific outlet types and incorporates controls for socio-demographic factors likely to covary with outlet density and domestic violence rates. The study is limited by its use of police data, which are likely to underestimate the true rate of domestic violence and may result in biases in the findings. In broad terms, the results of this study are consistent with previous research in this field. Domestic violence rates are highly correlated with levels of socio-economic disadvantage (Bartolo, 2001;Browning, 2002;Pearlman et al., 2003;McKinney et al., 2009). There is a significant and negative relationship between the proportion of residents from non-English speaking backgrounds and domestic violence, which has not been found in previous studies of domestic violence (most of which are from the US and are based on specific race data). This result does find support from a previous study of general assault rates in the same jurisdiction, with a similar negative effect found for ethnic diversity (Livingston, 2008a). There are a number of possible explanations for this relationship. While there is little data available on the prevalence of domestic violence in non-English speaking communities in Australia, evidence from the Women's Safety Survey suggests that these women experience lower rates of violence than the rest of the Australian population (The Women's Services Network, 2000). This may be partly due to lower alcohol consumption amongst these groups (Australian Institute of Health and Welfare, 2009). In addition, there is some evidence that people from minority racial groups are less likely to report crimes to the police (Shah and Pease, 1992), which may result in their under-representation in the data used in the current study. The findings for the alcohol availability variables were mixed, with a positive effect found for general licence density (hotels, pubs, etc.) and a negative effect for on-premise licence density (restaurants, bars, etc.). These associations were considerably weakened by the inclusion of a range of socio-demographic variables in the models, but remained significant. Hotels and pubs have previously been identified as particularly problematic venues for general levels of violence (Briscoe and Donnelly, 2001;Livingston, 2008b), while restaurant licences have been found to be associated with lower rates of general violence in other jurisdictions (e.g. Zhu et al., 2004). These findings provide support to Freisthler's theory of alcohol availability affecting people's routine drinking activities, with higher densities of alcohol-focussed venues (general licences) associated with increased domestic violence, while densities of less alcohol-focussed venues (on-premise licences) associated with reduced domestic violence. Traditionally, hotels and pubs have been the site of heavy drinking and drinking at these venues may be particularly likely to involve drinking to intoxication (Room, 1988). However, the lack of relationship between packaged liquor density and domestic violence was surprising in the context of this theory, which suggests that increased availability of packaged liquor leads to increased consumption within the home, which may increase the risk of incidents of domestic violence. The relationship between density of on-premise licences (restaurants, cafes, bars, etc.) and domestic violence is negative, with higher densities of onpremise outlets related to lower levels of domestic violence, which may be due to the more moderate drinking styles associated with these kinds of premises. However, it seems unlikely that this relationship is due to the actual outlets themselves exerting protective effects, and instead reflects unmeasured socio-economic or demographic factors that vary along with on-premise licence density. This then raises the question of whether the effect found for general licences is a similar artefact and whether there are unmeasured neighbourhood factors related to both general licence density and domestic violence rates. For this reason, along with the limitations identified in the measures used in this study, the results presented here can point only to tentative implications for policy. The finding that general licences are positively associated with domestic violence rates, combined with previous work which found strong links between general licence density and assault in Melbourne (Livingston et al., 2007;Livingston, 2008a) suggest that the trend towards increasing numbers of these licences in Melbourne needs to be curtailed. --- Conclusions The results presented here provide support to previous findings linking alcohol outlet density to domestic violence (Chikritzhs et al., 2007;McKinney et al., 2009). When these results are considered alongside previous work examining the broader impacts of alcohol outlet density (Livingston et al., 2007), the policy implications are clear: efforts to stem the ongoing increases in alcohol availability in Victoria are required. However, the specific policy approaches required depend on a deeper understanding of the mechanisms of the link between alcohol outlets and domestic violence. While this paper has provided a first step in that direction, there remain uncertainties about the precise nature of this relationship. Future studies making use of longitudinal data are required to better understand the influence of alcohol availability on domestic violence.

Studies have consistently found positive associations between the density of alcohol outlets and levels of violence in areas. Few studies have examined whether this relationship holds for domestic violence. This study assesses whether alcohol outlet density is related to domestic violence and whether this relationship is due to alcohol availability or to co-occurring economic disadvantage and social disorganisation. Cross-sectional data on family incidents, liquor outlets and socio-demographic characteristics were obtained for 217 postcodes in Melbourne, Australia. These data were used to construct models assessing the association between alcohol outlet density and domestic violence, both with and without controlling for socio-demographic factors. Models were tested for spatial autocorrelation, and spatial-error models were developed to control for its influence. Outlet density was significantly associated with rates of domestic violence, even controlling for socio-demographic factors. The density of hotels (pubs) was positively associated with domestic violence rates and the density of restaurants and bars was negatively associated with domestic violence. Socio-economic disadvantage was also associated with domestic violence rates. The density of packaged liquor outlets was not associated with rates of domestic violence. The results present a mixed picture, and further study is required to develop a clearer understanding of the links between alcohol availability and domestic violence.

T he prevalence of Alzheimer disease (AD) is rapidly increasing in the aging population and is projected to nearly triple in the coming decades. 1 Black and Hispanic individuals are 1.5 to 2 times more likely to be diagnosed with clinical AD or related dementias (ADRD) compared with other racial and ethnic groups. 2,3 In contrast, Asian American individuals (across subgroups) in the US may have lower age-adjusted incidence of all-cause dementia. 4 The increased risk of ADRD among Black and Hispanic individuals may be driven by dementia risk factors, including rates of cardiovascular disease and diabetes, 5 as well as social and structural factors (eg, lived experiences of discrimination and racism, economic opportunity, neighborhood disadvantage, and access to quality education). [6][7][8] Disparities in ADRD are further exacerbated by delayed diagnosis and misdiagnosis of AD, 9 lack of access to dementia specialist practices, 9 and biases in neuropsychological testing 10,11 impacting minoritized racial groups. The emergence of novel molecular therapies for AD (eg, the recently approved anti-amyloid monoclonal antibody aducanumab) highlight the importance of diagnosis at an early clinical stage and biomarker confirmation of AD pathology among patients who are potential candidates for diseasemodifying therapy. 12 Pathologically, AD is characterized by <unk>-amyloid and <unk> deposition in the brain with amyloid plaques representing a core feature of disease. 13 Presently, there are 3 positron emission tomography (PET) radiopharmaceuticals approved by the US Food and Drug Administration (FDA), namely fluorine 18 ( 18 F)-labeled florbetapir, 18 F-labeled flutemetamol, and 18 F-labeled florbetaben, for the in-vivo detection of amyloid plaques. Unfortunately, amyloid PET biomarkers have largely been studied in individuals identified as White, with minimal inclusion of racially and ethnically diverse groups. [14][15][16][17][18] Studies examining racial and ethnic differences in amyloid PET have yielded variable results and have primarily included cognitively unimpaired participants. [18][19][20][21][22] Given the importance of biomarkers like amyloid PET in the early and accurate diagnosis of AD, understanding racial and ethnic differences in amyloid PET positivity is crucial for improving the management of ADRD among diverse groups. The Imaging Dementia-Evidence for Amyloid Scanning (IDEAS) 23 study assessed the utility of amyloid PET in Medicare beneficiaries across the US with MCI or dementia in a large national network of dementia specialists. In the present secondary data analyses, we compare amyloid PET results across racial and ethnic groups from IDEAS and compare sociodemographic and comorbidity data to further explore racial and ethnic differences in ADRD clinical presentations and risk factors. --- Methods IDEAS 23 examined the association between <unk>-amyloid PET and changes in patient management and patient-oriented outcomes in Medicare beneficiaries with MCI or dementia. The practice-based, pragmatic study engaged 946 dementia specialists from 595 practices who recruited and referred Medicare beneficiaries for amyloid PET imaging at 343 imaging facilities across the US. Study design and report on the asso-ciations between amyloid PET and changes in patient management have been previously published, along with the full protocol and study materials. 23 --- Study Design Population The population in IDEAS consisted of Medicare beneficiaries 65 years and older diagnosed with MCI or dementia by a dementia specialist and in whom the cause of cognitive impairment was uncertain after a comprehensive clinical evaluation and knowledge of amyloid status was expected to impact diagnosis and management. 23 Of the 21 949 participants enrolled in IDEAS, 18 256 (83.2%) were considered for inclusion in this study. Exclusions included protocol violations (n = 319), not receiving an amyloid PET scan (n = 3337), and not having a positive or negative scan result (n = 37). Of 18 256 participants considered for inclusion, participants who were identified as multiracial (n = 31), Indigenous (n = 32), or had unknown or unreported race or ethnicity (n = 1086) were excluded owing to small numbers, resulting in 17 107 total participants available for analyses (Figure). Race and ethnicity in IDEAS were recorded during study registration by dementia specialists, and it is unknown whether race and ethnicity were ascertained by patient report (eMethods in the Supplement). Options for race included American Indian, Alaska Native, Asian, Black or African American, Native Hawaiian or Pacific Islander, White, not reported, or unknown. More than 1 race could be selected. Ethnicity was recorded as Hispanic or Latino, not Hispanic or Latino, not reported, or unknown. A single variable was created to summarize participant race and ethnicity for the purposes of analysis in the present study (eResults in the Supplement). Individuals identified as Hispanic or Latino ethnicity, regardless of race, were categorized as Hispanic. Individuals indicating multiple races were categorized as multiracial and were not included in the comparisons. Individuals identified as Indigenous (American Indian, Native American, Alaska Native, Native Hawaiian, and Pacific Islander) were not included in the comparison. All other individuals were categorized by the 1 race they selected: Asian, Black, or White. --- Key Points Question Do amyloid positron emission tomography (PET) positivity rates differ across racial and ethnic groups with mild cognitive impairment (MCI) or dementia? Findings In this cohort study of 17 107 Medicare beneficiaries with MCI or dementia, the proportion of amyloid positive PET scans was greater among White participants compared with Black and Asian participants. When racial and ethnic groups were matched by social and demographic factors, the proportion of amyloid positive PET scans was greater among White participants compared with Hispanic and Asian participants but not compared with Black participants. --- Meaning The results of this study showed differences in rates of amyloid PET positivity among racial and ethnic groups; these findings may reflect differences in underlying etiology of cognitive impairment between groups. --- Informed Consent in IDEAS The IDEAS study used a central institutional review board (Advarra, formerly Schulman Associates) and was managed by the American College of Radiology. Participants provided written consent to allow for their data to be used for future research purposes. PET results are reported according to standards for studies of diagnostic test accuracy in dementia. 24 --- Statistical Analyses Baseline participant and disease characteristics were summarized by median and range for continuous variables or by counts and percentages for categorical variables for each coded racial or ethnic group (eResults in the Supplement). The proportion of participants with a positive amyloid PET scan result and corresponding 95% Wilson confidence interval (CI) 25 were calculated for each racial and ethnic group as well as in subsets defined by impairment level. Because of differences in social factors (eg, educational attainment) and medical history across racial and ethnic groups, we performed optimal 1:1 matching to compare amyloid PET positivity between racial and ethnic minority groups and White participants (eMethods in the Supplement). Variables for matching 26 were selected based on their association with the outcome of interest (amyloid PET positivity) and included age (matching within <unk>3 years), sex, highest level of education attained, living arrangement (with whom do you reside; coded into living alone vs not living alone for purposes of analyses), history of hypertension, history of diabetes, family history of dementia, and level of impairment (MCI or dementia). For each racial and ethnic group, the proportion of amyloid-positive participants was calculated along with its corresponding 95% Wilson CI. 25 The amyloid PET scan results for each racial and ethnic minority group were compared to the amyloid PET scan results of White participants using the McNemar test. Because matching reduced the sample by 82% (from 17 107 to 3154), we were concerned the matching analysis was not powered to detect significant changes. Thus using all data available for analyses (the full analysis set), we also conducted a multivariable logistic regression model where age, sex, highest level of education attained, living arrangement, history of hypertension, history of diabetes, family history of dementia, level of impairment (MCI vs dementia), and race and ethnicity were added as covariates and amyloid PET scan result was used as the outcome. The association between each covariate and amyloid PET scan result was summarized by odds ratios (ORs) and 95% Wald CIs. Wald tests evaluating whether the odds of having a positive scan were equal were also performed. All reported P values are from 2-tailed tests at a significance level of.05. These analyses were exploratory in nature and were not adjusted for multiplicity. All statistical analyses were performed using SAS/STAT version 9.4 (SAS Institute). --- Results A total of 17 107 participants were included in the full analysis set, including 321 Asian individuals (1.9%), 635 Black individuals (3.7%), 829 Hispanic individuals (4.8%), and 15 322 White (89.6%) individuals. The median (range) age of participants was 75 (65-105) years; 8769 participants (51.3%) were female and 8338 (48.7%) were male. Additional sociodemographic characteristics and information about health comorbidities are listed in Table 1. Results of the 1:1 matching, which included 3154 participants (313 Asian to 313 White, 615 Black to 615 White, 780 Hispanic to 780 White), matched by age, sex, educational attainment, living arrangement, personal history of hypertension, personal history of diabetes, family history of dementia, and level of cognitive impairment, are included in the eResults in the Supplement. We were unable to identify matches for 8 Asian participants (2.5%), 20 The results of the logistic regression model including all participants in the full analysis set (N = 17 107) and adjusting for all matching variables are shown in Table 3. In this adjusted model, the odds of having a positive amyloid PET scan were significantly lower for Asian participants (OR, 0.47; 95%, CI 0.37-0.59; P <unk>.001) and Black participants (OR, 0.71; 95% CI, 0.60-0.84; P <unk>.001), and Hispanic participants (OR, 0.68; 95% CI, 0.59-0.79; P <unk>.001) compared with White participants. Increasing age (OR, 1.36; 95% CI, 1.30-1.44; P <unk>.001), female sex (OR, 1.20; 95% CI, 1.12-1.28; P <unk>.001), bachelor's degree educational attainment (OR, 1.24; 95% CI, 1.07-1.44; P =.004), master's degree educational attainment (OR, 1.24; 95% CI, 1.06-1.45; P =.009), living with another person (OR, 1.19; 95% CI, 1.09-1.29; P <unk>.001), and family history of dementia (OR, 1.36; 95% CI, 1.27-1.47; P <unk>.001) were all associated with increased odds of having a positive amyloid PET scan (Table 3). --- Discussion Our study of a large, national cohort of community-dwelling Medicare beneficiaries with MCI or dementia revealed lower proportions and odds of amyloid PET positivity among minoritized racial and ethnic groups. The proportion of amyloid PET positivity among Asian individuals and Hispanic individuals with MCI and dementia was 7% to 12% lower than matched White individuals and 4% lower among Black participants, although these differences were not statistically significant. Asian, Black, and Hispanic participants had lower odds (0.47-0.71) of amyloid PET positivity than White participants. With 1785 individuals who were Asian, Black, or Hispanic, this study includes one of the largest samples of individuals with cognitive impairment from minoritized racial and ethnic groups receiving amyloid PET imaging. [27][28][29] Despite disproportionately higher rates of dementia and clinical AD among Black and Hispanic populations, we found lower odds of amyloid PET positivity among Black and Hispanic participants compared with White participants. These findings may reflect differences in the etiology of cognitive impairment, such as underlying vascular disease or social factors that are impacting health. Black [30][31][32][33] and Hispanic individuals 30,34 have higher age-adjusted rates of hypertension [35][36][37][38][39] and diabetes, 40,41 which are associated with increased white matter pathology, cortical and lacunar infarcts, microinfarcts, and nonamyloid and nonvascular pathology. There are implications of our findings in the context of the recently approved disease-modifying anti-amyloid monoclonal antibody aducanumab. This drug received FDA approval for treatment of MCI or mild dementia due to AD based on lowering of amyloid PET signal as a surrogate biomarker. Lack of inclusivity of diverse populations in clinical trials for aducanumab (2 phase 3 studies included 10% Asian participants, 1% Black, and 3.4% Hispanic) along with our findings must be considered in the context of recommendations for novel therapies. 17 If diverse groups are less likely to benefit from amyloid-directed therapies and likely to experience considerable financial hardship from the associated cost, there is a risk these novel treatment options may exacerbate existing racial and ethnic disparities in dementia care. Additionally, we found more Black and Hispanic participants with dementia (vs MCI) in our study, which may have been the result of referral bias in the IDEAS study, but may also be representative of disparities in level of impairment at presentation. This has important therapy implications given those with more advanced impairment may not be eligible for novel therapies. Our findings are the result of 2 different analytic approaches, which were chosen to address group differences in sociodemographic factors and medical history (1:1 matching) and underrepresentation of minoritized racial and ethnic groups in the study (multivariable logistic regression). Be- cause there are known differences (ie, educational attainment, hypertension, and diabetes) and likely unknown or uncaptured factors (such as neighborhood disadvantage and structural racism) associated with the racialization of groups, we chose to match individuals based on variables collected that could potentially impact cognitive status and amyloid PET results. However, matching reduced our sample size by 82% (from 17 107 to 3154) leading to concerns that the matched sample size would be insufficient to detect differences. Hence we also performed a multivariable logistic regression to use all available data. Although sociodemographic and comorbidity factors are interconnected with race and ethnicity as a result of structural racism and systemic inequality within the US, 42,43 these were treated as independent variables in the logistic regression model. Although we found lower proportions and odds of amyloid PET positivity for Asian and Hispanic participants, Black participants had significantly lower odds but no difference in proportion. As noted above, this may reflect insufficient sample size after matching to detect a difference in proportion; however, it may also indicate that observed differences between racial and ethnic groups are likely associated with social factors and comorbidities. Our results are consistent with prior work demonstrating mixed findings (ie, lower rates of amyloid positivity among non-White individuals 44 vs no racial and ethnic differences in amyloid PET positivity [27][28][29] ). Our findings of lower odds and proportion among Asian participants may be difficult to interpret due to lower prevalence of clinical AD in this population and different social and cultural factors among Asian subgroups. Although rates of hypertension among Asian American individuals is lower in the US in general, 45,46 rates of hypertension are higher among some Asian groups (eg, Vietnamese and Korean individuals) 45,47 and hypertension may be underdiagnosed among Asian individuals compared to other groups. 48 Association of family history of dementia with positive amyloid PET scan may be indicative of shared genetic features that were unmeasured in this study. A family history of AD or dementia in a first-degree relative has previously been associated with increased risk of development of AD and amyloid positivity. [49][50][51] Genotyping of apolipoprotein E (APOE), the most common gene associated with risk of sporadic AD, was not performed systematically in IDEAS and is an important limitation in this secondary data analysis. The APOE <unk>4 allele has a known association with amyloid positivity, but evidence suggests this risk may vary by racial and ethnic background. 52,53 Among individuals in a study screening for an amyloid-lowering antibody, Deters et al 44 demonstrated lower rates of amyloid PET positivity and continuous amyloid levels among African American individuals with APOE4 and greater African ancestry on admixture analysis compared to APOE4-positive African American individuals with less African ancestry and APOE4-positive White individuals. Ethnic differences in single-nucleotide variants in the APOE region may also modify expression levels, thereby mitigating amyloid deposition and risk of AD in individuals with E4 positivity and Asian or African ancestral backgrounds. 54 --- Strengths Our study has several strengths. To our knowledge, it is the largest multisite study to date examining differences in amyloid PET positivity among Asian, Black, Hispanic, and White individuals with MCI or dementia. Previous studies examining racial and ethnic differences in amyloid PET positivity have largely focused on comparisons between smaller numbers of Black and White individuals, with greater representation of cognitively unimpaired older adults. [19][20][21]28,29,44,55 Recently, increased attention has been called to disparities in screening criteria that have precluded non-White individuals from being included in AD research and clinical trials, 22 which may partly explain lack of inclusion in prior research. Our study also incorporated multiple sociodemographic and comorbidity variables, allowing us to explore the association of these factors with rates of amyloid PET positivity. --- Limitations Our study also has limitations. First, despite the numbers in the study, Asian, Black, and Hispanic individuals were underrepresented in IDEAS. Although pragmatic studies like IDEAS are expected to reflect more real-world settings than traditional clinical trials, pragmatic studies also reinforce existing structural and systemic issues that limit care for minoritized racial and ethnic groups, such as lack of access and referral to dementia specialists and cost of copay for PET (PET scans in IDEAS were covered by Medicare under coverage with evidence development and copays were not covered). 9,[56][57][58] Other areas of potential bias include selection bias and reliance on appropriate-use criteria to determine study eligibility. Our study did not collect comprehensive social and structural determinants of health data, such as early life quality of education, neighborhood deprivation, and experiences with discrimination among other variables, which are associated with dementia. 7,[59][60][61][62] We matched racial and ethnic groups based on available demographic characteristics, social factors, comorbidities, and family history but were unable to match on other factors and also unable to match a small percentage (less than 6% within each group) of Asian, Black, and Hispanic participants. Although we matched on MCI or dementia, there may be differences in progression to AD across groups. We used medical history of hypertension and diabetes as matching variables in our analysis but did not have data pertaining to treatment and control, which may vary between racial and ethnic groups as a result of disparities in access to health care, biases in treatment within the medical system, and other social determinants of health. As noted above, we also did not have genotyping of APOE for any of the participants. Additionally, the IDEAS database does not include a cognitively normal comparison group and longitudinal data are not yet available. It is important to note that the categorization of individuals into racial and ethnic groups is based on social factors, not on biology or genetics. This categorization oversimplifies the tremendous heterogeneity (genetic and otherwise) that exists among members of these groups and our results should not be used to argue for some level of shared biology among members of a specific racial or ethnic group. Race itself is a sociocultural and political construct, and often serves as a proxy for social determinants of health, structural racism, and cultural and linguistic factors. --- Conclusions In this large multisite practice-based study, we found lower odds of amyloid PET positivity in older Asian, Black, and Hispanic adults with MCI and dementia compared with non-Hispanic White individuals. These results have important implications for the diagnosis, treatment, and prevention of ADRD in groups that are at the highest risk of dementia. Future research should include racially and ethnically diverse cohorts that reflect the burden of ADRD in the population at large. The recently launched New IDEAS 63 study is focused on addressing these gaps by using multipronged recruitment and community engagement strategies to evaluate the clinical utility of amyloid PET in a more diverse cohort of patients with MCI or dementia, with a specific focus on recruiting Black and Hispanic Medicare beneficiaries. --- Author Contributions: Drs Wilkins and Gatsonis had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: Wilkins, Dilworth-Anderson, Apgar, Carrillo, Gatsonis, Hill, Hillner, Siegel, Whitmer, Rabinovici Role of the Funder/Sponsor: The funders participated in the design and conduct of the IDEAS study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript. The Alzheimer's Association and American College of Radiology additionally participated in the decision to submit this manuscript for publication but did not have the right to veto submission or to require submission to a particular journal. The Centers for Medicare & Medicaid Services provided coverage for amyloid positron emission tomography scans under coverage with evidence development. Disclaimer: Dr Rabinovici is Associate Editor of JAMA Neurology, but was not involved in any of the decisions regarding review of the manuscript or its acceptance.

IMPORTANCE Racial and ethnic groups with higher rates of clinical Alzheimer disease (AD) are underrepresented in studies of AD biomarkers, including amyloid positron emission tomography (PET). OBJECTIVE To compare amyloid PET positivity among a diverse cohort of individuals with mild cognitive impairment (MCI) or dementia. DESIGN, SETTING, AND PARTICIPANTS Secondary analysis of the Imaging Dementia-Evidence for Amyloid Scanning (IDEAS), a single-arm multisite cohort study of Medicare beneficiaries who met appropriate-use criteria for amyloid PET imaging between February 2016 and September 2017 with follow-up through January 2018. Data were analyzed between April 2020 and January 2022. This study used 2 approaches: the McNemar test to compare amyloid PET positivity proportions between matched racial and ethnic groups and multivariable logistic regression to assess the odds of having a positive amyloid PET scan. IDEAS enrolled participants at 595 US dementia specialist practices. A total of 21 949 were enrolled and 4842 (22%) were excluded from the present analysis due to protocol violations, not receiving an amyloid PET scan, not having a positive or negative scan, or because of small numbers in some subgroups.In the IDEAS study, participants underwent a single amyloid PET scan.The main outcomes were amyloid PET positivity proportions and odds. RESULTS Data from 17 107 individuals (321 Asian, 635 Black, 829 Hispanic, and 15 322 White) with MCI or dementia and amyloid PET were analyzed between April 2020 and January 2022. The median (range) age of participants was 75 (65-105) years; 8769 participants (51.3%) were female and 8338 (48.7%) were male. In the optimal 1:1 matching analysis (n = 3154), White participants had a greater proportion of positive amyloid PET scans compared with Asian participants (

What this paper adds What is already known about this subject? Locked-in status refers to a workers' non-preference toward his or her job that is harmful to workers' wellbeing. However, current studies were conducted only in Western countries with the theory based on the value of individualism. --- What are the new findings? Contrary to the individually-oriented approach of interacting with the living environment (or life sphere) in Western countries, individuals in Eastern countries generally take a socially-oriented approach to interact with their life sphere. Furthermore, the concern of "face" is one of the more important concepts in an individuals' decision on their relationship with others in Chinese culture. The current study argues the need to consider a socially-oriented job preference in locked-in studies. Furthermore, study results demonstrated that participants' face needs and locked-in status were significantly associated with personal and work-related burnout. Participants with higher face needs had a significantly higher risk of developing personal and work-related burnout. Similarly, participants with a higher locked-in status had a significantly higher risk of developing personal and work-related burnout. How might this impact on policy or clinical practice in the foreseeable future? The current study recommends a culture-sensitive policy in workplace mental health program. Furthermore, there is an additional need to understand the impact of socially-oriented job preferences on job performance and workers' long-term health. --- Background Historically, being "locked-in" among employees has referred to a situation in which workers have difficulty transiting to another equitable job in labor studies [1][2][3]. The primary concept of such a situation is the nonpreference toward one's job, so the worker does not want to remain in the current workplace [3]. To remain in the non-preferred job, despite wanting to change companies or workplace, likely causes a strain on workers and leads to impaired well-being [2][3][4][5]. Though research on the health impacts of locked-in jobs is comparatively rare, the existing studies demonstrate that people in non-preferred jobs tend to report poor psychological well-being, including burnout [2]. Recently, Stengard added the element as "perceived employability" to being locked-in with the argument that a revolving situation of a non-preferred occupation frequently requires a complete career re-orientation, which is difficult. Therefore, perceived employability plays a role in the locked-in status of employees. Additionally, their studies, based on longitudinal designs, also showed the negative impact of locked-in status on well-being among workers [6,7]. However, previous studies were conducted only in Western countries with the theory based on the value of individualism. Studies have suggested that even though the benefit of a voluntary job transition such as increased salary or self-esteem was clear, employees would occasionally decide to stay in their current job despite the availability of other suitable jobs for several reasons [8,9]. With the exception of the satisfaction of their current job, the fit between work and private life and social relationships with co-workers were listed as the primary reasons for staying [10]. These findings suggest the broader work-related context, including that factors that are more indirectly connected to the job rather than the work tasks themselves, could be an important influence for remaining in a job. Contrary to the individually-oriented approach of interacting with the living environment (or life sphere) in Western countries, individuals in Eastern countries generally take the socially-oriented approach to interact with their life sphere [11][12][13]. The founder of Indigenized Chinese psychology, Kuo-shu Yang, proposed a four-part theory of the Chinese self to explain how Chinese individuals build their own identity and interact with others [14,15]. He argued that the Chinese self is composed of both the individually-oriented and sociallyoriented self, with the latter further dividing into the relationship-oriented, family (group)-oriented, and other-oriented self. According to the definition, "the individually-oriented self is a combination of a tendency toward high personal autonomy and low homonymy that emphasizes an individual's personal achievement, performance, uniqueness, and autonomy" [16], while "the socially-oriented self is a combination of a tendency toward high homonymy (with the surrounding environment) and low personal autonomy, which emphasizes that this type of self tends to attach importance in order to maintain harmonious interpersonal relationships, accountability, and responsibility, and requires appropriate personal behaviors such that individuals position themselves according to their relationship with others" [16]. As an individual raised in a collectivist culture, concerns of family and others are fundamental in a variety of individual decisions, including decisions related to work. The concern of face is one of the most important concepts in individuals' decisions concerning the relationship with others in Chinese culture [11,17]. Contrary to the concept of face in the Western culture as politeness, in Chinese culture it refers to a sociological concept that is linked to the dignity, prestige, and reputation that a person has in terms of their social relationships [18,19]. While job title and industry are related to the social recognition of workers in Eastern culture, the decision of taking a job or transferring to another job is highly connected with the concept of face of individuals [17]. In addition, Taiwan is a typical Chinese society and, as such, is strongly influenced by the traditional value of filial piety [20,21]. Many Chinese parents consider the job and career of their adult child to be an issue of face, which represents their achievement in raising their child [22]. They will, therefore, urge their adult child to take a job with a strong social reputation. With these strong cultural influences, we suggest that the socially-oriented concerns should be considered in locked-in evaluations, such that the individual's family's job preference also may play a role in workers' locked-in status in Chinese society. Therefore, we hypothesize that workers in Taiwan might take a job they do not prefer based on the concern of face of both themselves and their parents. Furthermore, the conflict of personal preference and family preference toward one's job represents the locked-in status of the individual and may lead them to develop burnout syndrome. In detail, the current hypotheses are: Hypothesis 1: The socially-oriented job preference plays an important role in workers' locked-in status. Additionally, the locked-in status with which the worker's preference conflicts with their family's preference will have a negative impact on workers' health. In other words, the higher the score of lock-in status due to conflicting preference, the higher the risk of burnout among workers. Hypothesis 2: Workers' face needs are positively related to their locked-in status, represented by workers' preferences and the preference of their family. In other words, workers with higher score of face needs have a higher risk of being locked-in job. In Taiwan, work in high-technology industries are considered a fashionable job with a potentially high salary and good career future [23]. Additionally, working for a big company is considered to be a good job due to the presumed job security and popular recognition among society, though the working conditions, including salary and work hours, might sometimes be worse than in smaller companies [24]. Therefore, we conducted the current study on high-tech company workers. --- Methods The current study was conducted using an anonymous self-report questionnaire given to workers at three hightech companies in Taiwan from October 2015 to January 2016 to assess their face needs, locked-in status of their job, and burnout. Those companies were all in electronic industry which producing computer-related products. And two companies had more than 500 workers while another one had around 300 workers. The questionnaire was delivered to 2341 workers and 1116 returned their questionnaire surveys (response rate of 47.68%). After excluding 14 questionnaires with more than 10 invalid response items, 1102 questionnaires were included in the final analysis. The research protocol was approved by the Institutional Review Board of Taipei Medical University (No. N201805098). While it is an anonymous survey, only participants who were willing to participate in the study would return the questionnaire. While the research involved survey without identifiers, answering the survey served as implicit consent. So informed consent was taken from all participants. All methods were carried out in accordance with Declaration of Helsinki. --- Measures --- Independent variables Work-related individual characteristics, including job content (engineer, administrator and others), age (from 22 to 64 years old), gender (male/female), education level (high and professional school/college/master and above), marital status (single/married or with partner/others), work experience (years), work level (administrative level or not), work hours per week, and company size (above 500 persons/ less than 500 persons) were all collected for the analyses. --- Locked-in status The workplace preference scale from Stengard's lockedin questionnaire was used in and modified for the study [6,7]. Workplace preference was measured by questions such as "Is your current occupation the occupation you expected?"; "Is your current occupation the occupation you wish to have in the future?"; "Is the company you work for today the company you expected?"; and "Is the company you work for today the company you want to work for in the future?" The socially-oriented questions of locked-in status was added as "Is your current occupation the occupation your family (parents, spouse, and relatives) expected?" and "Is the company you work for today the company your family expected?" The responses were evaluated on a 4-point Likert-scale, ranging from 1 = very much disagree to 4 = very much agree. Information of perceived employability from Stengard's locked-in questionnaire was also collected for analysis. The perceived employability was measured by a question as "I don't need to move or rent another additional house for finding another similar work. I can easily find another similar work like the current one." --- Face needs For evaluating the construct of face needs, we used the "face needs questionnaire" developed and validated by a Chinese psychologist [25][26][27]. Cronbach's <unk> was 0.84 for our study. The questionnaire of face covered two dimensions: (1) concern about face of one's self (14 items) and (2) concern about face of others (seven items). The detail information of the items was provided in the supplementary file. The responses were evaluated using a 5-point Likert-scale ranging from 0 = very much disagree to 5 = very much agree. The higher scores represent a higher face need of participants. --- Burnout Burnout was used as a health outcome indicator in the current study. The Chinese version of the 21-item Copenhagen Burnout Inventory (C-CBI) was used to measure the three domains of burnout: personal burnout, work-related burnout, and client-related burnout [28]. In the current study, only the personal (five items) and work-related burnout (five items) items were measured among workers. Per the CBI, personal burnout is defined as "the degree of physical and psychological fatigue and exhaustion experienced by the person", while work-related burnout is "the degree of physical and psychological fatigue and exhaustion perceived by the person as related to his/her work" [29]. The responses were evaluated using a 5-point Likert-scale from 0 = never to 5 = always. The C-CBI was validated by Taiwaness researchers [30]. And the Cronbach's <unk> was 0.93 for our study. We used expert review approach to obtain the content validity of all the variables and measure except standardized questionnaire as Face-needs and burnout. Six related experts were invited for the review. And the Item-level Content Validity Index (I-CVI) of each item was above 0.78 and used in the study. --- Data analyses Two approaches were used to calculate the locked-in status of participants. The scores of participants' preference and their family's preference of the job were summed to comprise the locked-in scores. Then, the locked-in scores were divided into three groups for analysis. Participants were also categorized by the conflict of the preference of the job between themselves and their family as "locked-in job conflict of preference between themselves and their family" status. Participants were divided into four groups as "both prefer", "self prefers but family does not", "self does not prefer but family does", and "both do not prefer". Three different approaches were used to perform the analyses. For summary statistics, a chi-square test was used to compare individual characteristics including age, gender, educational status, marital status, job title, work level, work year, working hours per day, and company size by "locked-in job conflict of preference between themselves and their family" status. An ANOVA was used to compare the means of face needs, including face of self and face of others, burnout, including personal and work-related burnout, and self-evaluated employability between "locked-in job conflict of preference between themselves and their family" status. A Pearson correlation analysis was used to evaluate the association between face needs and locked-in status. Personal burnout and work-related burnout were defined as dichotomous variables with the cut-point as the mean. A logistic regression was then adopted to estimate the relationship between face needs, locked-in status, and burnout after adjusting for confounding variables. Variables that were significantly associated with personal and work-related burnout in the bivariate analysis including age, gender, educational status, marriage status, job title, work year, number of hours worked per day, company size, and self-perceived employability were adjusted for in the regression model (supplementary). The regression model was also run to evaluate the associations between face needs, "locked-in job conflict of preference between themselves and their family" status, and burnout. The odds ratios (OR) and related 95% confidence intervals (CIs) were also calculated. The significance level was set at 0.05. All analyses were performed using SPSS, Version 18.0. --- Results --- Comparison of individual characteristics between "lockedin job conflict of preference between themselves and their family" status groups The comparison of individual characteristics between "locked-in job conflict of preference between themselves and their family" status groups using a chi-square test are shown in Table 1. In general, majority of the participants were single (57.17%), male (59.44%), aged 30 to 39 years old (53.36%), and had obtained an educational degree above college (97.83%). Moreover, 68.78% of the participants were engineers without an administrative position. While 48.55% of participants had only worked for less than 5 years, 54.72% of them worked over 40 h per week, and 83.58% of them worked for a large company with more than 500 employees. Among the 1102 participants, 18% had a job that they did not prefer but their family preferred and 9% of them had a job which neither themselves nor their family preferred. Educational status, marriage status, and company size were significantly different between "lock-in job conflict of preference between themselves and their family" status groups. Those with an educational level at Master's degree and above comprised the largest group in the "self does not prefer but family does" group (55.72%), followed by the "both prefer" group (54.23%), the "self prefers but family does not" group (44.12%), and the "both do not prefer" group (41.84%). Participants who were married were most represented in the "both prefer" group (43.56%), followed by the in "both do not prefer" group (40.82%), the "self prefers but family does not" group (38.24%), and finally in the "self does not prefer but family does" group (35.32%). The percentage of participants who worked in big company was the highest in the "self does not prefer but family does" group (88.56%), followed by the "both prefer" group (84.92%), the "both do not prefer" group (68.37%), and the "self prefers but family does not" group (67.65%). The other factors were not significantly different between the groups. Comparison of face needs, burnout, and self-perceived employability between "locked-in job conflict of preference between themselves and their family" status groups Comparisons of face needs, burnout, and employability between "locked-in job conflict of preference between themselves and their family" status groups using an ANOVA are shown in Table 2. Face needs, including concern of face of one's self and concern of face of others, burnout, including personal burnout and workrelated burnout, and self-perceived employability were all significantly different between groups. The mean of face needs was highest in the "both prefer" group (70.7), then in the "self does not prefer but family does" group (70), then in "both do not prefer" group (67.59), and finally in the "self prefers but family does not" group (66.68). The concern of face of one's self was highest in the "both prefer" group, then in "self does not prefer but family does" group, then in the "both do not prefer" group, and then in "self prefers but family does not" group. Contrarily, the mean of concern of face of others was the highest in the "self does not prefer but family does" group, then in the "both prefer" group, then in the "both do not prefer" group, then in the "self prefers but family does not" group. Personal burnout was the highest in the "both do not prefer" group, then in the "self does not prefer but family does" group, then in the "self prefers but family does not" group, then in the "both prefer" group. Similarly, work-related burnout was the highest in the "both do Table 1 Comparison of Individual characteristics between lock-in job for conflict between themselves and family preferences by chisquare test not prefer" group, then in the "self does not prefer but family does" group, then in the "self prefers but family does not" group, then in the "both prefer" group. Self-perceived employability was highest in the "both prefer" group, then in the "self does not prefer but family does" group, then in the "both do not prefer" group, then in the "self prefers but family does not" group. The Pearson correlation analysis showed that participants' face needs are not correlated with their locked-in status as evaluated by both approaches. --- Associations between face needs, locked-in job status, and burnout The associations between face needs, locked-in job status, and burnout when using a regression are shown in Table 3. Workers' marital status, job title, hours worked per day, and their company size, self-perceived employability, face needs, and locked-in job status were all significantly associated with personal and work-related burnout. For model 1, the tertile locked-in scores were used for the analysis and marital status, job title, company size, face needs, and locked-in status were all significantly related to personal burnout. Participants with a marriage status marked as other (divorce or widowed) and job title marked as other had a significantly higher risk of developing personal burnout than single participants or those whose job title was "other" (i.e. not an engineer or in administration) (OR = 10.93 and OR = 1.83, respectively). Participants who worked in a big company had a significantly lower risk of having personal burnout compared to participants who worked in smaller sized companies (OR = 0.54). Importantly, participants with high face needs had a significantly higher risk of developing personal burnout (OR = 1.91). Moreover, participants with a high locked-in job status had a significantly higher risk of having personal burnout (OR = 1.75), while participants with a middle locked-in job status had a four times greater risk of having personal burnout (OR = 3.96). Participants who worked for a big company had a significantly less chance of developing work-related burnout (OR = 0.23), while those with high face needs had a significantly higher risk of developing work-related burnout (OR = 1.68). Moreover, participants with a high locked-in job status had a significantly higher risk of having personal burnout (OR = 1.82) and participants with a middle locked-in job status had a four times greater risk of having personal burnout (OR = 4.49). For model 2, which used "locked-in job conflict of preference between themselves and their family" status for the analysis, marriage status, job content, company size, face needs, and "locked-in job conflict of preference between themselves and their family" status were all significantly related to personal burnout. Participants with a marriage status listed as other (i.e. divorced or widowed) and job content described as other had a significantly higher risk of developing personal burnout than single participants (OR = 10.01 and OR = 1.73, respectively). Participants who worked in a big company had a significantly lower risk of having personal burnout than participants who worked in smaller sized companies (OR = 0.55) and participants with high face needs had a significantly higher risk of developing personal burnout (OR = 1.88). Moreover, participants with a job status of "self does not prefer but family does" had twice the risk of having personal burnout (OR = 2.03), while participants with a job that both themselves and their family do not prefer had a four times greater risk of having personal burnout than those with a job that both themselves and their family prefer (OR = 4.10). Participants who worked in a big company had a significantly less chance of work-related burnout (OR = 0.25), while participants with high face needs had a Table 2 Comparisons of face need, burnout and employability between lock-in job for conflict between themselves and family preferences status by ANOVA significantly higher risk of developing work-related burnout (OR = 1.62). Moreover, participants with "locked-in job conflict of preference between themselves and their family" status of "self does not prefer but family does" had twice the risk of having work-related burnout (OR = 2.34) and participants with a job that neither themselves nor their family preferred had a four times greater risk of having work-related burnout (OR = 4.17). --- Table 3 Association between face need, lock-in job and burnout --- Discussion To the best of our knowledge, the current study is the first to aim at evaluating the impact of special concerns in the Chinese culture, namely the concept of face on locked-in status and burnout among workers. Also, this is the first study to suggest the need to consider the socially-oriented job preference in locked-in studies. The current results demonstrated that participants' face needs and locked-in status were significantly related to their personal and work-related burnout. Participants with higher face needs and, independently, higher locked-in status had a significantly higher risk of developing personal and work-related burnout. The study result confirmed our hypothesis 1 and hypothesis 2. As confirmed by the results of the current study, it is not surprising that participants with a higher locked-in status showed a significantly higher risk of developing personal and work-related burnout. And the study finding was similar with Stengard's and Fahlen's studies [5,6]. Interestingly, however, participants with a middle locked-in status had the highest risk of developing personal and work-related burnout. Their risk is even higher than workers with a high locked-in status. This phenomenon may be due to the fact that two thirds of our participants whose preference was in conflict with their family were categorized in the middle locked-in group. The analysis using "locked-in job conflict of preference between themselves and their family" status showed a more coherent result for the risk of developing burnout syndrome among workers. Participants with a job status as "self does not prefer but family does" had twice the risk of having work-related burnout, while participants with a job that neither they nor their family preferred had a four times greater risk of having workrelated burnout. These result suggest that workers' preference toward their job played the primary role in protecting their health. Even if the worker has a high socioeconomic status job that everyone admires, if s/he does not like the job, their feeling of being trapped at work will lead them to develop burnout syndrome. Additionally, these results support the notion that sociallyoriented concerns should be considered in locked-in evaluations in Chinese society. These finding may also generalize to other Asian countries with a patriarchal culture. And we suggest to conduct relevant comparative studies in multi-ethnic Western countries in the future in order to further understand the impact of conflict between individual and their family regarding their job choice on workers' health. In the meantime, we suggest the introduction of mental health program regarding the issue in the workplace in order to reduce the negative impact of such issue on workers' health. The current results demonstrated that participants with higher face needs had a significantly higher risk of having personal and work-related burnout. Although the correlation between face needs and locked-in status was not demonstrated, workers who stayed in a job which "self does not prefer but family does" were those who tended to have high face needs. There were 18% participants in our study who were locked into the job that they did not prefer but their family preferred. Given these participants had the highest concern of face of others, there is a high possibility that they stayed in a job they did not prefer in order to preserve face of their parents or families. In other words, those workers who had high face needs and a high concern of face of others were locked in job due to their concern of face. As such, workers who had high face needs demonstrated a significantly higher risk of developing personal and workrelated burnout. Interestingly, participants (69.78%) who reported that both themselves and their family prefer their job also reported similarly high face needs compared to the workers stayed in their job in which they did not prefer but their family did. Interestingly, even though the percentage of those who were married and the percentage of those who worked in smaller companies were higher in the "both prefer" group, the socio-economic background of those two groups was similar. The majority of those two groups were comprised of single young men with an educational level of a Master's degree or above who were working as engineers. In addition, nearly 50% of them had work experience of less than 5 years. It is plausible that participants in "both prefer" group had convinced themselves that they liked the job because they felt proud about their good socio-economic status and enjoyed the feeling of "having face" for themselves and their family. Additional research is needed to understand the reason behind this phenomenon. Several limitations were noted in the current study. First, the response rate was comparatively low. The nonresponders may be the workers with higher levels of work stress and there may be an under-reporting of the locked-in status among high-tech company workers. Secondly, relationships should only be considered as correlational rather than causal due to the crosssectional design. Finally, it is important to not discount a potential reporting bias due to the self-report study design. --- Conclusion In conclusion, the current study indicated the importance of considering the socially-oriented job preference in locked-in status evaluations in Chinese society. Both participants' face need and locked-in status were significantly associated with their personal and work-related burnout. Participants with higher face needs had a significantly higher risk of developing personal and work-related burnout and workers' locked-in status who had a conflicting preference compared to that of their family demonstrated a significantly negative impact on their health. Additional studies are recommended to understand the impact of the socially-oriented job preference on job performance and workers' long-term health. --- Availability of data and materials The datasets used and/or analyzed for the current study are available from the corresponding author upon reasonable request. --- Abbreviations High-tech: High technology; CBI: Copenhagen Burnout Inventory; OR: Odds ratio; CIs: Confidence intervals --- Supplementary Information The online version contains supplementary material available at https://doi. org/10.1186/s12889-021-11269-8. Additional file 1: Supplementary 1. Detail information of the "Face Need Questionnaire". Supplementary 2. Correlation table for variables by Pearson analysis. --- Authors' contributions FJT carried out the design of the study, performed the statistical analyses, and drafted the manuscript. RYC participated in the design of the study, data collection, and the drafting of the manuscript. HJC participated in the design of the study, data collection, and the analysis of the data. All authors have read and approved the manuscript. --- Declarations --- Ethics approval and consent to participate The current research protocol was approved by the Institutional Review Board of Taipei Medical University (No. N201805098). While the research involved survey without identifiers, answering the survey served as implicit consent. So informed consent was taken. All methods were carried out in accordance with Declaration of Helsinki. --- Consent for publication Not applicable. --- Competing interests The authors declare they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Studies on the health impacts of being locked in a job are primarily conducted in Western countries, with the theory based on the value of individualism. But the socially-oriented concerns should be considered in workers' locked-in status in Chinese society. So the current study aims at evaluating socially-oriented concerns on workers' locked-in status in Taiwan. Methods: Anonymous surveys were conducted with 1102 workers at high-tech companies in Taiwan from October 2015 to January 2016 to assess their "face" needs--a sociological concept linked to the dignity, prestige, and reputation that a person has in terms of their social relationships, locked-in status of the job, and burnout. In addition to being separated into three groups by lock-in score, participants were categorized by the conflict of preference of the job between themselves and their family. Chi-square, ANOVA, Pearson correlation, and regression tests were conducted. Results: Among the 1102 participants, 18% had jobs that they did not prefer but their family preferred. Participants with higher face needs and higher locked-in status had a significantly higher risk of developing personal and workrelated burnout. However, the analysis using "locked-in job conflict of preference between themselves and their family" showed a more coherent result. Participants with a job which "self does not prefer but family do" had twice the risk of having personal and work-related burnout (OR = 2.03 and 2.34, respectively). Participants with a job which neither themselves nor their family prefer had four times the risk of having personal and work-related burnout (OR = 4.10 and 4.17, respectively). Conclusions: The current study suggests an importance in considering a socially-oriented job preference in lockedin status evaluations within the Chinese culture. Workers' whose locked-in status preference conflicted with their family's preference showed a significantly negative impact on their health.

Introduction Telehealth has been a primary approach internationally for maintaining health care during the COVID-19 pandemic [1,2]. The UK implemented personal distancing measures on the 16 th of March, 2020, and National Health Service (NHS) mental health care providers responded with extensive adoption of telemental health, including video calls and phone calls, to maintain service delivery in conditions of lockdown, and then subsequently with social distancing measures in place. Telemental health adoption was somewhat piecemeal, with policies and platforms varying between services and provider Trusts, but most community services shifted to a majority of patient contacts being remote, but with some face-to-face contact continued where deemed essential, especially in crisis care settings (7) Huge increases in remote patient consultations have subsequently been seen in NHS primary care and mental health settings with a surge in patients' uptake of remote health services, including registrations for the NHS App, NHS login and e-prescription services [3,4]. As the pandemic continues, mental health services have faced severe disruption, with increased demand alongside reductions in capacity and infection control requirements [5,6]. It is in this context that the rapid and widespread adoption of remote delivery of mental health services (telemental health) has been central to maintaining assessment, treatment and support in community, hospital and residential settings [2,7]. Telehealth has been defined as "the delivery of health-related services and information via telecommunications technologies in the support of patient care, administrative activities, and health education" [8]. Before the pandemic, numerous research studies found evidence of telemental health effectiveness in reducing treatment gaps and improving access to care across a wide range of populations, settings and conditions [9][10][11][12][13][14][15]. Overall, the evidence suggests that synchronous modalities such as videoconferencing have appeared comparable, or better, in research contexts than face-to-face delivery in terms of quality of care, reliability of clinical assessments and treatment outcomes and adherence [10,[16][17][18]. The attitudes of clinicians who have experienced synchronous telemental health in research settings also appear to be largely positive, with professionals finding it both effective and acceptable [19]. There is also positive health economic evidence, with several studies suggesting telemental health is no anonymity of participants. If you wish to obtain access to this data, please contact the UCL ethics committee on [email protected] and/or the corresponding author. more expensive than face-to-face delivery and tends to be more cost-effective [17]. The use of remote consultations appears to be welcomed by many service users, who find it as satisfactory as face-to-face alternatives [2,15,20,21], and in some cases preferable [22], even where they have initial reservations or limited experience of using computers [23]. However, studies assessing telemental health care tend to report small-scale and carefully planned interventions with volunteer participants, rather than large-scale emergency implementations, as in the current crisis [21]. The rapid, highly variable and often piecemeal adoption of telemental health modalities during the pandemic [7] has highlighted a range of very significant challenges, risks and implementation barriers. These include deterioration in the quality of care received by service users with certain mental health conditions; digital exclusion of those with limited technological access or expertise, or those facing significant social disadvantage, potentially exacerbating inequalities that already exist; lack of technological infrastructure and clear protocols within services; difficulty in establishing and maintaining therapeutic relationships; problems with conducting high quality assessments; and service users who lack private space or find participating in sometimes intimate and distressing discussions at home intrusive [2,[24][25][26]. A range of other ethical, regulatory, technological, cultural and organisational barriers to wider implementation of telemental health and its integration with routine face-to-face care have also been identified, both before and during the pandemic [17,[27][28][29][30]. Following the rapid shift to remote care during the pandemic, many policy makers, service planners, mental health professionals and service users have advocated for further evidence to inform the continued use of these technologies in the longer term [7,[31][32][33]. In order to achieve and sustain effective, integrated and acceptable implementation of telemental health approaches, and to identify the situations for which these are not appropriate, we need to understand how to optimise remote healthcare with a population that presents unique relational challenges associated with mental distress. This requires a nuanced understanding of the impacts of the rapid shift towards telemental health on clinical practice, quality and safety of care, equitable access to services, and the experiences of service users, carers and staff. The current study focuses on the experiences of mental health service users already living with mental health conditions prior to the pandemic. While people already living with and receiving care for mental health problems have been active in writing personal accounts of the pandemic [2], few studies have systematically explored their experiences [26]. We report on a rapid collaborative framework analysis of service user interview data on views and experiences of telemental health during the pandemic, focusing especially on the following research questions: a. What are the experiences, from a service user perspective, of the switch to telemental health care? b. Which factors facilitate or impede people engaging with remote contact? c. What are the advantages and disadvantages of specific remote care tools, in different setting and for different populations? d. How do service users envisage the future for remote interactions in mental health services? --- Methods Our study is a collaborative framework analysis [34,35] of relevant material from interviews with people with pre-existing mental health conditions conducted as part of a broader study exploring loneliness and mental health during the COVID-19 pandemic [26]. Ethical approval was obtained from the UCL Research Ethics Committee on 19/12/2019 (ref: 15249/001) and an amended topic guide covering experiences of COVID-19, including telemental health, was approved on 14/08/2020. This paper reports findings from the second wave of interviews which took place in September-October 2020. Both the data collection and the current analysis used a participatory, coproduction approach [36]. The research team included mental health service researchers, lived experience researchers (LERs) with personal experience of using mental health services and/or supporting others who do, and mental health clinicians, many with two or more of these roles. We conducted weekly team meetings to discuss the analysis methods and results, and collectively wrote this paper. This work is part of the National Institute for Health Research Mental Health Policy Research Unit's programme in response to the COVID-19 pandemic, agreed in discussion with policy makers including Department of Health and Social Care and NHS England officials. --- Sampling and recruitment strategy Initial recruitment took place between 7 th May and 8 th July 2020. We included adults living in the UK who identified as having had experiences of mental health difficulties that had begun prior to the pandemic. We recruited conducted via relevant community and voluntary sector organisations and networks, and via social media, with support from the Mental Elf blogger. In this paper we report findings from the second of two waves of interviews which was conducted in September-October 2020: this second interview included detailed questions about telemental health, as it was identified as a high priority area for data collection following discussions with policy makers and other stakeholders. Purposive sampling, reflecting characteristics relevant to the research questions, guided participant selection (diagnosis, use of mental health services, gender, ethnicity, sexual orientation, and rural/urban setting). We reviewed our sample during recruitment and implemented targeted strategies to ensure diversity, such as approaching organisations with a specific remit to work with Black and Minority Ethnic communities and the use of targeted recruitment materials in an attempt to increase participation amongst these communities. Potential participants were sent an information sheet and given the chance to ask questions, then formal consent was taken and recorded before the interview. Those who gave informed consent and participated in an initial interview were approached again approximately three months later and asked if they would like to take part in a follow-up interview. Participants gave audio-recorded verbal informed consent to being interviewed a second time. --- Data collection Interviews were conducted by a team of LERs, including PS, PN and BC and KM (lived experience commentary authors, see Box 1), using videoconferencing or freephone options within the Microsoft Teams application. A second study researcher was present to support recording and save the interview in password protected university folders. Consent was audio-recorded before the interview. Interview audio files were then transcribed verbatim by a transcription company and all personal identifiers were removed for analysis. Interviewers were generally paired with the same interviewee for the first and second wave of interviews. Other than this, there was no relationship between interviewers and interviewees prior to the interview and they corresponded only in order to arrange a time for the interview and answer any questions about the study. University researchers provided training for interviewers on the practicalities of conducting interviews on Microsoft Teams, obtaining verbal informed consent, dealing with distress and content of topic guide through a 2-hour online workshop. Additionally, a weekly peer-facilitated reflective space provided LERs with emotional support and space to discuss the research process. Interviews were guided by a semi-structured interview schedule (see S1 Appendix). A previous study on loneliness, emerging literature on COVID-19 and mental health [2,7], and lived experience within the research team informed the topic guide. It included a detailed exploration of experiences and views about remote contacts and questions relating to acceptability, preference (current and future), and differential experiences across modalities and clinical settings. --- Analysis We conducted a two-stage collaborative framework analysis [34,35]. This was an iterative approach with team discussions, including LERs, clinicians, and researchers at each stage to Box 1. Lived experience commentary written by Beverley Chipp and Karen Machin on 24/01/2021 Our commentary encompasses observations from several Lived Experience Researchers. Common concerns were exclusion and choice. Person-centred care means everyone should be offered choices about whether to accept digital options, with shared decisionmaking, regularly reviewed without assumptions. Interviews commenced at a point in the pandemic when there were few alternatives for support. Gratitude for any contact under the circumstances, and satisfaction within this context should not be used to justify any narrowing of choice in future provision. The rapid adoption of digital shows how swiftly services can change, but any longerterm future policy should avoid making remote consultations a default option. We hear concerns that decisions are driven by budgetary, particularly estates, considerations, and that clients' needs may be secondary, with requirements to meet certain criteria before face-to-face options are offered. Hidden costs such as clinician time lost addressing technical challenges remain unaddressed. Future research is urgently needed on safety and risks of video calls, particularly with group-work addressing self-injurious events impacting on others, and complexities of support. The quality of therapeutic relationships should also be considered. Choice, including how it is supported and encouraged, needs scrutiny. Hybrid approaches could be explored. The needs of people who are digitally excluded and consequently unable to take part in this study need to be understood. Initiatives should not exacerbate existing digital inequalities such as incompatibility with assistive technologies, poor Wi-Fi in residential facilities or inequalities considered under the Equalities Act. Co-produced research can bring insights into emerging experiences from a grass-roots level. For example, community and peer support is under-represented in this paper, which focuses more closely on statutory services provision and existing literature emphasising 'psychotherapy' and 'treatment', whereas many people relied upon and provided support for their community over the pandemic. Survivor/service user led research may start with different questions ensuring studies are ahead of the curve at times when agile responses are required for unprecedented situations. enhance reflexivity and develop consensus on coding approaches and development of themes/ topics. In the first stage, interview data relevant to the research question was selectively extracted and indexed. We identified data from the interviews which referred to service users' experiences and perceptions of telemental health care, including data relating to: a. Remote provision of support through video-calls, phone and any messaging means. b. Support provided by, or facilitated by, any organisation (NHS, social services, voluntary sector or community organisation, excluding unmediated support such as apps). c. Support provided by both paid staff and volunteers, including peers (but excluding entirely informal support provided, for example, by family and friends) For this data extraction, a set of broad categories/codes relevant to telemental health care was developed through iterative team discussions. These categories were based on preliminary data screening, interviewers' familiarity with the data, and our and other groups' previous work on this topic. In this way, we generated a thematic coding framework which included deductively and inductively derived categories. A data extraction form was developed using Opinio [37] and an initial set of five interviews was coded to assess the specificity and adequate fit of the categories to the data. Based on this initial test, the coding book was refined through group discussion, and RA, PN, MS, PS, MB, UF, NL and LSR proceeded to index interview data under the pertinent categories. An "other" category was included to index relevant data that did not fit in any of the existing categories. The full list of categories is presented in Table 1. At a second analysis stage, a smaller group of researchers (RA, PN, PS, MS and NV) reviewed the data collected within each category and synthesised the key emerging topics/ themes. Emerging themes were identified through iterative cycles of group discussion and returning to the data, and related themes were grouped under headings/dimensions until a refined coding framework was developed. Codes were mapped across the data to report on variations between respondents. The results were then shared and discussed with the broader research team, giving each member a chance to reflect on key topics and assess whether the refined set of codes encompassed all the relevant data. Following this discussion, the smaller group of researchers further refined the results and selected quotes from interview transcripts. --- Results --- Sample Of the forty-nine people who participated in the first wave of interviews and agreed to take part in a follow-up interview, one did not take part due to work/home demands, one preferred not to take part, and three could not be reached. Forty-four participants took part in the follow-up interviews analysed in this study. The interviews took place between 1st September and 14 th October 2020. A majority of the sample was female (N = 28, 63%), aged between 26-55 years (N = 33, 75%), and living in urban settings (N = 35, 80%) across England. The main ethnic groups were White/White British (N = 28, 63%), Black/Black British (N = 6, 14%), and Asian/Asian British (N = 6, 14%). The majority (N = 33, 75%) reported current or recent mental health service use (this was during periods of lockdown or social distance regulations in the UK in 2020), mainly of NHS community mental health services (N = 22, 50%). Table 1 presents further characteristics of the participants. We report emerging topics under four headings: (1) Varying settings for telemental health, (2) What works for whom: experiences and preferences, (3) Patient safety and privacy, and (4) Participants had varying preferences for the method of remote communication depending on the context and nature of clinical contact. Some preferred phone calls to video calls because they felt less intrusive, whilst others preferred video calls because they allowed both the service user and the therapist to pick up on visual cues. A few preferred email or text as they found written forms of communication easier and sometimes thought it helped with managing privacy issues. I never saw the face of the person that I was chatting with, or even heard their voice, but we did build up a kind of professional relationship through messaging each other over the message service thing. And you could see the records of the conversation, which I think was helpful. [P31: female, White Asian, 18-25, town] Most favoured using text-based communication only for scheduling appointments or requesting medication repeat prescriptions. This indicated a distinction between 'functional' and'relational' appointments: contacts to exchange practical information with less room for ambiguity (functional), vs. contacts to express states of mind and interact (relational). With something like a therapeutic thing, there is so much that goes on-I don't know, with body language, eye contact, developing empathy, you know, being comfortable with silences, that doesn't work if you are using Zoom or something. [...] in a therapy type thing, there is something just very sort of special about being in that room. In a room like that, it has got to sort of... I mean, it sounds a bit ridiculous, and it is exaggerated to say sacred, but it is a veryit is a special thing. [P40: male, White British, 26-35, small town] Phone appointments are okay with the care co-ordinator, but not so brilliant with the psychiatrist, [...] because there's more to talk about with the psychiatrist, it's better to either see them in person or see them by video... [P25: female, African, 46-55, city] I think there are a lot of things, like medication repeats where they haven't been set up as an automatic repeat or dosage changes, discussions. A lot of things like that, that I think could just be dealt with, with a telephone call. [P22: male, White British, 26-35, town] Not seeing staff's body language and other social cues generally made participants feel less connected and overall has a negative impact on participants' experiences of relational appointments. For online support groups run without live video, this led to reduced engagement and a lack of connection with other group members. Because of autism, I sometimes find it hard to focus on what people say, unless there is some sort of visual connection to it. [...] The ideal would be if everybody could just have subtitles built in. But the second best to that is being able to see somebody talking as they speak. [P27: female, White British, 26-35, city] Participants who started mental health support or changed their therapist/psychiatrist during the lockdown said that the relationships with the new care provider often felt less personal compared to face-to-face. This was particularly the case for initial assessments over the phone, as participants found it difficult to convey how distressed they were and felt that this affected the care they were offered afterwards. Conversely, familiarity with health providers facilitated remote contact. Since I have always managed to get to see the same GP... there is an established relationship. So it doesn't matter quite as much if I am not able to see him face to face. [P40: male, White British, 26-35, small town] Subtheme 1.2: Barriers and facilitators to remote access. Participants appreciated having access to remote care during the pandemic because it allowed them to continue with their mental health care when there was no other option. Some also wanted this mode of care to remain in the future, for example, participants who struggled to travel due to mobility issues or anxiety. --- Yes, it's [video calls] better for me, because I've got mobility issues, it means that I don't have to travel. So, sometimes... I would miss appointments because I just didn't have the get-upand-go to leave the house, or I'd have a panic attack at the front door or something. [P9: female, other ethnicity, 36-45, big city] The switch to remote care facilitated access to different forms of support for some people who would have otherwise not been able to receive care, such as those admitted to mental health wards or participants living in remote areas that have only limited services available. I feel like it is only because of the fact that it is virtual now that the people can do this. A lot of members are signing on from hospital beds and retreats for people who are suicidal and things like that. [...] There was one member who was in a general hospital [...] It would not have been possible for him to come to the group had it been a face-to-face group. [P29: female, African, 26-35, city] I've actually made more friends in the lockdown, through groups that I wouldn't have joined otherwise. (...) Online communities, and the group therapy is a local group. So I guess I've got more of a sense of community there. But I just literally wouldn't be able to go to all these groups if they weren't online. [P8: male, White British, 36-45, city] However, some participants did not have the necessary technology, internet connection, or private space to receive remote care. These factors increased the stress and difficulty of engaging with remote care, and in some cases resulted in users having to terminate their therapy sessions completely, or not being able to access remote care at all, leading to some participants feeling abandoned by services. The private therapist I see on Zoom, which is actually okay, but when I'm in the crisis house, the Wi-Fi there is non-existent. [P42: female, Asian, 18-25, Additionally, there were multiple examples of contexts in which people reported not being able to cope with technology, such as, when they were feeling severely unwell; experiencing paranoia and/or distress; and when there was lack of trust and interpersonal connection with service providers. --- Because with accessing healthcare I find the phone very ineffective, and it brings a lot of its own problems, when I get unwell I really can't cope with technology anyway. I will just wrap my phone in foil and put it in the garden when I am not feeling well, just get rid of it until a later date when I feel able to unwrap it. [P44: female, White British, 36-45, town] Access to services was particularly impeded when service providers were inflexible regarding the type of remote technology used, for example, insisting on using platforms only accessible with laptops or smartphones rather than phone calls. --- Some people don't have the technology and, you know, are missing out, and nobody calls them. [P48: female, White/Mixed, 46-55, city] Flexible service delivery and adaptability were key to reaching different populations and providing personalised care. Those who were not offered a choice regarding the delivery of care or the opportunity to have video-call voiced disappointment and perceived care provision as inequitable. --- So twice a week phone calls, for as long as we couldn't see each other, but if I chose not to have one of my sessions, she would add it onto what I get face to face at the end. And, to be honest, that made it-I mean, I appreciate her flexibility in that. [P1: female, White British, 36-45, city] Again, it's like a postcode lottery, what area you live in [... influences] what support you get. To me, that's not okay. So, restricts people depending on what practice you use and what support you can get. [P4: female, White British, 26-35, city] Many participants reported not having received support or information to help them engage with remote care. This was an issue particularly for older service users who did not have the necessary devices or knowledge to use technological devices or were not offered the option to use services that they were familiar with. I wanted to get to use a computer [...]. And it said on the leaflet, "We run computer courses." So, I go along to the head office and it says, "No, we are not open. Go round the corner and walk half a mile and come to our other office." I walked all the way there and it said, "We are not open." [...] It has really put the kibosh on a lot of things that people would like to do. They just aren't really terribly available." [P13: male, White British, 66-75, London] I have certainly been in that kind of place where I have just wanted to text somebody, but I have never seen them available for my age group, to be honest. I suppose maybe the younger generations when they feel suicidal maybe they hit the computer. I don't know. Maybe that is the thing for them, but maybe it might not be so for me. [P13: male, White British, 66-75, city] There were examples of service providers helping service users to access and adjust to remote care. For instance, some service users received information sheets and support from their therapist on how to use technology and switching to remote care. Other examples of service strategies to facilitate access to care were appointment reminders and charities promoting their services and reaching out to people. --- A week before your appointment, she would text you and say, "We're due to meet next week at this time." She did it off her own back. [...] I think that's really good that she does that because it shows she's thinking ahead, "I'm due to see you," and for people that don't remember-[...] that shows to me she's aware of who she's going to see, when they're coming and she's got you in mind. [P4: female, White British, 26-35, city] 2. Varied settings for telemental health. Table 2 presents the range of remote care tools used by participants. These included video and phone consultations, phone helplines, text helplines, and email appointments. Table 3 presents the mental health care services that participants were in contact with, such as NHS community mental health services, inpatient services and private sector psychotherapy. Some participants described experiencing difficulties to stay connected with community mental health services. Many participants relied on services other than secondary mental health services, such as their General Practitioner (GP), charities, crisis helplines and/or online support groups which offered a variety of new forms of remote contact that were seen as broadening choice and service availability. Overall, participants reported having received high quality care, information, and resources from these service providers and identified them as providing valuable support. --- The community mental health team that I've been under haven't been brilliant over lockdown. I obviously know we can't see each other in person, but it's literally gone completely quiet for most of the six months [...]. I'm supposed to hear from my care co-ordinator every --- I actually think charities did a great job. (...) I felt like there was a lot of content and a lot of, like, helpful hints, and tips, and guides, and suggestions, and recommendations, and resources available. [...] I think charities did a particularly good job at pulling together a lot of information that could be disseminated, and people could pick and choose what they wanted to digest. [P21: female, White, 36-45, big city] Sustained contact via remote methods with GPs was particularly valued due to the possibility of keeping medication prescriptions up to date and assessing potential urgent need for physical health care. The reliability of GP services helped reduce the impact of worries about physical health on participants' mental health. Overall, interviewees welcomed the efficiency and support provided by GP surgeries during the pandemic and reported relying on this support when specialised health services were reduced or inaccessible. GPs' mental health support and active listening were very important in reducing participants' anxiety about potentially losing access to services. Examples of helpful GP strategies for providing remote care were scheduling weekly check-ups regardless of symptoms and offering email consultations. From GP I have done email consultation online form and then he emails me back. (...) It feels much more efficient at times. Like, just get to the point. And in some respects, I feel like be more honest like that. (...) the reception option is a bit scary (...) [I want to] avoid them. [P15: female, White British, age not specified, town] Some participants saw having to register on apps and fill out complicated online forms as an important barrier to accessing this care. Participants appreciated having the support of charities, voluntary sector organisations and crisis helplines. However, some participants perceived crisis helplines as dismissive, and call handlers less knowledgeable than mental health staff about mental health. One person additionally stated that NHS crisis helplines did not offer support in different languages despite being promoted in several languages. Needs not being met by helplines which were thought to be available 24/7 was perceived as a risk to personal safety. --- Sometimes I just needed somebody to talk to, and [the helplines] were always available. I think the [helpline] have helped me the most. Because they just listen and understand. I'm not in a crisis today, but if I was having a bad day, it would be very useful to phone the [helpline]. [P18: female, British Asian, 36-45, city] Not the mental health lines [...]. People are not well trained, just causing more distress than anything else. [P48: female, White/mixed, 46-55, city] 3. Patient safety and privacy. New situations occurring in the online space, such as inpatients joining community peer support groups described in Subtheme 2.2, had safety and privacy implications that had not been encountered in face-to-face care. Some people described feeling unsafe during remote consultations either due to a lack of privacy and safety at home, and this sometimes led to a pause in treatment. Text-based services were cited as being a helpful alternative where safety or privacy at home was compromised. My cognitive analytical therapy ended up stopping because I couldn't find somewhere that felt private and safe to have those conversations...Basically, if I was upstairs my boyfriend could hear me because he is working from home [P19: female, White British, 36-45, city]. [Remote therapy via text is better than via phone] for people that can't talk because it is not safe to.. ---.You can delete the texts if you are in that kind of situation, where you don't want people to know what you are saying or you are worried about your privacy. I would love to see more text-based support in the UK. [P19: female, White British, 36-45, city] There were also reports of people feeling that non-verbal signals of escalating distress and agitation were being missed by clinicians, particularly over the telephone, potentially leading to safety concerns. There was one incident in particular where she, towards the end of the session, said something that really upset me. And she didn't realise that I was upset because she couldn't see me. And I did try to, kind of, tell her and I thought she had heard it in my voice that I wasn't okay. [...] But I don't think she had any idea [...] And actually later that evening, I ended up in A&E because I was quite upset and not really coping with things, [...] I felt like that might have been different had we been face to face... [P1: female, White British, 36-45, city] 4. Views about the future. Several participants identified aspects of telemental health that they would like to see incorporated into their care in the future. Their comments suggested that a hybrid model of care delivery could combine the advantages of face-to-face care, including developing the therapeutic relationship, with the advantages of remote care, such as flexibility and reduced need for travel. If I was given a choice then I would probably say that I would want some of my appointments to be face-to-face, some of them to be virtual, and the remaining can be done over the telephone. [P49: male, British Asian, 46-55, city] I think the first couple of times I'm meeting s different had we been face-to-face [P1: female, White British, 36-45, city] Some participants reported incidents of severe distress during or after remote sessions. For example, a participant described risky or self-harming behaviours taking place during a video support group. This had an impact on all members of the group and new ground rules, risk management and support were put in place in subsequent sessions. Another participant raised concerns around attending online group therapy sessions where they were not allowed to have their cameras on and how this had potential implications for safety. Something big that happened in one of the groups that I don't think would have happened in a face-to-face group. There was this one guy, a member, [self-harmed] on camera. I think it is unlikely that he would have done that in a face-to-face group...an ambulance was called straight away.. ---. And it was quite difficult to watch, actually, because I have got personal experience of [self harm]. [P29: female, African, 26-25, city] A blend of face-to-face and digital support groups was also identified as beneficial to balance the advantages of both methods, with face-to-face perceived as facilitating stronger interpersonal connections, whereas online groups provided greater flexibility and anonymity, which was sometimes preferred. A participant explained that her online therapy group worked well because they had already met face-to-face before the pandemic. Conversely, a participant who started attending a new online group where no-one knew each other felt there was very poor engagement. Service users also identified other scenarios in which a blended approach would be beneficial, for example, receiving phone calls when on a waiting list for face-to-face therapy, or for "functional" appointments as described in Subtheme 1.1. --- A real-life appointment with PPE [Personal Protective Equipment] is massively preferable to a phone consultation... Unless it is to do with something purely factual like [...] explaining something to do with medication or something. Then, in which case, certain circumstances a phone call is fine. [P40: male, White British, 26-35, small town] Overall, however, many participants expressed a preference to return to face-to-face for relational appointments such as psychological therapy in the future, as technology is "not really a replacement for the real thing" [P3, female, White British, 36-45, city], but were happy to use remote contact methods for more functional appointments such as brief medication reviews with their GP or psychiatrist (relational and functional appointments are described in Subtheme 2.1). --- I don't think most of the appointments, especially with things like mental health medication where you have to speak directly to a GP while you're still increasing dosages, making sure you're tolerating them properly and they won't give you just a repeat prescription for some time as you're starting a new medication. That sort of follow up appointment can definitely be done just over the telephone. [P22: male, White British, 26-35, town] --- Discussion --- Main findings We analysed data about telemental health care from interviews with forty-four service users who experienced mental health problems with onset prior to the COVID-19 pandemic. A diagram mapping key findings is presented in Fig 1. Service users appreciated remote care options during the height of the pandemic when other forms of care were not possible. However, remote care was mainly seen as an option to allow access to care in extreme circumstances, rather than an alternative of comparable quality to face-to-face care. We found that participants' preferences for modality of care varied between participants and within participants. Factors that influenced preferences included contextual factors, such as having a private space for therapy, and individual factors, such as attitudes to technology, mediated human contact. Differences were identified between relational appointments such as those to participate in psychological therapy, and functional appointments to renew medication prescriptions or complete quick health checks. Other variables influencing remote care experiences were the relationship with the care provider, including whether they had met face-to-face in the past, and ease of use or access to necessary technology. Overall, participants stressed the need to provide alternatives for people who could not access or did not feel

The prominence of telemental health, including providing care by video call and telephone, has greatly increased during the COVID-19 pandemic. However, there are clear variations in uptake and acceptability, and concerns that digital exclusion may exacerbate previous inequalities in access to good quality care. Greater understanding is needed of how service users experience telemental health, and what determines whether they engage and find it acceptable.We conducted a collaborative framework analysis of data from semi-structured interviews with a sample of people already experiencing mental health problems prior to the pandemic. Data relevant to participants' experiences and views regarding telemental health during the pandemic were identified and extracted. Data collection and analysis used a participatory, coproduction approach where researchers with relevant lived experience, contributed to all stages of data collection, analysis and interpretation of findings alongside clinical and academic researchers.The experiences and preferences regarding telemental health care of the forty-four participants were dynamic and varied across time and settings, as well as between individuals. Participants' preferences were shaped by reasons for contacting services, their relationship with care providers, and both parties' access to technology and their individual preferences. While face-to-face care tended to be the preferred option, participants identified benefits of remote care including making care more accessible for some populations and improved

Discussion --- Main findings We analysed data about telemental health care from interviews with forty-four service users who experienced mental health problems with onset prior to the COVID-19 pandemic. A diagram mapping key findings is presented in Fig 1. Service users appreciated remote care options during the height of the pandemic when other forms of care were not possible. However, remote care was mainly seen as an option to allow access to care in extreme circumstances, rather than an alternative of comparable quality to face-to-face care. We found that participants' preferences for modality of care varied between participants and within participants. Factors that influenced preferences included contextual factors, such as having a private space for therapy, and individual factors, such as attitudes to technology, mediated human contact. Differences were identified between relational appointments such as those to participate in psychological therapy, and functional appointments to renew medication prescriptions or complete quick health checks. Other variables influencing remote care experiences were the relationship with the care provider, including whether they had met face-to-face in the past, and ease of use or access to necessary technology. Overall, participants stressed the need to provide alternatives for people who could not access or did not feel comfortable with telemental health care. Strategies used by GP practices and charities, such as scheduling regular phone calls and signposting to available resources, were mentioned as measures that facilitated remote care. Some new risks and challenges were also identified for example in online group sessions or in initial mental health assessments conducted over the phone. --- Relation of findings to previous literature We found that service users valued personalised, flexible options that include a combination of different types of remote and face-to-face contact. These preferences are in line with reports prior to the pandemic indicating people's desire for mental health services that offer choice and responsiveness to individualised needs and circumstances [38]. It was apparent in our study that experiences and preferences regarding telemental health were essentially personal. Acceptability varied greatly between participants in ways that could not be readily explained by their situation, thus suggesting it is not possible to make assumptions about participants' preferences. In line with previous research, participants mentioned beneficial aspects of telemental health care and innovative strategies that services adopted to address safety and technology challenges during lockdown [22,23,39,40]. These include improved accessibility and continuity of care, especially where difficulties such as physical mobility, social anxiety or paranoia impede travel and face-to-face contacts, increased convenience for those facing geographical barriers, convenience and communication within and between mental health teams [2,7,16,21,23,40]. However, participants also commented on the risk of digital exclusion of those lacking the necessary skills, resources, or privacy to engage with remote services. This is particularly notable given that our participants were sufficiently digitally connected to be able to take part in the study. Digital exclusion appears not to have been widely addressed in previous studies [2,7]. Groups who were already disadvantaged, such as older adults, people with sensory or cognitive impairment and minoritised groups are most at risk [41][42][43][44]. An example are people experiencing significant social disadvantage or severe mental health problems including psychosis have been reported to benefit less from telemental health [45]. Evidence is lacking both on the extent of digital exclusion and on how it might be overcome [21]. Barriers to access care such as those described in Subtheme 1.2 may help to explain why, despite its robust research evidence base and the strategic focus in England on more effective integration of digital technologies across the NHS [46], implementation of telemental health had remained very limited prior to the pandemic in England and in other countries with similar mental health systems [1]. Participants highlighted limitations of remote care that went beyond lack of access. Service users commented on different contexts in which remote care seemed more or less appropriate. Previous literature has varied in reports about how far video-calls can offer an authentic substitute for the connection made between service user and professional face-to-face. Regarding relational appointments, research involving female older adults [23] and veterans [22] has reported that service users tend to find video more impersonal than face-to-face due to reduced physical cues, and feel more comfortable talking to therapists in person, where possible. Conversely, a systematic review found that in terms of therapeutic alliance, only a minority of studies reported video-based interventions as inferior to face-to-face treatment [39]. Our study suggested people vary in the extent to which they feel rapport and therapeutic alliance can be of equivalent quality to face-to-face, but most seemed to feel it was to some extent inferior: it may be that the volunteer research participants in previous studies have been particularly open to seeing digital contacts as equivalent, or that better planning and preparation has improved experiences in these previous research studies. As with the present study, the evidence is mixed and potential negative impacts on rapport and therapeutic relationships, leading to more superficial therapeutic contacts, have been noted, including during the pandemic [2]. Many felt that remote care tools were inherently a less satisfactory way to form a therapeutic relationship, and some discarded remote care options altogether. However, we found potentially remediable barriers to delivering good quality mental health, such as, the service provider's ability and level of comfort to use technology. Specific barriers to engaging with remote care identified in our interviews and previous literature were lack of familiarity with and mistrust of relevant technology, low image quality on video calls, connectivity problems, and audio delays [22,23]. These concerns have been shared by mental health service staff in the UK [7] and professionals have tended to report a preference for face-to-face contact for both assessment and treatment [21]. Privacy and safety have also been emphasised in the literature as serious issues to consider in telemental healthcare provision [21,24,47,48]. Research has suggested that telemental health is potentially effective for group interventions and our participants described benefits from newly established online groups, [12], but the distressing situations described by some suggest a the need to develop guidelines to ensure service users' wellbeing and privacy during remote group care provision. While face-to-face care remains preferable in some situations, participants also identified strategies adopted by service providers to facilitate engaging with remote care. This suggests there is scope for extending digital access for those who wish to receive it [21] and develop strategies to prevent patients with limited access from being at a disadvantage [49,50]. --- Strengths and limitations Because of the pandemic context, recruitment for this study was mainly undertaken through social media, and interviews were conducted via Microsoft Teams. Although efforts were made to include participants who felt less comfortable with remote communication (recruitment via voluntary sector and community groups, providing a phone interview option), our paper may under-report problems with remote working. At the same time, remote data collection allowed us to reach a sample across England and attending a range of mental health services. Our findings provide a snapshot of a specific short period at a stage in the COVID-19 pandemic when social distancing restrictions were in place to varying degrees in England: some aspects of service provision and service user views and experiences may reflect this particular time and may have subsequently changed. However, the significant contributions of LERs and clinicians ground this research in real-world experiences and increases the applicability of our findings. An important strength of this study is the collaborative nature of the analysis, combining larger multidisciplinary group input with smaller group consolidation of findings, resulting in a detailed rapid analysis. Our group's experience and previous work contributed to the identification of important issues, while leaving room for the development of inductively derived themes from the data. --- Implications Remote care has been an important strategy to allow care to continue at a time of social distancing and is likely to remain a major modality for delivering continuing care in any similar emergency. There are also potential benefits in continuing some use of telemental health for future service delivery during recovery from the pandemic and beyond, including greater convenience and accessibility for some service users, as well as efficient service delivery. Our findings suggest that continuing some use of telemental health beyond the pandemic is feasible and acceptable from a service user perspective, but that further steps need to be taken to ensure that this is safe, high quality and in keeping with service users' individual needs. Guidelines equivalent to those in place for face-to-face care need to be developed to protect service users' wellbeing and privacy. At the same time, we found service users considered remote care was not an acceptable option in some situations and face-to-face care was required. We found that that how people experience remote care and what they find acceptable is very individual: modality of care offered should thus ideally be discussed on a case-to-case basis to find the best fit for service users' preferences and circumstances. This requires flexible and personalised hybrid models of care take advantage of positives of telemental health care, as well as offering face-to-face care when necessary or preferred: a mixture of types of contact may work well for many. Guidance around care options and access to telemental health care, such as those we found were provided by some charities and GP practices, should be further developed and routinely offered to service users in all mental health services. A shift to remote care also has organisational, training and technological implications, such as the need for recruitment of IT support staff and development of policies and technology to allow for staff to adopt flexible approaches. Our research suggests staff urgently require training in assessing remotely what is likely to work best for each individual service user and overcoming communication barriers associated with technology, culture and language. Different approaches for service providers to identify and deliver the most beneficial packages of hybrid models of care should be assessed. Promoting communication between staff and service users will reduce assumptions and anxieties about new ways of working. Research emerging from other countries has identified similar advantages and limitations to remote care, such as increased accessibility for groups who cannot travel, or difficulties resulting from not picking up on non-verbal cues during phone consultations [2]. This suggests our findings are also relevant to contexts outside of the UK. However, lack of health insurance coverage for remote care [51], or lack of access to medication prescribed during online appointments [52] are examples of context-specific challenges that were not identified in our sample. Further research is required into overall uptake of telemental health, the barriers and facilitators to engaging with it, and the unintended consequences and risks of exacerbating existing inequalities that may result from its use. We found relatively little evidence of innovative strategies to improve acceptability and reduce digital exclusion: potential approaches that may warrant further investigation include providing training in technology use and access to devices for service users. Questions for further research include impacts on therapeutic alliance and communication of telemental health adoption, and how best to implement telemental health in the longer term, including investigating the training needed by staff to become confident in using remote technologies and incorporating them in safe and flexible care pathways.

The prominence of telemental health, including providing care by video call and telephone, has greatly increased during the COVID-19 pandemic. However, there are clear variations in uptake and acceptability, and concerns that digital exclusion may exacerbate previous inequalities in access to good quality care. Greater understanding is needed of how service users experience telemental health, and what determines whether they engage and find it acceptable.We conducted a collaborative framework analysis of data from semi-structured interviews with a sample of people already experiencing mental health problems prior to the pandemic. Data relevant to participants' experiences and views regarding telemental health during the pandemic were identified and extracted. Data collection and analysis used a participatory, coproduction approach where researchers with relevant lived experience, contributed to all stages of data collection, analysis and interpretation of findings alongside clinical and academic researchers.The experiences and preferences regarding telemental health care of the forty-four participants were dynamic and varied across time and settings, as well as between individuals. Participants' preferences were shaped by reasons for contacting services, their relationship with care providers, and both parties' access to technology and their individual preferences. While face-to-face care tended to be the preferred option, participants identified benefits of remote care including making care more accessible for some populations and improved

as long as possible, few members of racial and ethnic minority groups know about EOL care options and use such services. 7,8 According to a report by the National Hospice and Palliative Care Organization, 9 of the approximately 1.4 million individuals who received hospice care in 2008, only 18.1% were nonwhite. Of particular note is the pronounced underuse of hospice by Asian Americans. 5 A series of studies with Medicare beneficiaries registered in the Surveillance, Epidemiology, and End Results program found that terminally ill older Asian Americans with cancer were less likely than their white counterparts to use hospice. 10,11 The disparities observed between older Asian-American and white patients with cancer persisted even after the effects of socioeconomic and clinical information were controlled for. Given the heterogeneity between and within Asian-American groups, the present study focused on Korean Americans, the fourth largest subgroup of Asian Americans. 12 The study examined factors that might increase or decrease willingness of older Korean-American adults to use hospice. Although willingness and use represent distinct constructs, studies have shown that when combined with skills, abilities, and minimal environmental barriers, personal willingness is a critical determinant of help-seeking behaviors in many health-related contexts. 13 The exploration of a broad range of predictors of willingness to use hospice was expected to provide essential information to help in the design of interventions to increase awareness and acceptance of services and eventually to lead to greater use of services. Andersen's behavioral health model, 14,15 which has proven to be a useful framework for studying use of various types of services, [16][17][18] including hospice, 19 guided the present investigation. Predisposing variables usually include demographic variables that may influence individuals' propensity to use services. Physical health constraints, such as chronic conditions and functional disability, represent potential needs for hospice use. Enabling variables consist of a variety of resources that may facilitate access to services. Studies have suggested that younger individuals, women, and those with a spouse and more education are likely to have more-favorable attitudes toward service use. [16][17][18] It was also expected that older individuals who currently had severe health constraints would be more favorably disposed to using hospice. Health insurance is a critical enabler of health service use, 20 and those with such coverage are likely to have a greater willingness to use services. Research on immigrant populations has shown that those who are more acculturated to the host culture have a more-positive perception of and greater use of services. 21 Finally, prior knowledge and familiarity may promote service use. 16,18 Taking the existing literature as a springboard, the goal of the present study was to explore factors that might increase or decrease willingness of older Korean-American adults to use hospice. An array of potential predictors, including predisposing variables (age, sex, marital status, educational attainment), potential health needs (chronic conditions, functional disability), and enabling factors (health insurance, acculturation level, prior awareness of hospice), was assessed. --- METHODS --- Participants The Mental Health Literacy Among Korean American Elders project provided data for the study. Data were collected from March through August 2008 in Tampa and Orlando, Florida. The surveys were conducted with Korean adults aged 60 and older who were able to understand and complete the questionnaire. The survey questionnaires were in Korean and were developed through back-translation and pilot-testing with 20 older Korean adults who were representative of the anticipated sample. Because immigrant populations are often underrepresented in public databases (e.g., Census data, telephone directories), a variety of sources were used for recruitment, including specific organizations and a telephone directory. The survey sites included 25 local Korean churches, two other groups with a religious affiliation, one Korean senior center, two senior apartments, and a local Korean elder association. Individuals for whom on-site contacts were not possible were mailed a packet including a set of questionnaires and pre-stamped return envelope. To recruit individuals who were not affiliated with the aforementioned groups, requests were made for referrals from respondents and other individuals associated with the primary data collection sites. The convenience sampling procedure was supplemented with a systematic review of a telephone directory of Korean residents provided by the Florida Korean American Association. Two thousand Korean residents in Tampa and Orlando were listed in the directory. After excluding those who had already been recruited through the convenience sampling efforts, all remaining individuals were called to ask whether there were age-eligible members in the household. Up to five phone calls were made to each number. When there was an eligible person in the household, a survey packet was sent. This step was designed to improve sample representation by including individuals who were not recruited through the convenience sampling efforts; this is often done with hard-to-reach minority groups. 22 The final sample consisted of 675 participants. Of these, 73.0% were recruited through visits and 27.0% through mail surveys. A series of analyses was conducted to determine whether there were any significant differences in sample characteristics according to recruitment method. Only one significant difference in a major demographic variable was found; participants recruited in visits were less educated those who responded to the mail surveys (chisquare = 10.3, P <unk>.01). This finding suggests that sole reliance on the mail survey approach might have a bias in favor of more-educated individuals. All respondents were paid $20 for their participation. --- Measures Predisposing Factors-Demographic information included age, sex (0 = male, 1 = female), marital status (0 = married, 1 = not married), and educational attainment (0 = <unk>high school, 1 = >high school). Potential Health Needs-For chronic conditions, individuals were asked to report, using a yes-or-no format, existing medical conditions from a nine-item list of chronic diseases and conditions commonly found in older populations (e.g., arthritis, stroke, heart problems, diabetes mellitus, cancer). A summated score was used for the analysis, with a high score indicating more chronic conditions. Functional status was assessed using a composite measure adapted from the Older Americans Resources and Services questionnaire. 23 The 20 items covered a wide range of instrumental and basic activities of daily living, including eating, dressing, traveling, managing money, carrying a bag of groceries, and reaching the arms above the head. Participants were asked whether they could perform each activity. The responses were coded as 0 (without help), 1 (with some help), or 2 (unable to do). Responses to individual items were summed for a total score. The possible range for functional status was 0 (no disability) to 40 (severe disability). Internal consistency for the measure was high in the sample (<unk> =.93). Enabling Factors-Health insurance coverage, level of acculturation, and prior awareness of hospice were conceptualized as enabling variables. With respect to insurance, respondents answered yes or no in response to a question about whether they had health insurance coverage. Acculturation was assessed using a 12-item acculturation inventory 24 that included items representing language, media consumption, food consumption, social relations, sense of belonging, and familiarity with the host culture. Each response was coded from 0 to 3. Total scores could range from 0 to 36, with a higher score indicating a greater level of acculturation to mainstream U.S. culture. Internal consistency for the scale was high in the present sample (<unk> =.92). To assess awareness of hospice, participants were asked whether they had heard of hospice. Outcome Variable-The following statement and question were used to assess willingness to use hospice: "Hospice is a program that helps people who are dying by making them feel comfortable and free of pain when they can no longer be cured of their disease. If you needed hospice services, would you use them?" To use a nontechnical definition of hospice, the wording was based on a consumer guide developed by the Hospice Association of America. 25 A binary code of yes or no was used. Affirmative response indicated a personal willingness to use hospice. --- Analytical Strategy Descriptive information and bivariate correlations were assessed to understand the underlying characteristics of the sample and study variables. Logistic regression models of willingness to use hospice were estimated by entering the constructs in an order determined using Andersen's model: predisposing factors (age, sex, marital status, educational attainment), potential health needs (chronic conditions, functional disability), and enabling factors (health insurance, acculturation, prior awareness of hospice). Because of the individual distinctness, each of the enabling variables was added in succession. Analyses were performed using SPSS version 17.0 (SPSS Inc., Chicago, IL). --- RESULTS --- Sample Characteristics and Descriptive Information As shown in Table 1, the sample consisted of 675 older Korean-American adults ranging in age from 60 to 96, with an average age of 70.2 <unk> 6.9. More than half (58.8%) of the sample was female, and approximately three-fourths was married. Approximately 45% of the sample had more than a high school education. All participants were foreign born, and the number of years lived in the United States ranged from 1 to 54, with a mean of 28 years. These demographic characteristics were similar to previously reported profiles of older Korean-American adults in Florida and in other U.S. states. 24,26 The sample was in reasonably good health, as evidenced by average scores for chronic conditions and functional disability of 1.35 <unk> 1.24 and 1.81 <unk> 4.47, respectively. Approximately 82% of the sample had health insurance coverage (39% Medicare only, 6.3% Medicaid only, 14.2% private insurance, 12.2% Medicare and Medicaid, 9.9% other combinations). Acculturation scores averaged 15.8 <unk> 7.6 out of a possible 36. More than half of the participants (52.8%) reported that they had heard about hospice, and 73.6% of the sample expressed a willingness to use hospice. --- Predictors of Willingness to Use Hospice Table 2 summarizes the series of logistic regression analyses on willingness to use hospice. Of the predisposing variables entered into the initial model, age and education made significant contributions. Those who were younger (odds ratio (OR) = 0.96, 95% confidence interval (CI) = 0.93-0.98) and who had more than a high school education (OR = 1.67, 95% CI = 1.13-2.48) were more willing to use hospice. Variables representing potential health needs were entered into the second model, and a significant contribution was made by chronic conditions (OR = 1.23, 95% CI = 1.05-1.03). As the number of chronic conditions increased, so did the likelihood of endorsing a willingness to hospice use. In the subsequent models, three enabling variables were entered in succession, and each was significant. The likelihood of endorsing a willingness to use hospice increased when individuals had health insurance coverage (OR = 0.59, 95% CI = 0.37-0.94), a greater level of acculturation (OR = 1.07, 95% CI = 1.03-1.10), and prior awareness of hospice (OR = 4.43, 95% CI = 2.85-6.90). During the process of adding the predictive variables, age, education, and health insurance coverage lost their significance, although the significance of chronic conditions, acculturation, and prior awareness of hospice persisted. --- DISCUSSION Responding to the demographically driven imperative for more research on EOL issues in racial and ethnic minorities, [3][4][5][6][7] the present study explored predictors of willingness of older Korean-American adults to use hospice. Andersen's behavioral health model, 14,15 which considered predisposing factors (age, sex, marital status, education), potential health needs (chronic conditions, functional disability), and enabling factors (health insurance, acculturation, prior awareness of hospice), guided the study. The generally low use of hospice by racial and ethnic minority groups [3][4][5][6][7] suggests that older Korean Americans might also lack knowledge of hospice and be unwilling to use such services. It was encouraging, therefore, to find that more than half of the present sample had heard of hospice and that nearly three-quarters were willing to consider using it. However, the rate of endorsed willingness to use hospice reported in the present study (73.6%) is still notably lower than the 94.4% and 89.8% found in whites and African Americans in a statewide survey of older Floridians. 27 Although higher than expected, this overall lower rate of willingness of Korean Americans to use hospice than of other ethnic groups is consistent with previous findings for Korean Americans. 28,29 A study that compared four ethnic groups in California found that Korean Americans had the lowest level of knowledge about advance directives and that a majority of those participants exhibited negative attitudes toward EOL care planning. 28 Focus groups of older Korean adults and caregivers living in Florida also found a lack of knowledge and misconceptions about advance care planning and hospice. 29 Of the array of predisposing variables included, age and education contributed significantly to the variance in willingness. Younger individuals and those with more education were more likely to endorse willingness to use hospice. This finding is in line with the gerontological literature showing that those who are younger or more educated are generally more open to service use. 16,18 The significance of age disappeared when prior awareness of hospice was entered into the model. The same pattern was found for education, whose contribution became nonsignificant with the entry of acculturation. In short, awareness of hospice and acculturation had a greater effect on the outcome variable than did age and education. Chronic conditions played a significant role in increasing willingness to use hospice. Functional disability did not contribute to willingness. The lack of a significant association may result from the fact that most of the sample were functionally intact and that hence variance was limited. The existence of health concernsFparticularly chronic conditionsFseems to prime older individuals on EOL issues. The significance of chronic conditions persisted through the final model. Each of the three enabling variables considered made a substantial contribution to the predictive model. Although it lost its significance in subsequent models, health insurance promoted willingness to use hospice. Numerous studies have reported that health insurance is a critical determinant of the use of health-related services. 20 As previously noted, those with higher levels of acculturation were more willing to use hospice. The positive associations between acculturation and service utilization have been reported in many studies with immigrant populations. 21,24 A greater level of adaptation to the host culture seems to be linked to greater knowledge about and acceptance of available resources and services. The present study highlights the role of prior awareness in shaping individuals' attitudes toward services. Older Korean-American adults who were aware of hospice were at least four times as likely as those who were not to endorse the use of hospice. This finding is consistent with other studies that have shown that prior exposure to services promotes openness to and acceptance of those services. 16,18 The fact that 28% of the sample had never heard of hospice but indicated an willingness to use such services when provided information in a questionnaire underscores the potential benefits of community education and outreach programs. Previous research has shown that simple educational interventions can dramatically increase hospice use. 30 A supplemental analysis also showed a positive association between willingness to use hospice and personal interest in having advanced directives (correlation coefficient = 0.37, P <unk>.001). This latter analysis suggests that there may be moderate overlap across domains of health decisions and further supports the importance of educational and outreach programs on EOL care. To maximize the effectiveness of the programs, particular attention should be paid to the unique characteristics and needs of the target group. Given previous studies that document the central role that family plays in EOL decision-making of Korean Americans, 29 programs that target not only older adults, but also their family members are recommended. In addition, cultural beliefs that may influence perceptions and attitudes toward EOL care such as filial piety and filial expectations need to be addressed in the programs. 29 Within-group differences in the level of acculturation also represent a challenge. For those who are less acculturated and especially for those with limited English proficiency, programs should be disseminated in the native language of the target population. Overall, the openness to and acceptance of hospice observed in the present sample of older Korean-American adults provides support for the idea of including the presentation about EOL care not only in a community setting, but also in a clinical setting. Some limitations to the study should be noted. Because a cross-sectional design and a geographically defined, non-representative sample were used, causal inference and generalizability are not warranted. The outcome criterion, willingness to use hospice, was measured using a simple yes-or-no format. Future studies need to incorporate continuous variables with multiple items and adopt a mixed methodology combining qualitative approaches to expand the knowledge base of the attitudes of older ethnic minority adults toward EOL services. In addition, research interests should be extended to issues of access and quality by addressing use of services and perceived satisfaction with service use. In the present assessment, the scope of predictive variables was limited to factors at the level of the individual; future studies need to consider a broader spectrum of variables, including structural and systematic barriers. The fact that participants' cognitive abilities were not systematically screened may add to the limitations. Despite these limitations, the present study expands researchers' knowledge about EOL issues in understudied and underserved older ethnic minority adults. The findings may serve as a basis for developing and implementing educational interventions to promote awareness and use of hospice in older minority populations.

Responding to an urgent need for more research on end-of-life concerns of racial and ethnic minorities, the present study explored predictors of willingness of older Korean-American adults (N = 675) to use hospice. Guided by Andersen's behavioral health model, the study considered predisposing factors (age, sex, marital status, education), potential health needs (chronic conditions, functional disability), and enabling factors (health insurance, acculturation, prior awareness of hospice). Nearly three-quarters of the sample answered yes to the following statement and question, "Hospice is a program that helps people who are dying by making them feel comfortable and free of pain when they can no longer be cured of their disease. If you needed hospice services, would you use them?" A greater willingness was observed in younger persons (odds ratio (OR) = 0.96, 95% confidence interval (CI) = 0.93-0.98) and those with higher levels of education (OR = 1.67, 95% CI = 1.12-2.48), more chronic conditions (OR = 1.23, 95% CI = 1.05-1.44), health insurance (OR = 0.59, 95% CI = 0.37-0.94), higher levels of acculturation (OR = 1.07, 95% CI = 1.03-1.10), and prior awareness of hospice (OR = 4.43, 95% CI = 2.85-6.90). The present study highlights the role of prior awareness in shaping individuals' attitudes toward services, calling attention to a need for community education and outreach programs for racial and ethnic minorities, with specific emphasis on dissemination of information and greater awareness of hospice services.As the U.S. population becomes more diverse, greater attention is being paid to racial and ethnic disparities in various aspects of health and health care. 1,2 One area that still has not received much attention is that of racial and ethnic diversity or disparities in end-of-life (EOL) care. 3,4 Disparities in EOL care are particularly important, because the beneficial functions of hospice in terms of symptom management and enhanced well-being for patients and their families have been well documented. 5,6 Despite the value of hospice in ensuring quality of life

to characterize the use or concurrent use of medications with somnolence as an adverse effect and to assess associations between their use and the prevalence of somnolence (excessive daytime sleepiness or sleep duration of <unk>9 hours) using data from the National Social Life, Health, and Aging Project (NSHAP) 2010-2011 (N=3,338). Adjusted prevalence was estimated from multivariable logistic regression models adjusted for socio-demographic measures and the number of medications without somnolence as an adverse effect. The estimated prevalence of somnolence was 54% for those reporting use of 3 or more medications with somnolence as an adverse effect vs. 36.5% for those not using such medications. The use of prescription medications with somnolence as a potential adverse effect was prevalent. Concurrent use of medications with somnolence as an adverse effect was associated with a higher prevalence of somnolence. These findings demonstrate the need for more research to understand the impact of concurrent use of medications with similar adverse side effects on OAs' health and well-being. The current academic literature extensively used linear logistic models with social, behavioral, and psychological status to predict mortality. However, few address the interdependency of predictors and the imbalance of targets that adversely bias the results. Using the National Social Life Health and Aging Project (NSHAP), we developed two machine learning models predicting the 10-year mortality of older adults in the US. We first used tree-based algorithms of Decision Tree (DT) that account for the interdependency of the social features and decide the splitting nodes and thresholds using entropy gain conditional on the previous splitting predictor to discern disposition status. Second, we used the Fuzzy Support Vector Machine (FSVM) that regards every sample as a node in high-dimensional vector space and splits the nodes with an optimum plane by finding the best linear combination of features to get an optimum prediction accuracy. Additionally, FSVM addresses the target imbalance problem by conducting a more delicate classification of samples with close predicted probabilities of being alive and deceased. Compared to the accuracy rates achieved by the Logistic Regression, our algorithms perform better on the entire population and the population near the class boundary. We also discussed the social and demographic characteristics of the cases whose disposition statuses were either wrongly predicted as deceased or alive by our algorithms. The findings serve important purposes for public health practitioners in accurately understanding the risk and protective factors of mortality in aging. --- SESSION 4210 (SYMPOSIUM) Abstract citation ID: igad104.1563 --- EFFECTS OF DISCLOSING DIAGNOSES AND BIOMARKER RESULTS FOR ALZHEIMER'S DISEASE AND RELATED DISORDERS Chair: Takashi Amano Discussant: Brian Carpenter Early diagnosis of mild cognitive impairment (MCI) and Alzheimer's disease and related dementias (ADRD) has been recognized as a key strategy to improve health-related outcomes. However, given that these conditions are highly stigmatized, receiving a diagnostic label or knowing about biomarkers related to MCI or ADRD may have a profound impact on the person's life. As the impacts of disclosure of diagnostic information related to MCI and ADRD on the well-being of people living with these conditions are not well understood, this symposium attempts to address this knowledge gap. The first presenter will describe a study that examined racial/ethnic variations in the effects of diagnostic labeling of ADRD on social aspects of the person's life. The second presenter will discuss findings from a systematic review of the effects of ADRD and cognitive impairment on social engagement. The third presenter will present the findings of the study on the effect of amyloid PET results disclosure on health-related behaviors of people with MCI. The fourth presenter will describe the results of an online vignette study on the outcomes of a preclinical Alzheimer disease diagnosis. The fifth presenter will discuss the use of blood biomarkers to aid in the diagnosis of ADRD in primary care (PC). At the end of the presentation, the discussant will highlight implications for future research and policy development to alleviate negative impacts and maximize the positive impacts of a diagnostic label of MCI and ADRD. Abstract citation ID: igad104.1564 Theoretically, knowing about one's diagnostic label of dementia may increase a chance to receive necessary social support. However, empirical evidence especially regarding the racial/ethnic variations in the effects of diagnostic labeling on provision of social support is lacking. Data from the Health and Retirement Study (HRS, 2000(HRS, -2018) ) were utilized to examine variations in the effects of knowing about a diagnostic labeling. A total of 7,192 person-year observations who had dementia were included in the analysis. Knowledge about a diagnostic label of dementia was measured by asking whether a doctor told the person that he/she had dementia. Formal and informal types of received social support were measured. Regression analyses with inverse probability weighting were performed. A moderating role of race/ethnicity was examined by conducting subgroup analyses. Knowing about a diagnostic label significantly increased likelihood of having more informal (OR = 1.58, p <unk> 0.001, 95% CI = [1.34, 1.86]) and formal support (b = 0.09, --- RACIAL AND ETHNIC VARIATIONS IN THE EFFECTS OF KNOWING ABOUT ONE'S DIAGNOSTIC LABEL OF DEMENTIA ON SOCIAL SUPPORT --- States, 3. University of Michigan, Ann Arbor, Michigan, United States Mounting evidence supports the clinical utility of amyloid PET. However, whether patients use knowledge of their amyloid status to alter health behaviors remains unclear. We describe health-related actions taken by Mild Cognitive Impairment patients following amyloid PET disclosure (n=34) vs. a comparison group (n=37). Patients were 92% non-Hispanic white, 59% male, 73<unk>8.61 years old with 16.25<unk>2.49 years of education. Over 12 months of follow-up, all participants reported at least one behavior change from baseline on a 14-item health behavior questionnaire. Amyloid positive patients reported the most behavior changes (mean=4.67<unk>1.83) while amyloid negative patients reported the fewest (mean=3.72<unk>1.58). Across groups, no significant differences were observed in: lifestyle, vitamin/supplement use, stress reduction activities, cognitive stimulation, or advance directive completion. Amyloid negative participants were significantly less likely than controls to consider long-term care insurance (63.6% vs. 89.2%; P=.025), and to endorse changes classified as "other" (36.4% vs. 64.9%; P=0.037). After adjusting for education level, gender, and MMSE, logistic regression showed that amyloid negative patients were 74% less likely than controls to report "other" changes (OR=0.26, 95% CI [0.08, 0.85], P=0.025), and 78% less likely to consider long-term care insurance (OR=0.22, 95% CI [0.06, 0.86], P=0.03). Qualitative analysis of supplemental interviews with scan group participants revealed "other" activities to include changes in employment, driving, and residential status, and engagement in other non-medical activities (pursuing bucket lists and mending strained relationships). Health-related behavior changes following amyloid PET disclosure may differ by scan result and encompass actions to enhance not only cognition but quality of life. Preclinical Alzheimer disease (AD) is characterized by abnormal levels of amyloid beta and tau proteins in the brain of a cognitively normal person. There are potential risks and benefits of communicating biological marker (biomarker) risk information for AD using the preclinical AD label. The current study used a vignette methodology to evaluate older adults' reaction to their hypothetical risk for developing AD. Community dwelling cognitively intact older adults (n = 300, M age = 64.1) completed an online survey, randomized to disclose biomarker results and risk information (with or without the preclinical disease label) pertaining either to heart disease or AD. Participants reported on their perceptions of risk and behavioral intentions based on the Health Belief Model. A series of ANCOVAs and moderated mediation models support the idea that, across disease conditions, a preclinical label does not influence perception of the disease or behavioral intentions. Both groups indicated a similar and

p < 0.001, 95% CI = [0.05, 0.13]). The significant effect of a diagnostic label on formal support was not observed among non-Hispanic Black (b = 0.07, p = 0.086, 95% CI = [-0.01, 0.15]) or Hispanic (b = 0.07, p = 0.152, 95% CI = [-0.03, 0.16]) individuals. Results suggest that knowing about a diagnostic label of ADRD may increase informal and formal social support. However, the beneficial impacts may not be salient for non-Hispanic Black and Hispanic individuals. Strategies to maximize the beneficial effects of diagnostic labeling should be identified to address racial/ethnic disparities.

Introduction Over the last few decades, tourism has seen massive growth, meaning it is now considered one of the major industries in the world (World Travel & Tourism Council, 2022). For continued positive development, however, new knowledge about what makes us want to travel to places and visit their attractions is constantly required. As tourists, we often look for something new, exciting, and unknown, while we also look for a feeling of recognition, familiarity, and security (Blom & Nilsson, 2000). We expect memories and knowledge but also acquire attitudes to the places and their attributes that we visit and experience with our senses (Kim & Kerstetter, 2016). In general, as tourists, we seek experiences that stimulate all our senses, and their mutual interaction creates a complete experience when we get to know places and cultures (see for instance, Urry & Larsen, 2011;Xiong et al., 2015) Our senses, corresponding with the material and narrative content of the place, construct a feeling of uniqueness and satisfaction through interaction with our knowledge, memories, and experiences. In tourism, aspects such as, for instance, bodily states, pseudo-events, and stimulation of our mental frames of mind are used to attract us to an activity or destination, which also has a cognitive function (Krishna, 2012;Poljanec-Bori<unk> et al., 2018). Acquiring a sense of place related to how our bodily senses interact with or have been integrated into places we visit is a vital part of the experience. Historically, the importance of vision concerning our lived experiences of the places we see has been foregrounded in tourism studies. This is not least obvious when the tourism researcher Urry presents an argument in the classic The Tourist Gaze (1990a) about the dominant role of seeing for the tourist's experience. Following Urry (1990a) and also the interventions by other tourism researchers such as Crang (1997), Lee (2001), Park and Kim (2018) and Szadziewski et al. (2022), the perceptions of our other senses related to places, landscapes, and encounters are seen more or less as complementary realities in the context. For them, the "significance of the gaze" speaks of the primary sense of creating an experience, but often complemented by the auditory through narrative (Pan & Ryan, 2009). Our sensory perceptions of a place are very much linked to what we bring when we encounter an area defined by tourism, the current place information available at the time, and what is present based on what we see, smell, taste, hear and touch. We contend that, in relation to the formation of our personal experiences of places and their attributes, narratives about haptic experiences should also be included and thus given far more attention. This is not least related to the fact that many place-related attractions play an important role as symbols of the value of a place and for which touch is a central aspect of the experience. Previous tourism studies that have touched upon tactile experiences have primarily focused on the development of Virtual Reality in tourism (Cheong, 1995;Alyahya & McLean, 2021) and accessibility for persons with disabilities (Martin-Fuentes et al., 2021;Cerdan Chiscano et al., 2021). Here we can see a lack of tourism research regarding a more explicit focus on our other senses and their importance for creating satisfaction for tourists and attracting and drawing visitors to a specific place while they are travelling as tourists. In the article "The more-than-visual experiences of tourism", Edensor (2018) emphasizes that tourism should be understood to a greater degree as a multisensory practice and contends that tactile aspects should be granted more weight in the tourist's experience of a place. In other words, it is more than just the visual content that contributes to an experience of the tourist location (Hughes, 1995;Echtner & Ritchie, 2003). At the same time, the cultural landscape conveys historical and geographical meanings to the specific place, seen as a destination (Relph, 1976). To avoid the ocularcentrism of tourism experience-scapes, we foreground the importance of the haptic dimensions and experiences associated with destinations. The aim of the present article is, therefore, to conceptually highlight the significance of tactile attractions. The tactile experience is spatially arranged in a place based on a tourist-specific experience of that place. Still, the emotional content is simultaneously challenging to categorize, so we have focused solely on destinations that offer tourists a chance to touch well-known attractions. Touch constitutes a vital aspect of the conditions for a visit, alongside a narrative component which is also essential for creating interest, expectations regarding actions, and the tourist's experience. A critical discussion at the general level has already engaged researchers regarding the problem of the relatively one-sided focus on seeing in tourism studies (Graburn, 1995;Jacobsen, 1997;Selänniemi, 1999;Edensor, 2018). For instance, Pan and Ryan (2009) stress that there should be a much greater focus on the visitors' five senses compared to what is discussed today to enhance the tourist experience. This line of reasoning is also presented by Agapito et al. (2013) and Mossberg (2007), emphasizing that a prerequisite for creating a positive experience is that all our senses must be stimulated and interact. However, the present article will not discuss the tourist's considerations, whether the tourist is an active or passive recipient in connection with the visit or the extent to which the event should be seen as an ordinary or extraordinary aspect of the visit to the place in question. We would instead like to regard this type of attraction as a form of play or "make-believe" and as a momentary pleasurable act for the tourist. Examples in our discussion about tactile tourism-related experiences include the sculpture "The Bronze Pig Fountain" in Florence, the statue of Everard t'Serclaes in Brussels, the statue of Juliet in Verona, attributes in the cathedral in Santiago de Compostela and a bronze statue in Bremen. 2. The importance of the senses for understanding the tourist-specific place --- The significance of place By paying attention to the place-related experiential space of the senses, we expose and contribute to clarifying the tradition and home of an individualized tourist attraction. In connection with a clear representation, both informed and emotional, the intrinsic relevance of the attraction for the tourist is revealed. For instance, Ley (1977) mentions that it is a social product associated with a particular place where our actions concerning this product -the attraction -are expected to signify various meanings to us. Further, in this context, Tuan (1977) has chosen to emphasize the importance of our feelings and sensory perceptions when they focus on our experiences, which confirm the meanings of the attractions when combined. Cresswell (2008) discusses this problem when he says that a place, in its most elementary definition, is a combination of location and the meaning of the place for a specific individual. The same argument is also presented by Aronsson (2020), who emphasizes that a place is fundamentally constructed by people's social actions and interactions and relationships with their material surroundings over time. This starting point indicates to us that, from a tourist perspective, it matters that there is, on the one hand, the symbiosis between the intentions related to the attraction and, on the other, between those intentions and the general demand that tourists desire for an attraction to be successful (<unk>orluka et al. 2021). An essential factor in this context is how the place, with its various attributes, is marketed to the tourists (Cherifi et al. 2018). Ward and Gold (1994), Morgan and Pritchard (1999), O <unk>Dell and Billing (2005), and others highlight that it is important for the place to generate a positive image among tourists and that its unique features are foregrounded to be able to create the identity in demand. Our experiences comprise a combination of impressions we receive through our senses, and the attraction should ideally be considered unique (Baloglu & McCleary, 1999;Tung & Ritchie, 2011). Our senses are given different weights depending on what attraction we visit as tourists. At the same time, it is difficult to assess the significance of the different senses concerning each other to establish which of our five senses is predominant in terms of making us feel satisfied and giving us a positive general experience in a specific context. It is likely that the combination of how our senses correspond with each other and to what extent some kind of synergy effect is present provides the basis for how we experience and receive satisfaction from visiting a landscape, a place, or a concrete attribute. Our previous experiences and memories also constitute important components in the context, which contribute to how our experience is construed (Tung & Ritchie, 2011). This discussion should be developed in the light of what Mossberg (2015) calls experiential space, where the material enactment, which includes encounters with other people, influences our general impression and, in a way, our experience of an attraction. This is also highlighted by Tussyadiah (2014), who claims that tourism is primarily about creating experiences that cater to the different senses of the visitors. Mossberg and Johansen (2006), as well as Helmefalk (2017), have also chosen to emphasize that the more senses are involved in connection with our experience of a tourist-related attraction, the more attractive the message that we expect will be conveyed will seem, which in turn might influence our inclination to wish to visit that attraction as tourists. From a tourism perspective, place, identity, and storytelling are critical concepts for attracting tourists. It is impossible to talk about the identity of a place in the singular, however, since areas have several different identities that are constantly changing based on people's various relationships with that particular place (Blom, 2017). The human geographers Tuan (1974) and Relph (1976) have also presented the same argument, and both focus on the importance of an individual's specific feeling in a particular place. This line of reasoning is developed by Tuan (1977), who suggests that our emotional needs enable us to identify with a specific place where our previous experiences provide a foundation for the new experience that we receive from it. This emotional need relates, then, to a combination of the different impressions that we receive through our senses, and we think that the role of the tactile aspect for the value of an experience has so far been relatively overlooked in this context. --- The place-related experiential space of the senses To better understand the significance of the senses in place-related experiential spaces, we would like to provide a brief overview of previous discussions in this field of research. It is clear that all our senses are cocreators, but we might give one or several senses a particular focus at different attractions. Narratives, visual experiences, smell, taste, and opportunities to touch create an interaction in us which constitutes experiences that we associate with places, attributes, and people we have encountered. In marketing places or a specific attraction from a tourism perspective, it is clear that seeing and hearing are the senses that usually receive attention. Smell and taste are primarily related to culinary experiences, while the experiential value of touch has so far attained relatively little attention in this context. As a starting point for our discussion, we ask how our senses are expressed in various ways in tourism. The significance of seeing for our experiences when we assume the role of tourists is important, and seeing is one of our five senses foregrounded in the tourist industry (Urry & Larsen, 2011). Urry (1990b) also stresses this fact in a discussion about the meaning of visual impressions when we consume what can be called attributes or features of landscapes, specific places, and individual attractions. The various attributes we receive visually at the same time construct a mental category for our experience and the expectations that we have of a place. Through literature, film, marketing materials, and social media, we encounter geographically situated attributes, or symbols, which generate an identification in our minds centred on the visual (Blom & Nilsson, 2000;Urry & Larsen, 2011;Huang et al., 2017;Salim, 2017). Travelling and visiting places and cultures different from our everyday environment have always been closely related to seeing in the field that can be defined as tourism. The attributes or symbols our gaze focuses on represent a place that offers the visitor an identity-related association rather than a focus on the attribute or character (MacCannell, 2013). In this context, MacCannell also highlights that as tourists, we do not only regard, for instance, San Francisco as a city in more general terms, but instead, we see Fishermans's Wharf, Golden Gate Bridge, and Union Square, which together form a whole which in turn create a complete experience through synergy in our minds (MacCannell, 2013). Hearing in the form of different narratives that are significant for our experiences is also an essential component of tourism (He et al. 2019). It often provides us with added value when we as tourists visit an attraction where we, in addition to the visual perception, receive verbally mediated facts, for instance, from a guide, and also have a chance to listen to narratives and anecdotes. This type of narrative -storytelling -is used increasingly as a value-enhancing strategy, both in tourism and in efforts to promote a company's brand and its products. Narrative meaning, seen as a tool for marketing a tourist destination or a specific attraction, is also discussed by Lichrou et al. (2008), who emphasize that the use of narrative constitutes an essential aspect of highlighting the immaterial dimension of an attraction, which thus also enhances its value for tourists. Gabriel (2000) stresses this process of value-creation as well, who contends that the purpose of storytelling is not only to convey facts but also to create meaning concerning the tourist destination (Mossberg & Johansen, 2006). This can, in turn, develop a sense of participation so that the tourist's imagination, supported by the dramaturgy that has been constructed around the attraction, increases the value of the experience (Stasiak, 2013). Mossberg (2008) moreover notes that a narrative can make it difficult for others to copy the attraction since place, identity, and storytelling create a specific identity. An example is the statue Manneken Pis in Brussels which also exists as copies in other locations worldwide (for instance, in Broxeele in France and Koksijde in Belgium). Another example is the more miniature replica of the Statue of Liberty on the island Grenelle in the river of Seine in Paris, which is approximately 11 metres high, compared to the original in New York, which is 46 metres high. These examples do not contribute to any considerable "tourist value" for the place, except as curiosities. However, storytelling also provides the visitor with a particular entertainment value, which invokes feelings, meaning, and a sense of context that enhances the product's tourist value (Prebensen et al., 2014;Mossberg & Eide, 2017). Rodaway (1994) and Porteous (1985Porteous ( /2006) ) point to the significance of smell as an essential aspect for creating emotions concerning places and establishing a clear identity. The associations that may emerge when confronted with a particular smell can give us emotional associations that relate to something that we experienced a long time ago and that can be reactivated by stimuli (Henshaw, 2013). At the same time, a smell which one person might experience as unpleasant or relatively insignificant may, for another person, initiate an emotional process which instead generates positive associations (Blom, 2016;Xiao et al., 2020). Xiao et al. ( 2020) call this place-related experience, for which smell is centrally crucial in terms of the associations that it inspires, a "smellscape", which is a concept introduced by Porteous (1985Porteous ( /2006)). At the same time, Agapito et al. (2013) stress the difficulty of distinguishing between smell and taste, as these two senses interact. Our sense of taste is usually related to food and drink experiences and is closely linked to creating tourist attractions where culinary experiences are in focus (Cohen & Avieli, 2004). Food and drink as the reason for visiting a region or a specific place are therefore significant factors in the tourist industry. In addition to the visual experience, tourists wish to taste what a destination offers and thus gain deeper insight into the regional and local (food) culture (Hall et al., 2004). However, Hall and Mitchell (2001) point out that it is not only the food and drink as such that constitute the main attraction here, but this type of offer may also include visits to food producers, food festivals, and restaurants, which contribute to the whole experience of the visitor. Franklin (2001) notes that this taste-and place-related expertise means that we are not generally looking for a touristic landscape but rather seek a "taste-scape", i.e., that which we perceive as genuine and the kind of food culture that exhibits a clear place-related identity (Sims, 2009). The significance of food and meals from the perspective of destination marketing is also discussed by Berg and Sevón (2014), who argue that places should consider sensory positioning to a greater extent by offering multisensory experiences to promote a positive brand image. To reach this kind of complete experience, tactile components are essential. Regarding touch or tactile experiences associated with tourist attractions, this has rarely been the focus of research relevant to tourism. Holbrook and Hirschman (1982) point out, however, that tourism needs fivedimensional brands that can represent the real experiences of the tourist. Furthermore, the multi-sensory experience is a crucial aspect of tourism. It generates sensory signatures that create presence, emotions, and identity, which Hirschman and Holbrook (1982) also touch upon in their discussion. Cheong (1995), at the same time, highlights the importance of being able to develop methods for stimulating our senses in connection with artificial "travels", which take place virtually using virtual reality technology (VR). Cheong (1995) further finds it crucial for us to receive a whole experience through VR technology, for users to receive simulation of all our senses, not just visual impressions, to have the full range of experience we demand. --- Method The study is based on a conceptual discussion highlighting a few empirical examples to illustrate our argumentation and contextualise the theoretical debate. The premise for selecting empirical material is that the attraction fulfils the well-known criteria and is supposed to be touched by the tourist. The choice of attributes has been made intuitively, against a background of the selected attractions frequently being highlighted in the marketing of each place and thus possible to regard as significant for that place. Contacts have been established with representatives of official European tourist organisations for each area. The material was collected and analyzed in autumn 2021 and spring 2022. The collection of empirical data has been carried out in three different ways to, include data collected from official webpages, data from questionnaires sent to representatives of official tourist organizations, and information about the places and attributes collected through social media to acquire an understanding of how the tactile content is communicated in those channels. Five places in Europe have been selected: Florence, Brussels, Verona, Santiago de Compostela, and Bremen, all of which have attractions which feature tactile aspects and draw a relatively large number of visitors each year. As the selection of study objects is limited, we have not aimed for the study to be comprehensive or to yield any quantitative results. Instead, the purpose is to highlight the significance of initiating a discussion of the role of the tactile dimension of tourist attractions. --- Discussion Historically, touch was not associated with tourism and tourist activities but rather healing. One of the most famous historical persons who practised touch in this sense was the Roman physician Galenos (130 BC), who used massage for medical purposes (Doktorn, 2022). This is just one example of touch being central and, in different ways, foregrounded in historical texts. The Bible is another example of a text in which personal touch is centrally important. To use the contact for various purposes, then, has been a practice for centuries. Above all, this action has been related to prayers for a cure and specific religious ceremonies. Still, it has also increasingly been part of a tourist-related practice. It can be seen as a ritual as the tourist pays a visit and is planned and determined in advance, including conditions and procedures. In Piazza del Mercato Nuovo in Florence, "The Bronze Pig Fountain" or "Il Porcellino", as the locals call the famous sculpture, is placed in the fountain (figure 1). The Bronze Pig Fountain was designed and constructed (1612) by Baroque master Pietro Tacca (1577-1640) after an Italian marble copy of a Hellenic marble original which Pope Pius IV donated to Cosimo I in 1560 when he visited Rome. Today, the original sculpture can be found in Museo Stefano Bardini in Florence, which means that the sculpture that tourists touch is a copy, but this is, in fact less significant in the context since here it is the place as such that matters (Destination Florence, 2021). Nevertheless, the tourist attraction value of the sculpture is substantial. Large crowds of tourists come to see the statue each year due to the historical narrative and myth associated with it. Tourists are supposed to place a coin inside the boar's mouth, hoping that the coin will fall through the grid and into the well. As the coin hits the water, the person who placed it is supposed to make a wish. There is also a story, related to the sculpture, which says that a woman who wishes to give birth to a son should rub her nose against the boar's snout. Another story has it that those who want to be sure that they will return to Florence should pat the boar's snout, which is visible from the wear on the snout of the sculpture (Florence Inferno, 2021). An important factor in the successful marketing of an attraction is that it is mentioned in literature and/or featured in films, social media, or in other ways visible, which will increase its status and thus create a clear place-related identity (Blom & Nilsson, 2000). The Bronze Pig Fountain has, for instance been made visible in The Metal Pig (1842/2020) by Danish writer H.C. Andersen (1805Andersen ( -1875) ) and is also present in The Chamber of Secrets and the Deathly Hallows: Part II in the Harry Potter series, and in the 2001 film Hannibal featuring Anthony Hopkins (Destination Florence, 2021). --- Figure 1 The Bronze Pig Fountain Source: Thomas Blom. The attraction for tourists to touch The Bronze Pig Fountain in Florence is significant from the tourism perspective since many tourists who visit the sculpture also buy souvenirs, go to cafés and restaurants nearby, and perhaps stay overnight in the city. There are no official figures for visits to the sculpture as such. According to statistics, however, Florence had 15.8 million overnight stays of tourists in 2019 (Statista, 2021a). Therefore, it can be assumed that many of the tourists who visit the city also visit The Bronze Pig Fountain. We think this assumption can be substantiated using the response that we received from Touristinfo Firenze (Touristinfo Firenze, personal communication, 2022) regarding our question about people's interest in visiting the statue: "I confirm: it is one of the most popular attractions" and "It has always been very popular. Still, it seems that it's constantly increasing". Furthermore, in response to the question of how the statue is used in marketing, the following information was provided: "We used it as a logo to promote the Museo Stefano Bardini (one of the Municipal museums, where the original version is kept)" and "It is probably one of the most famous statues of Florence" (Hidden Italy, 2022). As for the number of visits per year, the Florence tourist organization informed us that "we don't have statistics on this. But I confirm: it is one of the most popular attractions". Notably, the attraction has become increasingly popular recently, which is established as follows: "It has always been trendy, but it seems that it's constantly increasing". This is somewhat surprising, not least since the sculpture is not foregrounded in the general marketing of the city but is used more specifically for Museo Stefano Bardini, which can be seen in the following quote; "in some occasions, we used it as a logo to promote the Museo Stefano Bardini, one of the Municipal museums, where the original version is kept". Everard t'Serclaes (circa 1320 -1388) in Brussels is considered to have been significantly involved in expelling Flemish troops from the city and has been regarded as a liberator of Brussels in 1356. t'Serclaes was honoured with a statue sculpted by the artist Julien Dillens, which has been possible to view at the entrance to the street Rue Charles Buls from the square Grand Place (figure 2) since 1902. Over time, a narrative has developed that the statue brings luck to people who touch it and makes their wishes come true. During World War I, touching the statue's right arm through caressing it was seen as a sign of luck and success. This action has then developed into a practice for visitors who wish one year of luck to touch the statue by caressing it from its head to its feet. In addition, visitors should ideally find the three mice hidden in the wreath, enhancing their wishes further. The narrative also states that touching the arm of the statue verifies that the person doing so will, at some point, return to Brussels (Brussels Life, 2021). Based on the different stories related to the statue, these incentives create an almost religious experience among visitors since the proximity and sense of touch signify luck and provide each individual with a feeling of satisfaction and strength (Nilsson, 2018). Most tourists who visit another well-known statue in the city, Manneken Pis, tend to start on their walk to Manneken Pis from Grand Place, which means that they pass the statue of Everard t'Serclaes. According to official statistics, the number of visits to museums and other attractions in Brussels in 2019 was around 5 million (Statista, 2021b). Since Grand Place is one of the most visited tourist attractions in Brussels, it is possible to assume that a substantial proportion of the tourists who visit Brussels have also visited Grand Place and at the same time also seen the Everard t'Serclaes statue. Since touch is an essential aspect of the experience, alongside the visual impression, it is moreover likely that most visitors have also touched the statue (Planetware, 2021;World Packers, 2021) Many of the pilgrims who walk along the Way of Saint James visit the 12th-century cathedral in Santiago de Compostela to demonstrate their gratefulness and reverence to God for having been able to complete the pilgrimage. Many pilgrims also perform several rituals when they arrive at the cathedral. However, some of the rituals performed in the cathedral that focus on touch are no longer possible since the structure and artefacts need to be protected from visitors' touch. At the statue of Santo dos Croques, for example, not only pilgrims but also university students facing examinations used to be able to knock three times on the forehead of the statue to ask Santo dos Croques to share some of his great wisdom. After knocking, the procedure was to perform a so-called "coscorron" -to lay one's forehead on the forehead of the statue, as this is considered a way to open up the senses (Duperier's Camino de Santiago Blog, 2022). Yet another ritual centred on touch is inserting fingers into the marble pillar that symbolizes the Tree of Jesse (figure 3). After 800 years of touching, imprints from fingers are visible on the stone. Since 2008, to preserve them, you cannot touch either the Santo dos Croques or the Tree of Jesse (Duperier's Camino de Santiago Blog, 2022). Comments on TripAdvisor (2022) indicate that touch is a meaningful action, something that is expressed, for instance, in the following quotes: "When we arrived at the Cathedral in Santiago after walking 151 km. We found out that we would not be able to touch the Tree of Jesse" and "It is incredible to think how many entered at this location and took a moment to rest their hand on the statue and pray. You can see the handprint on the marble. Very significant in the pilgrim's journey" and "We were able to easily the Tree of Jesse which has a pilgrim hand imprint in the stone which has been naturally worn away over time (a testament to the millions of pilgrims who have passed thru here)." It is, however, still possible to embrace the sculpture of the Apostle Saint James in the cathedral to express gratitude to him for helping the pilgrim complete the challenging walk. The tradition is prevalent, and there is usually a long line of people waiting to approach the statue of the Apostle. Of all the rituals carried out by pilgrims, this is one of few that have been kept since the earliest days of pilgrimage, and it is the most wellpreserved tradition today. The embrace symbolizes the end of the pilgrimage and represents reunion with God, faith, or the pilgrim's self Santiago Ways (2021). The number of pilgrims who arrived in Santiago de Compostela was 347,500 in 2019, while the total number of tourists visiting the city was a bit more than 2.5 million (Statista, 2021c). It is possible to estimate that the cathedral is seen by most tourists who visit the city since it has a significant role in the marketing of Santiago de Compostela. Those who visit the cathedral without previous knowledge of the stories about the attributes receive information on-site about what they mean, including an emphasis on the importance of touch. --- Figure 3 --- Tree of Jesse Source: Mats Nilsson. Usually, in this context, attractions with a long history featuring a central narrative are foregrounded, but there are also examples of more recent types of tactile attractions. One of the more innovative tourist attractions is a bronze model, which has been on display in central Bremen since 2019. It is a designed attraction intended to allow visually impaired persons to experience the structure of nearby buildings through touch (Bremen Tourismus, personal communication, 2022). A response to a question via an e-mail about the value of the Tree of Jesse to the attraction of the place, the following is highlighted: "It is a wonderful addition to the traditional attractions around the Market Place which fits perfectly in its appearance to the historic town hall and Roland Statue (United Nations Educational, Scientific and Cultural Organization, 2022) as well as the bronze sculpture of the Bremen Town Musicians". A study by the local tourist organization shows that around 80 per cent of tourists visiting the city can be expected to see the model (Bremen Tourismus, personal communication, 2022). However, it is unusual to count tourists at these attractions since they are usually part of the general design of open cityscapes. Hence, it is difficult to establish the exact number of visitors. --- Concluding remarks Taking as our premise the place itself, with its unique identity, this article has aimed to highlight the importance of the haptic to destination attraction and place-related "value". Moreover, the paper seeks to identify and understand the significance of tactile attractions from a conceptual perspective. The field has been relatively unexplored in tourism research, both surprising and unsurprising. It is stunning that the tactile meaning in --- Mats Nilsson / Thomas Blom Tactile Tourism Vol. 71/ No. 3/ 2023/ 553 -567 An International Interdisciplinary Journal a particular context often constitutes a significant component in the total experiential value of an attraction. On the other hand, it is not surprising that tourist organizations in places where such interests can be found only marginally tend to promote this type of tourist destination in their marketing of the place. To a great extent, these attractions seem to take on "a life of their own", mainly through word of mouth and stories shared among tourists as well as communication on social media. Based on this study, we conclude that distinctly tactile attractions, like many other tourist-related attractions, are (re)formulated through their history, culture, and narrative components. The latter aspect, in particular, is incredibly significant for their value as attractions. At the same time, this type of attraction does not always have to have a long tradition or start from a myth or a religious narrative. Indeed, the attribute can be created in the present and then "invested" with identity through narratives, which can be fictive or based on reality. We also conclude that there is, in many cases, a clear "passage" between the myth, the visible elements, and the added symbolic meanings which mould tourist practices via narrative. This engenders a unique way of seeing and using the attraction (Cosgrove, 1984). In this way, an understanding based on shared frames of reference is created and confirmed socially affirmed through constant activity -the action itself -during the visit, for which a well-defined ritual is centrally important. It is also essential to keep in mind in this context that the value of an attraction for tourism is fundamentally staged and not natural, so it could instead be seen as a result of history manipulated to contribute added value in economic terms. We have also shown that tactile components, which allow tourists to acquire a multisensory experience, including a clear place identity, constitute a significant factor with the potential for further development. To sum up, this makes a plea for taking the haptic seriously in tourism studies. It also provides examples for understanding the interplays between tactile attractions, places, and tourists, especially in destinations where history, beliefs, and values comprise vital components. The result of this process in the tourist industry indicates that it is relevant to highlight and acknowledge the significance of tactile aspects when trying to attract tourists to a specific place. We can identify many other areas for further research in tactile attractions, such as studies on tourists and their motivation, the creation of tourist value from a strategic place development perspective, and models for the sustainable planning of these attractions.

The present article explores tourist attractions that include touch as a significant component of the tourist's place-related experience. We have chosen to label this phenomenon "tactile tourism". The tradition of touching attractions is not new but has been a central part of the experience of specific place-related interests for a long time. However, despite its frequent and often significant presence in tourist destinations, few studies in tourism research study the phenomenon. The article is based on a theoretical line of reasoning which starts from a few concrete examples of different places in Europe where we have identified tactile features relevant to tourism. The empirical material for our discussion has been collected partly through e-mail correspondence with tourist organizations in the places selected for the study and partly through articles in social media and other relevant secondary data. This tactile experience contributes a substantial share of the primary enticement of the attraction and simultaneously invests the place with meaning, which in turn attracts tourists.

Introduction Recent data from functional magnetic resonance imaging suggests that various people responding to the same naturalistic stimuli will exhibit overlapping brain hemodynamic responses (Dmochowski et al., 2012). The findings imply that when exposed to similar stimuli, the brains will respond in a similar manner. As they view a brief clip, Dmochowski et al., (2012) discuss the overlapping pattern of brain activity. The resulting components capture correlations with a one-second time resolution, demonstrating that peak correlations of brain activity can occur across viewings in striking congruence with emotionally charged scenes. These findings cumulatively imply that the synchronization is a reflection of attention-and emotion-modulated cortical processing, which may be deciphered with great temporal resolution by isolating the most highly linked neural activity components. This indicates that the brain's responses to any film episode include feelings and responses that could be similar to what a person would have gone through during the real action. Thus, watching movies provides not only an enjoyable viewing session but also a real-life experience where people are not able to insulate themselves from either their inner or outer worlds, but rather, are able to feel the stimuli more intensely and clearly. From this vantage point, it should be underlined that films that are in the public screenings such as TV broadcasts, cinema films, etc. have the potential to influence, transform, and inspire the societies through their screenings. Appadurai (2019), in his essay, points out the case that the practice of Bollywood audiences watching the same movie more than once deserves consideration as a distinctive cultural practice in which repetition and difference support and reinforce one another in the way that Gilles Deleuze suggests. According to him, the way that music and plot interact in these movies makes this relationship particularly possible since song passages add a repeating or percussion aspect that enhances the melodic and inventive elements of the story. However, most of the mystic fairy tales, specifically the oriental ones, feature repetitions, rhymes, and other literary elements that might give audiences a comparable taste in many other languages. Appadurai also underlined that not all movies are successful in drawing in repeat viewers, there should be something appealing in its text for the audience. Sometimes, the same text appeals to different audiences in different geographies, decades later and this kind of nostalgia brings another appeal factor. Due to her fame as a Canadian woman author, who lived from the late 1800s to the mid-1900s, Lucy Maud Montgomery was awarded the Order of the British Empire in 1935. She began writing "Anne of Green Gables", a story about the various adventures of an orphan girl, when she was still in her 30s, and eventually made it into a book series. The so-called "children's story" by Lucy Maud Montgomery, which served as the basis for the Netflix series "Anne with an E" in 2021, was turned into a Turkish television series under the name "Green Valley's Daughter" (Yeşil Vadi'nin K<unk>z<unk>). It's been noted that 195B people saw the series teaser by March 2022. This cyclical transformation of the narrative could be handled in the way that Deleuze proposes. The issue of whether the series is appropriate for children or similarities and differences between the original and adaptation arises as a result of the series' debut to the public and the widespread circulation of social media reflections on it. Finally, the Turkish adaptation of the TV series expanding internationally, which was initially sold to Chile and Uruguay, generates a great deal of attention. With the support of first literature, then TV series, and finally digitalized media, a 19th-century true story is transformed into a narrative that may influence millions of people and spark crucial discussions. It is discussed in this study as to why the label "For Children" was allocated to this narrative. Due to its moving story and unifying elements that evoked a kind of Christmas spirit, Montgomery's literature, which was accepted primarily as the basis of the feminist theory during that period, was able to hold readers' attention. However, from a contemporary perspective, the literary work involves many more completely opposite issues such as sexist attitudes, issues with justice and inequality, jealousy, educational issues, discrimination, and even sexual harassment. It is clear that recently, the narrative brought the concept of genocide as the time's pressing issue as well as fundamental ideas like collaboration, compassion, love, and understanding. --- Purpose and Methodology This study generates predictions based on social feedback, news, events, and new facts after monitoring, even though it is not based on qualitative and quantitative data collected in the form of questions and feedback from a specific sample group. The film raises the topic of whether fact and fiction can coexist, but it also focuses on how the audience's opinions on their own reality have altered as a result of realizing the narrative reality of the film after ten decades. The purpose of this study is to carry out a formal, functional, contextual, and discursive analysis of the narrative. It also attempts to highlight how the story is reshaped and spread across time utilizing oral, written, visual, and digital media. Yet, the main question is to reveal how closely the fundamental meaning of a film matches with the meaning that is uncovered after seeing it? What changes in a movie's significance when it is rewatched after a while? Under the heading of "meaning", this study addresses the film-audience interaction in two different approaches. The focus of the debate is the relationship between the meaning that viewers infer from the movie's material while watching it. The other one is the complementary meaning of the text through the social, psychological, or historical components. --- Findings From the first Holy Books to the oldest myths, from Gilgamesh to the Mahabharata and the Ramayana, all kinds of narratives convey humanity's destiny, aspirations, and values to us. According to Wood (2020), mainly, the stories are typically told from top to bottom, from the leading to the leaded ones. Folktales, however, and other narratives which were subsequent works of creative writing later on, began to stutter the tiny voices that had hitherto gone unheard, from the bottom to top. This might be the case for the Montgomery book. When considered the nearly 500 television channels and pay platforms in any country, including Turkey today, it can be concluded that people want to be able to experience not only what happens to themselves, but also what happens to others, in all these verbals, written and visual texts. People are intrigued by the tales because they may use them to gauge their level of knowledge, learn what they don't know, make comparisons, and choose which version of the events is accurate. Wood (2020) calls it invented history and fabricated power. Various dynamics should be considered while discussing literature, imagery, film, and cinema. What makes a person purchase a book? What emotion or thought may compel them to choose one and begin reading it? Or what feelings and concepts can inspire people to travel outside their native country, spend money, or interact in a different culture. These texts have spread, maybe intentionally or not, due to the early oral tale form's transformation into written and later graphic texts. Through plays, musicals, and movies, many literary works of art have been seen by a variety of audiences. --- Suitability of the Text for Children Most of the texts and narratives are somehow related to education since they provide lessons to be learned or are geared toward children because the protagonists are kids. From this vantage point, it is evident that many texts that seem to be created for children have in fact reached them, most often through their parents or the educational system despite the fact that they do not include content that is appropriate for children. It is true that many fairy tales and stories featuring kid heroes are inappropriate for children, are not written with children in mind, or are perhaps even undesirable for children to witness and consume. It is extremely challenging to produce a text that is novel, and distinctive nowadays. As a result, each phrase, character, and image-regardless of how true or clear they may be-functions as a new brick added to previous experiences, current culture, accumulation, and fantasies. It is crucial to stress at this point how technology has radically transformed into what we think of education in a modernized world. Learning outside of the classroom and at home is prioritized by the development of both general technical tools and educational technologies. Finally, it can be assumed that children's publications appeal to a large portion of the population and that the rate of exposure to these publications rises exponentially with the number of children given that 45.3% of households in Turkey have at least one child in the 0-17 age group. Yet, both educational tools and modern media of its all types has the power to "minimize or maximize," issues playing with the perceptions and opinions of the audience whose "media literacy skills" are lacking. This study aims to analyze the television serial "Anne with an E", adopted from the book series "Anne of Green Gables". It mainly aims at social reflections of the time it was written, comparing and contrasting them with the reflections of the present society. One of the topics to be explored and discussed is whether the series in question is appropriate for viewers over eight. Thus, one could first question the contents and ingredients of the children's texts. Children's media must, however, include works that take into account the cognitive, verbal, and operational facets of children's developmental and emotional states. For discourse activity, the generation and integration of at least two forms of information are required (Aksu-Koç & Aktan Erciyes, 2018). First, information on the general story structure at the conceptual level, including all its components (such as setting, aim, initiative, and outcome), is anticipated to be included in children's media. Second, it should combine knowledge of various linguistic elements like vocabulary, morphology, and coherence to make the narrative's events linguistically related to one another. It might be challenging, especially for young children, to generate a well-formed narrative when relying on both conceptual and language issues. Even if narratives can be found even in children's early language productions, it may take time to develop the necessary knowledge and the ability to successfully integrate different types of knowledge (Ko<unk>utar, Kramari<unk>, & Hr<unk>ica, 2022). At this point, it should not be forgotten that there may be differences between the narrative texts conveyed by the visual media and the real narrative ones. This means that there may be great differences between what children understand, perceive or filter from narratives and reality. Texts regarding the child shouldn't be too complex for them to understand. Here, it is expected that the text's meanings be not ambiguous and that its cause-and-effect linkages are well-defined. The linguistic structure of the series, however, makes it impossible for even the other cast members to fully comprehend what the protagonist Anne is doing or saying. Most messages are concealed inside the text and must be decoded using additional knowledge of culture, psychology, or sociology. Narratives for children are expected to have certain characteristics. If the narrative is verbal, it is vital to stress how crucial it is for children to be able to think in their mother tongue. It is important to be aware of the various language abilities available, including reading, reading aloud, meaningful reading, and quiet reading (Nas, 2002). It's also critical to assess the child's reading speed and determine whether the child has developed the necessary listening skills. It is also essential to take into account if the children can express themselves verbally or in writing after getting the information. Along with language characteristics, the quality, formality, and content of the images used in the book are also crucial. The younger children are exposed to reading, the more positive an impact it has on their mental growth. For this reason, it is suggested that children spend time together when in the pre-reading period with books that are visual, auditory, and contain a lot of pictures. Numerous field studies have demonstrated that high-achieving children frequently participate in street activities or play with modified toys when they are young, show early reading skills, have an animal-loving disposition, and have positive connections with their parents and other family members (Akkuş, 2020). Although it seems possible to reach children more closely, faster, and more effectively with the help of children's literature and to transfer cultural heritage education to them (Üremen & Dilidüzgün, 2022), especially when the importance of myths and narratives in the formation of material and spiritual cultural elements is taken into account, it is thought that the myths provide important contributions to the education of 12-15 age group children, the richness of imagination and the acquisition of some values (Yalç<unk>n & Aytaş, 2017). Sever defines children's literature as "the general name of products that enrich children's worlds of emotion and thought with linguistic and visual messages of artistic quality and increase their level of appreciation, in accordance with the language development and understanding levels of children, in a life phase starting from early childhood and including adolescence." (Sever, 2015). With the diversification of media, it is now evident that children are less concerned about their reading and library habits than people in the older generation (Y<unk>ld<unk>z, 2020). Additionally, it is evident that visual and even digital media products are displacing textual and printed media more and more. <unk>nan-Kaya highlights that the process of digitization, which influences daily life in every way, results in a change in the responsibilities and roles of parents and teachers in terms of the psychological development and education of children and adolescents (<unk>nan Kaya, 2021). The young generation should be encouraged to take full advantage of digital learning and development opportunities, but it is also important to make sure they are safeguarded from online threats. An incredibly quick rate of creation is required to maintain the target audience of children who are now considered to be "customers" while also capturing a 24-hour broadcast environment with movies that are continuously strolling on both television screens and digital platforms. This makes it possible to quickly develop a variety of goods without focusing on any of their possible negative consequences or long-term effects, as well as without taking into account any potential misunderstandings or unpleasant repercussions. Due to the difficulty of examining each product independently and evaluating whether it would have an immediate or delayed negative impact on children, it is also possible to come across numerous harmful issues. In the name of globalization, emphasizing both local and foreign productions may also present storylines that contain discourses, phrases, and ideas from cultures they have never heard of. When children internalize them in this way, they unintentionally assimilate into other cultures or become alienated from their own culture, language, and values over time. Here, actually, there seems to be no threat of becoming Canadian, yet, they may think of some other issue that might be not appropriate for them. Very young children are emotionally impacted by what they view due to their cognitive processes rather than the development of their creative thinking. It has been retrieved from the data that young children are able to retain their shortest and most memorable scenes in their memory, regardless of how long the media messages they watch, because their hearing and vision skills are still developing, and they are unable to make inferences at the level of abstract thought. According to a recommendation made by the American Academy of Pediatrics (AAP), "pediatricians should encourage parents to avoid television viewing by children under the age of two." (Brown & Council on Communications and Media, 2011). This statement highlights the potential negative health and developmental effects of media use by children under the age of two and specifically the detrimental effects of parental media use (background media) on children. It also emphasizes the lack of evidence for the educational or developmental benefits of media use by children under the age of two. The impact of watching television on infants' attention span varies depending on the program's content, but it is evident that the likelihood that children will experience attention problems rises, according to a study looking at the effects of early screen exposure (Christakis, Zimmerman, DiGuiseppe, McCarty, 2004). For older children, for instance, it is generally accepted that children over the age of 7-8 are capable of understanding longer media texts and have mastered abstract thinking and language skills enough to establish cause-and-effect relationships, which is why it is believed that the rate of negative media exposure and the rate of involuntary viewing is on the decline. However, it is insufficient for children's media to include infantile elements or depict characters who are children. It is crucial to consider how a narrative might influence kids in educational, practical, emotional, and cognitive contexts as well as how it can serve as an example for kids. However, at a young age when children's visual communication is still developing, it is also crucial for the infant to hear and form abstractions (Tokgöz, 1979). While their parents once watched silent films or the first animations and cartoons on black and white television just a few decades ago, children today watch more than 200 channels of color television via cable broadcast. Even young children who used to watch movies on their own PCs are increasingly giving way to moviegoers who use a smartphone or tablet. This technological development prevented children from making independent decisions in their world and instead made them more deeply connected to and quickly acclimated to the adult world. But exposing kids to media without considering their readiness in terms of intellectual, emotional, social, and cognitive development might be detrimental to them. These effects might become apparent straight soon or later. The nature of the shows indicates a huge change in the types of TV shows watched by the entire family, moving beyond Netflix, Disney, and Amazon and networks like Nickelodeon, which only airs instructional programming and cartoons for children. Furthermore, despite the fact that many of these serials have portrayed children as main characters, kids shouldn't be watching these stories. This study explores the process by which a story with a real-life topic becomes a novel, a series of novels, and then a Canadian television program. The story's impacts on real life, the planning of the story as a Turkish television series, and its transformation into a Turkish serial that is sold internationally are the next extremely significant steps. The relevance of this tale may actually be seen in the fact that it was able to influence millions of people, first via literature and then through a TV series based on a true story from the 19th century. In this story, the phrase "For Children" is referenced, but it is also explained why this is the case rather than how. The title "For Children" is connected to this tale, but it is also explained why this is so rather than how the fundamental issue is how the narrative is viewed differently across time, across space, and across cultures. Even with this part of her autobiography, the author can still appeal to a wide range of readers because, at the time, many children were either separated from or lost their family members due to diseases, wars, and migrations. According to the Globe Orphan Children Report 2022, there are currently an estimated 1 billion biological, legal, and social orphans living in the globe. In addition to the 400 million orphans, it is reported that around 10,000 kids become orphans each day. Children who have been displaced by war or disaster live in camps or on the streets. (<unk>içek, 2022). Such recollections of adolescence were thought to be not only instructive for children but also a re-expression of the experiences of the majority of adults. However, having an educational topic or harsh realities of life doesn't make that text suitable for children. The kids' emotional, psychological, and social development levels might not have been sufficient for them to handle the given material and comprehend that. In this serial, the protagonist experiences several instances of prejudice, injustice, bullying, harassment, identity crisis, rigid social norms, terrible teachers, and ineffective educational practices. Apart from them all, it is estimated that children are abducted by authorities, held away from their parents, and later reached in a manner consistent with actual life at their neighborhood cemeteries. There is a great deal of worry that these tendencies may cause children to have less faith in democracy, public opinion, public authority, and education. So, the first question is answered: It should be taken into account that determining that children eight and older can watch this series, that this series is promoted as a children's series, and that it is repeated several times in several ways, does a significant harm to children. --- Textual Overlaps The article tries to conduct a textual and historical review to determine whether literary narratives and current realities overlap with each other. In the narrative, there appear to be three distinct dimensions. The first thing to be questioned is if the author's biography overlaps with the realities of that particular time, secondly if the filmic narrative overlaps with the author's actual manuscript, and last, the audience's perception of reality. However, there seem to be many different remakes of the same narrative, which one should be taken into consideration to check this overlap. If we consider the film as a part of the documentation of real life in 1890's, as Beattie, (2008) mentioned, not all documentary films and videos are sober depictions of the real world. Documentary representations can present expressive, entertaining, and spectacular images. Yet, the narrative's structure has an impact on how it serves its purpose. Such functionality, for instance, is needed when transferring a person's life story into a narrative. A single person's story is suddenly thrust into the spotlight of a larger audience, where it can no longer be understood as the narrative of a single person but rather as the shared experience of many people. A new field of review is created as a result of a book's proliferation through subsequent works and its translation into numerous languages and cultural contexts, each with its own intended audiences and perspectives. By doing this, it will be possible for the book to be evaluated, criticized, accepted, or rejected by other people-possibly by people and groups who have never shared similar emotions and opinions. The fact that a book turns into a series and takes on a format that includes many books brings with it a polyphony, continuity, and the power of the book to be accepted. To the extent that a cultural history understands culture to be primarily about the production of meanings, identities, and representations, it explains Italian (or Canadian) modernity largely in terms of aesthetic conventions, signification, and textual practice. (Hay, 1983:106). Verevis (1997) indicates the remake as a commercial and critical genre product. Frow (1990:46) expands the discussion by stating that, in the past, when a production credit was missing, the focus shifted from a legal-industrial definition to a critical-interpretive one, where the remake was evaluated in relation to a "general discursive field that is mediated by the structure of the filmic system as well as by the authority of the film and literary canon." Yet, Verevis (2004) asks further questions like: What is film remaking? Which films are remakes of other films? How does remaking differ from other types of repetition, such as quotation, allusion, and adaptation? How is remaking different from the cinema's ability to repeat and replay the same film through reissue, redistribution, and re-viewing? These are questions that have seldom been asked, let alone satisfactorily answered. Verevis ends the discussion stating the remakes fall into three sections: the first, remaking as an industrial category, deals with issues of production, including industry (commerce) and authors (intention); the second, remaking as textual category, considers texts (plots and structures) and taxonomies; and the third, remaking as a critical category, deals with issues of reception, including audiences (recognition) and institutions (discourse). Morra (2002) refers to the authors' insistence that the representations of the land and ownership dynamics in their early 20th-century artistic endeavors were seen as manifestations of a colonial reaction and a refusal to fit in with the place is mentioned by Northrop Frye in his analysis of the specifics of recent cultural and art history texts' assessments of Canadian literature. Another viewpoint holds that early 20th-century artists were linked to the imperial center "there," outside the boundaries of the nation. A closer look at their work reveals how they used the land to create a sense of national identity and that they perceived it as a friendly force rather than a hostile one that was the source of a transcendent spiritual experience that led to the adoption of Canadianism. Perhaps similar things could be revealed from the narrative of Montgomery. However, Braun & Spiers (2016) state that alongside postcolonial and feminist studies, as well as recent trends in queer theory and ecocriticism, literary celebrity has offered a bridge to those scholars who want to think literature back into the bigger picture of society. Karim (2004) on the other hand would handle the situation as re-viewing the 'national' in 'international communication' through the lens of diaspora. --- Original Text and Remake In the remake, those that reconstruct the series from an artistic and aesthetic point of view do it from a standardized aesthetic standpoint. The critical approach, on the other hand, is far from liberal or pluralist and strives to challenge social beliefs. The critical method highlights the minute nuances in that story that not everyone can perceive and seeks to make them more obvious and tangible for everyone. He uses an interdisciplinary approach because, as Türko<unk>lu stated, the series care about cultural differences and emphasize that they should be highlighted. Social disparities play an important place in research (Türko<unk>lu, 2004:96). From the perspective of the story's transformation, turning a book into a picture book, a comic book, or otherwise simplifying or adapting it for young readers instead of adult readers will also result in the book putting things upside down. Sure, in that case, the book and the stories it contains are being read and discussed by more intense and active masses, yet, this might not be the right way. Additionally, the book might be made into a theater production to be performed on stage or into a film to be visualized. In this instance, it may be said that the narrative's target audience has altered, and the influence's range has grown or shrunk depending on the audience's access options. In particular, cinema is a form of art that successfully presents significant experiences through the magical worlds it offers, the past it reminds us of, the emotions it provokes, and the future it creates. Since its initial applications in history until today, this unique experience of participation, which has the power to add more to a film than a person may experience in his entire life, has faced quite various dynamics. For a few key reasons, the book-tomovie adaptation is becoming increasingly important in today's culture. People's fascination with the past, their desire to research historical evolution in order to better understand themselves, or their efforts to evaluate numerous factors in order to apply themselves to the present and the future are a few examples of this. Cinema uses visual texts and can demonstrate a relationship that supports both, but literature builds its language through written texts. This is true even if there are two different story genres represented by literature and film. Despite the fact that numerous literary works have been adapted for the screen, it is also clear that literary works have been affected by movies. Filmmakers and TV show producers are often drawn to literature for a variety of reasons, including financial concerns, artistic considerations, time constraints, basic difficulties in writing original scripts, a lack of creativity, as well as awareness, practicality, the facilitative effect of registered works, and the influence of literary circles. Getting permission is only one of several important factors. In a sense, it is thought that visual literacy significantly emphasizes a reading facilitation impact, such as benefiting from the background and experience of visual literacy. In the process of transforming an original work into an adapted work, the stage of oral or written narratives, and audio-visual texts taking on another form may cause the new work to appear with a completely different identity. The images in the visual world have no place to be found since each adaptation can actually be read in a variety of ways. The truth is that, just like an author makes a unique literary work, each director develops his or her own film in a distinctive style, due to the fact that the adaptation drastically departs from the original text in a number of areas. As seen in the instances of new media, social media, and transitional media, a shift of ownership can be discussed within the context of the transformation of the narrative. It is also conceivable to discuss a character shift because, for instance, a more moderate character in the original work could evolve into a crueler and greedy figure in the adaptation. Characters from the original story or historical fact could not be mentioned, yet characters from the fiction might be. In this case, it would be necessary to specify which characters were added or removed. A few changes might be made by the information source that makes it easier to transfer intercultural texts from one community to another. One might attribute a generalization or familiarization concept to this. Despite having been fed by the arts that came before it, cinema, the so-called seventh art, has a strong affinity for literature, especially novels, which are the most suitable literary form for it. After the 19th century, popular publications, which emerged in two distinct channels, elite and popular, became a crucial resource for the film industry. We frequently become intrigued and excited by the stories of people who lived hundreds or even thousands of years ago. Because they want to live in accordance with the society and circumstances in which they find themselves, people prefer to dream of the future over the past. However, documentaries and narratives taken from real-life stories always attract attention and such films can reach large audiences. It may even be argued that there is a growing prevalence of "standardized" interpretations and reflections that are based on false information. From the industrial point of view, the movie is a ready-to-watch product with its theme, actors, and story. The target market for this product's audience must approve of it and it must live up to their expectations. Only in this way are familiar stories retold by actors from many times able to spark interest and pique curiosity in a wide audience. The known and unknown components of identity have started to be predominantly developed through fiction during this time of development, while the aspects that conflict with Western culture have been sharpened, brought to the forefront by being revealed, and the like aspects have been rasped. In the digital and visualizing world of the 20th century, movies have been one of the best methods for establishing identities. The general public was exposed to identity and cultural codes in this way, and corresponding identities and codes were frequently employed. Just now, different narratives have given varied roles to the same groups, identities, and cultural norms. It's occasionally feasible to present historical tales in a fresh way by increasing their aesthetic elements. Within standardization of the original version or remake, there is a connection between representation, difference, and even diversification. Power can be used in this context not only as a means of economic advancement and physical supremacy but also as a means of larger cultural or symbolic representation of one another. However, it is important to correctly address the nature of this power. According to Said (1978), standardization incorporates connections between representation, difference, and power. However, the concept of power needs to be carefully considered. It shouldn't just be understood in the materialistic and physical meanings, but also maybe more broadly in the cultural and symbolic senses, as the capacity to stand in for someone or something in a certain way. Even though modernism develops new and distinctive story forms and brings various sorts of mobility to the audience, the dynamics produced by orientalism have never been lost, and this narrative form continues to be active without losing much of its content. A lengthy retelling of many narratives that are interwoven with the main narrative might be used as an example. Considering the topic of the narrative through the point of view of narrative, Montgomery's literature, which was primarily based on feminist theory, contained many more distinct sexist attitudes within the context of today's contemporary perspective, justice, inequality, jealousy, educational problems, discrimination, and discrimination during the time that it was able to hold readers' attention due to its heartfelt narrative and its unifying aspects carrying the Christmas spirit. In fact, it is evident that it combines fundamental ideas like cooperation, compassion, love, and understanding with the contemporary issues of the day, such as genocide. Bishop & Durksen (2020) indicate that utilizing the literature as providing a chance for policymakers, educators, researchers, and teachers to evaluate the personal qualities required to engage Indigenous kids in education and civic society would yield a lot. This approach of making use of literature emphasizes how critical self-reflection is essential to being a relationally responsive teacher. Generally speaking, for a literary work to be discussed, it is essential to take the period in which it was written as a basis. In this respect, it is useful to review the literacy rates. As is known, literacy is a basic skill and an important measure of the education of the population. From a historical perspective, literacy levels for the world's population have increased drastically over the past few centuries. For example, while in the 1820s only 12% of the world's people could read and write, today only 17% of the world's population is illiterate (https://ourworldindata.org/literacy). The core idea of the narrative has been portrayed to focus on childhood and adolescent memories and expressing identity issues in the evaluations made in the context of the Anne with an E series, which was taken from Montgomery's works centered around her autobiography. According to sources, Lucy Maud Montgomery was born in Crediton (now New London), Prince Edward Island, on the same day as British Prime Minister Sir Winston Churchill. Her mother, Clara Woolner Macneill Montgomery, passed away from tuberculosis when Lucy was just 21 months old (https://www.geni.com/people/Lucy-Montgomery/6000000002316026900). Montgomery won't be able to experience this fact, as she declares in her book, until after she completes high school. She spent a lot of time searching and trying to find out who her real family was, and she was finally able to find real records about them. Even this issue of neverending identity search could make it an adult narrative rather than the children. When considered in terms of the 'Topic' it is seen that the subjects such as social acceptance, modernization, social oppositions and inequalities are brought to the forefront. Considering Montgomery's autobiography, Lucy's grandparents Alexander Marquis Macneill and Lucy Woolner Macneill moved to live in the nearby town of Cavendish, as her father, Woolner Macneill Montgomery, left the province after his wife's death and settled in Western Canada. It also emphasized that growing up with a strict upbringing style. The sad life of a young girl is depicted in this component of the story, but it also touches on sensitive issues that many people can identify with, such as conflicts with family elders, intergenerational conflicts, and puberty issues that can arise even from ordinary events. It is crucial that such narratives have a component that is so significant and universal that it may be accepted with sympathy by significant portions of society when autobiography is turned into a book. It is important to note that at the time, such texts were in fashion. It is intended to modify any adaptation to the society in which it will be planted once more while also being faithful to the original. While some works are quite effective in this regard, others may only be rudimentary given the community or setting they are intended for. The fictional aspect of the work joins the picture at this point, and fiction can be effectively used to adapt the original work to a new language, culture, country, or historical period. For instance, versions that are not supported by features like 21stcentury narrative structures, the tempo of the narrative, and the rate at which it contains action are more likely to remain based on history and memory or reality. Yet, the idea of making use of the best-seller product in other countries where they have neither the profile, nor similar social, economic, or cultural setting would cause some kind of dangers. --- Overlapping Realities between the Narrative and Real Life Texts that fall short of current narrative fiction may therefore provide a window into the

In this study, there are three key topics are explored and discussed. It is first and foremost vital to examine whether the series in question is appropriate for viewers who are older than eight years old. In the second evaluation, it is examined whether contemporary realities and literary narratives have any overlap. This implies that issues related to history, society, psychology, and modernity are approached from a fresh angle. Last but not least, it has to do with the potential social effects of adjusting historical tales to the present. This entails reassessing the past or historical narratives from the viewpoint of the present. However, the key inquiry is: How closely does a film's core meaning align with the meaning that is revealed after seeing it? What about a movie's relevance alters when it's watched again after some time? This study addresses the film-audience connection in two separate ways under the category of "meaning." The relationship

significant and universal that it may be accepted with sympathy by significant portions of society when autobiography is turned into a book. It is important to note that at the time, such texts were in fashion. It is intended to modify any adaptation to the society in which it will be planted once more while also being faithful to the original. While some works are quite effective in this regard, others may only be rudimentary given the community or setting they are intended for. The fictional aspect of the work joins the picture at this point, and fiction can be effectively used to adapt the original work to a new language, culture, country, or historical period. For instance, versions that are not supported by features like 21stcentury narrative structures, the tempo of the narrative, and the rate at which it contains action are more likely to remain based on history and memory or reality. Yet, the idea of making use of the best-seller product in other countries where they have neither the profile, nor similar social, economic, or cultural setting would cause some kind of dangers. --- Overlapping Realities between the Narrative and Real Life Texts that fall short of current narrative fiction may therefore provide a window into the past for the younger generation. The only thing that counts in these tales, which are the common language of the 21 st -century media in that era, is what happens to us. However, these narratives, which are not appropriate in terms of sound, image, and movement, may not proceed beyond stories where this is the only thing that matters. The risks are obvious, as noted by Hornaday (2018): this level of audience and subject familiarity brought about by the reality-to-film translation is vulnerable to reductionist judgments and an easy editorial approach to the ambiguity and reflection required to find lasting value in the film. Within the text, the society of the past seems to be more stick to the rules of religion or society. Similar to the religious explanation of social events, the first communication studies comprised of theological interpretation. With factors like the spread of capitalism and the culture of fear, rhetorical propaganda, and persuasion have become more significant (Erdo<unk>an & Alemdar, 2010:74). The first overlapping seems to be at the level of including real-life aspects in the narrative. Montgomery's original work presents a variety of subjects by fusing them into a single narrative. The most effective of them exhibit strong ideas that encourage readers to connect emotionally with one another since they have gone through comparable things. For instance, topics like immigration, the search for ancestors, identity issues, societal adaptation, educational issues, discrimination, financial difficulties, making important life decisions, and choosing a job are sprinkled throughout the narrative like salt and pepper, and with the innocent narrative of that time, it is quite lovely in and of itself appears to have been successful in fostering a favorable atmosphere for growth. The second overlapping issue is handling the narrative as a kind of mission to be performed. The fact that Montgomery lives on an island imposes various limitations from the perspective of the task notion. For instance, the island seems a little distant from Canada or anywhere and there seem to be not many good examples around. So, the protagonist always questions her task in life, within society, at home, at school, etc. The main themes of her autobiography are issues like being caught between having an education and not having one, the intense pressures of social acceptance, being separated from classmates who will understand her, and being surrounded by elderly people. Her writings express the agony that she went through at the start of her childhood in a way that makes it possible for many readers to identify themselves with her. The streamof-consciousness technique is used to change the narratives in the book's television series to include more than merely concentrating on the writer's personal observations as well as improvisations toward others. Yet, in this case, we have a little girl always talking to herself instead of listening to her inner voice or analyzing her thoughts and feelings. One issue that the narrative might not correspond with real-life reality is deep knowledge of the literature cited in the book, having long passages, and attributing several emotional states and facts. Even if onscreen, Montgomery is portrayed by Anne, who is totally immersed in reading, as a little child who frequently moves as a result of what she is reading, speaks "like a book," and has developed into a young girl, this portrayal is different from what the book provides. The inner thoughts and ideas, the joy of creativity are all transferred into the self-dialogues that make a different portrayal of Ann speaking so fast or murmuring to herself only. The spectator comes across a character named "Anne" who has internalized reading to the point where she can recite exact passages verbatim from books she has read during the course of her experiences. This is both really amazing and extremely terrifying specifically for those new types of young audiences who lack a reading culture. Because none of what she reads is appropriate for her age, mental developmental stages, or physical state. Due to the reading selections being made without taking into account any criteria such as emotional or intellectual readiness, these readings could even put Anne into difficult situations causing ambiguous traumas as well as social misunderstandings and issues. However, despite not being particularly severe to prevent Anne from reading and writing studies on her own, these drawbacks play such a significant role in her life that they ultimately have an impact later on her professional choice and educational life. That means each episode of the story has a stitch to the other part that makes it easy for the reader to follow the text. Throughout her biography, it is also known that Montgomery lived with her father and stepmother in Prince Albert, Saskatchewan, after leaving home as an infant, but that he later moved back to Cavendish, the house of his grandparents, a year later (Montgomery, 1997). Given that it raises topics like migration, differentiation, and various cultural structures that individuals frequently face, it can now be said that the issue is much more significant and presents hints that are subject to evaluations and discussions. It is also stated that Montgomery continued her schooling at Prince of Wales College in Charlottetown after completing her education in Cavendish in 1892 and obtained a teaching certificate after enrolling in a two-year program there (https://internetpoem.com/lucy-maud-montgomery/biography/). Examples of injustices experienced at school, a view of gender inequity, and sowing the seeds of inventive and creative thinking can all be observed in the context. For instance, Anne's approach to the housefire of their neighbors was regarded as a behavior and style of thinking that society would value. It may be argued that this experience has given her a clear acceptance and predisposition. The character of Anne in the series has been pushed to the edge of society due to society's intolerance, rules, or unfair, unjust points of view, but she has been transformed into a hero who can conquer these challenges because of her strong character. Anne's intelligence, creativity, and capacity for problem-solving enable her to successfully navigate any challenge. When Montgomery turns this point of view into literature, she gives her the chance to talk about the significance of education in children's lives and inserts passages in her stories where readers can voice their opinions on the educational system. This is like how the hero in her writings is given the opportunity to do by Montgomery. Another issue that falls under the category of transitions, acceptances, and trends is that the show's creators were bold enough to back Anne's struggle in the show, and nearly the whole cast of the show is in favor of gender equality. The executive producer, some of the directors, and the whole writing staff are all women, demonstrating an extraordinarily attentive approach, in addition to the fact that the series has many female actors (https://www.imdb.com/title/tt5421602/fullcredits/?mode=desktop&ref=m_ft_dsk). To Yaylagül, ( 2006) the critical approach seeks to transform the society and the subject they study, even the obligation to transform it, whereas the current approach seeks to legitimize and maintain the current social order with an evolutionist attitude (Yaylagül, 2006:18). Thus, the text of the narrative could be functioning on the way to transform the society at that time, and even today. People are also very interested in and wish to experience Anne's working life, the protagonist of the stories. For this reason, Montgomery's narratives about her teaching experiences allowed her to offer a portion of her own life story while simultaneously serving as a role model for other young females who aspired to become teachers like her and helped them view the field from a different angle. The idea of tailoring can be used to convey this. The fact that the stories were well-received by Canadian readers at the time, that they were successfully turned into serials with their still-accepted themes even decades later, and even contributed to raising a similar voice in Turkey. Having the stories that were broken, both in their initial plane and later, readers reveal that it is well welcomed by the audience on various levels. For instance, Montgomery's narratives appear to be extremely successful in portraying women's participation in education and their unequal position in society, despite the fact that women have historically been highly represented as students and instructors in Canadian schools. According to Gaskell, (2014) prior to 1850, middle-class families in the area tended to employ governesses and teachers to teach their children at home. With the introduction of free provincial public schools in 1871, girls' enrollment in Ontario schools began to catch up to that of males. There were also private colleges providing further education for female students, special education only in personal, social, and domestic aspects. However, they were few in number and the curriculum in these schools was designed to educate women in the "arts and graces of life" and prepare them to become school teachers (Gaskell, 2014). Even if the verbal form of the narrative generally penetrates social circulation, the literary form of the story is equally widely accepted and leaves enduring traces. It is impossible to deny how changes in technology have impacted how stories are represented visually. It's amazing to see how certain countries advance in this area while others lag behind. However, technological advancements like film cameras and movie theater equipment are thought to have occurred concurrently almost everywhere in the world. Visual literacy rates, a crucial element of literacy rates, evolve differently, just as literacy rates do regionally. Many congregations and associations can be mentioned in the country, which were established in order to carry the position of women to a higher level at that time. One of them, the Sisters of St Anne, stands out as a women's religious community founded in 1850 in Vaudreuil, Qué by the Servant of God, Marie-Esther Sureau, dit Blondin (Anne Marie-Anne) for the education of young rural girls and some charity activities. Although both its manager and many employees had serious problems and had to hide in the first years of its establishment, it is known that they reached a total of 1369 branches in 1986, when it quickly spread to Canada and the USA (Thériault, 2013). It should also be emphasized that the supporters of the party life emphasized in the series are such "hidden communities" and that the information and action plans regarding how the actions related to gender discrimination against women find strength are carried out with such organizations. --- The Reflection of The Realities in Today's World In real-life reflections, nothing could be just black and white, positive or negative. There could be some blurred issues that one might not find the right answer. For example, there are so many harsh issues that children shouldn't get confronted with. One of them is the issue of gender. Children might be shocked to hear that Aunt Josephine, who occasionally appears in the television series Anne with an E and puzzles kids with her anti-social perspective, views, and attitudes but looks to be extremely rational to them, is actually a lesbian. Even their organized parties are off-limits to adults and conversing with her for an extended period of time is uncomfortable. The kids are happy for their creative buddy Cole when she stays with Aunt Josephine and finds that he can go to art school there. Cole, on the other hand, believes that he may be emancipated from not only his work but also his many sexual tendencies, which he keeps hidden from everyone. However, apart from what the narrative proposes, in real life dating someone of the same sex was punishable by up to 14 years in prison under the Canadian penal code in 1867 (https://www.thecanadianencyclopedia.ca/en/article/homosexuality). The attitude towards perceiving the narrative as if it is a kind of apology from the past or helping the past stay in the past. Anne with an E can be perceived as an initiative that makes it possible for the wounds of the past to be revealed and possibly even healed. Despite the fact that numerous diverse pieces of evidence have been presented, many discoveries have fallen short of what a series may do. The fact that it was a Netflix series that debuted to the world and inclined toward education in general, as well as the facts Montgomery conveyed during the time she lived, caused the series to have a greater impact and win recognition all over the world. Changes that shocked both spectators and administrators appeared in the third season of the show, and as a significant wound of Canada's past, the exhibition of proof of genocide in schools where minorities and indigenous cultures in education took place came to the fore. Perhaps, this part of the narrative was not considered that much serious even its own time. The history of the extermination of the indigenous women and children in that region is tied to how the narratives and tales relate to reality (https://www.bbc.com/news/world-us-canada-57325653). Despite the overwhelming evidence regarding the past, it's also vital to remember that today's losses are not smaller (https://en.wikipedia.org/wiki/Highway_of_Tears). Not stated in the narrative but in real life nowadays, Sir John A. Macdonald, Canada's first prime minister, is blamed due to his permitting the establishment of a boarding school system in 1883 to assimilate Canada's indigenous peoples on the basis of previous suffering, which might have been viewed as a mere power and culture battle (https://www.dunyabizim.com/mercekalti/kanada-nin-150-yillik-asimilasyon-politikasi-cocuklarin-icindeki-yerliyi-oldurun-h29215.html). His stated goal was to "Kill the Native in the Children!" Up to 150,000 First Nations, Metis, and Inuit children were all removed from their homes during the 1920s when boarding schools were made mandatory for kids between the ages of seven and fifteen. Many indigenous parents handed their children up to the police under penalty of prosecution because they were powerless to fight the contemporary, white Canadian overlords. Even if the entire story and its outcomes have not been implemented in the movie, a story of an Inuit boy attending such a school is involved in the film. First friendships are built and later trust is established with her but later on, everything turned upside down. It is evident that she isn't shown to her parents and that she is being tried to be isolated from her native language, beliefs, and clothing, despite the fact that her suffering includes agony similar to Anne's suffering from being separated from her parents and continuing her education across numerous geographies just for the sake of "education". When such occurrences are covered by the media and a sizable audience follows them, people might desire to learn more about history from a more skeptical and realistic standpoint. The boarding school system thus appears to be in danger of collapsing as the psychological and cultural repercussions of the schools come under increasing examination, despite the fact that the last of the previously established schools closed in 1996. It has been documented that more than 150,000 indigenous children were enrolled in a total of 130 boarding schools at the time, despite the fact that an estimated 6,000 children died in these institutions, according to the testimonies of various witnesses, recent findings, and Murray Sinclair, a former chairman of the Canadian Truth and Reconciliation Commission in the 1960s. His history of physical and/or sexual abuse is also well documented. According to research, the majority of child fatalities were caused by disease, neglect, or accidents. Parents are still trying to count the number of children who have passed away or locate many of their graves today. In 2008, Canadian Prime Minister Stephen Harper expressed regret publicly for the country's boarding school system, which includes more than 130 schools, and stated that "today we know that this assimilation policy is wrong, it does great harm, and it has no place in our country." (https://akra.media/Haber/HaberDetay/74656/kanadada-6-bin-cocukoldu/?share=h74656). The main principle guiding the system was referred to as "cultural genocide" in the final report on the legacy of boarding schools published by The Truth and Reconciliation Commission of Canada in 2015. Additionally, they provided cash to locate cemeteries and honor children who passed away far from their homes. A preliminary ground radar assessment employing ground radar revealed an estimated 215 unmarked graves on the site of a former boarding school by 2021, according to the Tk'emlps te Secwepemc First Nation in British Columbia. Other Canadian provinces are apparently still carrying out similar research (https://www.bbc.com/news/world-us-canada-57325653). In 2019, the government committed CAD 33.8 million ($28 million; $19.8 million) over three years to develop and maintain a school student death record and establish an online record of boarding school cemeteries. The Truth and Reconciliation Commission up to now emphasized that it has only gotten a small portion of the money, although "discussions continue". It is not always possible to observe as modern media audiences bring up a historical occurrence and patiently wait for anything to be clarified. For the media producers, performers, screenwriters, and even the public, this kind of tension and accusation poses some hazards. However, it can be argued that the show grasps this in a very reasonable and mature way, maintains its objectivity by putting a strong emphasis on reality and avoiding polemics, and serves as a catalyst. In an interview from 2016 about his 2015 documentary Our Sisters in Spirit, director, and producer Nick Printup stated: "To begin to understand the extent of the tragedy that Indigenous women face today, you must first understand history." He emphasized that settler colonialism must be understood in the context of history and that the issue of Indigenous women and girls who have disappeared or been killed in Canada is a longstanding one. Indigenous women have historically been sexually objectified and subjected to harmful cultural attitudes and stereotypes that permeate many facets of contemporary Canadian society (https://www.thecanadianencyclopedia.ca/en/article/missing-and-murderedindigenous-women-and-girls-in-canada). In 2015 The Truth and Reconciliation Commission of Canada, issued 94 calls to action, six of which were on burial sites and missing children. The vow to "fully implement" them all came from Prime Minister Trudeau. Millions of people have watched the show "Anne with an E" as a result of its brave decision to portray this reality, and it has earned honors for being the most popular and viewed series. Geronimo Henry, 84, spoke to an AA correspondent about the series' emphasis on indigenous people and oppression, particularly in the final season. He recalled his time at the Mohawk Institute boarding school in Brantford, Ontario, where he attended from 1942 to 1953 (https://www.aa.com.tr/en/americas/canada-can-be-charged-withgenocide-says-residential-school-survivor/2287671). Since 1972, the school has acted as a cultural hub, housing three art galleries, a history museum, and a library in a later-built structure. It is believed that the "child-oriented" idea of the narratives is undermined by the horrific cruelty that characterized the last episodes of the series. Given that the story deals with both real-world issues and hypothetical love situations, it raises questions about whether or not it is appropriate for young readers. We can figure out how much the audience participates in the narrative during the integration of the stories with the audience by looking at the participation and risk-taking issue. Examples of notions that can demonstrate the broad variety of participation in the topic include going to the locations where many TV shows are filmed, visiting movie locations as a tourist, and purchasing items from the movie when they are on sale. For instance, the show Anne with an E, which Netflix released to a wider audience, has been deemed unsuitable for viewers under the age of eight. It is claimed that children aged 8 to 13 can watch it with adult supervision despite the fact that the movie contains mild violence, moderate language, brief nudity, and occasionally sexual overtones. However, a serial that is set at the end of the 1800s and moves extremely slowly is not really aimed at children in the twenty-first century, nor are there any ideas or actions that kids can apply to their everyday lives from this series. On the other hand, it can be observed that the show attracted a sizable audience and was seen as a children's show because it featured a child character. Additionally, 97% of people rated the television show positively. At the Canadian Television Awards in 2017 and 2018, the show won the best drama series. The main character of the series Anne dealt with social issues like being an orphan, psychological traumas, gender inequity, racism, and freedom of expression. Ultimately, it's a story of a girl who is fiercely unique, independent, determined, and doesn't see any barriers to herself, said executive producer Miranda de Pencier (Ahearn, V., 2017). According to the idea of participation, it can be assumed that many viewers today take part in bringing the issues from 200 years ago to the present day, discussing them, evaluating them in light of crucial ideas like gender equality, homosexuality, the LGBT community, and genocide, or gathering information about them. The interesting thing is that these issues are still debatable today. When seen in terms of Crisis Solutions, Anne with an E series typically focuses on issues like educational solutions, women's rights, child rearing in line with the changing social structure, the search for the past, and the concern for the future. In actuality, each of these issues has a connection to both the fictional Anne from the film and the real-life Montgomery figure. However, rather than being personal concerns, these topics continue to be discussed as societal challenges and difficulties. This perspective highlights the fact that these issues are typical in both literary and visual contexts and explains why the narratives are so well-liked. The crises to which the era and the narrative both relate also have remedies, making the voice of the voiceless lauder. In reality, Montgomery expresses the injustices she has seen, experienced, and witnessed in this way in order to be able to address them even after the fact. Social scientists of today underline that society's structure have altered, creating a "risk society" as well as a "crisis society" that is going through a crisis. Contrary to the assertions of authors like Lyotard, Beck, and Wynne say that nations undergo a reflective or second modernization process, not a postmodern one and that this process takes societies from national industrial societies to the global risk society (World risk society) (Beck & Wynne, 1992). The story concentrates on the life of a very different era in terms of literary and cinematic qualities when analyzed in terms of Rules, Technological Possibilities, and Innovations. Both Montgomery's autobiography and the scenarios Anne invents in her stories are rife with these rules, technological possibilities, and breakthroughs at a period when notions are still novel and societal acceptance is incredibly uneasy and cautious. For instance, Montgomery worked as a teacher at various island schools while pursuing her literary studies at Dalhousie University in Halifax, Nova Scotia, in 1895 and 1896. In 1898, she then relocated to Cavendish to live with her widowed grandmother, where she remained from 1901 to 1902 (Morrison, June 20, 2018, https://www.dal.ca/news/2018/06/20/a-look-back-in-time-at-lucy-maud-montgomery-sdalhousie.html). She is said to have worked for the Chronicle and Echo newspapers in Halifax for a very brief period of time. It is crucial to stress at this point that Halifax is one of Canada's oldest publications. The earliest newspapers in Canada were printed in Nova Scotia and Québec in the early 1750s, followed by Upper Canada in the 1790s. These publications were the tools of the colonial administrations and were closely regulated and observed by the authorities who supported them (Yarhi, 2017). To ascertain when independent newspapers initially emerged, it would be necessary to wait until the years 1800-1850. It may be claimed that throughout this time, printing houses were less expensive to establish and run, literacy rates increased, and people got more interested in and knowledgeable about news and viewpoints. Although the social norms and legal framework enabling women to work in journalism were developed much later, 16 female journalists launched the first women's journalists' club (CWPC) in 1904 with the goal of "maintaining and improving the status of journalism as a women's profession" appears to provide professional growth and support (James & Dann, 2016). It should also be remembered that less than 60 women were listed as journalists in the 1901 Canadian Census, and women made up only 18% of society's overall labor during this time. As a result, it is clear from her propensity for this line of work that Montgomery also has a "different," "entrepreneurial," and "innovative" side. Montgomery's return to Prince Edward Island later in 1902 to live and care for his grandmother was the inspiration for her first works, even though all these aspects and her work are not often regarded as positive or approving by the general public. It may be argued that her desire to write in order to express herself and reach a larger audience stems from her experience as a journalist. It is seen that this well-known autobiography is transferred to the works of the author almost line by line. After her grandmother passed away, Montgomery moved to Ontario three years later and married Ewan McDonald (1870McDonald ( -1943)), a Presbyterian priest, where she lived in Leaskdale. She continued her life by remaining a member of the nearby Zephyr congregation (https://lmmontgomery.ca/about/lmm/her-life). Montgomery's first book, Anne of Green Gables, was published in 1908. In spite of the fact that the couple had three sons-Chester Cameron Macdonald (1912-1964), (Ewan) Stuart Macdonald (1915-1982), and Hugh Alexander-who passed away in 1914 at birth, Montgomery persisted in her writing Brennan, 1995). It is known that she composed the following eleven novels in the Leaskdale parsonage. Although the congregation ultimately sold this structure, its worth can be seen in the fact that it has been transformed into the Lucy Maud Montgomery Pastoral House Museum (https://www.historicplaces.ca/en/rep-reg/placelieu.aspx?id=7716). The family is believed to have relocated in 1926 to the Norval Presbyterian Charge, in what is now known as Halton Hills (Ontario), where the Lucy Maud Montgomery Memorial Garden can be viewed from Highway 7. Montgomery passed away in Toronto in 1942, and after being interred at the neighborhood Presbyterian Church, she was remembered in Green Gables and laid to rest in the Cavendish Community Cemetery. During her career, Montgomery is reported to have penned 20 novels, 500 short stories, and numerous poems and also received the Order of the British Empire in 1935 (https://www.cbc.ca/news/canada/prince-edward-island/l-m-montgomery-suiciderevealed-1.723426). Even if the character Anne in the serials was a positive character, Montgomery, the family members stated, had to cope both with her husband's mental illness and the restrictions of her life as a clergyman's wife and mother in an era when women's roles were highly defined. Flood revealed the secret regarding Montgomery's death referring to her overdose at the age of 67 which indicated mental illness affects people of all types, even famous authors (Flood, 2008). The TV series Anne with an E has been successful in telling the story of its creator to millions of people thanks to its release on Netflix's platform, just as Montgomery's courage to write her autobiography gave her the ability to reach millions of readers, Nilüfer Pembecio<unk>lu, Nebahat Akgün <unk>omak QUESTIONING IF THE LITERARY NARRATIVE AND REAL-LIFE STORIES OVERLAP WITH TODAY'S REALITIES: THE EXAMPLE OF "ANNE WITH AN E" TV SERIAL express herself, and announce her achievements to larger crowds. With its narrative strategies, immersive structure, and the important messages it presented overall, the series, which everyone followed with interest, succeeded in establishing a throne in almost everyone's hearts. Anne's bravery, frankness, and combative personality helped her become a role model for many people. The transmedia issue highlights an elegant approach in which narratives move across diverse geographies, cultures, communities, or media platforms without encountering any contact with each other or influencing one another. When considered from this perspective, it is possible to discuss an ideological and historical reality like colonialism, racism, and the transformation of women into sexual objects within the context of historical context, as well as to turn these narratives into a series that travels the world via Netflix and make them known to wider masses. Another example of such fluidity is an adaptation made in Turkey and sold to nations like Chile and Uruguay that has nothing to do with Montgomery's autobiography or the social structure. By focusing mainly on the visible indications, such as the reinterpretation of the anguish of a little girl living alone in a small cottage in the country, this type of de-essentialization tries to capture a portion of the current mass approval and universalize it. It should be remembered that while current adaptations are new, actual reality and the original story can never be replaced. Thus, the new adaptations of the serial might be interpreted as something having only a faint connection with the issue. --- Evaluation and Conclusion The children's "perceived sensory needs" become crucial when taking into account the movies they see. The children might not have the opportunity to pause the movie and consider other realities in the outer world since they choose to be only accessible in the audio-visual context while watching. This indicates that the child perceives everything reflected on the screen as real. Although it is anticipated that children's stories might develop their imaginations, when childhood is over, the child may come across with these stories once more and may not have the chance to reflect on the knowledge they have learned via these stories. Therefore, it is evident that the dominant sense will be impacted in the monitoring process given that kid views are reviewed in a way that creates cumulative perceptions rather than being dependent on individual perceptions. For the youngster, perceived relationship needs are equally crucial. The connections a child makes with other kids, teenagers, and adults as well as their personal and social interactions are crucial. Understanding and using various modes of communication at various levels is essential. Also crucial are perceived adjustment abilities and the length of the adjustment process. It is well-recognized that kids are quicker to pick up on situational, physical, emotional, and contextual differences. They can also learn new adaptation strategies. For instance, a child can intentionally distinguish between communicating with his mother, father, or grandmother by altering traits like word choice, language use, and voice tone. Depending on how they are impacted by the event, children's perceptions of what is different from themselves, how to defend their values, and whether to relinquish them can differ. However, it can act differently in real life in contexts where the process can be simulated, such as in games, theaters, and online environments. The family's habits of spending time together have evolved in the modern day. Since the 1980s, using the media has become a primary means of socialization for families that were unable to do so owing to immigration, cultural imbalance, socioeconomic issues, and other factors. Since everyone has access to a screen today, the families of the past who gathered around to watch television together have been replaced by lonesome people. According to some figures, 75 percent of people in today's culture find it difficult to interact with others. Digital media stands out as the most accessible, affordable, and simple form of social interaction and entertainment in this regard. Children constitute one out of every three internet users worldwide. Due to socioeconomic and cultural issues, it has been seen that people who are unable to socialize in a healthy way struggle with productive thinking, making wise decisions, and time management. They also find it difficult to identify the correct role models, which leads to a sense of worthlessness and social isolation. The new media, which has gained popularity, rapidly conveys information to people about events, facts, and thinking and decision-making processes while ignoring corrective measures. To discover out the truth, for instance, after someone lies at the beginning of a movie, you must watch 156 episodes. The way that women, men, and kids are portrayed in the media is unsuitable for the general public because it focuses on people who are mimicked or strange, who typically act and think with an emphasis on power, and who are unpleasant and sometimes problematic. The problem can be transformed into a less traumatic one by including media representations of people who think well, make healthy decisions, have goals, make sacrifices for their loved ones, and are gentle, kind and thoughtful. Being famous, mediatic, or unique shouldn't be offered in the media as a goal; instead, it should emphasize that there are also regular people and that every member of society is truly significant and respected. These individuals should also be recognized. Otherwise, it is possible to be surrounded by people who are obsessive having a unique appearance and create a fake personality for themselves in an effort to be exceptional. There is a more individualized aspect to the idea of literature and readership. The idea of watching television is entirely different, though. It should not be forgotten that everything that is watched happens in front of the children because they are the center of the family. The media shouldn't release stories that diminish the value of the family. When taking into account the daily flow, it is noteworthy that several news stories, television shows, and films might erode confidence in family members. However, there is a regulation in education that states that only the truth is said in class; mistakes are not written, reworded, or visually represented. However, it is observed that several errors are made again and again on television displays and in front of everyone, in order to visualize feelings of pity as well as feelings such as empathy, sympathy, and self-sacrifice. Filmic identities that appear in television shows, movies, and digital media are complicated and difficult personalities. Producers and filmmakers believe that such enigmatic or captivating personas will be successful. However, it should be emphasized that even for the sake of the movie, their egotistical, money-oriented, overlooking others and values, leading to inequality and injustice should not be allowed and that they do not go unpunished, especially when it is taken into account that the characters of the children on the screen are taken as role models. In other words, fairness and equality that are impossible to attain in reality should

In this study, there are three key topics are explored and discussed. It is first and foremost vital to examine whether the series in question is appropriate for viewers who are older than eight years old. In the second evaluation, it is examined whether contemporary realities and literary narratives have any overlap. This implies that issues related to history, society, psychology, and modernity are approached from a fresh angle. Last but not least, it has to do with the potential social effects of adjusting historical tales to the present. This entails reassessing the past or historical narratives from the viewpoint of the present. However, the key inquiry is: How closely does a film's core meaning align with the meaning that is revealed after seeing it? What about a movie's relevance alters when it's watched again after some time? This study addresses the film-audience connection in two separate ways under the category of "meaning." The relationship

t release stories that diminish the value of the family. When taking into account the daily flow, it is noteworthy that several news stories, television shows, and films might erode confidence in family members. However, there is a regulation in education that states that only the truth is said in class; mistakes are not written, reworded, or visually represented. However, it is observed that several errors are made again and again on television displays and in front of everyone, in order to visualize feelings of pity as well as feelings such as empathy, sympathy, and self-sacrifice. Filmic identities that appear in television shows, movies, and digital media are complicated and difficult personalities. Producers and filmmakers believe that such enigmatic or captivating personas will be successful. However, it should be emphasized that even for the sake of the movie, their egotistical, money-oriented, overlooking others and values, leading to inequality and injustice should not be allowed and that they do not go unpunished, especially when it is taken into account that the characters of the children on the screen are taken as role models. In other words, fairness and equality that are impossible to attain in reality should at least be depicted on a screen. The predetermined patterns of virtual life, however, correspond to people's ideas in the actual world. It is seen that the time spent in front of the screen changes and transforms the family and individuals only with the effect of its intensity, frequency, and negative content. The loss of value that permeates into the home from the screen, the promotion of alternative values, the tolerance intended to be created against them, the excessive use of foul language, and the so-called encouragement of celebrity are all things that need to be taken into account. Family members who imitate movie characters by repeating what they hear on screen result in people who lack the ability to talk on their own way. It should be underlined that, in both serials and other formats, the media is merely a tool and should not become an end in itself. Only then can audiences, kids, teenagers, and families, in general, be safeguarded against potential media harm. It may additionally occur with body image, which is attempted to be heroized or artificially beautified by the impact of the content from youth. Other negative media side effects include language use and devaluation. It is possible to discuss the variables that affect the audience's life expectancy, behavioral patterns, and orientations, as well as the young people who approve of and save money for various aesthetic procedures. In addition to these, it is seen that many subjects such as visiting TV series venues, vacations planned there, economic investments made according to TV series directions, and interior designs are influenced by the media. The family must spend quality time together for each member to develop a proper, healthy, and long-lasting link with the other family members. However, family members are frequently represented in the media as being indifferent to one another, if not outright hostile, and as being willing to give up, ignore, or make sacrifices for one another. It is obvious that we require aware parents, qualified instructors, a highly aware society, and an accepting family structure. According to Ba<unk>der, who is evaluating cinematic semiotics from Metz' perspective (1992), cinematic semiotics differs from narrative semiotics. Moving images, words, noise, music, and writing come to mind when speaking of cinema and a visual language of expression. The editing of the photos, merging the photographs with music, sound, and noise, arranging the colors and objects in a complementary way, arranging the receiver movements, and choosing the shooting angles and sizes are all included in the area of the form of narrative. The content's substance, on the other hand, includes events, ideas, and sentiments that can be imagined or actual, scientifically fabricated, or both. The narrative's shape and structure serve as a pathway for the transmission of feelings and thoughts (Ba<unk>der, 1999). The language, word, form and substance of expression are different from the language, word, form and substance of the content When it comes to adaptations, it's critical to understand exactly what has changed and been replaced from a semiotic perspective. The text's transitions between emotions and concepts must be controlled, and the messages it intends to convey to the audience must be set up in a logical and orderly fashion. Based on the series' real-life development, it can be observed that it began as a novel, then a series of novels, and finally a Canadian television series. Additionally, it is clear that the story spills over into real life, engenders social influences, and results in social, legal, and economic changes in a variety of contexts-from women having jobs and professions to the standard of education and schooling to gender equality. However, it is clear that the format adopted in Turkey fails to connect with the Turkish public in a meaningful way and does not provide a lens through which to uncover, examine, or organize any social reality. However, it is evident that it has the advantage of being a Turkish series that sells outside of Turkey. Furthermore, it is evident that the narrative has real-world implications, influences society, and leads to social, legal, and economic changes in a variety of circumstances, including gender equality, the standard of education, and women having access to jobs and professions. However, it is evident that the format used in Turkey does not effectively engage the Turkish community and does not offer a lens through which to identify, analyze, or organize any social reality. However, it is clear that the fact that it is a Turkish series that is popular abroad is a benefit. The ties Montgomery formed with her own culture, geography, and society, the wars she fought in this context, the courage she exhibited, and her attempts to alter the value system are what make her life story still accessible and valuable despite the centuries that have passed. The support, comprehension, and sharing of emotions of people who share the same stories and life experiences can be assessed as the sale of the books she published in her spare time. It is also conceivable to imagine that readers in later generations used the writings as a guide to comprehend the kind of lives their ancestors led. The fact that the works are still relevant today, despite all the time that has passed, is due to their emphasis on ideas that are not tied to time or place, as well as on people and the changes that even one person can bring about in society, in addition to events, situations, and commonplace occurrences, which results in the creation of heroes. We can explore the social, cultural, and economic challenges of that era with a nostalgic perspective that reveals the social structure of that era when we perceive the series just as the memories of a teenage girl. We can, however, have the opportunity to rebuild the past if we consider this narrative as a component of ethnographic art that offers data that can distinguish our perspective on the past. Similarly, we might regard the narrative as a motivating force that might shed light on the structure of today's society and help us transform society. The accepted function of the array might be determined by which of these choices we add. It holds true for various versions as well. Each society is free to attribute a series of different interpretations based on its own time period, perception, culture, or action focus. However, the most important thing is to be able to recognize the social change, to comprehend why and how it occurred, and to comprehend the part played by the story's protagonists in this change. For those who do not attribute a function of change, transformation and awareness to the series, the series will be no different from an ordinary one watched for fun, and the essence of the historical or social insights it contains will be lost. Even at that time, the main character Anne adopts a perceptive position, much like a Chicago school representative who is very intelligent and monitors societal movements and trends with the intention of changing them when she has the opportunity. When examining the writings of Chicago School intellectuals, it becomes clear that, much like in the series, the emphasis is placed on collective behavior and its relationship to social order and chaos, and that communication is constrained and occupies very little physical space. Anne cares about the same objectives as members of the Chicago School: the advancement of democracy and the smooth operation of the public sphere. For instance, in Park's view, interpersonal communication affects the question of harmony and reconciliation in society. In order for a democratic society to be developed, minorities and individuals with diverse ethnic identities need to come in contact and interact (Gungör, 2013:322). When mass media use is taken into consideration, it is important to note that the media were extremely potent and structured in accordance with the propaganda analyses of the time, which made it difficult for people to make reasonable, logical, and rational decisions. However, this state of power and organization presents a propaganda-based perspective as well as a censorship perspective that affects the development, information, and thinking of society. Later, multiple viewpoints that ranged from such conservative approaches to liberal pluralist perspectives evolved. These ideas and models are based on the working class that developed alongside industrial capitalism, which is one of the most significant forces for change, and the economic and political mass movements that were brought about by this class (Yaylagul, 2006:32). Shannon and Weaver emphasize the need to balance entropy and redundancy while describing the noise component. The transmitted and received message might not always be the same, claim Shannon and Weaver. In other words, the meaning of the communication transmitted by the source and the message received by the recipient could differ. Communication will break down if forwarders are unaware of this (McQuail and Windahl, 2005:31-32). Shannon and Weaver's view of the message is the same as in the stimulus-response model. If the presentation of the series from the social point of view is taken into account as an effect within the context of this evaluation, society should respond to this message. This response may center on praising the acting, the writing, or the aesthetics of the work of art. However, the action-reaction direction and purpose of the series change when the emphasis is on substance, history, research, and challenging the social system when the reactions are different. In his work The Effects of Mass Communication (1960), Klapper presents an inventory of studies questioning the impact paradigm and argues that the effects of mass communication are limited and that it has a function of reinforcing the current situation, among other factors affecting the society The impact of mass communication on society is illustrated using a two-stage flow model (<unk>zçetin, 2018:100). In such a setting, people are not alone; instead, they interact with one another as members of social groups. The fact that mass media messages are transmitted indirectly through social relations despite the fact that a series is highly valued by the public means that there is no immediate and direct response to them. Two processes that become significant in this context are mentioned. Receiving and being interested come first; accepting or rejecting influence and information interference come second. One other factor, reception, does not aim to generate a response to the anticipated effect, as it does not mean a response to the effect, nor does it entail not responding to reception. Individuals are not equal, they are not informed equally, and they do not internalize the messages in a similar way when it comes to the mass media and the messages that emanate from it. However, each individual has a different role in the communication process. At this point, for instance, the society inspired by Mother may begin to exhibit similar tendencies based on her experiences, may need to examine the issues she brings to light, individuals who have had similar experiences in the environments she exemplifies may become active in the society, and testimonies may emerge that suggest that she was right and they all have encountered similar events. Individual entertainment and happiness may not be as essential as satisfaction factors like the understanding of historical events, people's capacity to define their own identities and their status in society. The emphasis now specifically moves to how people make use of the media. The value of reading books, watching documentaries, or watching TV shows has decreased, and what the audience chooses to do with this knowledge now has a greater significance and has the potential to alter the power dynamics in communication. Although it is generally accepted that mass media is waning while social media is growing, this series demonstrates that mass media is not ineffective but rather powerful enough to pressure social/class constraints and power structures into action or even an admission of guilt. The audience is assumed to be made up of people from various societal groupings and classes as a starting point. Hence the influence of media messages being disseminated in bulk rather than individually as in the past. But not just mass emails are sent from this location. For instance, the introduction of ideas like the importance of the family and the importance of childhood carries additional didactic viewpoints in addition to societal messages like the importance of education, schooling of girls, and incorporating them into the economy. However, references made outside of education, awareness-raising, historical, and cultural viewpoints, as well as the recall of experiences, indicate that the media effect has very strongly emerged, and this effect is evaluated progressively and cumulatively. It is important to stress that this can be assessed as the "creation of perception" if desired, or as the "creation of a mental agenda" in the audience, using the agenda-setting hypothesis. It is clear that the days when political authorities could use the media to influence society in order to divert attention from their own acts are over. The public can now set their own agendas, employ the media for this purpose, and even pressure the government to act as a result of this new knowledge. Due to this, it should be noted that the presentation of this series aims to demonstrate framing, the questioning of a particular time period, acts, and their outcomes within the framework of a particular story. Gilman claims that in order to discover their "true" selves, people occasionally require stereotypes. They claim that the flaws or strengths of the individuals are what they see or cannot see in the stereotypes (Gilman, 1985). Because "identity" is now regarded as something that can be worn or removed, much like a piece of jewelry or clothing that can be worn from time to time. For a very long time, researchers have explored how media affects identity. Marketing and promotion initiatives, which are prominent at the same time as movies, assist viewers to internalize their fictional heroes even more and give them resources to help keep them as immersed in the cinematic narrative as possible. It's pretty intriguing that there hasn't been any Anne with an Erelated merchandise found since no notebooks, t-shirts, lunch bags, or school bags have been released on the market. In this case, it might be argued that there is no need for a different item because Anne's persona is crucial and impossible to replicate. However, the character Anne's identity is adored since she relates to many people as an orphan, a young girl going through the puberty transition, and someone who has a throne in everyone's hearts thanks to her thoughts and feelings. Furthermore, with a historyoriented identity, it appears to have met the criteria for universal acceptance both as a result of interdisciplinary approaches with the influence of postmodern structuralism and functionalism as well as with the aid of the concepts of polyphony, multilingualism, and multiculturalism supported by global policies. It would be quite suitable to title it "Mobility" if the character "Anne" had come to a stage where it required to be handled and explained with a single concept. Because mobility not only has an impact on the societies or individuals that respond to it, but it also alters them visibly. This assessment can be argued to apply to all the female characters in the series, whose viewpoints, actions, and behaviors develop alongside Anne's. As a heroine who prepares, presents, and realizes this transformation and shifts in society, Anne observes society and illustrates her firsthand experience to present us profoundly how society has changed. The more the message occupies and remains in memory, the more impressive it has become, in Jamieson's opinion (Jamieson;1996:25). According to Severin and Tankard (1997:408), an information gap theory is in fact at work in this situation. The theory contends that knowledge about the topic spreads among the social classes as long as it is a topic of discussion. The lower layers with information deficit (those who haven't seen/heard about the subject before or haven't voiced their opinion on the subject) will be more receptive to getting information, while the upper layers will act more reluctantly, and as a result, the information gap will be broader. However, somehow, the balance will be restored (citing from McQuail, 1993:107, An<unk>k, 2000:73, Severin and Tankard,1997:408). Generally speaking, different goals may employ various text analysis formats. To make 'things' visible, however, we must first consider what we are looking for and examine it with a'suitable' tool that allows us to perceive it. The findings we can observe will also depend on how accurate the measurement tool is. It is also obvious that if we look without the proper equipment, we won't see what we need to see. The movement that we experience in the narratives is handled in a somewhat different manner; it is described as the "I" becoming the "other" or the self-replacing the other. Instead of presenting a fresh viewpoint, the development of a new habit entails a relocation and a reevaluation of lifestyle, society, and expectancies of life. Building a separate universe in a dynamic environment is the aim. A new interpretation of the universal cultural codes "me" and "the other" arises out of this particular issue. A newly established persona is basically a new hat to wear. It presents idealism, nationalism, globalization, local and national ideals, and transformed identities in a novel way. We can also see how much this is reflected in the films given that 45.88% of the film productions are adaptations of books or short tales. Films based on books make up 70% of the top 20 grossing films worldwide. On average, films that are adapted from books generate 53% more revenue than original screenplays globally (Talbot, D., February 23, 2023, Impact of Book Publishing on Film Industry https://wordsrated.com/impact-of-book-publishing-on-film-industry/). It may even be argued that interpretations and reflections that are "standardized" and based on madeup information rather than actual knowledge become increasingly more widespread. With its subject, performers, and narrative, the movie is a packed product ready for the viewer. It has to measure up to and be accepted by the audience, who is the target consumer for this product. Only in this way can versions of well-known stories, performed by actors from various eras, nevertheless elicit excitement and curiosity and appeal to a broad audience. During this period of development, the known and unknown components of identity have begun to be primarily built through fiction, the aspects that conflict with western culture have been sharpened, brought to the fore by being revealed, and the like aspects have been rasped. The movies have been one of the most effective approaches for creating identities in the digitalized and visualizing world of the 20th century. In this way, identity and cultural codes were disseminated to the general populace, and related identities and codes were repeatedly used. The same groups, identities, and cultural norms have just been given alternative roles in various narratives. It's sometimes possible to tell ancient tales in an innovative manner by enhancing their visual components. There is a connection between representation, differentiation, even diversification, and power within standardization. Power can be utilized here not only in the sense of economic expansion and physical dominance but also in a broader cultural or symbolic sense of representing one another in a different way. However, the nature of this power needs to be properly addressed. Said argues that standardization contains linkages among representation, difference, and power (Said, 1978). But the idea of power needs to be given careful thought-not just in the economic and physical senses, but maybe more generally in the cultural and symbolic senses, as the ability to represent someone or something in a particular way. The dynamics created by orientalism have never been lost, and this narrative form continues to remain active without losing much of its content, even when modernism creates new and distinct narrative forms and introduces other types of mobility to the audience. To illustrate this, a long storytelling of various narratives embedded in the basic narrative can be mentioned. Years later, as a television series, the reality that began as a true story and was later turned into a novel and a serial novel, connects with new viewers and readers. Here, actual life's impact on the story is first brought to light, and it is then clear that the impact of the story on real life is greater. The subject is freed from the content of all societal patterns and motifs peculiar to that time when it is handled by a Turkish teenager in the context of Turkey. The issue is almost reduced to the fact that an orphan girl can find a roof to live in. Even in this condition, the narrative can become widely recognized by referring to itself with the "Turkish Series that sells abroad" label. It can be stated that a far more composed, thoughtful, and inquisitive audience has emerged as a result of Montgomery's work being turned into a TV series. The series quickly succeeded in claiming a throne in the hearts of the viewers and attracted a large audience by winning awards. It was accompanied by numerous national events and phenomena in the Canadian landscape. Netflix's global reach made it possible for the series to travel to Turkey as well as other parts of the world. Turkish producers decided to develop a Turkish version of a similar series in response to the series' global success as well as the Turkish public's acceptance and appreciation of it. Even if the Turkish adaptation of the series appears to have been a success, it is not possible for it to convey the message of the original series despite the use of competent performers and precise and planned shots. The Anne with an E series' protagonist, who lived during that time period and carries the traces of their social life, accomplishments, and failures, is hard to duplicate and use now. There is no one in Turkey who shares his passion for writing and his early career as a writer in Canada's first newspaper. Or, the adventure of teaching in their own tiny town will present unique adventures that will never correspond to Canadian reality. Reconstructing the conceptual equivalents of today such as gender equality, education problems, understanding of benevolence, and orphanhood of that period will not be out of the question. However, it goes without saying that new and modern productions are required within the framework of policies developed and put into place on topics like sector development, observing global trends and adapting them to our nation, increasing employment, accelerating regional development, and ensuring income justice. Additionally, one may favor adaptations for a variety of reasons, including public diplomacy initiatives including Turkish TV programs, language promotion, and spreading Turkish culture. Films that have been adapted from books, true stories, or other movies are always a possibility, but it's important to take care when transferring the social transformation, time, place, and values that underlie the adaptations to the adaptations and to shield the characters and dialogue from their historical setting and unique context. In conclusion, this study aimed to question if the literary narrative and real-life stories overlap with today's realities in the example of "Anne with an E" TV serial. In this study, there were three key topics to be explored and discussed. Depending upon the data collected through the explorations, the main question of if the series in question is appropriate for viewers who are older than eight years old is replied as negative. Even if it seems to be too naive for youngsters, this serial is found to be too much for kids at the age of +12 or +14 could be more appropriate. Usual restrictions might involve, abuse, sex, drugs etc. but this one could even cause harm as much as these. The study was questioning if contemporary realities and literary narratives have any overlap and this requires a positive reply. Most of the realities mentioned in the narrative had something to do with the real realities. Having its roots as an autobiography at the beginning and covering most of the events, and perceptions of those times, the narrative couldn't be accepted as a documentary but a narrative yielding valuable and neutral information of the era. This implies that issues related to history, society, psychology, and modernity are approached from a fresh angle. With its positive qualities not causing harm or violence, the narrative was also questioning the past of a nation, revealing certain facts into the light even if it is about ten decades later. This was the potential of the present time, to hear the whispers of the past, to maintain the cure for the old wounds. Thus, all the Montgomery books provided a good setting in the TV serial addressing the film-audience connection in the best possible way revealing the text's complementary meaning through its social, psychological, political, and historical elements. --- Conflict of Interest Statement The authors declare no conflicts of interest. --- About the

In this study, there are three key topics are explored and discussed. It is first and foremost vital to examine whether the series in question is appropriate for viewers who are older than eight years old. In the second evaluation, it is examined whether contemporary realities and literary narratives have any overlap. This implies that issues related to history, society, psychology, and modernity are approached from a fresh angle. Last but not least, it has to do with the potential social effects of adjusting historical tales to the present. This entails reassessing the past or historical narratives from the viewpoint of the present. However, the key inquiry is: How closely does a film's core meaning align with the meaning that is revealed after seeing it? What about a movie's relevance alters when it's watched again after some time? This study addresses the film-audience connection in two separate ways under the category of "meaning." The relationship

Introduction Health is complex and defined by the World Health Organization (WHO) as not just the absence of disease but a state of complete physical, mental, and social well-being [1]. Access to appropriate health services including preventive health or routine visits is an important heath determinant. It may be helpful to characterize the WHO components of mental and social well-being by looking at associations between health care utilization and potentially modifiable measures of social-emotional support, life satisfaction, and mental health. Social support includes positive relationships, encouragement, and emotional support that is often provided by a network of family, friends, colleagues, and others [2]. Social support is closely tied to effective coping [3] and has health-protective properties [4,5]. Social support has been associated with reduced morbidity and mortality [6] and been found to aid successful aging [7]. Social support may contribute to these effects through many mechanisms including promoting better adherence to diet [8] and physical activity recommendations [9][10][11], as well as utilization of medical treatments and screenings [12,13]. Life satisfaction, a subjective self-evaluation of an individual's overall contentment with life and well-being [14][15][16], is a Healthy People 2020 marker of quality of life [17]. Associations have been documented between dissatisfaction with life and increased risk for all-cause morbidity, chronic conditions, injury, and mortality. Life satisfaction is associated with positive health behaviors, higher levels of self-reported health, decreasing prevalence of obesity, as well as smoking, heavy drinking, and physical inactivity [18,19]. Mental distress can be intermittent or chronic and refers to a wide range of experiences, including stress, depression, and emotional problems. Frequent mental distress (mental health problems experienced on 14 or more of the past 30 days) has been found to be associated with unhealthy and high-risk behaviors such as smoking, alcohol and drug use [20-22, 26, 27]. Mental distress is also associated with medical conditions such as heart disease, asthma, and arthritis, as well as health outcomes such as mortality and hospitalizations. [23][24][25]. Routine checkups, during which clinical services are performed for a person who may not have any signs or symptoms of an illness [28], are an important aspect of preventive health care. The routine checkup also provides an opportunity for primary caregivers to offer personalized behavioral counseling to patients [29], provide recommended screenings [30,31] and promote healthy behaviors. Those who regularly see doctors for health examinations develop a trusting relationship with physicians, and are more likely to receive preventive screening than those who only see doctors for treatment episodically [32]. The routine checkup is a cost-effective means of reducing morbidity and mortality caused by chronic diseases [30], since much of the morbidity and mortality related to chronic disease in the US is attributable to modifiable health behaviors [25,31,33] that could be addressed during a routine checkup. Routine periodic health exams provide opportunities for promoting healthy behaviors and chronic disease prevention. Previous research on the effects of social, emotional, and mental health on the utilization of health services such as a routine checkup has not yet yielded a clear picture of these relationships. One previous study found that social support was positively associated with service use and annual physical examination use in older adults [34], which may suggest that social networks aid users in gaining access to care [32]. Contrastingly, other studies have found that lower social support was associated with higher health services utilization in a general population of adults [32,35]. Variability in the conceptualization and categorization of social support [36] and the utilization of health services may have led to these mixed findings. Little research has examined the association between life satisfaction and health behaviors [19,37]; with most existing studies conducted in the 1970s [19,[38][39][40]. An association has been shown to exist between increased distress and reduced utilization of health services [35]. In a study of older adults, frequent mental distress was shown to decrease the likelihood of practicing positive health behaviors and utilizing preventative health care services [31]. Overall, women are about twice as likely to receive a regular medical examination than men [28,32]. However a recent study using BRFSS data demonstrated that women aged 25-44 were significantly less likely than women older than 44 to have had a recent checkup [28]. In addition to benefits provided to the women themselves, counseling on medical and behavioral risks provided during a routine checkup can improve pregnancy outcomes through the provision of anticipatory guidance [41]. Furthermore, women are more likely to experience life dissatisfaction [19] and frequent mental distress [31] than men, and these factors may have stronger associations with health services utilization for women. For example, in a study of Canada's general population from 1994-1995, social structural and psychosocial determinants of health including social support was shown to generally have a stronger effect in women whereas behavioral determinants were generally more important in men [42]. No comparable examination of social and mental health on the utilization of health services for the US population could be found. The main objective of this study was to examine the associations between social-emotional support, life satisfaction, and mental health with not having a routine checkup among women of reproductive age (18-44 years) in the US. Identifying relationships among these psychosocial measures among women, variation across sub-populations, and how they relate to receipt of routine checkups may contribute to focused strategies to increase access to clinical services and improve overall health outcomes. --- Methods --- Data Source The Behavioral Risk Factor Surveillance System (BRFSS) is a state-based cross-sectional telephone survey of adults 18 years of age and older conducted each year by state health departments in collaboration with the Centers for Disease Control and Prevention. The BRFSS uses a standardized questionnaire to determine the distribution of demographic variables, health practices, risk behaviors, and health care access among the noninstitutionalized adult population in the US. The Council of American Survey Research Organization (CASRO) response rate reflects both telephone sampling efficiency and the degree of participation among eligible respondents who were contacted. In 2009, the BRFSS median CASRO rate was 53 % and ranged from 38 to 67 % among states. The cooperation rate reflects the proportion that completed an interview among eligible persons contacted. The median cooperation rate for BRFSS in 2009 was 75 % and ranged from 55 to 88 % among states. Detailed information on the sampling methodology, survey weighting procedures, quality assurance of the survey, and other aspects of this survey is available online at http://www.cdc.gov/brfss/index.htm. Relative to other surveys, data from BRFSS have acceptable reliability and validity [43]. --- Study Participants In 2009, a total of 67,730 women of reproductive age (18-44 years) completed the BRFSS. Women who responded "don't know/not sure", who refused to respond, or who were missing data for the question on time since last routine doctor visit (n = 776) were excluded from analysis. The final analysis sample included 66,954 women. --- Dependent Variable Having a routine checkup in the previous year was the primary outcome and was grouped into a dichotomous variable: had a checkup in the past year or no checkup in the past year from the available response options (Table 1). --- Independent Variables In this study, there were three primary independent variables: social-emotional support, life satisfaction, and mental distress derived from the available response options for each question (Table 1, Fig. 1). Social-emotional support was grouped into three categories: Always, Most of the Time, and Sometimes or Less. Life satisfaction was also grouped into three categories: Very Satisfied, Satisfied, and Dissatisfied. Frequent mental distress was grouped into a dichotomous variable: of No for less than 14 days versus Yes for 14 or more days of mental distress in the past 30 days. This dichotomization is consistent with literature examining frequent mental distress in BRFSS analyses [31]. For each independent variable, respondents who refused to answer, did not know the answer to a question, or had a missing response were excluded from analysis of the specific question. Missing responses for each variable ranged from 1 to 7 %. These three variables were highly correlated with each other and were analyzed separately in order to avoid the impact on the variance of parameter estimates that might result from collinearity. --- Covariates Age was categorized into six groups: 18-24, 25-29, 30-34, 35-39, and 40-44. Selfidentified race/ethnicity included non-Hispanic white, non-Hispanic black, Hispanic (any race), Asian, Native American, or all others. For this report, the terms white and black refer to non-Hispanic whites and non-Hispanic blacks, respectively. Native American was used for both American Indian and Alaska Natives. The "other" race category included the heterogeneous group of Native Hawaiians, Pacific Islanders, other racial groups, and those who indicated more than one race. Education level was classified as: less than a high school diploma or graduate equivalency degree (GED), high school diploma or GED, some college, and college graduate. Annual household income was classified into quintiles as follows to account for the distribution of income in BRFSS: less than $10,000, $10,000-$19,999, $20,000-$34,999, $35,000-$49,999, $50,000 or more, and refused or not known. Health care coverage status was categorized as those that were covered by health insurance, prepaid plans, or government plans compared to those without any health care coverage. Employment status was dichotomized with employed for wages and self-employed considered "Employed," and being out of work, homemaker, student, retired, or unable to work considered "Not employed." Marital status was dichotomized into married and "Not married." Participants provided subjective ratings of general health status based on the question: "Would you say that in general your health is excellent, very good, good, fair, or poor?", and these were dichotomized into "good or better" and "fair or worse". Finally, participants were asked how many children less than 18 years of age live in their household. Their responses were dichotomized into children at home (number of children living at home > 0) or no children at home. --- Statistical Analyses To account for the complex multistage, stratified sampling design of the BRFSS, we conducted the analysis using SAS-callable SUDAAN version 10.0.1 (RTI International, Research Triangle Park, NC, USA). We calculated weighted prevalence estimates and corresponding 95 % confidence intervals (CIs) for not having had a routine checkup in the year before the survey by socio-economic status and demographic characteristics among reproductive age women. We also calculated weighted prevalence estimates for all levels of the three independent variables by socio-economic status and demographic characteristics. However, only data from those with the lowest level of social-emotional support, the lowest level of life satisfaction, and those with frequent mental distress are presented in the third table for clarity. Significant differences in the distribution of these factors were determined by <unk> 2 test with a p value <unk>0.05. Three separate models were developed using multivariable logistic regression analysis to assess relationships for each of the three measures (socialemotional support, life satisfaction, and mental distress) with not having a routine checkup within the past year. After including all covariates in an initial model, a manual backward selection procedure was used to produce the final models. A change of 10 % in the log likelihood ratio between models was used as the criterion. The final model for each of the three main independent variables (social-emotional support, life satisfaction, or mental distress) included four covariates: age, race/ethnicity, income, and health care coverage. --- Results Among US women 18-44 years old, the majority were in good or better health, had health care coverage, and had at least one child under 18 years old living at home. More than onethird of the women were of racial/ethnic groups other than white, were college graduates, and had annual household incomes over $50,000 (Table 2). Approximately one-third of the women in this sample reported they had not had a routine checkup in the past year (Table 3). Women aged 18-34 were less likely to have had an annual checkup compared to women aged 35-39 and 40-44. The prevalence of not having had a routine checkup in the past year was significantly lower among women who were Black compared to all other race groups, college graduates compared to those of lower education, those with health care coverage compared to those without, and among those with an annual household income of $50,000 or more compared to all other levels of income. In addition, the prevalence of not having had a routine checkup in the past year was significantly higher among women who were not employed, not married, and reported a fair or worse general health status, compared to those that were employed, married, and reported good or better health status, respectively. Overall, nearly 20 percent of women reported receiving the lowest level (i.e., sometimes or less) of social-emotional support, while 5.6 % reported being dissatisfied with their lives, and 13.1 % reported experiencing frequent mental distress (Table 3). There were also several differences in prevalence for each of these three measures (e.g., social-emotional support, life satisfaction, and frequent mental distress) according to levels of other examined characteristics. Women with an education of high school or lower, the lowest incomes, without health care coverage, not employed, not married, and who reported a fair or worse general health status had the highest prevalence of all three measures, compared to their respective counterparts. For the social-emotional support measure, there were distinct differences among groups in the prevalence of not having a routine check up in the past year. Among those who reported that they received the social-emotional support they needed sometimes or less frequently, 40.9 % did not have a routine checkup in the past year (Table 4). Among those who reported they received the social-emotional support they needed most of the time, 35.4 % did not have a routine checkup in the past year. Among those who reported that they always received the social-emotional support they needed, 29.9 % did not have a routine checkup in the past year. The significant differences in the crude measures persisted in the final model adjusting for age, race, education, and health care coverage with women who reported receiving social-emotional support most of the time (AOR = 1.29, 95 % CI 1.20-1.38) and women who sometimes or never received social-emotional support (AOR 1.47, 95 % CI 1.34-1.61) more likely to have not had a routine checkup compared to women who always received the social-emotional support they needed. There were differences in the life satisfaction measure among those not having a routine check up in the past year. Among those who reported being dissatisfied with life, 44.7 % did not have a routine checkup in the past year (Table 4). Among those who reported being satisfied with life, 36.5 % did not have a routine checkup in the past year. Among those who reported being very satisfied with life, 29.3 % did not have a routine checkup in the past year. The significant differences in the crude measures persisted in the final model adjusting for age, race, education, and health care coverage, with women who reported being satisfied (AOR = 1.27, 95 % CI 1.19-1.36) and those that were dissatisfied (AOR = 1.65, 95 % CI 1.43-1.91) more likely to have not had a routine checkup compared to women who reported they were very satisfied with their lives. Among those who reported frequent mental distress, 39.7 % had no routine checkup in the past year compared to 32.8 % for those who did not report frequent mental distress (Table 4). The significant differences in the crude measures persisted in the final model adjusting for age, race, education, and health care coverage, with women with frequent mental distress (AOR = 1.19, 95 % CI 1.09-1.30) more likely to have not had a routine checkup compared to women who experienced less frequent mental distress. --- Discussion Over one-third (approximately 18.8 million) of US women aged 18-44 years had not been to the doctor for a routine checkup within the past year in 2009. This study found that women who experience lower levels of social-emotional support, low levels of life satisfaction, and frequent mental distress were more likely to have not had a routine checkup even after adjusting for factors related to socio-economics and health care coverage. The routine checkup is considered the ideal occasion for screening for health behaviors [29,32,33,36]. Previous research has shown that a recommendation by a physician has an impact on the use of screening services, suggesting that those who have not developed a trusting relationship with a physician may be even less likely to undergo screenings, or initiate and maintain behavioral change [44][45][46]. Women with low levels of socialemotional support, low levels of life satisfaction, and frequent mental distress maybe most in need yet are less likely to have a routine checkup where they could be screened for and treated for appropriate physical or mental health conditions. It may be helpful for other patient-provider interactions, such as family planning and prenatal care visits, to also include screening for social-emotional and mental health to provide more comprehensive care to women of reproductive age. Screening results indicating a lack of social-emotional support or poor mental health could then be used to refer patients to community resources to address these issues when available. Other factors that may influence health care utilization among reproductive-aged women include age, race, education, type of health care coverage, and accessibility of providers [28,29,32]. Our study identified similar results among the available variables, with women who had lower education, lower incomes, and those without health care coverage being less likely to have had a routine checkup. Many of these same factors also are increasingly being referred to as Social Determinants of Health, are complex in nature, and may result from long-term health inequalities [47]. A recent study of women by Mosca et al. [48] showed that awareness that cardiovascular disease was the leading cause of death in women nearly doubled from 1997 to 2005 and was associated with positive action to reduce risks (added more physical activity, avoided more unhealthy foods, quit smoking, and lost weight). The most frequently cited barriers were not related to health care access, but were to confusion in the media (49 %), the belief that health is determined by a higher power (44 %), caretaker responsibilities (42 %), and didn't perceive themselves to be at risk (36 %). Thus, it is important to develop programs that can address multiple factors such as these, other Social Determinants of Health, and some of the more typical barriers to care (i.e., health coverage, availability of providers, costs, and adequacy of health coverage) in order to promote effective utilization of routine health care services. Strengths of this study include a large sample size of women of reproductive age to assess these three measures of social-emotional and mental health. Another strength is that the estimates are generalizable to the US population of women aged 18-44 years in all states. However, this study is subject to some limitations. First, BRFSS data is based on self-report and subject to recall bias. For example, previous research has demonstrated that self-reports of health services utilization tend to overestimate the actual extent of screening and doctor visits [49,50]. Additionally, some individuals may have reported receipt of a routine checkup for what was in fact an illness-related visit [30]. However, there is no reason to believe that over-reporting of routine checkups affected any group more than another in this study. Thirdly, BRFSS is limited to community-dwelling adults with a landline telephone and does not include institutionalized adults nor those without access to a landline telephone. Therefore, the social-emotional and mental health measures presented in this study thus may be underestimates and not representative of populations not surveyed. A fourth limitation could result if women who were pregnant within the past year accessed health care services for prenatal care when many preventive primary care services are provided and may or may not have considered it to be a routine checkup as listed in the BRFSS question. Unfortunately, this level of detail about pregnancy in the past year is not available in the data and a differential response to this question based on pregnancy history is not anticipated. There is a question about being currently pregnant in the survey and it was decided to those that were pregnant in the analysis in order to be representative of the age group. Lastly, the cross-sectional nature of BRFSS only captures information on the measures in this study at a single point in time and does not allow examination of temporality. Therefore, the study cannot determine causality; but does identify potential areas for additional investigation into the relationships among social emotional support, life satisfaction, and mental distress with having a routine checkup. These limitations notwithstanding, the results of this study provide useful information on the prevalence of low levels of social-emotional support, low levels of life satisfaction, frequent mental distress, and their associations with the utilization of preventive health services by women of reproductive age after controlling for potential confounding factors. These measures are important health constructs that help identify women less likely to utilize health services. This information can be used in conjunction with established risk factors to develop focused strategies that provide appropriate support so these women access clinical services in order to increase exposure to preventative health opportunities and improve overall health. --- Author Manuscript Willet et al. Page 13

To examine the associations among social-emotional support, life satisfaction, and mental health with not having a routine checkup among women of reproductive age in the US, data from the 2009 Behavioral Risk Factor Surveillance System, a population-based telephone survey of health behaviors, were analyzed among reproductive aged (18-44 years) women in the US. Prevalence estimates were calculated for not having a routine checkup in the past year with measures of social-emotional support, life satisfaction, and mental distress. Independent multivariable logistic regressions for each measure assessed not having a routine checkup within the past year with adjustment for age, race/ethnicity, education level, and health care coverage. Among women of reproductive age, 33.7 % (95 % CI 33.0-34.4) did not have a routine checkup within the past year. Factors associated with not having a routine checkup included: having social-emotional support most of the time (AOR = 1.29, 95 % CI 1.20-1.38) or sometimes or less (AOR = 1.47, 95 % CI 1.34-1.61) compared to those who reported always having the social-emotional support they need; reporting life satisfaction as being satisfied (AOR = 1.27, 95 % CI 1.19-1.36) or dissatisfied (AOR = 1.65, 95 % CI 1.43-1.91) compared to being very satisfied; and frequent mental distress (AOR = 1.19, 95 % CI 1.09-1.30) compared to those without. Women who report lower levels of social-emotional support, less life satisfaction, and frequent mental distress are less likely to see a doctor for a routine checkup. Targeted outreach that provides appropriate support are needed so these women can access clinical services to increase exposure to preventive health opportunities and improve overall health.

In a speech addressed to the United Nations in 2008, Finland's first female president Tarja Halonen drew attention to the long tradition of Finnish maternity packages as "the nation's gift to every new-born". The Finnish state began to distribute maternity boxes in 1938 to counteract the declining birth rate and the increase in infant mortality; but from 1949 onwards, they were given to all newborns regardless of income or social status (Näsi and Koskenvuo, 2022). The box, issued by the Social Insurance Institution of Finland (Kela), contains baby clothes and other child-care essentials such as a sleeping bag, a towel, a feeding bib and a cuddly toy. 1 The box itself comes with a mattress and can be used as the baby's first bed. As Ella Näsi and Karoliina Koskenvuo (2022, p. 411) note, the baby box is still a very successful and popular social policy in Finland, where the benefit "can be obtained as a maternity package [...] or an alternative cash benefit. [...] The baby boxes are provided free of charge [...] to all pregnant mothers or adoptive parents residing in Finland, and they are accepted by 95 per cent of first-time parents." Over the years, the contents of the box have mirrored current views of Finnish children, child-care practices, and reading. From 1982 onwards, the package has included a baby book. Since these books reach an average of 30 000 Finnish families every year, 2 they have a vast societal impact. Despite this state support of early literacy, the baby box books have received surprisingly little scholarly attention. In the baby box, prevailing attitudes and norms on health, well-being, population, society, nation, individuality, welfare and childhood converge (Taskinen, 2013;Haataja and Koskenvuo, 2017;Koivu, et al., 2020;Näsi and Koskenvuo, 2022). The maternity package is both a unique phenomenon internationally and a significant part of the Finnish history of childhood. Each year, the launch of a box with a new collection of baby clothes is a matter of lively debate covered by media, both at home and abroad (Tierney, 2011;Lee, 2013;Pells, 2016;Haataja and Koskenvuo, 2017;Smirnova and Selänne, 2018;Finnwatch, 2019;Koivu, et al., 2020). So far, research has focused on the social and didactic aspects of the maternity box. We, however, want to focus exclusively on the included baby books. For this article, we have chosen to take a close look at the baby box books distributed during the 2000s. The maternity package has not included a new book every year, and some titles have been offered more than once. The following 16 titles were included in the maternity packages distributed from 2000 to 2022: add temporalities deviant from the ones parents have been involved in before having a baby. Our article will begin with a brief statement of our view of baby books in the light of previous research, followed by an exposé on how our term infant temporality emanates from and builds on previous temporality studies. We then move on to recap previous research on the maternity box and its baby books, before presenting and exploring specific cases of infant temporality in the selected baby box books from the 2000s. --- The Aesthetics of Baby Books In line with Lian Beveridge (2017, p. 19), we define baby books as picturebooks for babies and toddlers up to 2 years of age. Regrettably, these books are often treated as objects rather than literature. Hence, few literary scholars have taken a serious interest in them (Kümmerling-Meibauer and Meibauer, 2005, p. 325;Hurme, 2006, p. 14;Kümmerling-Meibauer, 2011, p. 1;Kümmerling-Meibauer and Meibauer, 2018, p. 149). Perry Nodelman (1988, p. 36) is, however, among the first children's literature researchers to point out the complex skills needed to read picturebooks, including baby books. Bettina Kümmerling-Meibauer andJörg Meibauer (2005, 2018) have likewise convincingly argued that the visual grammar that infants must possess in order to interpret images in early concept books is quite complex. They also raise the important point that baby books support literary competence, and should therefore be regarded as a form of "pre-literature", since they constitute "a preliminary narrative form. Through oral storytelling, which establishes a connection between the pictures or hints at the relation between an image and a real object, young children will be introduced to a form of narrative" (Kümmerling-Meibauer and Meibauer, 2005, p. 342). A few studies also discuss the reading of baby books. Whereas Betsy Hearne (1992) introduces chewing as reading, Lian Beveridge (2017) sees the sensuality of chewing as part of a toddler's reading process associated with early sexuality. We have also set out to study the baby books distributed with the Finnish maternity package from a literary point of view. We share the idea that baby books have the potential to be a form of pre-literature that constitutes an infant's first meeting with visual and textual art and narrative. Yet, we acknowledge that the aesthetic and literary qualities of these books vary greatly. For example, some years the maternity box has contained mass-produced Central-European baby books of poor quality, such as Look What I Can Do! and Nice to Watch!, which present a cavalcade of rather odd objects that infants are supposed to find interesting. According to Vuokko Virtanen (2006, p. 32), the reason behind the import of these bland mass-market books was economic, since foreign originals were cheaper and quicker to produce. Clearly, the kinds of books included in the maternity box are a result of conflicting artistic, pedagogical, ideological, and economic interests that, at times, have resulted in mainstream and bland books. Still, the baby box books deserve to be studied as preliterature, since they reach such a large audience and may well be many new parents' and their infants' first encounter with books aimed at the very youngest. As we will show, one of the many discourses converging in the baby box books specifically produced for the maternity package is temporality. Since there are no earlier studies of temporality in baby books, we want to draw attention to how these seemingly simple books are part of a larger discourse of temporal normativity. As James Reid and David Swann (2019, p. 319) note: "All versions of the baby box convey symbolic messages." In our view, temporality is such a "message". --- Time, Temporal Normativity and Infant Temporality Time and temporality are fundamental aspects of all literature, as well as of the study of literature. Thomas M. Allen (2018) in fact identifies a "temporal turn" in literary criticism, when time is re-actualised as an analytical category within Literary Time Studies. More specifically, our understanding of temporality in the baby box books builds on queer temporal literary studies, which have critiqued late modern normative temporalities. 4 Since the notion of linear time builds on the idea of aging and growth as progression and reproduction, linear time -from a queer perspective -is considered "straight" (Freeman, 2010, p. xii). Jack Halberstam (2005, p. 1) argues that normative timelines -which typically include maturation, marriage and reproduction -can be interrupted by "strange temporalities, imaginative life schedules, and eccentric economic practices" emanating from queer subcultures. Other ways to disrupt normative timelines include shifting focus onto depictions of, e.g., anachronisms, non-maturation, and no future (Edelman, 2004;Halberstam, 2005;Freeman, 2010). Queer temporal literary studies make us aware that bodies age in various ways and that norms concerning maturation and development vary across time and space (Stockton, 2009;Pugh, 2011;Joosen, 2018;Malewski, 2021). Within children's literature studies, Kathryn Bond Stockton (2009) has coined the concept growing sideways in order to capture such non-normative timelines when growing up queer. Similar to queer temporality, we argue that the concept of infant temporality constitutes a non-normative temporality with the potential to disrupt prevailing normative timelines. Elizabeth Freeman's (2010) concept of chrononormativity or temporal normativity underpins our upcoming analysis, as a technique through which institutional forces enact, regulate, and discipline individuals with clock time, biological time and reproductive time. Freeman (2010, p. 3) argues that everyday life is divided into imagined time slots, which give those who follow the given norms an advantage. We argue that there is a need for a nuanced methodology for uncovering temporal normativity in literature aimed at babies. Equally important for our understanding of temporality in the baby box books is German sociologist Hartmut Rosa's widely influential theory of time. Rosa has introduced the linked concepts of social acceleration and deceleration in order to demonstrate how time is speeded up in late modern contemporary societies. According to Rosa (2019, p. 64), contemporary temporality consists of the following three dimensions: techno-logical acceleration, social acceleration and the acceleration of life rhythm. None of these competing forces is positive or negative as such. Instead, what matters is whether they foster resonance, which is understood as a feeling of deep connection with others. Although time and temporality are of great importance in all literature, they have a special significance within children's literature studies and keep recurring in discussions of what children are and what children's literature is. 5 Emma Uprichard (2008) emphasises how childhood studies and children's literature studies tie temporality to conceptualizations of childhood. According to her, temporality is the defining feature of both children and childhood. Clémentine Beauvais (2015) develops this train of thought further by studying how adult power and child empowerment have been debated for decades. Beauvais (2015, p. 24) concludes that for children, power lies within possessing more future time, and hence potential to act, in relation to adults who may have power, but less time left to exert it: "[a] child growing up is not subjected to the passing of time; its elongating limbs and developing existence are the passing of time". Like Uprichard, Beauvais concludes that children and childhood are not only defined by time, they are time. Queer temporal literary studies have drawn our attention to how non-normative temporalities can interrogate chrononormative timelines in children's literature. Hence, studying time and temporality may give us a new and deeper understanding of the baby box books. Queer temporal studies' view of time helps us address the books' deviant and competing timelines, which are primarily associated with age, understood as infancy and parenthood. Consequently, we aim to show that the books studied contain non-chrononormative temporalities that define babies' bodies and timelines as different from those of adults. In other words, queer temporal studies inform our concept of infant temporality in order to address the age-related gap in how babies and their caretakers perceive time. --- Studying the Maternity Package The Finnish maternity package was introduced in order to compensate for the lack of essentials during wartime and to counteract the decreasing birth rate and increasing child mortality rate. The contents of the box have over time mirrored contemporaneous attitudes towards childhood and child-care. For example, the name "maternity package" was recently criticized for being gendered and too exclusively tied to motherhood, hence ignoring other vital caregivers such as fathers; and today, the package is colloquially referred to as a baby box or parent package (Hanhinen, 2018;Näsi and Koskenvuo, 2022, p. 421). The box contains clothes, essentials, information material and -since 1982 -a baby picturebook. The maternity package as such has received plenty of scholarly interest, mainly by researchers within childhood studies, social studies and pedagogy. These studies hardly ever mention the books in any great detail. The issues raised and debated tend to focus on the success of this national policy in terms of welfare and health, as well as ideological issues, such as gender equality and decolonisation (cf. Suvilehto, 2014Suvilehto,, 2015Suvilehto,, 2018;;Reid and Swann, 2019;Koivu, et al., 2020;Näsi and Koskenvuo, 2022). A recent research project called "Baby Box as an Emotional Object: the Sociomaterial Experiences of the Finnish Welfare State from the 1930s to the Present" (2019-) studies the intergenerational meanings and memories related to the maternity package from a socio-cultural-material perspective. On their project website, the researchers emphasize that "the baby box is extremely rich with subjective and collective experiences that are multilayered, even sometimes contradictory". 6One prominent trend within the study of the baby box is to study the launch of the concept in other countries. In addition, Finnish social institutions closely monitor the reception of the maternity box. One such survey shows that astoundingly 92% of those who received the box in 2006 acquainted themselves with the included book Tumble Time (Miettunen, et al., 2006, p. 10). Some of the recipients praised the book for its bright colours and for supporting early interaction between caregiver and child, whereas the didactic elements and the bilingual text received both praise and reproof (Miettunen, et al., 2006, p. 10). When studied, the baby box books are rarely regarded as children's literature or picturebooks. The books combine tactile, visual and auditory repertoires in order to introduce the infant to rhymes, singing games and visual storytelling. Often the books have been produced in cooperation with a group of child-care experts. Pirjo Suvilehto (2014, p. 18), who has studied Finnish baby books published from 1970 to 2010 including the baby box books, concludes that these books -especially the ones published during the decades around the turn of the millennium -are characterised by prevailing pedagogical and psychological trends within education. Hence, the books' didactic agenda is overt and invites research on children's language development, early interaction and (gender) education, rather than on literary and aesthetic merits (cf. Suvilehto, 2014). We would like to add that this trend is especially visible in the books from the 2000s, which combine a normative instructional text solely directed to the reading adult, with playful content -such as images, rhymes and singing games -addressed to both infants and their co-reading adults. Consequently, these books are more or less awkward, hybrid mixtures of parental manuals and picturebooks, characterised by a striking double address (cf. Wall, 1991). The hybrid baby box books are predominantly illustrated with simple figures and forms in strong colours, which lend them a uniform appearance. None of them are socalled "early concept books", i.e. books presenting a sequence of objects. Still, they resemble classic infant picturebooks in that they seldom tell linear stories (cf. Kümmerling-Meibauer and Meibauer, 2005;Hurme, 2006, p. 9;Kümmerling-Meibauer, 2011, p. 4). As part of a state welfare project, they show a growing awareness during the 2000s of infants' cognitive abilities to perceive images and colours, as well as the need to portray a wider ethnic range of babies. From 2004 onwards, no imported mass-market baby books have been distributed with the maternity packages. A further typical feature of these books is that they are bilingual. All books include texts in the two national languages, Finnish and Swedish. 7 The books, therefore, stage an infant temporality that comprises an ideal, living, multilingual society where different languages -at least Finnish and Swedish -co-exist on the pages. In the following chapters, we will analyse how time is written into the baby box books and show how they, thereby, stage infant temporality. We will argue that these widely distributed books are sites of negotiation for contesting timelines attached to age. The baby box books simultaneously embody infant temporality and societal norms on how time is best or most healthily, and hence most economically, spent with the baby. --- In Resonance with the Rhythm of Baby Time Although babies are at the very beginning of their life span, temporality is a vital part of the baby box books studied here. First, these books are made to be read several times in a manner that encourages repetition and circularity. Secondly, as books produced, distributed and institutionally sanctioned by the Finnish state, they have a strong didactic purpose. In Elina Pulli's and Mikko Hiltunen's Together with the Baby, infant temporality is at centre stage. Baby time is made all the more visible as it is depicted in sharp contrast to adult temporality, which is regulated by clock time and characterised by galloping haste. The book comprises rhythmical verses that emphasise time and interactive games together with the baby, and informative text snippets addressed solely to the adult caregiver and placed in boxes at the bottom of the right-hand page of each spread. The information box on the very first spread tells adult caregivers that: "During the first weeks, the baby needs no toys. The adult's gaze, lap and familiar voice are enough" (Pulli and Hiltunen, 2014, p. [1-2]). On the following spread, (see Fig. 1) a baby is depicted resting in the lap of an adult caregiver. The scene is calm and serene, and the baby lies relaxed with eyes closed. While we can see the whole baby, the adult is only visible as an arm and hand holding the infant. Significantly, the adult wears a wristwatch with an empty dial. The image suggests that while measuring and following clock time is part of the adult timeline, baby time sets in when in the presence of the newborn, and clock time is banished in favour of focusing on the baby here and now. The verse to the right on the spread also alludes to time: Stroking, stroking your back, you are safe in my lap. Softly touching your leg and thigh, as you settle down with a sigh. Nestling in my arms, we belong together. Pulli and Hiltunen, 2014, p. [4] Both texts in Finnish and Swedish describe how the adult caregiver touches the child's body in a calm and safe manner that emphasises their togetherness. This book, along with others in our material, depicts baby time as a discursive, many-layered timeline of its own. Throughout Together with the Baby, baby time is explicitly depicted as a time of closeness and pleasure: "In you I see a new world, / in me you see warmth and safety. / Together we create the future." (Pulli and Hiltunen, 2014, p. [2]). Staying physically close to the newborn baby is presented as ideal parental behaviour. Baby time is, therefore, also body time. Mara Lee (2012, p. 8), who has explored the relations between queer temporality and time subversive pleasure, has shown that breaks, pauses and delays are intertwined with pleasure. She imagines that the feminist icon Luce Irigaray would express such pleasurable interaction and intimacy as "pleasure is time" (Lee, 2012, p. 16). The symbiotic and pleasurable baby time suggested in the baby box books can further, drawing on Rosa's theory of time, be understood as a pocket of resonance, i.e. a nick in the late modern accelerating timelines of adult life characterised by deceleration, presence, reassurance and bodily contact. The books' baby time is both institutional and democratic, yet not necessarily only part of a deceleration. Also acceleration can create resonance, based on the thrill and allure of the fast and action-packed. A prominent trait in the studied baby box books is, however, that the infant temporality staged is predominantly characterised by pleasure, presence, bodily touch and happiness. The books studied contain hardly any time markers, which creates a feeling of there being an infinite amount of time available for being together: "You are in my arms, always beside me. / My little baby. / We kiss, we dance. / My little baby" (Pulli and Hiltunen, 2019, p. [2]). Baby time thereby suggests a break from the chrononormative timeline, since it advocates repetition, deceleration, slowing-down, reproduction, affirmation, care and small-scale living in opposition to adult normative timelines defined by clock time, regulation and accelerating rush. Freeman (2010, p. 3) suggests that bodies incorporate time. We argue that baby time is expressed as time during which bodily tempo, such as the baby's rhythm and movements, organise existence. Research on and theory of temporality emphasise that rhythm is a central aspect of time. Rhythm is also at the heart of the baby box books. More subtly in terms of how the ideal time spent with the baby is orchestrated based on the baby's pace, but also overtly in that many of the books actively promote an oral-based tradition of rhymes and singing games. --- Baby Time Comes First! Presences and Absences of Infant Temporality On average, Finnish citizens welcome their first child between the ages of 29-31 (Official Statistics of Finland, 2021). Hence, the gap between me-time before the baby is born and you-time when the baby has arrived is considerable. This explains the striking conformity of the baby box books. They all centre on promoting time together with the baby as a temporality of its own and at odds with adult societal timelines. These books tell us that the Finnish state promotes deceleration and a reproductive, repetitive timeframe -a baby time -at odds with accelerated, adult timelines. 8 We find the baby box books interesting precisely because of this tension within their temporal structures. Baby box books from the 2000s put the depiction of babies' interaction with their caregivers first. This is often executed subtly without explicitly naming how the two are related. Especially in Rhyming in my Arms, Together with the Baby, Fun and Games with the Baby and The Baby in my Arms, the illustrations focus on depicting the baby's body, whereas the adult's presence is signalled only by visible body parts, as in Fig. 1. The consequence of including the baby in the visual narration, while excluding the adult from it, serves to strengthen the focus on infant temporality as a time of meeting the baby's needs by engaging in joyful and interactive play, such as affirmation exercises, word play, singing, touching and moving together. In order to put baby time first, some of the books go as far as to leave out all literary content altogether. For example, Adventures in my Arms, Dance Magic and Tumble Time consist of short, instructional texts for the adult on how to spend time with the baby, and bright, simple images meant to stimulate the baby. Most baby box books, however, use rhymes and playful, interactive elements to stage an ideal, happy infant temporality. Togetherness, but on the baby's terms, is at the heart of the books. In many of them, the personal pronouns "I-you" and "mine-yours" are repeatedly linked together: "You are soft and sweet, / so strong and able. / I give you strength, / you show me the way. / I draw the map, / around and around your body." (Pulli and Pertamo, 2016, p. [5-6]). In Rhyming in my Arms, words referring to the adult's lap keep being repeated. From a picturebook point of view, Rhyming in my Arms is one of the most intriguing titles. The illustrator Bast has chosen not to depict a realistic every-day setting, but a fantasy landscape inhabited by chubby toddlers dressed up as berries and animals. 9 In the background, the observant reader can discern adult hands drawn in white lines or soft colours, which hold, stroke or cuddle the toddlers (Fig. 2). The hands perform the movements proposed in the informative text. In addition, the cover illustration visualises subtly the adult presence in the background. The cover features two children. One child stands behind an orange elephant, whose shape resembles that of a large hand; whereas another large hand, almost invisibly drawn in a white and yellow line, lifts the acorn-shaped hat from the other child's head (Fig. 3). The choice to drastically tone down the adult presence in the illustrations and to depict the child characters as hybrid animals and plants wearing caps with animal ears or headgear that make them look like berries has a profound impact on the infant temporality. In analogy with the Romantic, dehumanising practice of animalising young children (see e.g. Higonnet, 1998;Lassén-Seger, 2006), childhood is separated from adulthood as a different species associated with an idyllic closeness to nature. The effect is further strengthened by the text referring to the children having animal body parts, such as "snouts" and "paws". As this example demonstrates, the infant temporality advocated in the baby box books can be imperilled, for instance by depicting the child characters in a manner that makes them less human and essentially different from the adult caregiver. The infant temporality communicated in the books is, overall, predominantly characterised as ideally idyllic. Time spent together with the baby is depicted as harmonically interactive and conspicuously free from technology, bad timing, and any feelings of displeasure. How to soothe a crying and screaming baby may be mentioned in the instructional texts, but the imagery only shows pleasantly smiling, sleeping, playing and laughing babies. 9 Children dressed up as animals also appear in The Happy Rhyming Caterpillar. Figure 2 1 3 The books' combination of sound advice to caregivers and material meant to entertain and stimulate the baby presupposes a temporality that stretches into the future, since the adult is encouraged to act in such a manner that the baby's needs are fulfilled so that s/he can grow and develop in the best manner possible. The books keep returning to scenes from a baby's everyday life, such as nap time, bath time, playtime and bedtime. Still, not every intimate aspect of baby life is portrayed. Most obviously, there are no baby box books from the 2000s that depict or even mention breastfeeding or nursing as part of infant temporality. The lack of breastfeeding scenes is all the more striking as Finnish mothers tend to nurse their babies at least for a year, if not longer. Näsi and Koskenvuo (2022, p. 418-419) confirm this when they note that feeding bottles in the baby box from 1998 to 2000 were removed when "it was established that breastfeeding has various positive health benefits for both the mother and the baby". Instead, nipple creams and breast pads for the mother were introduced to encourage breastfeeding (Näsi and Koskenvuo, 2022, p. 419). Still, the lack of nursing scenes is not unique for the Finnish baby box books. B.J. Epstein (2017) shows that breastfeeding is a taboo topic in all Western picturebooks, in part because the depiction of breasts is too closely connected to a sexualised discourse, but also because of the widespread use of breastmilk substitutes. Since the Finnish health authorities have kept recommending that mothers nurse their children throughout the 2000s, we suspect that an anxiety about depicting bare breasts lies behind the lack of nursing scenes in the baby box books. Figure 3 1 3 --- Whose Baby Time Is It? A conspicuous trait of the majority of baby box books from the 2000s is that their paratexts -typically placed on the back cover of the books -refer to academic expertise within developmental psychology and early childhood education. The back cover of, for instance, Dance Magic, presents a long list of experts consulted when making the book, ranging from specialists in child psychiatry, sexual health, choreography, graphic design to ophthalmology. The list even includes personal expertise of parenting. The absence of any children's literature or picturebook expertise is, however, glaring. Moreover, the sheer length of the paratext verges on being overwhelming -even intimidating -in its eagerness to persuade caregivers of the need to fill the shared baby time with particular, societally sanctioned content. The books hereby stage an infant temporality that is paradoxically defined by both deceleration and resonance understood as closeness and intimacy in the private sphere, and a strikingly pressurising, future-bound discourse on how to be the best possible parent in order to give one's baby the best possible chances in life. The infant temporality advocated by the books studied here is, furthermore, deeply rooted in attachment theory, without any acknowledgement of this theory having been scrutinised and criticised by feminist research (see Burman, 2007). The texts on the back covers of several of the books from the 2000s explicitly mention the importance of attachment. For example, the text on the back of Rhyming in my Arms claims that "The attachment between child and parent is strengthened when the child is held, when you hug and play together", and on the back of The Baby in my Arms it says that playing together supports "the interplay and attachment among parent and child". The choice of words manifests how strongly these books are rooted in developmental psychology, and how their purpose as parenting manuals for adult caregivers easily overshadows their purpose as entertaining and aesthetically enjoyable baby books. The discrepancy between the expert advice, which tends to be expressed in forceful clichés, and the basic, practical activities presented in the books -such as holding, caressing and hugging one's baby -is obvious. As a result, the infant temporality advocated is a junction where many, sometimes conflicting, institutional discourses meet. The baby box books juggle two purposes simultaneously that are hard to unite. One is to advocate a carefree baby time filled with deceleration and resonance. The other is to promote an ideal, institutionally sanctioned "good" baby time that demands something from the adult caregiver and is only superficially private and intimate. We therefore presume that these books differ from other baby books, since their institutional and didactic agendas raise the question of whose baby time is depicted: the baby's and caregiver's private, decelerated time together, or an idealised situation dominated by normative advice from experts advocating governmentally sanctioned resonance and deceleration? When the baby box books in 2006 begin to acknowledge the importance of fatherhood, the agenda also permeates the depiction of infant temporality. Tumble Time sets out to strengthen an ideal of fatherhood sanctioned by a long list of medical and psychological experts listed on the back cover. The core message, that the child should be at the centre of attention, whereas the father should direct all his attention onto the child, is clearly signalled on the cover which depicts a child hugging, and being hugged by, an adult with traditionally male traits, such as short hair and blue clothes. The two smiling characters embrace, but whereas the child looks straight out of the picture establishing an intimate eye contact with the reader, the adult character's gaze is turned towards the child (Fig. 4). The normative text on the back cover confirms that: "from the adult's gaze, the baby can discern that s/he is affirmed". Tumble Time is influenced by experiences from two national networks on fatherhood and family life organised by The Family Federation of Finland (Cacciatore, 2006, p. 2;Suvilehto, 2014, p. 15). The book features the father as the only adult character present, which shows an enlightened awareness of families being diverse and not necessarily complying with the traditional model of a nuclear family. The text on the back cover stresses again the importance of kinship and resonance in order to "support, not only infants' sense of self-esteem, but also the unity between child and adult", but goes on to claim that a different set of rules apply to fathers engaging with their children: "Men touch children in a different manner than women, and the child needs contact with both". We find that the gendered view of infant temporality proposed is astoundingly biased and, hence, undermines the air of scientific expertise behind the book that the paratext wishes to uphold. Tumble Time is one of the baby box books that contain no literary text, only instructions for the adult caregiver. The baby time depicted is explicitly described as a phase of attachment based on touch. The claim that men touch children in a different manner surfaces in the instructive text, which not only proposes quiet cuddling, but also more forceful interaction and rough play. In contrast to other baby box books, which implicitly advocate a supposedly softer, "female" approach to interacting with children, fathers are encouraged to "wrestle" and "romp with the baby" and help the baby "climb increasingly higher" (Cacciatore, et al., 2005, p. [9-12]). The latter suggestion comes, however, with a warning: "Don't let go!", which reveals a hidden fear and distrust in male caregivers being able to control such vigorous play (Cacciatore, et al., 2005, p. [9-10]). The appeal for control shows that the gendered infant temporality proposed to fathers in this book, despite its intentional progressiveness, is not only stereotypical but also Figure 4 1 3 infused with a fear that the male manner of interacting with the child may get out of hand. As we have argued throughout this article, most of the baby box books from the 2000s are -like Tumble Time -encumbered with an intent to teach rather than entertain their audience or provide them with enjoyable reading experiences. The appeal and address is so didactic that the books' intent to be playful occasionally implodes. Out of sheer benevolence, many of the books fail to create an enjoyable reading experience. They do, however, succeed in establishing a variety of infant temporalities. There is a strong connection between how temporality is narrated in the baby box books and how temporal normativity is promoted by the Finnish state. Infant temporality, prominently at odds with chrononormativity, becomes, in the hands of a welfare state's support programme, an expected, desired norm. In this way, baby box books tell us something about timelines and normativity in Western welfare states; in fact, these books make visible how temporal orders are negotiated on different structural levels, from overarching governmental initiatives, such as the maternity package, to individual levels, where families use the baby books. --- Conclusion: Institutionally Sanctioned Pockets of Resonance The baby box is still part of the Finnish welfare state's agenda to promote

Maternity packages, introduced in 1938 by the Finnish state to counteract declining birth rates and infant mortality, have reached all newborns regardless of income or social status from 1949 onwards. The box is a unique phenomenon internationally and a significant part of the Finnish history of childhood as its contents mirror contemporaneous views on childhood. In 1982, a baby book was added to the box. Research has, however, kept focusing on the social aspects of the box, ignoring the baby books. In this article, we make up for the lack of scholarly interest in the baby box books by studying how they depict temporal normativity. Drawing on queer temporal studies (Halberstam 2005; Freeman 2010), we introduce the terms infant temporality/ baby time to discuss the interplay of aesthetics and normative timelines. We claim that these books use bodily tempo, rhythm, movement, as well as sensory and tactile aspects to depict non-normative timelines, understood as infant temporality. Drawing on Hartmut Rosa (2019), we argue that the interplay between adult caregiver and infant encouraged by these books aims to achieve resonance, understood as presence and connection. In contrast to adult acceleration, we claim that these books, as part of their unique hybrid aesthetics, present infant temporality as deceleration. Reaching a vast number of Finnish families yearly, the baby box books provide a strong temporal discourse while promoting a resonant reproductive time. However, we find that while the books studied deviate from adult temporal normativity, infant temporality is still included in a normative reproductive timeline, sanctioned by the Finnish state. Hereby, the baby box books expose temporal complexities of our era.

Tumble Time -encumbered with an intent to teach rather than entertain their audience or provide them with enjoyable reading experiences. The appeal and address is so didactic that the books' intent to be playful occasionally implodes. Out of sheer benevolence, many of the books fail to create an enjoyable reading experience. They do, however, succeed in establishing a variety of infant temporalities. There is a strong connection between how temporality is narrated in the baby box books and how temporal normativity is promoted by the Finnish state. Infant temporality, prominently at odds with chrononormativity, becomes, in the hands of a welfare state's support programme, an expected, desired norm. In this way, baby box books tell us something about timelines and normativity in Western welfare states; in fact, these books make visible how temporal orders are negotiated on different structural levels, from overarching governmental initiatives, such as the maternity package, to individual levels, where families use the baby books. --- Conclusion: Institutionally Sanctioned Pockets of Resonance The baby box is still part of the Finnish welfare state's agenda to promote the health of newborns and their caregivers and is considered both a social and a political success (Näsi and Koskenvuo, 2022). It is also considered "a shared experience connecting generations" (Koivu, et al., 2020, p. 28). Approaching the baby books distributed in the package from the point of view of temporality has enabled us to uncover previously unnoticed aspects of the books. Whereas many health effects of the maternity package are followed up and reported on, the baby books are regarded as less influential on the health of babies and their parents; hence, they are not evaluated as frequently. We, however, have aimed to reveal how much they do tell us about how having a baby in a late modern, post-industrial society, means adapting to a different time order with other schedules and timelines, since the baby's life depends on the care given. As we have shown, time well spent, deceleration, adaptation to the baby's pace and timelines are at stake. Our take on temporality has been informed mainly by queer temporal studies linked to normativity, age and social theory. In our attempt to show how baby box books utilize temporality, we have drawn especially on Freeman's concept of chrononormativity or temporal normativity, and on Rosa's sociological theory of time. As we have demonstrated, the baby box books from the 2000s stage the time adult caregivers spend with their infants as a form of infant temporality. This baby time can be perceived as a non-normative timeline, since it -from the perspective of the adult caregiver -advocates deceleration and resonance in the form of an uninterrupted focus on presence and us-time with the baby. The advice shared on how to bring about such an ideal baby time together with a content and happy infant abound, yet only one of the books studied 10 encourages reading together or shows any books in the illustrations. In fact, the baby box books favour an oral tradition of rhymes and singing games ahead of reading and are, as such, related to non-normative timelines in that they presuppose a slowing down. In our analyses, we have argued that infant temporality -understood as pockets of resonance when adult caregivers set aside their own timelines or shift them to suit those of their babies -is characterised by competing temporal discourses. The baby time depicted in these books is so obviously framed by institutional discourses of the perception and use of time, as well as what it means to be a child or an adult, that while it is a decelerating non-normative timeline, it is simultaneously incorporated in normative temporality, that is, in sanctioned reproductive timelines. However, such timelines are inherently deviant. Examining what contemporary baby box books are currently addressing via temporality reveals a number of things. Firstly, infant temporality is enforced as a norm, in all its non-normative declarations. Apparently, there is a place for infant temporality in Finnish society. Finland recently implemented new legislation on shared parental leave, and has a history of extensive leave in comparison to other countries. This context is of considerable importance, since the Finnish state makes sure that there is paid time to spend with one's baby. The baby box books studied here portray infant temporality as a time when the baby is cherished, stimulated, kept safe and affirmed. As we have shown, queer studies ascribe queer temporality the potential to be a means of resistance (Lee, 2012, p. 8), which harmonises with Rosa's notion of democracy as time consuming (Lijster, et al., 2019). As a result, the baby box books exist in the intersection between institutional proposals and a time of pleasure. We have shown that temporality is at the heart of the unique aesthetics of the baby box books from the 2000s. These books are part of the welfare state's concern with equality for every Finnish baby. Infant temporality is, in many ways, promoted in contradiction to how Finnish society is organized, striving towards acceleration and productivity. When it comes to newborn babies, the Finnish state hands out baby box books that promote a temporality at odds with other societal temporalities. Thus, what would be expected to be a non-normative time, infant temporality, is in fact the inflicted norm. In our examples, we discern a certain kind of infant temporality: a period which creates a non-synchronism in baby box books as babies inherently belong to reproductive, repetitive timelines. Is not reproductive time already part of normative timelines? one might ask. Surely reproductive time is central to heteronormativity and, thus, included in normative timelines. However, read through the parameter of age, babies' timelines deviate from adult on-rushing temporalities. The shift from such accelerating adult time to infant temporalities is what most parents experience when having an infant. The feeling of being torn between competing timelines might be profound. However, there is no single mark of such a dilemma in the material; on the contrary, adults are absorbed by their babies and by the infant temporality. Nota bene, the baby time spent together is strikingly an affair between the adult caregiver and child. In our understanding, infant temporality as a concept is influenced by queer theory. Babies make ruptures in normative time from an adult point of view. They 1 3 disturb adult timelines. What baby box books seem to be doing is to convince and train the parent more than the infant. Infant temporality emanates from the baby's needs. At the same time, the concept of baby time is far from unambiguous, but fundamentally drenched in acceleration understood as the baby growing. The maternity package also contains baby clothes in the smallest sizes, which the baby is expected to grow out of quite quickly. The baby box's content thus signifies a governmental culture of care that sees infant temporality as both a time of moving growth and stagnant safety. --- Conflict of Interest The authors have no financial or proprietary interests in any material discussed in this article. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Maria Lassén-Seger Associate Professor of children's and young adult literature at Comparative Literature, <unk>bo Akademi University, Turku, Finland, is a senior researcher at The Society of Swedish Literature in Finland (SLS) in the research project Swedish-Language Children's Literature Criticism and Research in Finland 2022-25, and Service Manager at <unk>bo Akademi University Library. Her research is on Nordic and English-language children's literature, especially picturebooks and fantasy. She is the author of the study Adventures into Otherness (2006) and the co-editor of several scholarly anthologies, e.g. Nordic Dystopias and Utopias (2022) and Empowering Transformations: Mrs Pepperpot Revisited (2014). She is a member of the editorial board of BLFT: Nordic Journal of ChildLit Aesthetics and a former jury member for the Astrid Lindgren Memorial Award (ALMA, 2009(ALMA, -2021)). --- Mia <unk>sterlund --- Authors and Affiliations --- Maria Lassén-Seger 1 • Mia <unk>sterlund 1 Maria Lassén-Seger [email protected] Mia <unk>sterlund [email protected] 1 Comparative Literature, <unk>bo Akademi University, Tehtaankatu 2, Turku 20500, Finland 1 3

Maternity packages, introduced in 1938 by the Finnish state to counteract declining birth rates and infant mortality, have reached all newborns regardless of income or social status from 1949 onwards. The box is a unique phenomenon internationally and a significant part of the Finnish history of childhood as its contents mirror contemporaneous views on childhood. In 1982, a baby book was added to the box. Research has, however, kept focusing on the social aspects of the box, ignoring the baby books. In this article, we make up for the lack of scholarly interest in the baby box books by studying how they depict temporal normativity. Drawing on queer temporal studies (Halberstam 2005; Freeman 2010), we introduce the terms infant temporality/ baby time to discuss the interplay of aesthetics and normative timelines. We claim that these books use bodily tempo, rhythm, movement, as well as sensory and tactile aspects to depict non-normative timelines, understood as infant temporality. Drawing on Hartmut Rosa (2019), we argue that the interplay between adult caregiver and infant encouraged by these books aims to achieve resonance, understood as presence and connection. In contrast to adult acceleration, we claim that these books, as part of their unique hybrid aesthetics, present infant temporality as deceleration. Reaching a vast number of Finnish families yearly, the baby box books provide a strong temporal discourse while promoting a resonant reproductive time. However, we find that while the books studied deviate from adult temporal normativity, infant temporality is still included in a normative reproductive timeline, sanctioned by the Finnish state. Hereby, the baby box books expose temporal complexities of our era.

Despite overall high levels of solidarity between family members across two or more generations throughout contemporary Europe and the United States (cf. Kalmijn, 2014;Seltzer & Bianchi, 2013), studies also indicate considerable cross-national variation in the strength of adult intergenerational ties (e.g., Brandt & Deindl, 2013;Hank, 2007;Silverstein, Gans, Lowenstein, Giarrusso, & Bengtson, 2010). These geographic differences have been suggested to be fairly stable, reflecting longstanding variations in cul-tural and welfare state characteristics, social norms and preferences. Little is known yet, however, about possible changes in adult parent-child relationships within countries over time. Ever since Parsons' (1943) claim of the emergence of an "isolated nuclear family" and the subsequent discussion, whether this claim really holds true (e.g., Litwak, 1960), many studies (implicitly or explicitly) argue in favor of or against the "family decline" hypothesis (e.g., Popenoe, 1993). Even though historical census record linkage nowadays provides some opportunities to study kin relations beyond the household over longer historical periods (see Ruggles, Fitch, & Roberts, 2018), suitable repeated cross-sectional data for more recent time periods are still rarely available. The body of empirical research investigating temporal dynamics of intergenerational family relations around the turn to the 21st century-and the extent to which these may be associated with contemporary societal changeshas thus remained fairly small. Analyzing data from four rounds of the German Ageing Survey (DEAS; see Klaus et al., 2017), our study contributes to this sparse literature in two ways: First, covering a period of nearly two decades from 1996 to 2014, our data allow us to monitor stability and change in the relationship between parents and up to four adult children over a longer and more recent period of time than previous studies. Second, the DEAS data provide information on multiple dimensions of intergenerational solidarity (Bengtson, Giarrusso, Mabry, & Silverstein, 2002), which allows us to examine possible differences in the temporal dynamics of adult intergenerational relations by specific outcomes, namely geographic proximity, frequency of contact, emotional closeness, as well as the exchange of material and instrumental support. The German example seems well suited for our investigation, because it represents-in several ways-an "average" case with a pattern of intergenerational solidarity in-between the extremes of the (Western) European continuum of family ties, characterized by weaker ones in the Nordic countries and stronger ones in the Mediterranean countries (e.g., Brandt & Deindl, 2013;Hank, 2007;Silverstein et al., 2010). --- Previous Research on Stability and Change in Intergenerational Relations Alterations in adult parent-child relationships within countries over time might result from behavioral and/or compositional changes in the population (e.g., Kalmijn & De Vries, 2009;Treas & Guberskaya, 2012). Behavioral changes may be triggered by period effects, such as macroeconomic crises (e.g., Fonseca, Cunha, Crespo, & Relvas, 2016;Preoteasa, Vlase, & Tufă, 2018), as well as by secular trends, such as declining norms of family obligations and rising values of individualism (e.g., Gans & Silverstein, 2006;Roberts & Bengtson, 1999). Compositional changes might be driven by population shifts in sociodemographic factors correlated with intergenerational relations, such as family structure or education (e.g., Kalmijn, 2006;Steinbach & Hank, 2016). However, not all dimensions of the adult parent-child relationship-as described, for example, in the solidarity-conflict model proposed by Bengtson and colleagues (2002)-need to be affected in the same way by such shocks, trends, or shifts. It is therefore important to distinguish different aspects of this relationship. Previous research investigating temporal changes in parent-child proximity (structural solidarity) tended to focus on the decline of intergenerational coresidence during the 20th century (e.g., Grundy, 2000;Ruggles, 2007). This trend, however, appears to have reversed in recent years (Fingerman, 2017). Moreover, van der Pas, van Tilburg, and Knipscheer (2007) showed for the Netherlands that further declines in coresidence were paralleled by increasing intergenerational proximity between 1992 and 2002. Along the same lines, Shelton and Grundy (2000) found that in Britain the proportion of non-coresident adult children living farther away from their mothers was significantly lower in 1999 than in 1986. Also in contrast to notions of rising individualism and family fragmentation in Western societies, there is no clear indication of a trend toward reduced intergenerational contact frequency (associational solidarity) during the mid-1980s through the early-2000s; some studies even suggest an increase, especially if contacts other than face-to-face visits are considered (e.g., Grundy & Shelton, 2001;Kalmijn & De Vries, 2009;van der Pas et al., 2007;Treas & Gubernskaya, 2012; also see Fingerman, 2017). Emotional closeness (affectual solidarity) has been shown to bear a mutually reinforcing association with intergenerational contact (e.g., Hogerbrugge & Komter, 2012). Even though one may expect a trend toward greater emotional closeness between older parents and their children-resulting from declining generational differences in values and more liberal child-rearing styles-"[i]t is not well known if intergenerational relationships have [actually] become more or less positive over time" (Kalmijn, 2014, p. 387; also see Fingerman, 2017). Finally, so far barely any research explicitly investigated temporal changes in intergenerational transfers of time or money (functional solidarity) between adult family members. Van der Pas and colleagues (2007), however, found that somewhat more emotional and instrumental support was exchanged between Dutch parents and children in 2002 than in 1992. Fingerman (2017) suggests that recently increasing rates of coresidence in some industrialized nations might indicate growing flows of in-kind transfers of material resources from parents to adult children and vice versa (especially in Southern Europe; see Albertini Kohli, & Vogel, 2007). Moreover, Kornrich and Furstenberg (2013) showed that from the early 1970s to the late 2000s U.S. parents' spending on their children shifted from expenditures predominantly dedicated to teenage offspring toward young adult children in their mid-20s. Even though the authors explicitly focused on parental investments in children rather than on financial support, "it appears that parents are reconciled to the reality that it takes longer for their children to reach economic maturity than it did a halfcentury ago" (Kornrich & Furstenberg, 2013, p. 3; also see Henretta, Van Voorhis, & Soldo, 2018). Despite some indication of shifts in population composition, their net effect on observed change in intergenerational relations has been proposed to be small (e.g., Kalmijn & De Vries, 2009;Treas & Gubernskaya, 2012). Rather, van der Pas et al. (2007, p. 269), for example, "believe that the changes that have taken place in attitudes towards the family have had a more profound effect on parent-child relationships than social developments such as the increase of female participation in the labor market or an increase in divorce and remarriage." Greater autonomy in intergenerational family relations, where individual commitments partially replaced fixed obligations (Gans & Silverstein, 2006), thus seems to have contributed to closer rather than more distant ties. Moreover, advances in communication technologies and transportation facilitated proximity and contacts between parents and their non-coresident children (Fingerman, 2017). These developments in Western societies, together with parents' prolonged financial responsibilities for children (resulting from delays in the transition to adulthood), have created a situation, in which one may find some trend toward overall stronger adult intergenerational solidarities. Thus, if we were to expect any changes at all, there should be an increase in geographic proximity, frequency of contacts, emotional closeness, and material as well as instrumental support among parents and adult children in Germany during our observation period (that is, between 1996 and 2014). --- Method Our analysis draws on data derived from the DEAS (Klaus et al., 2017), a nationally representative study of noninstitutionalized adults aged 40-85 years. Cross-sectional samples were drawn in 1996, 2002, 2008, and 2014, covering birth cohorts from 1911 to 1974. The samples have been disproportionally stratified into three age groups (40-54, 55-69, and 70-85 years), gender, and region (East, West). The oldest age group as well as men and East Germans were oversampled. Because the first round of data collection was limited to German citizens, we exclude interviews with non-German respondents conducted in subsequent rounds. As DEAS does not provide any information on intergenerational contact and the exchange of material/instrumental support with coresiding children, these parent-child dyads were excluded from our analysis (but see Supplementary Table A2 in the Online Appendix). The response rate declined from initially 59% to 27% in 2014, following a common trend also observed in international surveys, but remaining in the same order of magnitude achieved in other German studies. Whereas there is no indication of increasing sample selectivity in the DEAS over time, we observe a general pattern of slightly lower participation rates in large cities, among women as well as in the middle-aged (40-54 years) and oldest (70-85 years) age groups. Overall, sociodemographic characteristics in the survey have been shown to match well the respective distribution in the population (Klaus et al., 2017; for more detailed analyses see Klaus & Engstler, 2017). The pooled sample size across all waves is 19,746. Excluding respondents without any living children (2,524), relationships with minor and/or coresident children (3,749), as well as cases with missing values on any of our dependent and control variables (367) leaves us with an analytic sample of 13,106 parents reporting on 24,450 dyadic relationships to a maximum of four adult children (see Table 1 for detailed descriptive sample statistics). Our final data set thus has a hierarchical structure, where multiple parent-child dyads (Level 1) may be nested in one respondent (Level 2). We therefore estimate two-level random intercept hierarchical linear models (see Gelman & Hill, 2007: Part 2A) for each dependent variable. Despite their binary or ordered outcomes (see latter), we opted for linear probability models, because these have been suggested as a suitable alternative to, say, logistic models, if the comparison of coefficients across models with different independent variables in a sample is a primary concern (e.g., Mood, 2010, p. 78), which is the case in our analysis. The dependent variables are operationalized as follows: (a) Geographic proximity has four categories, namely "living in the same town" [4], "living in another town within a radius of 2 hours" [3], "living farther away in Germany" [2], and "living farther away abroad" [1]. (b) Frequency of contact (comprising all contact modes) has seven categories, namely "never" [1], "less often than several times a year" [2], "several times a year" [3], "1-3 times per months" [4], "once a week" [5], "several times a week" [6], and "daily" [7]. (c) Emotional closeness has five categories, namely "not close at all" [1], "not very close" [2], "moderately close" [3], "close" [4], and "very close" [5]. (d) Exchange of support is assessed by a set of four binary variables allowing us to distinguish (i) which kind of support (material or instrumental) the parent may have (ii) provided to or received by a specific child in the 12 months preceding the DEAS interview. "Material support" comprises regular financial support as well as larger monetary and nonmonetary gifts, whereas "instrumental support" refers to help with housework, such as cleaning, small repairs, or shopping. Because we are primarily interested in assessing whether intergenerational relations have changed over time, the main "explanatory" variable of interest in all models is survey year (represented by a set of dummies, where "1996" constitutes the reference). Starting out from baseline models without any control variables, we estimate further models including sociodemographic characteristics of the respondents which previous studies have shown to be associated with our outcome variables (e.g., Kalmijn, 2014), namely: age (40-85 years), sex (where 1 indicates that the respondent is a father, 0 otherwise), education (represented by three binary indicators for "low" [ref.], "medium", and "high" levels of education), the number of children, partnership status (where 1 indicates that the respondent has a partner, 0 otherwise), self-rated health (five categories ranging from "very bad" to "very good"), and region (where 1 indicates residence in West Germany, 0 in East Germany). Moreover, several adult child characteristics enter the regressions, namely: relationship to parent (where 1 indicates a biological relationship, 0 otherwise), sex (where 1 indicates a son, 0 a daughter), marital and parental status (as binary outcomes), as well as employment status (represented by three binary indicators: "working" [ref.], "in education", "other"). We thereby aim to account for possible changes in population composition and a remaining effect of "survey year" would thus indicate behavioral changes in adult parent-child relationships. --- Results The multivariate results for our first set of dependent variables-proximity, contact, and emotional closeness-are presented in Table 2; those for the second set-indicating intergenerational exchange of material and instrumental support-are displayed in Table 3 (see Mahne & Huxhold, 2017, as well as Klaus & Mahne, 2017, for more detailed descriptive accounts). Table 2 (column (a)) shows that parent-child geographic proximity continuously decreased between 1996 and 2008, whereas we observe no statistically significant further increase in distance thereafter. A supplementary multinomial logit analysis (based on a sample that includes coresident adult children) suggests that the observed decrease in parent-child geographic proximity is particularly due to increasing risks over time of living farther than 2 hr apart (in Germany or abroad); see Supplementary Table A2 in the Online Appendix. Intergenerational contacts (column (b)) were most frequent in 2002 and emotional closeness (column (c)) was weakest in 2014, but there is no indication of any kind of trend, that is, systematic change in parent-child frequency of contact or closeness between 1996 and 2014. Turning to the intergenerational exchange of support in Table 3, columns (a) and (c) show that the propensity to support a child materially or instrumentally tended to decrease in the early 2000s, with some signs of recovery in 2014. Whereas parents' receipt of material support from children remained stable-at a very low incidence level-over time (column (c)), their probability to receive instrumental support continuously declined between 1996 and 2008, but not any further thereafter (column (d)). 3.4 (0.9) 3.4 (0.8) 3.4 (0.9) 3.4 (0.9) 3.4 (0.8) Controlling for parents' and children's sociodemographic characteristics in Models 2 did not have any substantial impact on the initially estimated associations between "survey year" and any of our dependent variables in Models 1 (see Supplementary Table A1 in the Online Appendix for a display of all control variables' coefficients employed in Table 3). The coefficients of the sociodemographic controls generally turned out to be as expected from previous research and shall therefore not be discussed in greater detail. --- Conclusions Using multiple rounds of data collected by the DEAS, this study aimed to investigate stability and change in various dimensions of adult parent-child relations across two decades, from 1996 to 2014. Four main conclusions can be derived from our analysis: First, if we observe any changes over time at all, these appear to result from behavioral rather than compositional changes, which we account for by controlling for an array of parent and child sociodemographic variables. This finding is consistent with, for example, Treas and Gubernskaya's (2012) study of changes in maternal contact. Second, different from our expectations derived from previous research, we observed a decline in adult parentchild geographic proximity and the propensity to exchange material/instrumental support in the early 2000s. Both trends, however, appear to be levelling off (or even reversing) in the most recent DEAS collected in 2014. Even though we controlled for a number of characteristics potentially affecting parents' and children's capacities and needs (such as their health or partnership status), we cannot fully determine the extent to which temporal variation in these factors mattered here. One might also speculate, whether the most recent financial and economic crisis played a role (e.g., Preoteasa et al., 2018), but this would clearly require further research. Third, if we look at other important dimensions of intergenerational solidarity, namely frequency of contact and emotional closeness, our results indicate that they remained stable, at high levels, throughout the entire observation period. This supports Litwak's argument-put forward as early as 1960-that increasing geographic mobility of family members need not be paralleled by declining family cohesion (e.g., because of improvements in communication technologies; see Peng et al., 2018). Fourth, and finally, even if we observed statistically significant temporal dynamics in some of the aforementioned dimensions of intergenerational solidarity, these are not necessarily substantively, that is, socially relevant (see Bernardi, Chakhaia, & Leopold, 2017, for a general discussion of this issue). The lack of support for our initial expectation of an increasing quality of adult parent-child relations over time should therefore not be any reason for concern, but might rather be interpreted as indication of some kind of "ceiling effect," where already high levels of solidarity may leave relatively little room for further improvements in the German context (which has been suggested to be characterized by lower preferences for, say, intergenerational proximity and contact than Mediterranean contexts (e.g., Hank, 2007). Moreover, our study shows that temporal patterns of intergenerational solidarities within countries might be characterized simultaneously by stability and change, where increasing geographic mobility, for example, is paralleled by continuous family cohesion. Family members appear to react to variations in social and economic circumstances with behavioral changes allowing them to maintain high levels of overall intergenerational solidarity. Next to having provided these novel insights, the present study also had several limitations: First, similar to the study by van der Pas and colleagues (2007), our results may reflect both period and potentially confounding cohort effects, controlled for age effects (also see Henretta et al., 2018). Second, the time period covered by our study is purely determined by the availability of DEAS data. That is, we do not know whether-if seen from a longer-term historical perspective-the patterns observed here are typical or rather exceptional. Third, and finally, our study focused on parent-child relationships, excluding family relations across more than two adult generations. Especially grandparent-grandchild relationships have received growing attention in the recent literature and are likely to exhibit temporal patterns of stability and change similar to those described here (e.g., Hank, Cavrini, Di Gessa, & Tomassini, 2018;Mahne & Klaus, 2017). These limitations, however, do not challenge the main substantive conclusion drawn here: In the two decades around the turn to the 21st century intergenerational family relations in Germany generally exhibited continuously high levels of solidarity. Our findings thus provide no indication of substantial changes in adult parent-child relationships over time, which might have lent support to previously expressed concerns regarding a possible decline in the importance of primary family ties in contemporary Western societies. --- Supplementary Material Supplementary data are available at The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences online. --- Author Contributions A. Steinbach performed all statistical analyses, helped planning the paper and to revise it. K. Mahne and D. Klaus helped planning and revising the paper. K. Hank wrote the paper, helped planning the paper and to revise it. --- Conflict of Interest None reported.

Objectives: Using high-quality data from Germany, this study aims to contribute to the yet little knowledge about possible changes in adult parent-child relationships within countries over time. Method: Analyzing 13,106 interviews from four rounds of the German Ageing Survey (DEAS), covering the period 1996-2014, we monitor stability and change in four dimensions of adult parent-child relationships, namely geographic proximity, frequency of contact, emotional closeness, and exchange of support. Results: We observed a continuous decrease in parent-child geographic proximity between 1996 and 2008, but no further increase in distance thereafter. There was no change in intergenerational frequency of contact or emotional closeness between 1996 and 2014. Parents' propensity to support a child tended to decrease in the early 2000s, with signs of recovery in 2014. Whereas parents' receipt of material support from children remained stable, their probability to receive instrumental support declined between 1996 and 2008, but not any further thereafter. Discussion: Temporal patterns of intergenerational solidarities within countries might be characterized simultaneously by stability and change, where increasing geographic mobility, for example, is paralleled by continuous family cohesion. Family members appear to react to variations in social and economic circumstances with behavioral changes allowing them to maintain high levels of overall intergenerational solidarity.

Background Lack of diversity in trial samples is a moral, ethical and scientific issue [1,2]. Homogenous groups can skew findings and impact generalisability to the wider population [3]. Greater inclusivity could result in more robust data to inform decisions in healthcare and treatment innovations, potentially reducing disparity in health outcomes [4]. Health inequalities have come to the forefront during the COVID-19 pandemic, where older adults, people with existing health conditions, and people from minoritised ethnic backgrounds in the UK continue to be disproportionately affected [5]. Despite being the gold standard of research to determine effectiveness/efficacy, randomised controlled trials (RCTs) often struggle with participant recruitment, engagement and retention, particularly for mental health interventions [6]. There are a number of reasons why recruitment to mental health RCTs may be particularly challenging. Clinicians may have concerns about the perceived vulnerability of their patients resulting in issues with decision making because of lack of equipoise or burden of research or concerns regarding consent and the lack of capacity to provide informed consent [7,8]. Patients may also have concerns about cultural stigma and stigma surrounding mental health [9], or concerns regarding the treatment itself [6]. It is also challenging to recruit a diverse population to mental health RCTs when the socioeconomically disadvantaged population are less likely to attend mental health services in-person [10,11]. Indeed, mental health service use is proportionately lower for males [12], people from ethnic minority backgrounds [13], older participants or those living in more rural areas [14]. In recognition of the need to reduce disparities in participation in research, the National Institute for Health Research (NIHR) Clinical Research Network commissioned the INCLUDE project to provide a framework for researchers and funders when developing research protocols and includes examples of how to broaden inclusivity [15]. The NIHR also developed their Equality, Diversity and Inclusion (EDI) Strategy 2022-2027 to ensure the implementation of inclusive practice in research, culture and systems [16]. Whilst RCTs of mental health conditions have been traditionally conducted in a clinical face-to-face setting, since the late 1990s, there has been a trend towards online or 'digital' RCTs in the field of mental health [17]. Online recruitment strategies, by which we mean any use of Internet technologies to recruit research participants, such as social media, Google search engine advertisements, and other website campaigns [18] offer researchers the opportunity to modify their recruitment materials and strategies based on feedback and engagement with the adverts to allow a targeted strategy to reach specific audiences [19,20]. In comparison to offline recruitment methods such as in-clinic recruitment, soliciting subjects through mail and telephone using health records and registers, media campaigns, newspaper advertisements, and input during radio and television talks [18], online recruitment may reach communities who are not currently engaged with specialist mental health services. This is likely to be particularly important for conditions where specialist care is only offered at centres typically in large cities [21], or when recruiting ethnic minority communities [22]. Despite these advantages, there is notable concern about the "digital divide", which in its simplest terms reflects those connected to the internet and those who are not, but more recently is considered to reflect differences in usage and internet skills [23]. For example there is evidence ethnic minorities, compared to white groups, access technology more outside the home and less frequently suggesting digital exclusion through availability [24]. In addition lower socioeconomic status has been associated with greater digital exclusion [25]. It is unclear therefore whether a move to solely online delivery of trial procedures (such as recruitment or the intervention itself ) will impact (positively or negatively) on health inequalities. This study, which is part of a wider project, REcruitment in Mental health trials: broadening the 'net', opportunities for INclusivity through online methoDs (RE-MIND) [26], aims to identify and provide considerations for use of online methods in the recruitment of participants into mental health RCTs, with a focus on whether online methods can enhance inclusivity. The aim of this qualitative sub-study was to explore the experiences and perspectives of key partners on the use of online and offline methods to recruit trial participants. knowledge about recruitment methods in RCTs would be of value. A qualitative approach was therefore used to investigate the experiences, opinions and ideas of key partners on the use of online and offline recruitment methods in the UK. Focus groups (FGs) and semi-structured interviews with participants were conducted by the researchers (KS, MI and CLH). FGs were used for several reasons, (1) to maximise time and resources, (2) to identify and clarify views in relation to others who may have a similar lived experience and (3) to support sharing of these ideas and similar or different opinions [28]. As FGs are more difficult to schedule than individual interviews [29], the research team decided to offer interviews as an alternative for practical reasons [30] to capture key partners who found it difficult in terms of availability/preference to attend FGs. FGs and individual interviews have been found to be very similar in their ability to generate unique items [29]. In this study, FGs and interviews drew on a similar technique for collecting data by using openended questioning with inductive probing of responses. The triangulation of both methods was used to ensure the completeness and richness of the findings [31] and to have a comprehensive understanding of the accounts of key partners' knowledge pertaining to recruitment methods. This study followed the reporting guidelines set out in the consolidated criteria for reporting qualitative research (COREQ) [32] (Additional file 1). --- Sample Typical case sampling [33] of research staff working on mental health research and Patient and Public Involvement (PPI) partners were approached. Selection variables [34] included past/current experience working in mental health research, expertise in designing and conducting mental health RCTs, age, gender and ethnicity. We identified research staff via the NIHR Clinical Research Network (CRN), and the UK Trial Managers Network (UKTMN). We approached potential PPI participants via existing groups including the "Sprouting Minds" Young Persons Advisory Group (YPAG) specialising in digital mental health research; the Deep End group, a patient participation group which inputs into study design in order to make research more accessible; and the NIHR Research Design Service (RDS) who offers specialist advice and support on research design and methodology to researchers as well as their own professional contacts from experience working on mental health RCTs. --- Recruitment and data collection An initial invitation was circulated by email to individuals and groups to share with their members. Those expressing an interest were then sent a copy of the participant information sheet outlining the aim of the study and a consent form. All those who expressed an interest in the study were included. FGs and interviews were conducted in person, as well as online using Microsoft Teams videoconferencing software. A topic schedule (Additional file 2) was designed to help guide the discussion by eliciting the understanding of the factors that may affect the recruitment approach in clinical trials and how this could impact the inclusivity of the trial participants. We used an exploratory, contextbased approach to develop the interview schedule. This included identification of core topics from the literature [21] as a starting point which were then discussed with the project management group who have experience in digital research, design and delivery of online and offline RCTs and equality, diversity and inclusion. In addition, the research team also had input to the interview schedule from two young people PPI with lived experience of mental health issues. The FGs and interviews were conducted between July 2022 and January 2023. The inperson FGs and interviews were audio recorded and the virtual ones were audio and video recorded via Microsoft Office TEAMS, with consent, for later transcription and analysis. Participation was voluntary and participants could withdraw at any time. --- Analysis An inductive and deductive approach was used to analyse the FG and interview transcripts [35] to develop a coding framework which drew on Braun and Clarke's thematic analysis [36] in which data are familiarised and coded into themes as part of a review process (see additional file 3). This approach was taken to provide rich data through generation of new codes from the data itself (inductive) as well as identification of codes within the data from known context of the interview schedules [37]. In-person recording was transcribed by KS and Microsoft Office TEAMS produced automated transcripts. Transcript validation was undertaken by two researchers (KS and MI) who read and re-read the transcripts for accuracy and to ensure familiarization with the data. The transcripts were imported into NVivo 12 Pro for coding. The researchers anonymised participants' information by removing identifying information and allocation of a unique identifier. To ensure validity and reliability of data interpretation, two researchers (KS and MI) independently analysed the transcripts using the topic schedule sections as an initial coding guide and open coding for additional themes. Overall codes were largely consistent between coders, where contradictory coding was apparent, the coders refined this through discussion. Similar codes were clustered together, and a framework of themes identified from the data was developed by KS. The framework was discussed between the research team (KS, MI and CLH) and refined by two researchers (KS and MI) and the final framework was developed when a consensus was reached [36]. The coding process was done in parallel with conducting the FG's/interviews to inform the interview schedule going forward and to monitor for data saturation. Although we were restricted by available resources to conduct a maximum number of interviews/focus groups, this was not deemed an issue by the interview research team (KS, MI and CLH) and although it can never be certain, it was agreed that no new themes were emerging by the end of the process [38,39]. --- Results Twenty-three people who volunteered were invited to participate in the study. A total of three FGs and three interviews were conducted with a total of 23 participants (15 PPI partners and eight research staff ). Two of the FGs were conducted with 11 PPI partners (three participants in one FG and 8 in the other) and one FG with eight research staff. The three interviews were conducted with four PPI partners. One of these was a joint interview as two participants opted to be interviewed together. Our sample included people who identified as male, female and non-binary as well as representation from white, Asian and Black ethnic groups and were aged between 20 and 70 + years of age. Four broad themes were identified from the data: [1] recruitment reach; [2] demographic factors that affect selection of recruitment method; [3] safety of technology, and; [4] practical challenges to online recruitment. Supporting quotation represents Staff (Research staff ) and PPI (Patients and Public). --- Theme 1 -recruitment reach Participants commented on three main areas; accessibility and convenience, impact on quality and managing expectations and flexibility. --- Accessibility and convenience Benefits of using online recruitment methods were their potential for reaching wider communities and greater accessibility to people who may feel stigmatised by their mental health condition or those living with symptoms that may impact their ability to participate. --- "I'd like to say that benefits of online are that you can target quite a wide range of people at once, all at one go" (PPI 11). "Some groups don't have good access to like health or social care services so it [online] could help to target different communities. " (PPI 14). "...it's [online] more flexible, it allows that better reach as well. We were able to recruit people in [different geographical areas], which otherwise would have been quite a challenge to do, if you had to travel. " (Staff 3). Some PPI and staff participants perceived that online methods could be more convenient for people with mental health conditions and those who care for them. "It costs people to travel as well, you know to come to meetings, time and money, or you've got kids to pick up, there's a thousand and one reasons for [online]..." (PPI 3). --- Impact on quality and managing expectations A common view amongst PPI participants and staff was that online methods required closer data monitoring for quality of responders, particularly their eligibility to take part in a trial, where they cited experience of dealing with hoax data or people lying about eligibility to be involved in a study. --- "...we've got 1000 hits [social media] and it looks so fantastic and then you whittle it right down and find at the end you're not that much different to what you got from your clinic. So I would say it works, but it needs a lot of safeguards and in-built procedures, so you can find a way to screen people because people will make up stuff. " (PPI 11). --- Flexibility In order to meet the individual needs of those living with mental health disorders (e.g. condition, age, digital literacy, transport), all participants agreed that offering a range of options was crucial for better recruitment. Participants agreed that researchers should not have a predetermined recruitment method but should explore preferences. "...its not one size fits all, so for some individuals they want that distance, they don't want you there, but for others it works better...others they want that two-way conversation, that contact, that hand on the shoulder, you know showing some empathy, some compassion..." (PPI 2). --- Theme 2 -demographic factors that affect selection of recruitment method Five main factors were identified; age, complexity of mental health, cultural and ethnicity differences, stigma and digital divide. --- Age PPI participants reported mixed opinions on recruitment preferences for older people. Some stated they would prefer offline methods over online, whereas others noted people are working longer and retiring later and therefore remain proficient in using technology and social media. "...you need to build up trust and I think face to face and touch is really important. But that might be because I'm old and I'm not in touch with modern ways..." (PPI 1). "...people are working longer and longer and they are still using those technologies at work so it will come into their social world. " (PPI 2). PPI participants felt that older people may have limited access to technology and therefore recommended a mixed methods approach to ensure opportunity and inclusivity. --- "Well, I'm an older person, and yes, struggling with technology, but some older people that really don't access technology at all. So they could be at risk of being missed out. So if you were wanting to target a group of older people, you probably wouldn't go down the social media route so much, would you? "(PPI 12). Age was also a factor in PPI and staff views of which media platforms would be most appropriate to use. For instance, Facebook was proposed for those aged 50 + whereas for younger people, TikTok or Instagram were recommended. --- "So if you're trying to recruit younger people, you probably want to go for Instagram...Facebook, brought in a lot of interest, but I think it does pay to just think about you know lots of different strategies to try and catch a wider variety of population..." (Staff 7). "You also have to be aware in social media, what the demographics are for social media. So 50 + use Facebook, Instagram are more younger people, would you use TikTok. " (PPI 5). --- Complexity of mental health and stigma Both PPI and staff identified the type and stage of mental health condition, as important in the selection of recruitment method. For example, some participants expressed the belief that online methods may reduce the stigma felt by people with mental health disorders through provision of privacy and a virtual space to engage with research from the safety of their own home or safe space. --- "... social anxiety and stuff, they don't want to talk where somebody can see them, so they might prefer something that was...over the phone, rather than face to face or even on Zoom. " (PPI 4). "...there were specific groups for posting research and people found that good because some of them had comorbidities that meant they didn't get out of the house as much to be able to see the research advertisements, and they had social differences that meant that they were less secure coming into a place to do research, so being able to do a lot of it online was a preference. " (Staff 4). However, two key disadvantages to online methods were also acknowledged in relation to the experience of mental health. One, related to vulnerability and online methods, particularly regarding privacy and disclosure of their condition. The second was loss of personal cues, particularly body language was thought to play a significant role in communication in this population. "...you're in a job and somehow it gets out that you have a mental health issue because you're on a study. Will it get round someway to someone who then shares it with employer or something, and then your vulnerable, aren't you?" (PPI 15). "I think that some things can be lost online...we read people's body language and you know how they are as a person and some of that can be quite difficult to pick up online, you know. " (PPI 11). Participants elaborated on the idea of vulnerability of people living with mental health conditions and a sense of heightened risk of cyber scamming or data fraud as well as potential for online bulling if confidentiality was not maintained appropriately. "...there is a stigma against mental health and it's like, how much do you share that could be accessed by other people, I have been scammed twice. So I realise I am so vulnerable...And if you're sharing information about someone with a mental health condition, they could be targeted..." ( PPI 15). "So one challenge of recruiting people online... younger people who are living with their families... it might not be safe or appropriate for them to take part online, so it might be better to do that in person. " (PPI 14). --- Cultural or ethnic differences Acceptance and understanding of mental health conditions, as a concept and a clinical disorder, was considered important to encourage engagement in research. For our participants, acceptance was still strongly informed by social and cultural factors which they felt needed to be addressed at a societal level, however they also considered positive generational changes in some cultures were now in evidence. "There is that political dimension that almost everyone is aware of, a sense of why is it that a particular profile of say a young Black man with mental health issues and the way he is treated is different to other communities. --- " (PPI 7). "I think if you're talking about mental health there's still a taboo around mental health and it's like you tell somebody in an Asian community you've got mental health problems, so what's wrong with your brain what's wrong with your head...so for that reason people don't say it. " (PPI 4). Language was still considered as a barrier that could affect recruitment regardless of method used, online or offline. --- "I'm thinking more from the 'BAME' community, you know the older generation, they can't read and write, they can't go on the internet, if they're living on their own they wouldn't know how to get on to Zoom and things like that..." (PPI 4). Trust was raised by several PPI participants as a significant barrier to diversity in research. This was described at a political level and within individual communities that may see themselves as marginalised, e.g. Black African and Black African Caribbean communities. Community gatekeepers were proposed as a way to develop trust within ethnic communities to improve diversity. --- "And we went to the Asian lunch clubs to talk to them...So it is targeting, isn't it? It's knowing your community and targeting your delivery really... Rather than throwing something out there and just hoping you'll get the numbers. " (PPI 15). --- Digital divide There were mixed opinions on access to technology between staff and PPI participants. Whilst staff experience was that the divide was reducing, PPI felt this was still a significant issue for those living in poverty or who were from underprivileged backgrounds including young people. Availability and access to technology was thought to potentially bias recruitment for online methods. --- "We were very worried about the digital devices because you had to have a smartphone to be able to take part. But actually what we're finding is that... everybody has a smartphone...we're recruiting par-ents and even parents from really struggling areas have got a smartphone..." (Staff 6). "One of the things that concerns me is the assumption that everybody's got a smart phone or computer. I volunteer...for people with mental health problems and virtually all of them are in a poverty situation and they haven't got a computer and they find it hard to live day to day and keep a roof over their heads so there's no way they can afford broad band etc. " (PPI 1). Whether online methods could better reach rural communities than offline methods also produced mixed opinions from participants. For example, lack of infrastructure such as broadband was seen as a disadvantage, however where this was present online methods had the potential to open geographically distanced communities to research opportunities. The correlation between the pandemic and confidence in using technology such as Zoom was interesting because some participants agreed that post-pandemic, there was a real need and desire for people to reconnect in person. However, others discussed how more people had embraced technology because of the pandemic and now see it as a benefit. --- "...no one would have imagined my elderly parents using YouTube five years ago and now they are quite savvy at using it because they get a lot of benefit out of it [through COVID]. Your mum started using WhatsApp groups. " (PPI 5). "I think it worked well, particularly because we had to recruit during COVID. And so initially we were going to offer assessments face to face, but we had to take all of that out...most of our consent was online and then all the assessments were online as well...it's more flexible, it allows that better reach as well. We were able to recruit people in [ --- different geographical areas], which otherwise would have been quite a challenge to do, if you had to travel. " (Staff 3). --- Theme 3 -safety of technology The majority of participants commented on two main areas encompassing confidence and personal interactions. --- Confidence Although some participants felt people are now more internet aware, issues regarding confidence and security were prominent in the data. All participants agreed that offline methods were considered more trustworthy than online due to existing relationships with health providers and felt it could be hard to ensure the authenticity of online communications. There was contrasting PPI opinions on the safety of data storage, whether people were now more used to providing, holding and sharing data online versus that mistrust and concern over privacy. --- "...you have to think about that ethically, don't you? It's OK saying your information is stored and protected, and we look after it but if it's online, you can't fully honestly say we have got this secure. Can you? I'm not saying even if it was face to face, you could be secure. But it's like there's probably less material that would be shared publicly online. " (PPI 15). "I don't think in this day and age, the majority of people...going to be too concerned if they agree to do a study, if they are assured that the personal data is anonymised, that anything that they put in, is logged under a number..." (PPI 9). Participants also felt that older people would be more mistrusting and reluctant to use online methods. --- "...if I look at my daughter and me, I wouldn't go, for example shopping online...because I don't know what that site is, whether it's a trusted site or not, but with the younger generation they don't really think, they just click..." (PPI 4). Many of the participants felt that, if the trial recruitment wasn't targeted well or demonstrated trustworthiness and good data protection, it could be viewed as spam or, worse, a phishing email, and therefore just deleted. --- "I don't open emails because I don't know who they're from. There's a lot of spam, saying click on this and this happened, so unless I know the email, from somebody I know I don't open them. " (PPI 4). Interestingly, staff participants reported different ways of managing online data safety. Some universities initiated a routine transfer onto the university's safe haven and clear of website data every day. Other universities used very secure platforms and a strong data management plan, so they don't have to delete data regularly. --- Personal interaction Credibility of the recruitment method was an important consideration for PPI participants. The source of the contact had to be recognisable, perceived as trustworthy and safe. Offline methods were deemed safe since they were typically through direct contact with what were perceived as a "trusted" individual such as a health professional, whereas online methods were seen as less so. --- "If you e-mail asking me to join a trial, it's in spam, I'm far less likely to take any notice which takes us all the way back to in the clinic [in-person], somebody who is appropriately dressed, talking appropri-ate language..." (PPI 9). "I think it does help coming from somebody that you know, someone that you trust, because I think you feel more secure in taking part. Whereas online there might be some uncertainty". (PPI 14) In their discussions surrounding trust, both staff and PPI participants stated that the participant/researcher relationship could be built up more easily using offline methods compared to online and if using online then a follow-up contact via telephone, or video conferencing or in-person, was still considered important for retention in RCTs and to build rapport and trust with participants. --- "...creating the rapport with the families is very important and that blended approach...I'm seeing that in the trial we're doing right now, it's really, really good and I don't think it would work as well if you just did everything online. " (Staff 6). "I think it can be quite impersonal online. So it might help to build a better report if it's if it's not online. " (PPI 15). --- Theme 4 -practical challenges to online recruitment Practical challenges emerged mainly in relation to staff knowledge and experience, organisational support and technical considerations. --- Staff knowledge and experience Experience of the wide range of digital platforms (e.g. Facebook, Twitter, Instagram, TikTok) in a research setting was limited, with staff participants citing lack of training and awareness as a significant issue, particularly around participant safety and data protection. Collection of individual data using digital platforms informing eligibility in the recruitment phase was of specific interest. --- "I have to say, when we started this, none of us had in the trial team had any experience online advertising. So it was quite a steep learning curve..." (Staff 7). --- Organisational support and funding Appropriate resourcing including funding, staff time and organisational support were highlighted as essential. A lack of understanding as well as fear of digital safety at the organisational level to engage with social media platforms were barriers to the use of digital platforms. "I do find our clinical trials unit is hyper cautious about everything like that [collecting data], to the point sometimes of being a bit debilitating. It's almost like you're not allowed to store any data anywhere. " (Staff 7) "One of the most complicated things we had was how to pay for it. You know, we work in a university. How on earth do you pay Facebook?" (Staff 7). --- Technical considerations For online communication (via Facebook, Twitter, emails) in terms of information, conciseness, presentation and empathy, are key factors. Participants felt this style of communication, due to its typically limited capacity, should have a clear purpose and be from a respected source to address trust and safety issues. There was also concern from PPI participants that sometimes digital communication could be seen as a bombardment of information which could just get lost. --- "We can't absorb all this stuff that's thrown at us and it's just thrown at you in a way, isn't it? It [emails] sort of comes at you and, you know, unin-vited..." (PPI 15). --- Discussion This study set out with the aim of exploring the perspectives of key partners on the use of online recruitment methods and comparing their pros/cons with offline ones in mental health RCTs. --- Summary of principle findings Four key themes were identified from the data analysis including recruitment reach, demographic factors that affect selection of recruitment method, safety of technology, and practical challenges to online recruitment. Overall, the study findings indicated the need to offer a flexible and multifaceted approach to participant recruitment to support trial teams to recruit a participant population that is reflective of the people that stand to benefit from the intervention being tested. --- Recommendations for recruitment practices Using a range of recruitment methods could improve inclusivity by expanding opportunities for participation The study participants overall identified the importance of using both online and offline methods in parallel in recruiting individuals into mental health RCTs. This corroborates the idea of Dawson et al., who recommended the use of an inclusive approach to improve recruitment and retention in RCTs [40]. Whilst integrating online methods is seen as progressive, there are still key challenges researchers should consider, for example online methods are attractive to reach wide geographical populations to improve inclusivity [41], however, it is crucial to consider the risk of losing sight of who is responding [42]. In addition, we found implications for resource allocation to manage digital platforms, the need for welldefined monitoring and screening processes and management of individual expectations in terms of eligibility. --- Selection of methods should be based on your target participant population It was agreed that one size does not fit all as an approach to recruitment. People with mental health issues, of all ages, and their carers where present need methods that support their situation both mentally and physically and offer convenience. This finding is consistent with that of a systematic review which found that researchers needed to consider potential participants' preferences and beliefs that could influence both health provision and willingness to participate in RCTs [43]. Age was a significant factor for consideration of recruitment methods. Although technology was historically seen as a barrier for older people, the problem is likely to reduce with greater generational exposure to technology (smartphones, social media, etc.) [44]. It was clear from the findings that stereotypes of age and technology should be avoided, both for older and younger people and that access to technology remains a significant barrier across all age ranges, specifically affording technology and access to WiFi. Therefore, to be age inclusive a multi-method approach may be preferable and is at present consistent with the literature [19]. Severity of mental health issues and confidentiality has previously been identified as a barrier to participation in mental health research [45][46][47]. This study found that it is crucial to use an individualised approach where researchers should consider the type of mental health issue, its stage, participants' feelings and carers' responsibilities when selecting recruitment methods. For example, using face-to-face methods with people who would benefit from direct contact (low mood) and using online ones for people with conditions where face-to-face interaction might be more challenging (anxiety, obsessive thinking, autism). Using a multi-method approach to recruit people with mental health issues may therefore improve inclusivity and representation. This suggestion is in accordance with other studies indicating that using a balanced recruitment approach is more effective in attracting people to participate in RCTs [40,48]. There is difficulty in recruiting people from certain ethnic communities who remain underrepresented in mental health research, threatening the generalisability of the trial results and health equality. A strong relationship between cultural/ethnic background and poor participation in mental health research has been reported in the literature [47,49]. The stigma attached to mental health experienced by ethnic minorities, such as South Asian communities and Black African and Black African Caribbean communities, was perceived to be a strong barrier to participation in mental health research. This stigma has been widely researched and proven to be a factor affecting recruitment, engagement and trust in mental health services and research [47,50]. Attempts to address stigma and mistrust may increase recruitment. Some participants suggested building relationships with trusted groups and community centres was deemed crucial to facilitate recruitment and achieve ethnic diversity. Community outreach work as well as identifying trusted individuals in these communities have been found to enhance recruitment and engagement efforts [51]. Although some participants mentioned that using the internet has become more 'normalised' due to the impact of the COVID pandemic, some people with mental health issues still do not have access to the internet (first-level digital divide) and some do not have the skills to use it (second-level digital divide) [23]. According to recent research at the University of York, people with severe mental health conditions, such as schizophrenia or bipolar disorder, were more likely to lack digital skills and were at greater risk of social isolation due to the digitalisation of health and social care services and research [52]. Digital poverty was raised as a factor affecting recruitment in this study, interestingly opinions were divided between our staff and PPI participants. Staff felt that smartphone access was less of an issue based on their recent experience recruiting successfully to mental health trials. This perspective may also be reflected in the evidence that suggests in 2021 in the UK 88% of all adults had access to a smartphone [53]. When broken down by age, 96% of those aged 16-24 owned a smartphone device compared to 78% aged 55 and above. However, our PPI participants from their own experience working with community and youth groups indicated economic poverty including digital poverty was still an issue. This evidence suggests that to be inclusive, methods of recruitment in RCTs should be varied, accessible and not discriminatory based on access to technology. --- Online methods should be safe for participants In recent years, especially post-pandemic, it has been reported that wider society has become more internet aware and trusting of digital platforms through necessity. However, in our study, predominantly our PPI participants felt that there remained an element of fear and scepticism in using online methods compared to offline methods, which they described as more trusted and secure. This is despite evidence suggesting participants being potentially better protected [54]. It is therefore recommended that where online methods are to be used, information on data protection is clear, concise, and readily accessible. Within our sample we found that vulnerability and discomfort sharing personal information could be a disadvantage of online methods because of concerns regarding confidentiality and anonymity [45,55]. However, this was offset against the advantages of in-person contact which provided opportunities to read body language and person cues deemed important to support mental health populations during the recruitment process. Verbal and nonverbal communication has a vital role in the process of meaning generation through understanding body posture, tone of voice, and rate of speech. This in turn builds trust and motivation required for participants to understand the aim of the intended trial and improve recruitment [56]. Our staff participants also suggested a range of amelioration strategies that match those observed in other studies recommending the implementation of strict screening procedures and the use of restrictive software features for online methods [57,58]. Several studies have noted and how mistrust may negatively affect willingness to participate in research [9,59]. It is interesting to note that in our study there were concerns about solely using online methods, as offline methods were described to give participants that feeling of trust, comfort, engagement and empowerment where people can ask questions. In addition, online methods were considered a limitation for personal interaction thereby potentially reducing understanding and engagement. This is in line with Balfe et al., (2012) who found that use of online recruitment methods could alter the researcher-participant relationship and bring challenges in creating and maintaining engagement and trust between researchers and participants [60]. --- Ensure the recruitment methods selected are appropriately resourced and staff adequately trained Staff knowledge, organisational support and technical considerations were the most common challenges discussed by the staff participants. Several reports have shown that researchers could inadvertently contribute to underrepresentation through lacking the experience and skills required to recruit and engage a diverse population [61,62]. This highlights the urgent need for training researchers on various recruitment methods to improve the quality of the included sample, the inclusivity of RCTs and the generalisability of data. Despite advances in online recruitment methods, technology of pre-screening and initiatives to train researchers, many of these advances will remain underutilised without enough resources. Constrained or low funding remains a barrier to the use of online methods or social media platforms to safely and effectively recruit potential participants [58]. Although the process of targeting potential participants using online methods is considered easy, time-efficient and cost-effective [18], a major challenge still exists of how to differentiate trustworthy and legitimate recruitment invitations

Background Barriers to mental health research participation are well documented including distrust of services and research; and stigma surrounding mental health. They can contribute to a lack of diversity amongst participants in mental health research, which threatens the generalisability of knowledge. Given the recent widespread use of the internet in medical research, this study aimed to explore the perspectives of key partners on the use of online (e.g. social media) and offline (e.g. in-person) recruitment as an approach to improving diversity in mental health randomised controlled trials (RCTs). Methods Face-to-face and online interviews/focus groups with researchers working in mental health and Patient and Public Involvement partners in the United Kingdom. Recordings were transcribed and analysed using a combination of inductive and deductive thematic analysis.Three focus groups and three interviews were conducted with a total N = 23 participants. Four overarching themes were identified: (1) recruitment reach; (2) Demographic factors that affect selection of recruitment method; (3) safety of technology, and; (4) practical challenges. Five main factors were identified that affect the choice of recruitment method: age, complexity of mental health problem and stigma, cultural and ethnicity differences and digital divide. The use of online methods was considered more accessible to people who may feel stigmatised by their mental health condition and with a benefit of reaching a wider population. However, a common view amongst participants was that online methods require closer data monitoring for quality of responders, are not fully secure and less trustworthy compared to offline methods that enable participants to build relationships with health providers. Funding, staff time and experience, organisational support, and technical issues such as spam or phishing emails were highlighted as practical challenges facing online recruitment. All participants agreed that using a hybrid approach tailored to the population under study is paramount.This study highlighted the importance of offering a flexible and multifaceted recruitment approach by integrating online with offline methods to support inclusivity and widening participation in mental health research.

et al., (2012) who found that use of online recruitment methods could alter the researcher-participant relationship and bring challenges in creating and maintaining engagement and trust between researchers and participants [60]. --- Ensure the recruitment methods selected are appropriately resourced and staff adequately trained Staff knowledge, organisational support and technical considerations were the most common challenges discussed by the staff participants. Several reports have shown that researchers could inadvertently contribute to underrepresentation through lacking the experience and skills required to recruit and engage a diverse population [61,62]. This highlights the urgent need for training researchers on various recruitment methods to improve the quality of the included sample, the inclusivity of RCTs and the generalisability of data. Despite advances in online recruitment methods, technology of pre-screening and initiatives to train researchers, many of these advances will remain underutilised without enough resources. Constrained or low funding remains a barrier to the use of online methods or social media platforms to safely and effectively recruit potential participants [58]. Although the process of targeting potential participants using online methods is considered easy, time-efficient and cost-effective [18], a major challenge still exists of how to differentiate trustworthy and legitimate recruitment invitations from spam and fraudulent misinformation on the internet [63]. --- Strengths and limitations To the best of our knowledge, this is the first study to explore from the perspective of key partners, the advantages and disadvantages of online and offline recruitment methods in RCTs involving mental health populations. This study is grounded in the views and experiences of patients, the public and clinical researchers who work in trial design, conduct and delivery. Although we had input from our experienced multidisciplinary team and two young PPI partners to develop the topic schedule, we acknowledge that having input from a wider group of PPI would have been preferable, but due to funding restraints this was not possible. Another limitation of this study could be the relatively small number of participants, especially in the research staff group, however, the purposive sampling and the diversity in the study sample that included various partners, age groups, ethnicity and genders, could mitigate this risk. We also acknowledge that respondent validation checks were not sought from participants for interview/focus group transcripts due to time and resource constraints although recognised to improve validity in qualitative research. The results of the research have however been shared with those who participated in the interviews and focus groups to which we have encouraged comment to inform future research in this area. Lastly, there may be some drawbacks of having two coders from the same research team. However, interrater reliability varies widely depending on the pair of coders [64] and in this study this may have been mitigated by the different backgrounds of the two coders and there experience in qualitative research and thematic analysis. --- Conclusion This qualitative sub-study aimed to explore the experiences of key partners on the use of online and offline recruitment of participants into mental health RCTs. Despite perceptions that COVID-19 may have increased the use of digital technology across age and population groups the major finding of this study was the general agreement of using hybrid, balanced and targeted recruitment approaches. As a result, this study highlighted the importance of integrating online with offline methods and considering the preferences of the population under study and their carers, the type of mental health issue and its severity, which would enhance recruitment and inclusivity in mental health RCTs. The findings from this study will be used to develop evidence and considerations for researchers designing RCTs to improve recruitment and engagement in mental health research. --- Data Availability Data are available on reasonable request. The unpublished data used and/ or analysed during the current study are available from the corresponding author on reasonable request. --- Abbreviations --- Declarations Competing interests The authors declare no competing interests. --- Ethics approval and consent to participate The RE-MIND study received University of Nottingham Ethical approval (FMHS 13-0422) on 13th June 2022. Informed consents were accepted as a return of email (to the invitation email) stating the participant had read and understood the consent form and agreed to participate. This study was carried out in compliance with Ethical Principles for Medical Research Involving Human Subjects outlined in the Helsinki Declaration [65]. The researcher(s) explained to participants that entry into the study was entirely voluntary and that they could withdraw at any time. In the event of withdrawal, it was clear to the --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background Barriers to mental health research participation are well documented including distrust of services and research; and stigma surrounding mental health. They can contribute to a lack of diversity amongst participants in mental health research, which threatens the generalisability of knowledge. Given the recent widespread use of the internet in medical research, this study aimed to explore the perspectives of key partners on the use of online (e.g. social media) and offline (e.g. in-person) recruitment as an approach to improving diversity in mental health randomised controlled trials (RCTs). Methods Face-to-face and online interviews/focus groups with researchers working in mental health and Patient and Public Involvement partners in the United Kingdom. Recordings were transcribed and analysed using a combination of inductive and deductive thematic analysis.Three focus groups and three interviews were conducted with a total N = 23 participants. Four overarching themes were identified: (1) recruitment reach; (2) Demographic factors that affect selection of recruitment method; (3) safety of technology, and; (4) practical challenges. Five main factors were identified that affect the choice of recruitment method: age, complexity of mental health problem and stigma, cultural and ethnicity differences and digital divide. The use of online methods was considered more accessible to people who may feel stigmatised by their mental health condition and with a benefit of reaching a wider population. However, a common view amongst participants was that online methods require closer data monitoring for quality of responders, are not fully secure and less trustworthy compared to offline methods that enable participants to build relationships with health providers. Funding, staff time and experience, organisational support, and technical issues such as spam or phishing emails were highlighted as practical challenges facing online recruitment. All participants agreed that using a hybrid approach tailored to the population under study is paramount.This study highlighted the importance of offering a flexible and multifaceted recruitment approach by integrating online with offline methods to support inclusivity and widening participation in mental health research.

Introduction The cervix is the lower end of the uterus, and cervical carcinoma is an abnormal rapid growth of anomalous cells of the cervix [1]. It is the second most frequent condition in females under 50 years of age and the fourth most prevalent cancer in women across all age groups worldwide [2][3]. In 2018, the World Health Organization (WHO) reported the incidence to be 570,000, constituting 7.5% of cancer-related female deaths, indicating that cervical carcinoma costs a life every two minutes all over the world [3][4]. South Asian countries constitute one-third of the total disease burden, while 85% of all deaths are reported in developing nations [3,5]. In Pakistan, the situation is even worse since it is one of the top ten countries with the highest female mortality rates, and 20 women are diagnosed with cervical cancer every day [6]. Moreover, the major age drift in the last few decades has increased incidence to 40% in the younger female population, with women aged 30-39 years being affected the most [5,7]. Cervical carcinoma may be asymptomatic in its early stages; nevertheless, patients can experience foulsmelling vaginal discharge and abnormal bleeding such as intermenstrual bleeding, post-coital bleeding, or postmenopausal bleeding [3]. The various risk factors that are linked to cervical cancer are high parity, early age at marriage, multiple sexual partners, smoking, low socioeconomic status, poor personal hygiene, and long-term estrogen exposure in the form of oral contraceptive pills [5]. Human papillomavirus (HPV) infection is found to be an essential causative factor in almost all cases of cervical carcinomas throughout the world, with 70% of the cases attributable to HPV oncogenic subtypes 16 and 18 [5]. It is quite alarming that more than one-third of all patients diagnosed with cervical carcinoma ultimately die when it is preventable and curable at an early stage, and around 50-90% of females who develop or die due to cervical cancer have never been screened [3,[8][9]. Cervical cancer can be effectively controlled through primary, secondary, and tertiary preventive measures, which include prophylactic HPV vaccination, screening, diagnosis, and treatment of pre-cancerous and invasive cervical cancer [3]. Papanicolaou cytological testing (also known as Pap smear test) is used as a screening tool to identify precancerous lesions of the cervix, effectively lowering its incidence by 75-90% [10]. Although routine screening with Pap smear test has substantially reduced the incidence in the developed world, the scenario is entirely different in low-and middle-income countries, where dearth of screening facilities and HPV vaccination can be ascribed in part to lack of resources, but mainly to the serious lack of knowledge and attitude among population regarding early detection of cancer and its high mortality rates [5]. Likewise, the introduction of the HPV vaccine is limited by many factors such as its high cost [11]. Owing to these barriers, less than half of the participants of a study conducted among university students of Lahore, Pakistan knew about the prophylactic HPV vaccine [12]. Comparably, the awareness level among females in the most populated province of Pakistan was found to be low (29.9%) and surprisingly, only 37% of health professionals recognized Pap smear as a screening test [13][14]. This highlights that in addition to a lack of a mandatory screening practice in place for cervical cancer in our country, the poor knowledge and ineffective approach of healthcare professionals also act as major factors in contributing to the lack of awareness of the masses and subsequent poor prevention practices. Socioeconomic status is highly associated with increased cervical cancer risk [15]. Hence, it is generally perceived that education level and socioeconomic conditions can influence the attitudes and awareness regarding risk factors and screening practices of females, which can eventually modify the potential of HPVinducing cervical carcinoma. Thus, this study was conducted to assess the various variables that may affect the knowledge and perception of female population belonging to different socioeconomic settings of Karachi, Pakistan regarding cervical cancer, its major symptoms and risk factors, its relation to HPV, its early detection and prevention through screening and vaccination, and practice and attitude towards screening and prophylactic HPV vaccination. --- Materials And Methods A descriptive, cross-sectional study was conducted in the waiting area of the out-patient department (OPD) of a tertiary care hospital in Karachi, Pakistan between June 2019 and November 2019. The study population consisted of females aged between 17-65 years and were recruited using the convenience sampling technique. Questionnaire-based interviews were conducted after obtaining informed consent from all participants, and their confidentiality and anonymity were maintained. Females with a history of cervical cancer, females with any gynecological disease, and those who did not give consent were excluded from the study. The sample size was calculated through the OpenEpi sample size calculator [16]. Keeping a confidence interval (CI) of 95% and a 5% degree of precision, the estimated sample size was 317 using an anticipated frequency (p) of 29.1% [13]. In order to get the maximum response, we increased sample size and approached 450 females, out of which 396 consented to be part of the study. After excluding incomplete questionnaires, the response rate was found to be 86.22%. The questionnaire was formulated based on objectives of the study after going through the relevant data available on the topic [4,9,[12][13][14]. The questionnaire was pretested on 30 females before the survey to validate and modify the questions accordingly. The final questionnaire was divided into four main sections. The first section inquired about the socio-demographic data of respondents. The second section consisted of 22 items and was further divided into six sub-sections that assessed the knowledge regarding cervical cancer, its symptoms, its risk factors, HPV, cervical cancer screening, and prevention respectively. The third and fourth sections of the questionnaire assessed the attitudes and practices of respondents related to cervical cancer screening and prevention. The attitude was assessed by asking if they would be willing to get a Pap smear test and HPV vaccination. They were also inquired whether they would be interested to know more about the screening and prevention of cervical cancer. Practices were assessed by inquiring if they had undergone the Pap smear test and if they were vaccinated against HPV. The modified Kuppuswamy socioeconomic scale updated for the year 2018 was used to assess the socioeconomic status of respondents [17]. Family incomes mentioned in the modified Kuppuswamy socioeconomic scale in Indian rupees were converted to Pakistani rupees (PKR) using an online converter and were rounded off to the nearest 500 in PKR [18]. Using relevant data, respondents were classified according to this scale as upper (i.e. upper), upper-middle and lower-middle (i.e. middle), upper-lower (i.e. poor) and lower (i.e. very poor) classes. A 30-point scale was used to assess the knowledge of cervical cancer. Each sub-section of knowledge was assigned five points so that respondents were expected to score between 0-30 points. Every correct answer was assigned one point and wrong answer zero point. Only those respondents who answered "Yes" to the first question of the questionnaire, "Do you know about cervical cancer?", were given a knowledge score. While evaluating the knowledge score, the knowledge level was based on the original Bloom's cut off points [19]. Respondents who scored between 24-30 points were considered as having good knowledge with 80-100% correct responses, those who scored between 18-23 points as having moderate knowledge with 60-79% correct responses, and those who scored <unk>18 points as having poor knowledge with <unk>60% correct responses. Knowledge scores for each sub-section were further grouped on the basis of three or more correct answers as "good" and two or less than two correct answers as "bad" scores respectively. Those participants who answered "yes" in at least two out of three attitude questions were considered as having a positive attitude, while those who said no in two out of three questions were categorized as having a negative attitude. Practices were analyzed by classifying respondents as having "regular practice" for those who have had Pap smear test and vaccination both done or any one of them done, and "irregular practice" for those who have had neither Pap smear test nor vaccination done. Data were analyzed using Statistical Package for Social Sciences (SPSS) version 24.0, (IBM Corp., Armonk, NY). Descriptive statistics, e.g., frequencies, percentage, and mean and standard deviation were used for categorical and continuous variables respectively. The chi-square test was performed to determine the association between socioeconomic and demographic variables with knowledge, attitude, and practice (KAP) levels. A p-value of <unk>0.05 was considered statistically significant. --- Results Table 1 Those 199 females were also questioned regarding their knowledge of symptoms and risk factors associated with cervical cancer (Figures 12). Most of the women were aware of lower abdominal pain and weight loss as being the predominant symptoms of cervical cancer. Similarly, a preponderance among the responses regarding the risk factors "unprotected sexual practices" and "HPV infection" was observed. The participants of the study were scored in each sub-section based on their knowledge regarding HPV and the epidemiology, screening, prevention, risk factors, and symptoms of cervical cancer; 49.2% of the people obtained a good score regarding knowledge of cervical cancer, 64.8% for symptoms, 55.8% for risk factors, 40.2% regarding HPV, 28.6% for screening, and 36.2% had a good score regarding prevention of cervical cancer. The total knowledge score that each participant obtained in all sub-sections was added up and classified based on the original Bloom's cut-off point (Figure 3). --- FIGURE 3: Knowledge score of the participants based on the original --- Bloom's cut-off point The participants were enquired about the sources of information regarding cervical cancer and the major sources were found to be family/relatives, doctors, and TV/radio/social media respectively (Figure 4). --- FIGURE 4: Source of information about cervical cancer All women (388) were interviewed to assess their willingness to undergo screening and undertake prevention practices regarding cervical cancer. In our study, the majority of participants had a negative attitude towards HPV vaccination and Pap smear test, whereas a majority was eager to know more about the screening and prevention practices (Table 3). The respondents were further interviewed regarding what they would like to know more about screening and prevention practices of cervical cancer. Most of the population was inclined to know more about the efficacy of these practices (Figure 5). --- FIGURE 5: Information regarding screening and prevention of cervical cancer that the respondents wanted to know The participants of the study were questioned regarding their practices related to screening and prevention of cervical cancer. The majority were seen to observe "bad practices" as only eight people out of 388 had a Pap smear test done during their lifetime, and only seven were found to be vaccinated against HPV. The reasons for these unsatisfactory practices were asked, and the most prevalent was the lack of awareness regarding these practices (Table 4). --- N (%) Have Table 5 shows the assessment of knowledge based on participants' level of education and their socioeconomic status. There was a statistically significant relationship between the socioeconomic status of participants and their knowledge regarding the term 'cervical cancer', whether it is a communicable disease, Pap smear test, and HPV vaccine. Participants belonging to the upper class were observed to have better knowledge as compared to those in other socioeconomic classes. Similarly, females who were university graduates were observed to have better knowledge than those with comparatively lower levels of education. We observed that there was a significant association between the participants' age, marital status, education level and positive attitude regarding cervical cancer screening and prevention (Table 6). Participants, who were of younger age group, were married, and those who had higher education were observed to have a positive attitude than their counterparts. There was no statistically significant relationship found between positive attitude and socioeconomic status of participants. This table also reveals that there was a statistically significant relationship observed between positive attitude and good knowledge scores regarding HPV, cervical cancer, and its symptoms and screening. --- Discussion Cervical cancer is one of the most preventable cancers among women [13]. Despite that, it is the third leading cause of mortality among the female population worldwide, especially in countries with stringent healthcare budgets, such as Pakistan [13]. This can be attributed to the lack of outreach by the government and media in order to raise awareness about cervical cancer. National screening and vaccination programs in developed countries such as Australia and the United Kingdom have reduced cervical cancer burden up to 90% [12,20]. However, in developing countries such as Pakistan, there is a dearth of such resources and taboos associated with this disease. Owing to this, women are unable to undertake proper measures towards screening and prevention of this deadly disease, contributing to the huge mortality rates associated with cervical cancer [12,21]. This study aimed to assess the knowledge of the general population about cervical cancer epidemiology, etiology, risk factors, symptoms, and their attitudes and practices related to screening and prevention. Moreover, the association of socio-demographic parameters, especially economic stability, with knowledge and attitude was investigated as there was a positive correlation observed in previous studies conducted globally [20,[22][23]. Our study has highlighted an overall lack of knowledge regarding HPV and the epidemiology, screening, prevention, risk factors, and symptoms of cervical cancer, with 64% of the respondents securing a total knowledge score of less than 60% and only 10% being able to score above 80%. A review by Raychaudhuri and Mandal shows that the same situation prevails worldwide as studies conducted in high-income countries like Japan and Korea emphasize a lack of awareness regarding the disease [5]. Studies conducted in low-and middle-income countries like South Africa, Ethiopia, and Indonesia also demonstrated subpar knowledge [5,24]. Furthermore, only 25.5% of women obtained a poor knowledge score in the neighboring country India, which is substantially better than the findings of our study, underlining the status of substandard knowledge among women in Pakistan [19]. The current study depicted that only half of the women visiting the tertiary care hospital of the most populated city of Pakistan were aware of the term "cervical cancer." This finding contrasts with the studies conducted in the Democratic Republic of Congo and India where the awareness regarding cervical cancer was much higher [19,25]. On the contrary, another study carried out in Pakistan showed that the awareness level of women was below average [13]. Similarly, a survey done in Karachi among health workers revealed that only 1.8% of the participants did not know about cervical cancer as a disease [14]. This depicts that there is an evident gap of knowledge between the general population and healthcare workers, which needs to be bridged. Our study showed that the awareness regarding signs and symptoms of cervical cancer and risk factors associated with it was only 34.2% and 19.6% respectively among the participants, which was drastically below par when compared with the regional study of Narayana G et al. where twice as many women were familiar with the presenting complains and risk factors respectively (Table 7) [19]. This lack of knowledge usually contributes to the high mortality associated with cervical cancer because women do not take proper preventive measures if they are unenlightened about the risk factors. Likewise, late presentation or failure of recognition of the symptoms may lead to delayed seeking of medical attention, leading to poor prognosis and increased mortality. In our population, 40.7% of women were aware that cervical cancer was one of the most common gynecological cancers in contrast with the knowledge of healthcare workers of Pakistan reported in a study, in which only one-fourth of the participants were found to be aware of the seriousness and virulence of this disease [14]. Despite being aware of the prevalence of this disease, only 54.3% knew that HPV was transmitted via sexual contact. This is similar to findings by a study conducted in Lahore, Pakistan where only about half of the population was aware of the sexually transmitted nature of HPV [12]. This can be attributed to the social and cultural taboos associated with sexually transmitted diseases (STDs), which means that information is not widely shared due to the stigma surrounding this topic and, hence, ignorance prevails [13]. Furthermore, among those who had heard of cervical cancer, around one-fourth presumed it to be a communicable disease, which is comparable to a study conducted in the Maldives where 15.5% believed the disease to be infectious in nature [26]. This depicts a lack of knowledge regarding the etiology of the disease and can potentially further stigmatize the disease, causing isolation and preventing women from seeking medical treatment [26]. Furthermore, in our study, only 34.2% were aware that Pap smear is a screening test for cervical cancer, and only 2.1% had gone for Pap smear test during their lifetime, whereas in Nigeria, another developing country, 55.1% were aware of the Pap smear test and 22.9% had undergone this screening test [27]. The lack of knowledge and unsatisfactory practices were not unexpected in our study given that a recent research study carried out in Pakistan showed that only 35.4% had heard about the Pap smear test and only 5.9% had opted to go for it [21]. A study by Imam et al. conducted in Lahore, Pakistan reported that about 95% of women were never advised to go for a Pap smear test by any doctor [28]. Similarly, one of the main barriers towards the Pap smear test was the non-recommendation by physicians as found in studies conducted in Nigeria, Kuwait, and Iran [27]. Hence, it is evident that cervical cancer will continue to impose a toll globally unless health workers counsel women regarding the available screening and treatment options. Moreover, there has been a significant reduction in the rates of cervical cancer morbidity and mortality in the United States (US), Canada, and the majority of European countries due to the highly efficient screening programs put into place [5]. Whereas, developing countries like Pakistan are lagging behind and will continue to bear the burden of this disease unless they establish such national screening programs. Prevention of HPV infections is essential to bring down the prevalence of cervical cancer, which is possible by the use of HPV vaccination; therefore, this important preventive tool should be made easily available to the general population. However, our study found that only 40.2% were aware of HPV vaccine's ability to prevent cervical cancer and only 1.8% were vaccinated against it, which is similar to the findings of a study conducted among university students of Karachi, Pakistan where only 19.3% of the participants were aware of this important information and only 1.3% had undertaken the vaccination [29]. Furthermore, 10% were under the assumption that there was no need to be screened if they were vaccinated, which is comparable to the findings of Khan et al. where 14% were found to be under this false belief [12]. This is largely due to lack of knowledge and the presumption of "not needing it right now" as found in our study and another study conducted in Karachi [29]. Furthermore, developing countries such as Pakistan are restrained by their limited health budgets; hence, the administration of these vaccines in large-scale populations is difficult because they are cost-intensive. Another common basis for this unsatisfactory practice and the prevalence of insufficient information was the unacceptability and non-recommendation by the physicians as shown in the study conducted in Lahore where health workers' approval played a critical role in influencing the students' decision to get vaccinated [12]. The attitude regarding undergoing the Pap smear test was found to be negative with only 30.2% willing to be screened in the future. As shown in Table 7, this contrasts remarkably with the study conducted by Ali-Risasi et al., where almost 80% of people were willing to undertake the screening test [25]. However, a study conducted in Pakistan showed similar results in that only one-fourth of the population was willing to go for a screening in the next three years, with the most predictable barrier being the feeling of embarrassment among the females [21]. This depicts that there is a dire need for the creation of a screening program, which will help females become accustomed to these tests without any social barriers. In a study conducted in Indonesia, almost all of the participants were willing to go for HPV vaccination with the most encouraging factor being the belief in the efficacy of the vaccine [24]. However, a study conducted in Karachi showed that only 37.5% were willing to accept the vaccination for themselves, among which almost half of the women believed that the cost of vaccines should be subsidized by the government [30]. These findings are strikingly similar to our study where 49% of women consented to go for vaccination if it was provided free of cost. Around 80% of women in our study wanted to know more about this disease, its screening, and prevention, which shows a strikingly positive attitude among people towards gaining awareness. However, in a survey conducted by Khan et al., around 93% of women stated that media was not fulfilling its role in raising awareness about this deadly disease and suggested that more information regarding this disease should be made available via campaigns, talk shows, and health bulletins [13]. This shows that a great responsibility lies on the government and health sector of our country in utilizing media as an informative tool for the propagation of awareness regarding cervical cancer, its transmission, prevention, screening, and other relevant information. Furthermore, there was a significant relationship between knowledge of participants and their socioeconomic status; however, no association was found between socioeconomic status and the willingness of women to undergo Pap smear test and take HPV vaccination. Women from the upper and upper-middle classes had better knowledge than those who belonged to a lower class. This is comparable to a study conducted in India where women with higher household income and those living in urban areas were found to have adequate knowledge [19]. This can be attributed to women with a higher socioeconomic status being able to access media and better healthcare facilities. Knowledge of the participants regarding cervical cancer was also found to be significantly associated with education, with those with a university-level education having better knowledge scores than the women from other education levels. This is comparable with the findings of two studies conducted in the Maldives and Congo in which women with higher education had increased knowledge [25][26]. A study conducted in Pakistan showed the same association, which signifies that women who are highly educated appear to have access to health information and resources to gain more awareness [21]. The current study also showed that the positive attitude of the females to go for cervical screening and prevention was significantly linked with age, marital status, and education level. A study conducted among Australian women showed a similar correlation: women aged between 30-49 years, who were married, and with higher education were more likely to have had Pap smear tests than their other counterparts [20]. As seen in this study, no correlation was found between attitude and socioeconomic status. A similar relation was established by Lin et al. regarding a higher willingness to get vaccinated with age, higher education levels, and being married [23]. Hereby, it can be concluded that older women who have had more gynecological visits are more aware of malignancies; hence, they are more willing to go for screening tests. Likewise, women who are currently married tend to visit healthcare facilities more frequently and are more prone to have any genital tract infections than those who are not currently in any relationships; thus, they are more open towards screening and preventive measures. Moreover, in Pakistan, there is a cultural trend that the majority of the women, especially from the lower socioeconomic groups, only visit a gynecologist after they are married; therefore, this positive attitude seems to prevail more in that group compared to others. In a study conducted by Jia et al., higher knowledge scores regarding cervical cancer and a family history of cervical cancer were significantly linked with the willingness to go for screening tests [22]. Likewise, another study conducted in China showed that women who were aware of HPV and its morbidities were more inclined to go for HPV vaccination [23]. The current study found the same correlation between good knowledge scores and a positive attitude to go for screening. This shows that willingness to go for screening and undertaking preventive measures are correlated with higher awareness rates. According to the participants in our study, the most common source of information regarding cervical cancer was family and relatives while doctors, social and print media played comparatively lesser roles. Hence, we feel that social and print media should proactively work in providing information regarding this disease. One such way to do this is by making use of public service messages that are currently commonly employed for breast cancer and family planning awareness. This will help in reducing the stigma surrounding this disease and encourage more women to undertake screening and preventive measures, consequently lessening the burden of this disease. Moreover, 91% of respondents of another research study conducted in Pakistan felt that the government was not fulfilling its role in spreading awareness regarding this disease [13]. Hence, there should be more focus on spreading awareness by the government authorities in order to get favorable outcomes, that is, decreased cervical cancer morbidity and mortality. The foremost limitation of this study is that it was a single-center study, and we relied on convenient sampling instead of random sampling; hence, the outcomes of this study cannot be generalized to the diverse general population of Karachi. Moreover, due to cultural barriers, this study did not include males who have a pivotal role as family heads in influencing women's decisions to undertake screening and preventive measures. Future studies should include this population so that an overview at a household level can be obtained. Despite taking a large sample size, this study did not include any rural population. Therefore, future researchers are advised to conduct larger-scale surveys in rural areas where awareness and practices are even more unsatisfactory. --- Conclusions The findings of this study highlight a dire need for spreading awareness among women of Pakistan regarding HPV and the epidemiology, etiology, screening, and prevention of cervical cancer. Similar to the findings from other studies conducted in Pakistan and other developing countries, unsatisfactory practices and negative attitudes were observed among participants regarding Pap smear screening and HPV vaccination. Furthermore, a significant link was found between knowledge and attitude scores and socioeconomic and demographic parameters. This brings attention to the lack of established national screening programs and large-scale government-subsidized vaccination induction programs in Pakistan, which should be considered and implemented by public and private health sectors in order to decrease the burden of this disease. --- Appendices --- Questionnaire Please note: all responses provided will remain completely anonymous and confidential. By filling this form, you voluntary give consent to be a part of the research "Knowledge, Perception, and Prevention Practices of Human Papillomavirus (HPV) Based Cervical Cancer and its Socioeconomic Correlates Amongst Women in Karachi, Pakistan" The information provided will be solely used for research purpose(s) only. A -Sociodemographic profile --- Additional Information Disclosures Human subjects: Consent was obtained by all participants in this study. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue. Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.

Cervical carcinoma is a widespread disease of the female genital tract, for which human papillomavirus (HPV) is an utmost risk factor. Of the total global burden, the majority is endured by the developing nations of the world, mainly due to inadequate knowledge regarding the disease and ineffective measures taken for its prevention, early detection, and screening. Hence, our study aimed to determine the level of knowledge, general attitudes and perception, and prevention practices related to HPV-based cervical cancer and its socioeconomic correlates among women in Karachi, Pakistan.A cross-sectional, questionnaire-based study was conducted by approaching 450 females in the out-patient department (OPD) of a tertiary care hospital in Karachi, Pakistan from June 2019 to November 2019. The modified Kuppuswamy socioeconomic scale 2018 was deployed to assess the socioeconomic status of participants, while the knowledge score of the participants was determined based on the original Bloom's cut off point. The analysis was conducted using Statistical Package for Social Sciences (SPSS) version 24.0 (IBM Corp., Armonk, NY). Descriptive statistics were used to present the knowledge, attitude, and practice level of respondents. The respondents' knowledge, attitude, and practice scores were compared across socioeconomic and demographic variables using the chi-square test.Of the 388 females interviewed, 199 (51.3%) were aware of the term cervical cancer, and 68 (34.2%) knew about Pap smear as a screening test; only 80 (40.2%) women were familiar with HPV vaccination as prophylaxis against cervical cancer. The practice of screening and prevention was found to be remarkably low (2.1% and 1.8% respectively). Socioeconomic status and education level had a significant association with knowledge of cervical cancer. Although around 64% of participants had poor knowledge, 308 (79.4%) demonstrated a favorable attitude as they were willing to know more about screening and preventive practices regarding cervical carcinoma.The majority of the participants had insufficient overall knowledge about cervical cancer, HPV, Pap smear test, and HPV vaccination, highlighting the need for mass education through health professionals and media. In addition, the government authorities should provide screening services and vaccination against HPV free of cost to promote early detection of lesions and prophylaxis against this deadly disease.

INTRODUCTION Tobacco, alcohol, substance use and drug misuse constitute a major public health concern in the world and in our country. It is important to recognize prevalence of use of these substances and socio-demographic factors associated with their use in order to attain reliable and comparable data. Tobacco smoking is one of the leading causes of preventable mortality and morbidity (1). According to the World Health Organization (WHO) data, nearly 1 billion people in the world smoke tobacco. Smoking is a great economic burden on health systems and families of smokers, especially in developing countries with limited resources (2). According to the Global Adult Tobacco Survey (GATS) data of Turkey, the smoking prevalence decreased among adults from 31.2% (16.0 million) in 2008 to 27.1% (14.8 million) in 2012. In Turkey 41.5% of men and 13.1% of women currently smoke tobacco (3). Although tobacco control effort has made significant progress in recent years, tobacco use still remains as an important health issue in Turkey. Periodic implementation of surveys is important to monitor and evaluate components of the tobacco control program. Alcohol dependence is one of the most severe and most prevalent mental disorders. The World Health Organization estimates that 76.3 million people worldwide suffer from disorders related with alcohol use (4). In the National Epidemiologic Survey on Alcohol and Related Conditions, the 12-month prevalence ratios were 4.65% for alcohol abuse and 3.61% for alcohol dependence (5), with corresponding 12-month prevalence ratios in the National Comorbidity Study of 3.1% and 1.3% respectively (6). In Turkey, Ar<unk>kan et al. (7) reported that the prevalence of alcohol consumption and alcohol dependence were 14.1% and 0.9% respectively. Greater risk in males for alcohol dependence is well-documented across countries and cultures (8,9). According to analysis of the National Epidemiologic Study of Alcohol Related Condition, onset of alcohol dependence peaks at age 18, with rapidly declining onsets after age 25.2 (10). Alcoholism is accepted as a crucial health problem and a burden on national budget in developed countries and therefore analyzed through nationwide projects (11). Although, data reveal that alcohol consumption in Turkey is lower compared to Europe and USA, we believe that alcoholism is still an important health issue in our country and more prevalence studies are necessary in order to understand the extent of the problem. Medicines such as analgesics, opioid substitution drugs, sedatives and hypnotics are increasingly being misused to induce psychoactive effects or to alter the effects of other consumed drugs. United Nation' s International Narcotics Control Board predict that worldwide misuse of prescription drugs will soon exceed illicit drug use (12). Women are more likely to misuse prescription drugs (13). Individuals from lower social class have a higher misuse of prescription drugs (14). Sedatives and hypnotics encompass a large group of substances that are generally used to treat symptoms of anxiety, stress and sleeping disorders. Benzodiazepines and Z-drugs (zopiclone, zaleplon, zolpidem) are two of the most prescribed groups of sedatives/hypnotics. The elderly and patients with a history of substance misuse or dependence have an increased risk of sedative/hypnotic misuse (12). Correcting mood and behavior through controlled drugs is becoming ever more widespread (15). Our clinical observations show that licit drug misuse is an important problem in Turkey and this data should be supported by objective survey studies. According to WHO data, by the year 2008,155 to 250 million people (3.5% to 5.7% of the world' s population) aged 15-64 in the world, used psychoactive substances such as cannabis, amphetamines, cocaine, opioids, and non-prescribed psychoactive prescription medication. The use of psychoactive substances causes significant health and social problems for the users, families and communities (16). Globally, the two most widely used illicit drugs are cannabis (global annual prevalence ranging from 2.6 to 5.0%) and amphetamine type stimulants (0.3-1.2%) (17). At least 85 million adult Europeans have used an illicit drug at some point in their lives, representing around a quarter of Europe' s adult population. Most report having used cannabis (77 million), with much lower estimates for lifetime use of other drugs: 14.5 million for cocaine, 12.7 million for amphetamines and 11.4 million for ecstasy (18). A study conducted in 72 provinces re-vealed that 1.3% of the population used substance at least once in lifetime. Prevalence of substance use was higher in males compared to females and in 15-24 age group compared to above 25 years (19). Results of the health study conducted by Turkish Statistical Institute (2010) revealed that 0.93% of the population used drugs at least once in lifetime, 1.26% of males and 0.61% of females (20). Studies about the epidemiology of substance abuse in Turkey are usually conducted with the young population. In a study conducted with 1720 students, 6.4% reported having used a substance; 2.8% used one within the past year. Prevalence of cannabis use at least once during life-time was 5.9% (21). Akkaya et al. (22) stated that committed drug-related crimes were to a very great extent associated with cannabis. We observe that substance abuse is progressively becoming an important problem in a country like Turkey with a young population. In Turkey, scant information is available on prevalence and socio-demographic determinants of tobacco, alcohol, drug and substance use in general population. This is particularly worrisome, given that there is a progressive increase in the burden of disorders related with the use of these substances. The aim of this study is to determine prevalence ratios of tobacco, alcohol, drug and substance use in a large scale sample of general population and recognize related socio-demographic factors. --- METHODS The study was based on a cross-sectional survey conducted with a nationally representative sample. The universe for the sample was household members aged between 15 and 64 living in 25 provinces of Turkey. The provinces were determined by TUBIM (Turkish Drug Addiction Monitoring Centre) General Population Survey study group. The study was applied in concordance with the Declaration of Helsinki. 18.319 (92%) households specified by TUIK (Turkish Statistical Institute) were visited and successful interviews were conducted with 8045 individuals residing in these addresses. 10.274 household members could not be interviewed for various reasons. Written informed consent was obtained from each subject. Due to the nature of the survey, it presents various challenges to attain In this study we report inappropriate benzodiazepine and antidepressant use by the term "drug misuse" and use of psychoactive substances by the term "substance use". --- Statistical Analysis The study data was analyzed and reported by TUBIM GPS Study Group consisting of one expert epidemiologist, one psychiatrist and one statistician. The Statistical Package for the Social Sciences (SPSS Inc; Chicago, IL, USA) v.15.0 software program was used to construct the databases and perform the statistical analysis. Quantitative variables were reported as mean and standart deviation and qualitative variables were reported as percentages. Categorical variables were compared using Pearson' s chisquared test. P-values <unk>0.05 were considered statistically significant. --- RESULTS The survey was conducted with 8045 respondents of which 49 Life-time alcohol use was found as 28.3%, while use of alcohol in the last 12 months and in the last 1 month were 14.3% and 10.1% respectively. 1.5% of those who used alcohol in the last 1 month take alcohol 1-3 times a week, while 1.3% consume alcohol weekly. 75.3% of users take 1-2 pieces of standard alcohol drink (a bottle of beer, a double rakia, a cup of wine) while 19.6% take 3-4 pieces. 2.5% of the participants (n=203) regularly take alcohol. Their primary choice is beer by 82.9%. Association of socio-demographic characteristics with alcohol use is given in Table 3. Participants aged between 15 and 24 use significantly more alcohol not for life-time but during last 12 months and during last 1 month (p=0.001). Males use alcohol significantly more than females (p=0.001). Alcohol use increases with increasing education level (p=0.001). According to the marital status, being single or polygamous is associated to lifetime alcohol use (p=0.001). There seems to be a positive correlation with alcohol use and income levels. 10.5% of the participants (n=843) misused drugs. The mean age of firsttime use of these drugs is 34.66<unk>14.42 with a median of 33.00. 68.7% of the individuals who misused drugs in the last 30 days indicated that they used the drugs daily. 98.3% use the drugs orally. 61.3% used drugs without a doctor' s recommendation and 94.6% indicated that they attained the drugs with prescription. Table 4 shows the association of some socio-demographic characteristics with drug misuse. Drug misuse increases with increasing age (p=0.001). Females misuse drugs more than males (p=0.001). Education level is significantly associated with drug misuse. Uneducated participants misuse drugs more than the educated ones (p=0.001). Drug misuse is significantly less in single participants compared to others (p=0.001). Income is not a determinant of drug misuse. The life-time prevalence of substance use was found as 2.8% with 0.7% life-time prevalence of marihuana as the most commonly used substance. (24). It is important to note that those studies previously conducted in Turkey are not intended for the general population. Prevalence of smoking any tobacco product was generally much higher for men than women in every GATS country in concordance with our study. Most smokers smoked manufactured cigarettes like the ones in our study. Waterpipe use and use of hand-rolled cigarettes were also found to be common in our study (25). Prevalence of smoking in men generally increased into young adulthood or middle age like in our study; but differing patterns were noted in women. Regarding our study, the significantly lower ratios of prevalence in younger ages and in women should alert us about the importance of prevention policies covering these groups. In our study it was found that uneducated participants use tobacco products significantly less compared to other groups. Highest ratios of tobacco use are reported for university graduates. GATS study results for Turkey regarding education level are in concordance with our study (26). Smoking is used as a tool for psychosocial coping with stress. Individuals whose work has greater time demands, or demands higher levels of concentration and extra work -suffer higher stress levels and are more likely to smoke (27). We found that participants with higher studies smoke more in our culture where smoking is perceived as an indicator of higher socioeconomic status, contrary to what was found in other countries such as the USA. In our study, income level seems to have no effect on tobacco use contradicting with GATS data of decreasing prevalence of tobacco use with increasing wealth (28). Generally, the pressure and support of a spouse is significant to reduce smoking (29). In our study, however, lifetime tobacco use of single participants was found to be less compared to married ones. Marriage itself is a social and economic burden for many in our culture and stress related with marriage and marital problems might increase smoking behavior. Alcohol is the oldest and most common psychoactive substance used in the world. Alcohol consumption and disorders related with alcohol use are observed more in males than females (30). Studies about prevalence of alcohol use covering general population are scarce in Turkey. In our study, life-time alcohol use was found as 28.3%, while use of alcohol in the last 12 months and in the last 1 month were 14.3% and 10.1% respectively. Ar<unk>kan et al. (7) reported that the prevalence of alcohol consumption and alcohol dependence were 14.1% and 0.9% respectively. In a study done with 1550 participants aged between 12 and 65 in <unk>stanbul, prevalence of alcohol use was determined as 33.5%. 25.6% indicated that they still use alcohol, while 7.9% told that they quitted. 12.6% of the alcohol users took alcohol more than once in 2-3 days. The most common age for the firsttime alcohol-use was found to be between 16 and 19 (31). In all of these studies, males use more alcohol than females like in our study. In the present study, 1.5% of those who used alcohol in the last 1 month take alcohol 1-3 times a week, while 1.3% consume alcohol weekly. In a study with 707 participants living in 24 districts of <unk>stanbul, 54.7% indicated that they used alcohol at least once in life-time. This rate was 73.4% for men and 35.0% for women. Using alcohol at least once every week in the last 12 months was found as 17.9% for men and 2.3% for women. For the last one month, this rate was found to be 18.4% for men and 2.3 % for women. Within the last one month period, using alcohol every day was most common in the 36-45 age group. No significant difference was found between education levels unlike our study where alcohol consumption was positively correlated with education level (32). Although it is difficult to compare our study with other studies conducted in Turkey due to the variations in method and population, most of the results are observed to coincide. According to WHO data (2011), Europe has the highest alcohol consumption in the world. Turkey is at the bottom of this list. However, data show that alcohol consumption is increasing progressively in Turkey (33). While alcoholic beverage consumption in Turkey was 867.9 million liters in 1997, it increased up to 968.9 million liters in 2006 (34). Increased beer consumption is noteworthy as it reflects increased alcohol consumption in young population (35). In our study, primary choice of alcoholic beverages was found as beer and participants aged between 15 and 24 use significantly more alcohol during last 12 months and during last 1 month which should alert us about increasing alcohol consumption in young population. The more income people have, the more educated they are and the higher their socio-professional category, the more likely they are to drink alcoholic beverages (36,37,38) In our study, alcohol consumption is reported to increase with increasing education and income levels. Association of marriage with decreased levels of alcohol consumption is rather a consistent finding in literature (39). In our study, likewise, being single was significantly related with alcohol use. Interestingly being widow, divorced or separate from spouse did not have a significant association with alcohol use. Being polygamous and living together without marriage was also correlated positively with life time alcohol use. In Wisconsin Longitudinal Study, it was reported that men' s alcohol use was tempered by being married but exacerbated by divorcing, women' s alcohol use was promoted by marriage and declined in marital dissolution (40). Influences of marital status on alcohol consumption are complex (41) and intercultural differences may exist. In our study, 10.5% of the participants misused drugs. Comparable with our results, Madianos et al. (42) reported that 6.5% of males and 14.9% of females used licit psychotropic drugs without a doctor' s order. It is noteworthy that 94.6% of drug misusers get the drugs with prescription. The first reason for this may be over-prescription of these drugs for psychiatric conditions in the primary care setting. The second reason is the prescription of especially antidepressants for non-psychiatric conditions. As 61.3% report that they took the drugs without a doctor' s recommendation, we believe that correct and responsible prescription conditions need to be further evaluated in our country. Gender, age and socioeconomic status seem to have a high impact on drug misuse. Age increases the risk of receiving prescription drug treatment (14). Women are more likely to misuse prescription drugs (43,44). Individuals from lower social class have a higher misuse of prescription drugs (45,46). These data from literature are in concordance with our results. We also found that drug misuse increases with increasing age. Elderly individuals usually begin to misuse psychotrophics because of anxiety and sleeping problems related to old age. They are also more prone to receiving prescriptions for these ailments. Other factors that precipitate drug misuse among the elderly include social isolation, being female, and having a history of mental disorders. Misuse patterns for the elderly include doctor shop-ping, receiving medicines from family and friends, and medicine hoarding (47). According to our study, females use more drugs than men. Females are generally more concerned about health issues and seek health care earlier than males In our society, especially females tend to use drugs adviced by their friends, neighbors, etc. We found that education level is significantly associated with drug misuse. Former studies have shown differences in healthcare seeking behavior with higher consultation rates among individuals in lower social classes with lower education levels (48). It is possible that uneducated individuals do not act upon information concerning health risks and are prone to use more drugs both by consulting multiple doctors and by advice from the others. Single participants were found to misuse less drugs in our study. The main reason for this is probably the younger age of single participants. According to our study, income level does not seem to be an important determinant of drug misuse. In our study, the life-time prevalence of substance use was found as 2.8% with 0.7% life-time prevalence of marihuana as the most commonly used substance. It catches attention that the drugs defined as "other" are more common than all identified drugs. Considering the manner of asking this question in our survey, this implies that individuals hesitated to name the drug they use. Most of the prevalence studies about substance use in Turkey were conducted with students. Marihuana as the most commonly used substance is a rather consistent finding. Prevalence ratios of substance use at least once in life-time range from 0.3% to 15.7% (49). This variation may result from differences in sampling and methodology. For example, benzodiazepines were included among used substances in some of the studies. Number of studies conducted with general population are scarce. There is only one study involving face-to-face interviewing. This study, conducted in 2002 covering 72 provinces, found that the prevalence of using any substance at least once in life-time was 1.35%. Substance use was found to be more common in men compared to women and in 15-24 age group compared to older ages (19). In our study, although the prevalence ratios were found to be higher in 15-24 age group and males, the figures did not reach to statistically significant levels. Epidemiological data provides evidence to suggest that married individuals are much less likely to use illicit drugs. Married individuals were found to be significantly less likely to use cocaine than unmarried individuals (50). Long-term committed relationships like marriage, provide the primary form of social support for many people in the world and in our country. In line with literature, being single was found to correlate positively with substance use in our study. In the literature, there are inconsistent findings about the impact of income level on substance use. Higher family income may be associated with substance use because of the increased ability to access substances. On the other hand, lower income may be related to substance use as a coping mechanism because of increased stress and less access to alternative activities (51). In our study, we found that participants from lower income levels used substance more than the high-income participants (p=0.001). In our country there are problems of migration from rural to urban areas, unemployment, familial problems related with economic troubles and lack of future prospects which could lead individuals from low socioeconomic status to substance use. Our study has several limitations. Firstly, because of the cross-sectional nature of this study, we were only able to examine an association and not establish a causal relationship. Secondly, it was a challenge to reach sufficient level of response although several approaches were utilized to increase the rate of response. In addition, subjects may not have responded properly because of confidentiality issues. A significant effort was put forward to minimize response bias. Multiple visits were made to the selected addresses. Only trained interviewers were engaged for this study. A helpline was formed to assist the participants and interviewers in managing the participants' questions, refusals and information requests as well as the appointments. Explanation letters were sent to the potential participants by bearing the signatures of the academic members. These letters provided additional assurance that the survey is an actual study. According to this study tobacco, alcohol, substance use and drug misuse are significant health concerns in Turkey. Use of tobacco and alcohol follows rather a consistent course. Licit drug abuse is considered as an important issue which should be studied more. Although far below the levels in western countries, the prevalence of substance use shows a slight upwards tendency and implies that protective and preventive measures against illicit substances must be implemented. Socio-demographic status is an important determinant of several aspects of tobacco, alcohol, substance use and drug misuse and should therefore be evaluated carefully. Socio-demographic findings may be an important tool to develop effective protective and preventive drug policies. --- Ethics Committee Approval: Authors declared that the research was conducted according to the principles of the World Medical Association Declaration of Helsinki "Ethical Principles for Medical Research Involving Human Subjects", (amended in October 2013). Informed Consent: Written informed consent was obtained from patients who participated in this study. --- Peer-review:

The aim of the current study is to determine prevalence ratios of tobacco, alcohol, substance use and drug misuse in a large scale sample of general population and recognize related socio-demographic factors.The study was based on a cross-sectional survey conducted with a nationally representative sample of 8045 individuals with face-to-face, in-house interviews implemented by a trained team of interviewers. Results: Life-time prevalences of tobacco, alcohol, substance use and drug misuse are 51.8%, 28.3%, 2.8% and 10.5% respectively. Age, gender, education level and marital status are important socio-demographic determinants for tobacco, alcohol use and drug misuse. Income level is an important determinant for alcohol and substance use.Tobacco, alcohol, substance use and drug misuse are significant health concerns in Turkey. Socio-demographic status is an important determinant of several aspects of tobacco, alcohol, substance use and drug misuse and should therefore be evaluated carefully to develop effective protective and preventive strategies.

Introduction HIV sero-status disclosure to sexual partners is an important component of HIV prevention and treatment efforts because it facilitates informed decision-making before sexual contact (Lan, Li, Lin, Feng, & Ji, 2016; Przybyla et al., 2013). Improvements in the health and longevity of people living with HIV (PLWH), due to advances in medical science, may have important implications for both individual and public health outcomes (Lan et al., 2016;Shacham, Small, Onen, Stamm, & Overton, 2012). Disclosing HIV status to sexual partners has been generally linked to safer drug-and sex-related practices (e.g., consistent condom use, lower frequency of drug use) (Crepaz & Marks, 2003; Li, Luo, Rogers, Lee, & Tuan, 2017; Parsons et al., 2005) and reduction in HIV transmission (Pinkerton & Galletly, 2007;Przybyla et al., 2013;Shacham et al., 2012). In recent decades, increased attention to transmission within sero-discordant couples has highlighted the potential role of disclosure as a way to encourage prevention approaches including the use of pre-exposure prophylaxis (PrEP) and HIV treatment-as-prevention (TasP) (Brooks et al., 2011;Cohen et al., 2011;Do et al., 2010). Furthermore, disclosure can also help PLWH garner better social support, reduce psychological distress, and encourage them to access comprehensive medical care and support services (Li et Despite its many potential benefits, disclosing one's HIV-positive status may have unintended negative consequences, like conflict with a partner, elevated stigma, depression, lack of social support, breach of confidentiality, rejection, and even violence, and therefore place significant burdens on PLWH (Brown,. Due to individual and societal attitudes towards HIV and the possible negative consequences, disclosure is a sensitive issue and is often difficult to negotiate for many PLWH. For HIV-infected opioiddependent individuals -a group of people who use drugs (PWUD) -who are often socially marginalized, it may involve an even more complex decision-making process (Go et al., 2016;Li et al., 2017;Parsons, VanOra, Missildine, Purcell, & Gomez, 2004). Given the complexities surrounding HIV disclosure, they may be reluctant to disclose their HIV status and avoid disclosure altogether. The decision to not disclose one's HIV status (i.e., non-disclosure) is multifaceted and influenced by structural, relational, and personal considerations. In the broader literature, several factors have been associated with non-disclosure (Adeniyi et Quinn, 2013). Despite significant research in this area, prior studies have not systematically investigated non-disclosure practices, nor have they explored theoretically informed correlates among HIV-infected opioid-dependent patients within drug treatment settings. An evidence-based understanding of the scope of non-disclosure can inform future interventions specifically tailored for this population. In this paper, we therefore sought to explore the factors associated with non-disclosure of HIV status to sexual partners among HIV-infected opioid-dependent patients. --- Methods --- Study setting and procedures The data reported here are derived from the baseline assessment of the Holistic Health for HIV (3H+) project, a randomized controlled trial to improve HIV risk reduction and medication adherence among high-risk HIV-infected opioid-depenent patients. The study design and procedures of the parent study has been previously described ( & Copenhaver, 2015). Briefly, participants were recruited from community-based addiction treatment programs and HIV clinical care settings within the greater New Haven, Connecticut. Participants were recruited through clinic-based advertisements and flyers, word-of-mouth, and direct referral from counselors. Screening was conducted by trained research assistants either by phone or private room. Individuals who met inclusion criteria and expressed interest in participating provided informed written consent and were administered a baseline assessment. All participants were reimbursed for the time and effort needed to participate in the survey. The study protocol was approved by the Institutional Review Boards at the University of Connecticut and Yale University, and received board approval from APT Foundation. Clinical trial registration was completed at www.ClinicalTrials.gov (NCT01741311). --- Participants Between September 2012 and January 2018, 133 HIV-infected, opioid-dependent individuals were recruited. Additional inclusion criteria included: 1) being 18 years or older; 2) reporting drug-or sex-related risk behavior (past 6 months); 3) being able to understand, speak, and read English; and 4) not actively suicidal, homicidal, or psychotic. --- Measures Participants were assessed using an audio computer-assisted self-interview (ACASI). Measures included socio-demographic characteristics, health status, HIV-related stigma, drug-and sex-related risk behaviors, and sero-status non-disclosure. Key measures are described below. The dependent variable was non-disclosure of HIV status, which was defined as having any sex without disclosure of HIV-positive status to the partners in the past six months. Serostatus non-disclosure to partners was measured as "yes"/"no" by asking, "In the past six months, did you have sex with anyone that you did not tell your HIV status sometime before you had sex?" Health status variables including length of time since HIV diagnosis, whether the participant was currently taking antiretroviral therapy (ART), and baseline viral load (VL) and CD4 count were abstracted from their medical record. Adherence to ART in the past month was assessed using a empirically validated, self-report visual analog scale (VAS) approach (Giordano, Guzman, Clark, Charlebois, & Bangsberg, 2004). Using standardized cut-off, adherence of 95% or greater was considered optimal adherence (Paterson et al., 2000). Viral suppression was defined as clinic-recorded HIV-1 RNA test value <unk>200 copies/mL and high CD4 count (<unk>500 cells/mL) (Bowen et HIV-related stigma was measured using a validated 24-item HIV stigma scale (Earnshaw, Smith, Chaudoir, Amico, & Copenhaver, 2013) with items rated on 5-point Likert-type scales with higher scores indicating greater stigma. Items were averaged to create a composite score (<unk>=0.93). The HIV risk assessment, adapted from NIDA's Risk Behavior Assessment (Dowling-Guyer et al., 1994) was used to measure several aspects of HIV risk behaviors in the past 30 days, including a measurement of "any" high risk behavior (sexual or drug-related) as well as measurements of event-level (i.e., partner-by-partner) behaviors. --- Data analyses Covariates included in the analysis were based on prior research as well as findings from other studies conducted within drug treatment settings. We computed descriptive statistics, including frequencies and percentages for categorical variables, and means and standard deviations for continuous variables. We conducted bivariate analyses for significant associations with the dependent variable (i.e., non-disclosure of HIV status). Additionally, we included the interaction term (variables from the main effects model), one at a time, in the model containing all the main effects to determine the interactive effect on nondisclosure. We then conducted multivariable logistic regression analyses on bivariate associations found to be significant at p<unk>0.10. Stepwise forward entry and backward elimination methods both showed the same results in examining the independent correlates (p<unk>0.05) expressed as adjusted odds ratios (aOR) and their 95% confidence intervals. Model fit was assessed using a Hosmer and Lemeshow Test (Hosmer, Hosmer, Le Cessie, & Lemeshow, 1997). Collinearity between variables was assessed using the variance inflation factor (VIF). Estimates were evaluated for statistical significance based on p <unk> 0.05. All analyses were conducted using SPSS version 23 (IBM Corp., 2015). --- Results The baseline characteristics are described in Table 1. The mean age of participants was 49.3 (<unk>8.3), and 41.4% were living with family/friends. The mean duration of HIV diagnosis was 14.1 (<unk>9.6) years and were maintained on a stable methadone dose (Mean: 64.5 mg). Of 121 (91.0%) individuals who were taking ART, 57.9% had achieved optimal adherence and 80.4% had a high CD4 count. HIV-related stigma scores ranged from 1.0 to 4.4, with a mean score of 2.0 (<unk>0.7). Self-reported HIV risk behaviors were highly prevalent among study samples. Almost half of the participants (46.6%) reported to have injected illicit drugs in the past 30 days. Of those, 58.1% reported having shared injection equipment. Similarly, 21.1% of the participants reported having sex with more than one sexual partner and only 14.3% reported to have always used condoms with their sexual partners in the past 30 days (Figure 1). Nearly a third (23%) of the participants reported not disclosing their HIV status with a sexual partner, whereas, 39.8% of them reported to know their sexual partner's HIV status. Table 1 describes the bivariate comparisons of those reporting nondisclosure. Of note, there was significant difference based on participants' knowledge of sexual partner's HIV status (p=0.002) and whether they had multiple sex partner (p<unk>0.001), which is portrayed in Figure 1. Other factors in bivariate analysis associated with non-disclosure were being older (p=0.005) and heterosexual (p=0.013). Table 2 shows the independent correlates associated with HIV-positive status non-disclosure. Participants who were virally suppressed were less likely to withhold disclosing their HIV status to sexual partners (aOR=0.189, p=0.041). Whereas, participants with a higher degree of perceived HIV-related stigma (aOR=2.366, p=0.032) and having multiple sex partners (aOR=5.868, p=0.040) were significantly more likely to not disclose their HIV status. Furthermore, we also found a significant interaction between stigma and living with family/ friends on non-disclosure (aOR=7.792, p=0.020). --- Discussion Several important findings were gleaned with regard to non-disclosure practices among high-risk HIV-infected opioid-dependent individuals, and these may have significant implications for future HIV prevention efforts in the clinical settings. A substantial proportion of participants in our study reported not disclosing their HIV status to any sexual partner. This finding underscored the complexities and challenges surrounding HIV serostatus disclosure among high-risk opioid-dependent patients in drug treatment settings. The higher rate of non-disclosure in our sample may be partially explained by a longstanding experience with drug use (e.g., mean duration: 24.7 years). One potential explanation is that those with longstanding drug use may have been subjected to discrimination before HIV treatment could potentially render PLWH non-infectious to others and now continue to perceive that the risk of sero-status disclosure outweighs the potential benefits (Li et al., 2017;Valle & Levy, 2009). Furthermore, the dual veils of stigma derived from both addiction and HIV creates synergistic jeopardy that reduces their willingness to disclose their status. Findings from this study support previous studies demonstrating that the HIV sero-status disclosure process is difficult and complex, especially for high-risk HIV-infected opioid-dependent individuals in drug treatment. We found that self-reported HIV risk behaviors (both drug-and sex-related) were highly prevalent among this sample, which is consistent with findings from prior studies with similar risks ( 2016). A significant proportion of participants reported sharing of injection equipment, having multiple sex partners, and inconsistent condom use during sexual intercourse. This is especially concerning given that they are continuing to engage in risky behaviors with most of them not disclosing their HIV status, and thus may be transmitting HIV to sero-discordant partner. These findings highlight the importance of HIV status disclosure and the need for additional evidence-based HIV prevention strategies. As such, the delivery of integrated PrEP and ART may be the most pragmatic HIV prevention strategy among HIVserodiscordant couples (Brooks et al., 2011;Cohen et al., 2011;Do et al., 2010). In this study, the odds of non-disclosure of HIV sero-status was lower among individuals who were virally suppressed. Although being virally suppressed has been shown to prevent sexual transmission of HIV (Cohen et al., 2016), it is encouraging that these individuals report a willingness to disclose their sero-status. Feelings of responsibility and the desire to protect one's sexual partners from potential HIV infection may have enhanced motivation to disclose their HIV status, and thereby overriding concerns about negative consequences (Parsons et al., 2004). Furthermore, we found that greater HIV-related stigma was associated with non-disclosure of HIV status to sexual partners, which is consistent with the literature on stigma and disclosure (French, Greeff, Watson, & Doak, 2015;Ojikutu et al., 2016;Ostrom, Serovich, Lim, & Mason, 2006;Overstreet et al., 2013;Przybyla et al., 2013). It is possible that negative beliefs around one's HIV status and the associated damaging consequences may reduce the likelihood that HIV status is disclosed in a sexual context (Overstreet et al., 2013). HIV-related stigma remains a considerable barrier to ending the pandemic, necessitating effective strategies that directly provide access to information, community support, and advocacy. One strategy that has been rapidly gathering momentum in the recent years is the Undetectable = Untransmittable (U=U) campaign (Prevention Access Campaign, 2018; The Lancet HIV, 2017). It synthesizes scientific data from the TasP literature (Günthard, Saag, Benson, & et al., 2016;Volberding, 2017) and places PLWH as being responsible for HIV transmission by caring enough to optimally adhere to HIV medications, rendering themselves unable to transmit HIV irrespective of ongoing sexual risk. Such strategies remove the absolute need to disclose their HIV status and markedly reduces the consequences to PLWH through the disclosure process. The pro-social U=U campaign empowers PLWH so that by protecting themselves, they protect others even when HIV disclosure is not addressed. Additionally, participants who reported having multiple sex partners were more likely to not disclose their HIV status. With multiple sex partners, the complex dynamics of relationship (Mbonye, Siu, Kiwanuka, & Seeley, 2016) may increase fears of rejection, and thus, lead to non-disclosure. Our findings further demonstrated that there is a complex interplay between HIV-related sigma, family/friend support, and nondisclosure. As an extension of prior findings, our results showed an interactive effect of stigma and living with family or friends on individuals' non-disclosure practices. That is, those living with family/friends were more likely to report not disclosing their HIV status when faced with a higher degree of perceived stigma. Social support is an important psychological factor that can promote HIV status disclosure (Jorjoran Shushtari, Sajjadi, Forouzan, Salimi, & Dejman, 2014; Lee, Yamazaki, Harris, Harper, & Ellen, 2015), but situational variables such as HIV-related stigma may override the positive influence of social support in decisions about disclosure. This buffering incluence can help explain how disclosure practices among PLWH changes in the presence of social support and how stigma impacts long-term social support and PLWH's willingness and/or patterns of disclosure. From a prevention standpoint, this highlights the importance of precisely targeting the impact of perceived stigma and increasing social support, while developing interventions to improve disclosure practices among HIV-infected opioiddepenent patients. The findings from this study are not without limitations. First, the sample was drawn from individuals enrolled in MMT, potentially limiting generalizability of findings to HIVinfected opioid-dependents individuals not enrolled in the methadone program. Second, we utilized a dichotomous measure of HIV sero-status non-disclosure status obtained from a single-item question. Furthermore, we did not assess non-disclsoure status at a partner specific level. We are therefore unable to fully capture the complexity and circumstances of non-disclosure to sexual partners. Third, much of the data in this study came from selfreport and is thus subject to both social desirability and recall biases. Fourth, the data in this study were cross-sectional in nature, thus limiting our ability to infer causation from the associations found. Fifth, this study included relatively small sample size which may have limited our ability to detect differences with smaller effect size. Last, the current study was focused on non-disclosure to sexual partners, potentially limiting our ability to assess HIV transmission risk through sharing of injection equipment among injecting partners. Despite these limitations, the findings from this study significantly contribute to the literature to date, in which there is little research investigating non-disclosure patterns among this underserved population. --- Conclusions HIV sero-status disclosure to sexual partners is an important component of HIV prevention and treatment efforts (Lan et al., 2016;Przybyla et al., 2013). Findings from this study underscore the complexities surrounding HIV sero-status nondisclosure/disclosure among high-risk HIV-infected opioid-dependent patients, as highlighted by the relatively high rates of HIV non-disclosure. Our findings are unique given the relative dearth of research on HIV non-disclosure practices and associated factors among this risk group. In the contemporary era of TasP, interventions that reduce the complexity of disclosure by reducing risks to others, like U=U, are crucial for providing the foundation for allowing the disclosure process to evolve over time. Given high prevalence of HIV status non-disclosure (23%) in this highrisk population, future interventions should consider the specific needs of the population (e.g., harm reduction, overcoming stigma, improving social support) that better address the impact of perceived stigma and HIV disclosure as it relates to risk behaviors among opioiddependent enrolled in treatment. --- Author Manuscript

HIV sero-status disclosure among people living with HIV (PLWH) is an important component of preventing HIV transmission to sexual partners. Due to various social, structural, and behavioral challenges, however, many HIV-infected opioid-dependent patients do not disclose their HIV status to all sexual partners. In this analysis, we therefore examined non-disclosure practices and correlates of non-disclosure among high-risk HIV-infected opioid-dependent individuals. HIVinfected opioid-dependent individuals who reported HIV-risk behaviors were enrolled (N=133) and assessed for HIV disclosure, risk behaviors, health status, antiretroviral therapy (ART) adherence, HIV stigma, social support and other characteristics. Multivariable logistic regression was used to identify significant correlates of non-disclosure. Overall, 23% reported not disclosing their HIV status to sexual partners, who also had high levels of HIV risk: sharing of injection equipment (70.5%) and inconsistent condom use (93.5%). Independent correlates of HIV nondisclosure included: being virally suppressed (aOR=0.19, p=0.04), high HIV-related stigma (aOR=2.37, p=0.03), and having multiple sex partners (aOR=5.87, p=0.04). Furthermore, a significant interaction between HIVrelated stigma and living with family/friends suggests that those living with family/friends were more likely to report not disclosing their HIV status when

of family responsibilities shifts. Her childcare load could lessen as children grow older, while a transition to grandmotherhood may increase her caregiving duties later in life. This is especially true in settings where women tend to become grandmothers at a younger age, during a time when they may still be actively participating in the labor force. In a context where cultural expectations of taking care of grandchildren is high and coresidence is common, she could once again be subject to a high level of work and family demands. This study is set in Cebu, the Philippines, a province where the pace of fertility decline has been slow. Total fertility rate still remains well above 3.0 and the mean age at birth is 23.1 (Population Reference Bureau 2014). Women thus often become grandmothers at a much younger age compared to most women in low fertility settings. A 2009 report indicated that 1 in 5 older adult Filipinos take care of grandchildren, regardless of coresidence status (Cruz et al. 2009). The percentage would be even higher if younger women were included in the estimate. High rates of international migration and the increasing trend of overseas Filipino workers translates into additional demand for grandmothers to participate in childrearing (Parre<unk>as 2000). While we know Filipino grandmothers take a major role in childrearing (Cruz et al. 2009), we have little knowledge about the extent to which they experience a double burden by also being engaged in work and household activities. In this paper, we take a step toward capturing the diversity in women's work and family responsibilities using rich time use data from a cohort study of mothers and children: the Cebu Longitudinal Health and Nutrition Survey (CLHNS) conducted in metro Cebu, the Philippines. Using latent class analysis (LCA) and latent transition analysis (LTA), we first profile time use allocation and classify women into distinct groups, followed by analysis of transition in time use group membership. Given data that spanned a period of eighteen years (1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012), during which these women were transitioning from midlife to old age, we are able to examine the association between changes in household composition and time use patterns, paying particular attention to the effect of presence or addition of grandchildren in the household. We are motivated by the following questions: 1) To what extent are women in Cebu, the Philippines shouldering the "double burden" of work and family tasks, as evidenced by their time use profiles? 2) As women transition from midlife to old age, how do their alignment of work and family responsibilities change? 3) Does the presence or addition of grandchildren in the household translate into a difference or shift in work and family responsibilities? Our paper makes a significant addition to recent literature on time use at older ages, which are often cross-sectional in nature (Gauthier andSmeeding 2003, 2010;McKenna, Broome and Liddle 2007). Our longitudinal study sheds light on the changes in women's time use patterns across the life course. The answers to the above questions allow us to gain a concrete understanding of competing demands that many mothers and grandmothers are facing as they age. Although our study does not directly measure the extent to which work and family responsibilities are in conflict with each other, by accounting for time allocation devoted to work, caregiving, household chores, and personal time respectively, our study provides a necessary step in documenting the multiple burdens women commonly face in a developing economy, where the boundaries of work and family are often not as clear cut as those in developed countries. --- Work and Family Responsibilities for Women over the Life Course Despite dramatic increases in female labor force participation in recent decades, particularly for working mothers, women around the world shoulder greater responsibilities for caregiving and housework in general, giving rise to the so-called "double burden" (Hochschild and Machung 2012;Mattingly and Bianchi 2003;Milkie, Raley and Bianchi 2009;Sayer et al. 2009). Existing literature on maternal employment and childcare abounds with discussions on role incompatibility and conflict that is caused by the dual demands of work and caregiving (Chen, Short and Entwisle 2000;Desai and Jain 1994;Oppong 1983;Tiefenthaler 1997). Does a woman face similar challenges later in life when she becomes a grandmother, in a context where caregiving for grandchildren is normative? For example, a study in China showed the average level of a grandmother's care to be equivalent to that of a mother in caring for preschool children (aged 0-6) except for children under the age of 1 (Chen, Liu and Mair 2011). In the Philippines, it is often common for parents to live with their adult children and grandchildren and an overwhelming majority of the grandparents, grandmothers in particular, provide regular care for their grandchildren (Agree, Biddlecom and Valente 2005). Research on grandparents' caregiving for grandchildren has extensively investigated the implications of grandparental caregiving for grandparents' well-being (Chen and Liu 2012;Chen et al. 2015;Hughes et al. 2007;Ku et al. 2013). One of the primary concerns of the grandparental caregiving literature is whether these grandparents are overburdened with the physical and psychological stress of work and caregiving activities (Baker and Silverstein 2008;Leopold and Skopek 2014;Wang and Marcotte 2007). For example, recent studies in the U.S. suggest that about 36% of the grandparents who are caring for their grandchildren were actively employed in 2010 (Scommegna and Mossaad 2011). Due to the zero-sum nature of time, the potential for work, caregiving, housework and time for self care activities to crowd each other out is high. However, researchers are often limited to the inclusion of work status, household structure and intensity of caregiving in the analysis (Baker and Silverstein 2008;Chen and Liu 2012;Wang and Marcotte 2007). A direct examination of actual time allocation for work and caregiving is lacking in the literature. In addition, in a developing country setting, to characterize work status is often not as straightforward as that in industrial societies. Work boundaries are often fluid, meaning that women may not engage in work in the formal sector but rather they are involved with a wide range of informal income generating activities, such as working in a family store, making and selling handicrafts, or other service activities (Desai and Jain 1994;Donahoe 1999;Lloyd 1991;Short et al. 2002). Some work is more compatible with childcare than others due to time flexibility. For example, agricultural fieldwork or handicraft work can be more easily combined with childcare, while a nine-to-five wage job at a factory makes it challenging for one to assume primary caregiving responsibility. Thus, adjustment in work activities and other chores could be gradual and subtle, compared with the settings where market work is formal and therefore exiting and entrance into the labor market often involves drastic reconfiguration of other activities. For example, a study using the China Health and Nutrition Survey demonstrates that birth has only a modest effect on mothers' wage employment but the shifts in work patterns are much more salient when using a typology that distinguishes different combinations of wage work, work in household businesses, and agricultural field work (Entwisle and Chen 2002). There is also the possibility that no apparent adjustment is made in either work or caregiving when role transitions occur. Analyses using time diary data from nationally representative studies in U.S. show that mothers' time with children has remained stable in the face of dramatic increases in female labor force participation in the last few decades (Bianchi 2000;Sayer, Bianchi and Robinson 2004;Sayer 2005). Instead, what was reduced was women's time for household chores and leisure. While this pattern has been documented for the mothers, it is not known whether the same is true with grandmothers in developing countries, who often assume primary care responsibilities and are often still relatively young in age (Entwisle and Chen 2002). Recent time use studies show a decrease in paid work and physically demanding leisure activities, together with an increase in personal time associated with ageing (Gauthier andSmeeding 2003, 2010;McKenna, Broome and Liddle 2007). In order to better understand different demands in a woman's life, it is imperative that we look beyond work and caregiving and include other activities that impact on time allocation, including household chores and leisure time. We take a life course perspective by positing that women's time use patterns are likely to shift with changing life circumstances, particularly as family roles evolve. Following the principle of life-span development (see Elder, Johnson and Crosnoe 2003), we expect both continuity and change in time use patterns as one moves from younger to older adulthood. For example, as children grow up and transition into adulthood, there would be an easing of caregiving duties for some woman and a subsequent increase in leisure time. For others, it could be accompanied by an increase in market work. As women become grandmothers, especially with grandchildren living in the household, they could get involved in childcare and scale back work activities. However, for some grandmothers, if they are engaged in home based work, then it may be relatively easy for them to combine childcare with work without making much adjustment in work activities. Based on a close examination of women's time use allocation, the current study expects to develop "profiles" of work/family combinations as well as transitions in profiles, as women transition from midlife to old age along with changes in roles and household composition. --- Data The CLHNS (Cebu Longitudinal Health and Nutrition Survey), established through collaboration between researchers at the Carolina Population Center at UNC Chapel Hill, and the Office of Population Studies Foundation (OPS) at the University of San Carlos in Cebu, follows a cohort of mothers and an index child born in 1983-84. Using a single stage cluster sampling procedure, 17 urban and 16 rural barangays (local administrative units) were randomly selected from the 255 barangays in Metropolitan Cebu. The 33 barangays, representing about 28,000 households, were surveyed to locate all pregnant women. Women were first interviewed in 1983 and were followed up in 1991in, 1994in, 1998in, 2002in, 2005in, 2007in, and 2012in. In 1983, the women ranged in age from 14 to 45, with a mean age of 26. These women were recruited into the survey by virtue of having given birth within a one-year period (May 1983-April 1984, n=3,237), therefore making the sample selective of high fertility and lower socioeconomic status (Adair et al. 1997;Gultiano 1999). There are no comparable large samples of women in developing countries followed from the reproduction years to older adulthood. For detailed information on the sample design, see http:// www.cpc.unc.edu/projects/cebu/about. For this paper, because our focus is daily time use activities, and since time use data has only been collected in 1994, 1998, 2002, 2005, and 2012, we start our analysis with a sample of women in 1994. With low missing rates ranging from 0.17% to 2.05% on the dependent and/or independent variables, we did not impute any missing data and used listwise deletion (see recommendation by Lynch, Brown and Mustillo 2016). The analysis starts with a sample of women in 1994 (n=2,279), with a mean age of 38. All together, 1,771 respondents were observed across the five waves of data. Additional analyses suggests the pattern in the dependent variable is not predictive of loss to follow up at the following wave of data. The final sample size for each follow up wave is 1,989, 2,102, 2,018, and 1,815 (see Table 1). --- Daily Time Use Activities in CLHNS The CLHNS collects data using 24-hr activity diaries (reported for a typical weekday). We can calculate the amount of time usually spent daily on activities such as food preparation, housekeeping, caregiving, working at home, working away from home, leisure, and sleep. We collapse daily time use into five broad categories: --- 1) working outside home for pay (including traveling time); 2) working at home for pay, examples include getting wares ready for selling, opening/closing store, watching store located at home, doing other people's laundry; 3) household chores, including activities such as food preparation, marketing, cooking, washing dishes, housekeeping, gathering firewood, doing household repairs, making beds, sewing, mending, tending animals/garden; --- 4) caregiving, including activities such child/parent care (bathing/cleaning, dressing up, putting to sleep/waking, bringing to and from school, helping with homework, taking for a walk, feeding child); --- 5) personal time, including personal hygiene, recreation (napping, reading, listening to the radio, watching TV/movie), schooling. Mean hours of daily time use activities for women in the CLHNS sample change over the 18-year time span. For example, the mean daily hours of working outside the home decreased from 3.47 to 2.85 hours from 1994 to 2012, while hours of working at home slightly increased (see Table 1). We observe decreasing average time in caregiving and household chores, but a considerable increase in personal time over the years (from 4.66 in 1994 to 7.25 hours in 2012). This is consistent with other time use studies (Gauthier andSmeeding 2003, 2010;McKenna, Broome and Liddle 2007). It is important to note that there exists tremendous variation and skewness in the distribution of these time use variables. Thus, it is not ideal to treat these variables as continuous. After carefully examining the distribution of these time use activities, we collapse the daily time use in each of these five groups into three categories: zero-low intensity, moderate intensity and high intensity (identified as response 1, 2, 3 in later latent class analysis). While the loss of detail is inevitable in transforming the continuous variables to ordinal variables, we believe that it is in line with our substantive interests in capturing the level of burden that women may have in these different dimensions of activities. Table 1 provides the frequency distribution of the newly created categorical variables in addition to the mean and standard deviation of the original continuous variables. Our choices of the cutoff points are substantively meaningful and are informed by the original distribution of the variables. We use overall means as reference points in defining the moderate categories and use the percentile distribution as our guide in picking up the cutoff points. The results are insensitive to small adjustments (e.g. using a five or six hour as the cutoff). For example, for work to be considered "high intensity", the work hours are longer than eight hours a day, the standard work day hours. If one does not work at all, then they are grouped into the "zero-low intensity category". To qualify as "high intensity" caregiving, caregiving hours are longer than an hour a day. As for personal time and household chores, longer than six hours a day is classified as "high intensity". If one does less than two hours of housework, we classify that as "low intensity". Table 1 shows that many women were heavily engaged in work. For example, about 23.5% of the women worked more than eight hours a day outside the home in 1994, and the proportion of women in high intensity outside work remained high in 2012 (19.3%). Fewer women were engaged in high intensity work at home, but over the years, a steady proportion of them worked at home with moderate intensity (<unk>=8 hours a day, from 16.5% to 19.5% over the years). Substantial changes are observed in the time spent in caregiving over the years. About 40.1% of the women provided care more than one hour a day in 1994, while such a percentage decreased to 20.9% in 2012. There was an even larger decrease in the proportion of women who were involved in moderate intensity caregiving (<unk>=1 hours a day). Women's involvement in moderate intensity housework was steady over the years. More than half of the sample performed household chores with moderate amount of hours (between 2 to 6 hours a day) across the years. A considerable number of women (from 20.6% to 34.4%) spent more than six daily hours in household chores. Finally, it seems women were spending more time on leisure on average over time, with more than half of the women spending more than six hours daily in personal time in 2012. --- Latent Class Analysis and Latent Transition Analysis The above description of these time use variables depicts the diverse activities in which Cebu women are engaged on a daily basis over a span of eighteen years. Although they are helpful, the statistics do not readily reveal how women are juggling among different activities everyday, that is, the combination of activities in which women typically engage. For example, for those women who spend long hours working outside home, do they also engage in caregiving activities? For those who spend more than six hours a day on household chores, are they also involved in any market work? A closer examination of the different combination of time use activities is warranted to understand how women align work and family responsibilities and how they adjust them over the life course. We use latent class analysis (LCA) to assess the time use patterns of these women in a systematic way. We posit that women's observed time use in different categories is a multidimensional manifestation of women's overall work and family responsibilities, an underlying latent class structure. The LCA approach offers an advantage over the standard cluster analytical technique in that it makes no assumption about the distribution of the indicators (Vermunt and Magidson 2002). In our analysis, we make use of the above described three-category classification of the five time use variables as our categorical indicators (or items), i.e. zero/low to high intensity (ranging from response 1 to 3) work outside and at home, caregiving, household chores, and personal time. We use SAS PROC LCA for the analysis. Parameters are estimated by maximum likelihood using the EM logarithm (Lanza et al. 2007(Lanza et al., 2015)). The basic LCA model estimates two parameters, latent class membership probabilities (<unk>) and item-response probabilities (<unk>). The latent class membership probabilities describe the distributions of classes of the latent construct. All the classes are mutually exclusive and exhaustive and add up to a probability of 1 (McCutcheon 1987). We start by computing a latent class model with only a single latent class (no relations between the observed indicator variables) and then add one class after the other. The "best" model is selected based on the goodness of fit measures, including the likelihood ratio test statistics, AIC (Akaike Information Criterion), and BIC (Bayesian Information Criterion), with smaller AIC and BIC values suggest a better fit of the model. Because the distribution of the likelihood ratio test is unknown, we use it as a rough guideline and do not conduct a formal chi square test (Lanza et al. 2007). The item-response probabilities are an estimation of the association between each observed indicator variable and each latent class. They vary from 0 to 1, with a number closer to 1 suggesting a high to perfect association between the variable and the latent class, and a number closer to 0 suggesting no or weak association between them. These probabilities allow researchers to identify the defining characteristics of each class so that they can label each class substantively and therefore distinguish one class from the other. The basic latent class analysis can be extended further to include estimation of <unk> coefficients when covariates are added to predict class membership. The coefficients are interpreted the same as those in multinomial logistic regression, predicting the likelihood of being in one class versus a reference class. In our analysis, we use household composition as our key covariates, with particular attention paid to the presence of grandchildren. While latent class analysis (LCA) effectively identifies latent class membership in time use for women in our sample at each wave, it does not readily capture the changes in time use patterns over time. Thus, in our next step of the analysis, we introduce a dynamic element by applying Latent Transition Analysis (LTA), which allows us to assess the probabilities that latent class memberships in time use change over time (Lanza and Collins 2008;Velicer, Martin and Collins 1996). We first use LTA to develop a cross classification table that shows how membership in each class changes or remains stable over time. We then follow up with multivariate analysis that explores transition in and out of certain classes (logistic regression) as a function of change in household composition, particularly, the addition of grandchildren to the household. --- Results from Basic Latent Class Analyses We present the goodness of fit statistics in Table 3 shows the item-response probabilities associated with each class in a six-class model in 1994. Each class is distinguishable from another based on the distribution of these probabilities and thus allows us to come up a clear substantive label for each class. In Table 3, we highlight the <unk> that centrally defines the class in bold which is also what informs us to create the labels. In the summary below, we describe each class in detail and make note of the commonality and differences across these different classes. The first class we identify is labeled as "high intensity worker away from home." They have a 1.0 probability of being high intensity (response 3) in working outside home. Meanwhile, women belonging in this group are likely to engage in moderate household chores (a probability of 0.617 in response 2, much larger than that in the other two response categories), low to moderate caregiving activities (probability of 0.427 in response 2 and 0.378 in response 1), and low level of personal time (probability of 0.536 in response 1). The second class is labeled as "high intensity worker based at home." They have a 0.999 probability of being in high intensity (response 3) for working at home. They are also likely to engage in moderate household chores (probability of 0.739 in response 2), moderate caregiving activities (probability of 0.544 in response 2), and low level of personal time (0.794 in response 1). Both high intensity workers away from and based at home are obviously characterized by their high level of involvement in work (more than eight hours a day). What is also striking is that they maintain a moderate level of family responsibilities, both in terms of housework and caregiving. Clearly these two classes of women are shouldering a heavy "double burden". We identify the third class as "moderate intensity worker away from home." They have a 0.999 probability of being in response category 2 for working outside home. They are likely to engage in moderate household chores (probability of 0.641 in response 2), moderate caregiving activities (probability of 0.467 in response 2), and moderate level of personal time (probability 0.465 in response 2). The fourth class is identified as "moderate intensity worker based at home." They have a 0.999 probability of being in response category 2 for working at home. They are likely to engage in moderate household chores (probability of 0.653 in response 2), moderate to high caregiving activities (probability of 0.417 in response 2, 0.439 in response 3), and low to moderate level of personal time (probability of 0.461 in response 2, 0.395 in response 1). Classes 3 and 4 are distinguished from 1 and 2 primarily by the intensity of their work load. Women in Classes 3 and 4 engage in a moderate level of work activities while those in 1 and 2 engage in a high level of work activities. There is some difference with respect to intensity of family responsibilities of caregiving and household chores. Those in Class 1 and 2 are unlikely to be in the high intensity group when it comes to caregiving and household chores. Those in Classes 3 and 4 are more likely to be in the high intensity group with caregiving and household chores, and equally likely to be in the moderate intensity group. All of this means that women in all these four classes are subject to a heavy double burden, but the high intensity workers are more likely to shoulder a heavier burden in the work arena while the moderate intensity workers have a heavier demand from family responsibilities. In terms of personal time, women in Classes 1 and 2, who are high intensity workers, suffer from the highest deficit, reflected by a very low probability of having high intensity personal time. In comparison, the moderate workers in Classes 3 and 4 are somewhat more likely to be spending a moderate or high level of their time with personal pursuits. The next two classes (5 and 6) are labeled as "high intensity caregiver" and "homemaker". What distinguishes these women from the above four classes is the high load of family responsibilities and the lack of work responsibilities. Those identified as "high intensity caregiver" have very high probabilities of being in the zero to low intensity category for working for pay outside the home or at home (0.846 and 0.998 respectively). For Class 6, labelled as "homemaker", the probabilities are 1.000 and 0.989 respectively. These two classes are strikingly similar in their probabilities of having a high housework load. Women in both groups are very likely to be in the high intensity category for household chores (item probability of 0.665 and 0.655 respectively). But, the two classes are also distinct in two important ways. "High intensity caregivers" have a 0.970 probability of being in the category of high intensity (response 3) for caregiving. For "homemakers", the same item probability is only 0.361. Women in the "homemaker" group are the only ones that are likely to enjoy a high level of personal time (item probability of 0.793 in response 3). "High intensity caregivers" are likely to enjoy a moderate amount of personal time (item probability of 0.655 in response 2). The patterns of item response responsibilities are quite similar in 1994, 1998 and 2002 (results not shown). The amount of caregiving decreases for women in every class except for Class 5 ("high intensity caregiver"). Given that the sample is getting older at later waves of data collection, this may be reflecting a decrease in mothering activities with increasing age. We run the six-class latent class model for year 2005 and 2012, the same as what we did for the earlier three waves. Based on the results of item response probabilities, we find an exact match between five of the classes identified in earlier waves. However, we did not find a class that matches with the description of "moderate intensity worker based at home" (Class 4), in terms of item response probabilities. Instead, in 2005, the sixth class has the item response probability distribution resembling the class of "high intensity worker away from home". For 2012, the sixth class has item response probabilities similar to the class of "homemaker." In order to be consistent with analyses in earlier waves, we did not create a new class for 2005 and 2012, rather we collapsed the sixth class with the already identified class in the previous waves which its item response probabilities resemble. We believe that a five-class latent class structure is sufficient to describe time use patterns in 2005 and 2012. The above estimates on item response probabilities (<unk>) help us to distinguish six classes of women apart from each other (five classes in 2005 and 2012), with the first four classes subject to heavy "double burden" from the combinations of work and family responsibilities, while the last two are mostly responsible for household duties only. The second estimate from the basic LCA model, class membership probabilities (<unk>), identifies the probability distribution of class membership in the six classes (adding up to 100%). This measure is extremely useful in helping us to document how prevalent "double burden" is among the CLHNS women and whether there exists any trend over time (see Figure 1). About a quarter of women were identified as "high intensity workers away from home" in 1994 (modal category) and the proportion remained steady over time but dropped to 19.1% in 2012. There is a slight increase in "high intensity worker based at home" over time. This suggests not much "scaling back" for women in the high intensity worker categories over time. At the same time, we observe a steady decline in the moderate intensity worker categories. For those who are considered "moderate intensity workers away from home", the proportion dropped more than half over eighteen years. As for "moderate intensity workers based at home", this category completely disappeared in 2005 and 2012. Accompanying the decline in moderate intensity workers, we see a dramatic rise in the "homemaker" class, from 22.4% in 1994 to 53.3% in 2012. The "high intensity caregiver" class fluctuates over time, with decreases early on and then with increases. Because caregiving is broadly defined, this could reflect a declining responsibility for caring for one's children and parents and increasing responsibilities for grandchildren care. What could drive the fluctuation in time use class membership? We hypothesize that they correspond to life course transitions. In the next step of the analysis, we explore household composition as a key predictor of time use class membership, and also include other covariates such as age, education, rural/urban residence. We measure household composition by 1) presence of a spouse; 2) presence of a grandchild; 3) presence of a child; 4) presence of a parent; and 5) presence of an other person. Table 4 shows the descriptive characteristics for these variables. What stands out in the changes in household composition over time is the dramatic increase in grandchildren's presence. While only 8.3% of the women had coresident grandchildren in the household in 1994, by 2012 the majority of these women were living with grandchildren. --- Latent Classes Analyses with Covariates The results of the multinomial logistic regressions predicting class membership are voluminous, because we have five models for each wave of the data, with each model containing a contrast between a class and the base category ("homemaker"). In order to facilitate our interpretation, and because the presence of grandchildren is of key interest, we calculate predicted probabilities of class membership as a function of the presence of grandchildren based on the estimated models, with all other variables held at their means (for continuous variables) and modes (for categorical variables), and present them in Figure 2a-d (results for the 1994 wave is not presented because only a few households had grandchildren and the effect of the presence of grandchildren was not statistically significant). We present the full results of the multinomial logit models in 2002 as an example in Table 5. We illustrate this year since it is the wave when grandchildren presence in the household became more prevalent (full results for other years are available upon request). Figures 2a-d clearly illustrate a difference in the high intensive work type (Class 1 and 2, both high intensity workers based at and away from home), that is, women with coresident grandchildren are less likely to engage in long hours of work, although the difference is significant in some years but not the other. For example, in 2002, the probability of being a "high intensity worker based at home" is 0.131 for those without grandchildren present in the household, as compared with 0.088 for those with grandchildren present in the household. The effect of grandchild presence is less consistent for the moderate intensity work type (Class 3 and 4). Having coresident grandchildren in the household slightly reduces the probability of being a "moderate intensity worker away from home" (Class 3) in 2002, but the difference is only 0.013 (see Figure 2b). Women with coresident grandchildren are also less likely to be "moderate intensity workers based at home" (Class 4). In 2012, those with grandchildren present in the household were much more likely to be "high intensity caregivers" than those without them (a probability of 0.134 versus 0.084). The effect of other household composition variables is less consistent across years but there exists some commonality. As shown in Table 5, for example, the presence of children in the household increases likelihood of being in Class 1, 2 and 4 ("high and moderate intensity workers") as relative to Class 6 ("homemakers"). This could indicate economic pressure faced by multigenerational households, as well as more opportunities for family members to cooperate with each other. As expected, age has little effect on time use patterns when household composition variables effectively take into account life transitions. Having more years of education is associated with an increasing probability of being a "high intensity worker away from home", while living in rural areas decreases one's likelihood of being a homemaker relative to the others, with significant effects in some contrasts. What could have prompted these transitions in time use membership over all? Our crosssectional LCA shows that household composition matters, particularly the presence of grandchildren in the household. As Table 4 shows, the presence of grandchildren has increased from 8.3% to 57.9% of the households. Do we observe any dynamic change in latent classes of time use membership? We now turn to our results from latent transition analysis (LTA). With five to six classes, potentially we could examine 30 to 36 types of transition/non-transitions in-between adjacent waves of data. With four survey intervals, the results could be overwhelming. Instead of taking an exhaustive approach, our LTA analysis thus uses results from cross-sectional analysis as well as our substantive interests as a guide. First, we model the transition into the "high intensity caregiver" (Class 5) in logistic regression models for four survey intervals, given our interest in grandparental caregiving. --- Results from Latent Transition Analyses The results are presented in Table 7. The key covariates for the analysis are dynamic changes in the household composition. The most notable effect in the table comes from the changes in the presence/absence of grandchildren. For example, compared with those households with no grandchildren in either 2005 or 2012, those women who experienced a transition from no grandchild to having grandchildren living in the household were twice likely to become "high intensity caregivers" (translating the log odds of 0.701 into odds ratio). The effect is marginally significant in 1998-2002. Interestingly, in 1994to 1998,, while the majority of the women did not have grandchildren in the household yet, those who have already had grandchildren in the household in both survey years were 3.13 times more likely to become "high intensity caregivers" than those without grandchild in the household (translating the log odds of 1.141 into odds ratio). We also note a couple of other interesting findings. For those with spouse present in 1998 and 2002, women were more likely to become "high intensity caregivers" than those with no spouse in the household in both years. For those who lost parents in the household from 1998 to 2002, they were less likely to transition into "high intensity caregivers". Because we do not distinguish among spousal care, elderly care and childcare in the measurement of caregiving, these transitions capture the diversity in the women's caregiving experience as they age over the life course. We note that, because the presence of children in the household is very steady across waves and without much changes (e.g., almost all household have children in 1994 and 1998), some dynamic effects could not be estimated and are therefore absent from the table. We also ran a series of logistic regression models of transition out of class 1 through 4 respectively but did not find much effect in the dynamic change of grandchildren presence (results not shown). We interpret the non-findings in two possible ways. First, women do not scale back work activities even when caregiving duties potentially increase. The moderate and high intensity workers actually are engaged in a significant amount of caregiving and household work, and are subject to a great deal of double burden. The other possible reason for non-findings is due to the lack of power as sample sizes are quite small in each category. We estimated another logistic regression model that combined Class 1 through 4 together (the "double burdened categories") and therefore modeled the transition out of work category. We find that the change from absent to present in grandchildren had a significant positive effect in 1998-2002, that is, women are more likely to transition out of the worker category if they experienced the addition of grandchildren in the household. We also find that from 1994 to 1998 and from 1998 to 2002 having grandchildren consistently present in the household was associated with a higher likelihood of women transitioning out of both moderate or intensity work category, suggesting that the continued demands from caregiving for grandchildren may make the woman completely leave the workforce and become a devoted caregiver or homemaker in the family. --- Discussion and Conclusion Women

we utilize latent class analysis (LCA) to develop time use class membership to characterize the degree to which women in Cebu are subject to the double burden of work and family responsibilities in mid and later life. Results suggest that close to a third of the sample are engaged in high intensity work for pay (either outside or home-based) over eighteen years, while combining it with a substantial amount of household chores and with low level of personal time in a span of eighteen years. Our latent transition analysis (LTA) also shows that, with the addition of grandchildren into the household, some women experience a shift in time use class membership by becoming high intensity caregivers or by completely transitioning out of work arena, while others remain double-burdened with active involvement in both work and family responsibilities. The term "double burden" is often used to characterize the challenges a mother faces when balancing employment and domestic responsibilities (including household chores and caregiving) (Moen 1992). Studies on the "second shift", that is, how working mothers juggle amongst family and work obligations are abundant in the literature (Hochschild and Machung 2012;Massey, Hahn and Sekulic 1995;Milkie, Raley and Bianchi 2009). However, absent from the literature is how the extent of a "double burden" changes over the life course. For example, as a woman transitions from younger to older adulthood, the nature Paper presented at the 2015 annual meeting of Population Association of America.

sults not shown). We interpret the non-findings in two possible ways. First, women do not scale back work activities even when caregiving duties potentially increase. The moderate and high intensity workers actually are engaged in a significant amount of caregiving and household work, and are subject to a great deal of double burden. The other possible reason for non-findings is due to the lack of power as sample sizes are quite small in each category. We estimated another logistic regression model that combined Class 1 through 4 together (the "double burdened categories") and therefore modeled the transition out of work category. We find that the change from absent to present in grandchildren had a significant positive effect in 1998-2002, that is, women are more likely to transition out of the worker category if they experienced the addition of grandchildren in the household. We also find that from 1994 to 1998 and from 1998 to 2002 having grandchildren consistently present in the household was associated with a higher likelihood of women transitioning out of both moderate or intensity work category, suggesting that the continued demands from caregiving for grandchildren may make the woman completely leave the workforce and become a devoted caregiver or homemaker in the family. --- Discussion and Conclusion Women are overwhelmingly responsible for housework and caregiving responsibilities around the world in spite of their increased entry into the labor market. Although the term "double burden" is often invoked in the discussion of women's work and family roles, our understanding of these competing demands remains inadequate, particularly in the context of developing countries and transitional economies, where the informal work sector is large and the distinction between work and home environment is often blurred (Desai and Jain 1994;Donahoe 1999;Lloyd 1991;Short et al. 2002). Time use data is an excellent source of information to help us to understand work activities that standard surveys often fail to capture, and to document the associated patterns between work and other unpaid household work and caregiving work (Floro and Komatsu 2011;Hirway and Jose 2011). While the gendered pattern of time use is no news, we also know little of the process during which women's time use patterns change over the life course (see review by Sayer, Freedman and Bianchi 2016). Time re-allocation is inevitable when one moves from younger to older age, and transitions from one role to the other. The results from our latent class and latent transition analysis of time use allocation data for the Cebu women in CLHNS data reveal diverse patterns of configuration in work and family responsibilities over time. Close to a third of the sample are steadily engaged in high intensity work for pay either away from home or at home. It is important to note that although we label them as "high intensity worker," the labels do not readily convey the sense that these women's involvement in household chores is still extensive (2-6 daily hours). At the same time, those who are moderate intensity workers often carry a heavy load of household responsibilities. Thus, for those groups of women, the double burden of work and home responsibilities is tilted in a different way. Without such detailed time use information, we would not have been able to capture the nuance in the alignment in work and family responsibilities for these women. The above results are not surprising by any means. Just like women in other parts of the world, women in the Philippines are disproportionately heavily involved in family responsibility despite their increasing contribution to the family economy. What is striking is how incredibly consistent are the classes of "high intensity workers" during an eighteen-year time span from 1994 to 2012. This means that these Filipino women have been shouldering the double burden of work and household responsibilities for a long life span from mid life to early old age. Although the proportion of the sample that is classed as "homemaker" is growing over time, those classed as high intensity workers decrease very little over time in their proportions. Indeed, our latent transition analysis shows that "the high intensity worker away from home" is relatively stable over time. In addition, as many women become grandmothers and live with grandchildren, the life transition undoubtedly affects their time use patterns. Both our LCA and LTA results clearly suggest that being a coresident grandmother significantly increases the likelihood of being a "high intensity caregiver" and becoming a coresident grandmother increases the transition probability into a "high intensity caregiver" across survey intervals. At the end of the eighteen-year span, the majority of the women in the sample still do not have the luxury of enjoying a substantial amount of personal time. For these women, the increase in caregiving responsibilities often crowds out the time for self care, which can be critical as individuals transition into old age (Arora and Wolf 2014;Offer and Schneider 2011). This stands in sharp contrast with experiences of older adults in developed countries, whom often experience a much more noted increase in leisure time (Gauthier and Smeeding 2003). Interestingly, we also observe that for a large proportion of the women in the sample, during the transition from midlife to old age or into grandmotherhood, work responsibilities remain an important defining characteristic for their time use patterns. Addition of grandchildren into the household leads to transition out of workers' role for some women but not all. Therefore, these women who remained in Class 1 to Class 4 are still heavily double burdened, without enjoying a surge in leisure time. While the current paper is a fruitful use of the time use data in CLHNS, it is not without limitations. For example, changes in time use patterns could also be attributed to period effects, apart from life course effects. The economy of the Philippines is gradually becoming more industrialized, given its GDP of 272.1 billion USD, being the 15 th largest in Asia (World Bank 2013). Therefore women's participation in market work during the 18-year span could be potentially affected by the economic development. In terms of generalizability of the analysis, we caution that the survey covers the second largest metropolitan area in the Philippines and is not a nationally representative survey. Further, our analysis starts in 1994, but close to one third of the sample has lost to follow up since 1984-1985, with outmigration being the most common reason. As a result, women in our sample are less educated, have a higher number of children and are more likely to live in rural areas, compared with the baseline sample. Another limitation is that we characterize work by work place and work hours but do not take into account other work characteristics such as occupations, types of job or work sectors. The latent classes do not directly capture work and family conflicts, but the differentiation between work place (away from home vs. home based) is a proxy to highlight the compatibility of work and family responsibilities. Finally, our definition of role transition, specifically the transition into grandmotherhood, is limited by the household structure. For those women who become grandmothers but do not coreside with grandchildren, we are unable to capture whether such role transition entails any shifts of their time use. Despite these limitations, we believe our study takes an important step in understanding women's time use patterns and the changes over time in a developing country setting and in illuminating the reconfiguration of work and family responsibilities in transitioning from mid to old age. --- Author Manuscript Chen et al. --- Author Manuscript Chen et al.

we utilize latent class analysis (LCA) to develop time use class membership to characterize the degree to which women in Cebu are subject to the double burden of work and family responsibilities in mid and later life. Results suggest that close to a third of the sample are engaged in high intensity work for pay (either outside or home-based) over eighteen years, while combining it with a substantial amount of household chores and with low level of personal time in a span of eighteen years. Our latent transition analysis (LTA) also shows that, with the addition of grandchildren into the household, some women experience a shift in time use class membership by becoming high intensity caregivers or by completely transitioning out of work arena, while others remain double-burdened with active involvement in both work and family responsibilities. The term "double burden" is often used to characterize the challenges a mother faces when balancing employment and domestic responsibilities (including household chores and caregiving) (Moen 1992). Studies on the "second shift", that is, how working mothers juggle amongst family and work obligations are abundant in the literature (Hochschild and Machung 2012;Massey, Hahn and Sekulic 1995;Milkie, Raley and Bianchi 2009). However, absent from the literature is how the extent of a "double burden" changes over the life course. For example, as a woman transitions from younger to older adulthood, the nature Paper presented at the 2015 annual meeting of Population Association of America.