Datasets:

mimiklee

/

masterthesis_clean_text_20k

like 0

Introduction to love Sociological theory, with some exceptions, generally does not address romantic relationships 1 , instead referring them back to more individualizing academic disciplines such as psychology . Broadly speaking, this originates in the prerequisite that love is either an emotion, a sentiment, a passion , a neuro-biological constant resulting in an affective state of the body , or an interwoven mixture of both . The assumption seems to be that it is better to leave the discussion of love to the respective experts of the relevant disciplines. This is not to say that the cultural and social content of love is unexplored. This would ignore recent discourse and advances in sociology, which places "emotions" in the context of its cultural foundation, social order, and biographical experience of the actor . However, sociological discourse is largely referring to love in its transformation and iteration as part of a modernity . While often brilliant and outstanding in their analysis, these works sometimes seem to speak more to the process of commodification of love in modern capitalism than the phenomenon itself. In this paper, I do not attempt to add yet another aspect to this discourse. Rather, I pose the hypothesis that love is not an emotion in the first place. It rather is an effect of actors distinguishing themselves from the public sphere and committing to certain institutionalized engagements in an interactional, communicational, sexual, and affective perspective. Tying these media of experience together is a "private universe of meaning." Universes of meaning are culturally founded, interconnected systems of predetermined interpretations of the world . Experiences become tangible by being contextualized within the "cognitive logic" of a universe of meaning, self-referentiality contributing to its maintenance. --- Love as a private universe of meaning In this initial chapter, I will elaborate on love as a private universe of meaning relying on Peter Berger's concept of the social construction of the individual actor's reality. I will concentrate on four pivotal aspects: The structure of the actor's reality and the process of constituting and maintaining a universe of meaning, as well as the peculiarity of privacy and the question of participation in its plausibility structure. These four aspects are interconnected, but for the sake of analytical clarity, I will try to highlight their idiosyncratic features and identify connections. --- The structure of reality and its maintenance Cultural products, both immaterial and material, and meaning-bearing social objectifications as well as institutionalizations constitute an undeniable external facticity . It is only through confrontation with this external reality that the human organism transcends its biological nature and becomes a true actor, both interdependent and independent . Thus, structure simultaneously limits and enables freedom . Against this backdrop, the process of producing and maintaining man's objective facticity, as well as its relation to the internal reality of the actor, can be conceptualized more clearly. From the perspective of the individual actor, internalization occurs when, in the course of socialization, the objectivations of external reality are absorbed into the actor's consciousness and become a subjective and meaningful reality. From the perspective of structure, the actor externalizes individual experiences using preexisting objectifications, maintaining and, in some circumstances, gradually transforming the stocks of cultural and social products of society . This dialectical approach mitigates the distinction between external and internal reality: one cannot exist without the other. If there is an inner reality of an actor, there must also be a corresponding accumulation of objectified and socially valid knowledge-and vice versa. This model is very dynamic and adequate to explain both traditional and modern societies in their various iterations. It also accounts for the fact that the social structure of modernity is quite complex and differentiated. For the actor, this amounts to more freedom, but also the challenge of constantly dealing with an "order of existence" that is fragmented into a multitude of "universes of meaning" . Universes of meaning exist twofold. Firstly, they are an aspect or a section of the actor's order of reality . They provide a connecting and unifying "style of existence," in which raw sensory data is coherently bestowed with meaning . They are immersive in the sense that their relation to the world is experienced in a unique and definable way, as they endow the actor's experience a particular "accent of reality" or "cognitive style" . They imply, suggest, and often demand a specific interpretation of the world that is not up to the actor but is supra-individual and grounded in historically, socially, and culturally validated knowledge . Secondly, they are distinct communicational, interactional, emotional, and esthetic forms of the objective world , deeply embedded in the social and cultural foundations of society. The relation between subjective constitution and objective construction is far from arbitrary, free floating, or negotiable: courses of action and, thus, experience always derive from culture . Henceforth, I use the term "universe of meaning" to refer to both interwoven aspects, the constitution of the actor's experience, and its construction by external objective facticities. It is peculiar for love as a universe of meaning that what has been public becomes private, once it is enacted by the new lovers. While actors constantly use socially validated set pieces, they make them their own in the process of being in love. This conceptualization of love warrants further explanation. "Love" is a universe of meaning in the same way the "American family," the "state," or "capitalist economy" are . As such, love is highly symbolic, more a guiding principle or narrative than a concrete course of action. Understood in this way, there can be little doubt that in Western modernity one of the highest symbolic values is attached to love, romance, and intimate relationships. One could argue that the legitimization of love even justifies violations of other universes of meaning, such as law or religion. It is significant that revenge against a perpetrator of violence against a loved one, be it the partner or a family member, is associated with justified anger and might be assessed differently by the law than violence against strangers-or even mitigated altogether . Similarly, in contrast to "love" as a cultural principle, "a love" is a set of concrete and specific social agreements, institutionalizations of typical courses of action, objectifications of meaning, behavioral scripts, artifacts, and roles. In both cases, love constitutes an "interpretive scheme" that denotes the effects of "social/collective meaning and action-organising schemata" . One side of this universe of meaning refers more to an actual and pragmatic framing of experiences and, thus, to choosing specific courses action, the other side more to legitimizing these experiences in the first place. So far I must admit to universes of meaning remaining an abstract concept. I have yet to explain how they are constituted and how they function as an intermediate between the social and its cultural foundations as well as the individual actors and their reality. Universes of meaning are formed by a multiplicity of interconnected and interlocking objectifications and institutionalizations that relate to each other on the basis of, or under the canopy of, an overarching symbolic legitimization. Both solve one of man's most profound problems: how to create time. Like all organisms we live perpetually in the here and now. Only in actual situations other actors are fully present to us. To become truly social, we must employ a mode of communication and interaction that transcends beyond the limitations of immediacy. Objectifications and institutionalizations solve this problem. Objectifications are relatively stable and robust thought objects that contain socially and culturally validated meaning and interpretation. Human products, both immaterial and tangible, from language, numbers, writing, money, tools, and architecture, to the designs of cities and landscapes would essentially be meaningless in themselves. They would not even exist if they were not to appresent meaning or to be used in a habitualized way which then again is perceived in a certain, predetermined understanding. Only if the object is endowed with meaning, if a signifier is coupled with objectivated experience, it appresents and makes the transcendent present. In other words, over time and by the externalization of subjective experience an object starts to "mean" something to others as well . When action reciprocally defines act and actor it becomes institutionalized and regulates behavior . These institutions can be conceptualized as "constraints" in social exchange . Institutionalizations regulate and shape interaction , communication , and emotions . Talking about love, privacy becomes essential as it is its defining feature. To be clear, "privacy" neither implies that objectifications entail no relation to the public stocks of knowledge in regard to romantic relationships nor that lovers are completely free, anarchic or unbound to enact their love. Instead, privacy denotes a greater latitude to transform objectifications and institutionalizations of love into one's own by gradually superseding common knowledge with lived experience. Meaning becomes exclusive to those who produce and maintain it. I will now elaborate on privacy as a concept of love and the process by which it is constituted. --- Privacy, agency, and plausibility structure Universes of meaning are generally historical and public. Actors are born into a society and become participants by internalizing its objective reality, essentially making it their own. This is how knowledge, order, and structure are upheld. Of course, in order for actors to enter this process of world maintenance, reliable objectifications and universes of meaning must already exist. Accordingly, while participation itself is essential, the actual effects in terms of transforming the appresented meaning of an objectivation are negligible. Generally, in the public sphere meaning is upheld by a resilient "plausibility structure" . Plausibility structures are not comprised of actors themselves. They are the "social base" of a universe of meaning, its cognitive logic, objectifications and institutionalizations . This includes institutionalized courses of action, language or lingo, stocks of knowledge, products and processes, even architecture and organizations pertaining to a universe of meaning, as well as its idiosyncratic approach to the world. The plausibility structure of "the law" is based on a variety of elements, including but not limited to a formalized legal code, a corresponding public knowledge of enforced and prohibited acts, formalized language, publications, courts with special jurisdiction and hierarchy, academic training, staff, a diversified law enforcement machinery including prisons, structural couplings with the press, policymakers, representations in film and television. These element are embedded with a certain symbolic legitimization. The plausibility structure of the law is "rigid" or culturally "settled" , and thus, its corresponding universe of meaning is valid in its entirety without the need of any further legitimization but its existence in the first place . An individual actor will subscribe to the cognitive logic of a universe of meaning and its plausibility structure, but will not change its meaning in any significant way, shape or form. The privacy of love, a specific and tangible love, structures the universe of meaning differently. Whatever becomes of the relationship, to a certain degree depends on the partners to agree on content, form, meaning, and process. This is not to be understood as indicating that romantic relationships are arbitrary. Lovers must adhere to the cultural repository that found and legitimizes love and provides a framework for the newly emerging universe of meaning. Which elements or set pieces lovers choose however, is for them to decide-tradition, social status, education, religion not withstanding. One could even say that courtship is the process of matching and coordinating exactly this: the careful agreement on the elements the new universe of meaning is to be constructed from. While still consisting of familiar elements, objectifications, and institutionalizations, but it will be unique to the lovers. In the private sphere, mutual objects of thought can be developed, not just internalized. It is feasible to think of this as "agency" . In the course of a relationship, the public meanings of phenomena are superseded by personal, biographical, and, above all, mutually shared experiences that constituted interconnected objectifications and institutionalizations. In this universe, things have meaning primarily intelligible to the partners. I would argue that this distinction, which is essentially privacy, is what constitutes the "form" of a romantic relationship in the first place. It is the foundation and reason of a very unique approach to experiencing the world and essentially excludes all other cognitive options , pp. 16-23). Privacy then is the answer to the question of the idiosyncrasy of love, understood as a conceptual process in which biographically shared experience becomes a specific cognitive approach. Still, the specific love remains part of the social order and founded in culture. Without a socially recognized, publicly shared, and historical system of expression and content, the private could not be conveyed to another actor. If anything, culturally established objectifications serve as intermediates between actors, enabling them to share personal experiences by using public means of communication. Otherwise, there would not be any private universe of meaning in the first place. But even here privacy must be treated with caution: In accordance with the Schützian assumption that socially validated, relevant knowledge is distributed unevenly , love as a cultural principle and as a set of social predispositions is, while public, only in theory accessible to every member of a society. The empirical world of love is not coherent, without being disjointed. Actors deviate from predetermined guidelines, interpret scripts and roles in their own ways, and act and experience relationships on the fringes of what is barely recognizable as love from the outside. Not only is relevant knowledge not shared by all, i.e., experienced and understood within-and sometimes beyond-a particular frame of acceptance, but the uneven distributed amounts to different cognitive approaches . Since actors are a product of social structure, class, ethnicity, gender, education, health status, personal biography, and their intersections, different stocks of material, social, and knowledge resources associated with these social domains must lead to different interpretations of reality. Based on this theoretical framework, I will now discuss the forms of experience that actors have in their romantic relationship. It goes without saying that what follows must be an abridgment of the lived experience of the individual actor. As a conceptualization of the self-referential nature and functionality of love, it nevertheless remains valid. --- The experience of love Love is experienced as a totality. I propose that love is constituted by being experienced in different media. These are as interrelated as they distinct regarding their properties. Actors do not break down how they are addressed by their beloved or how a relationship is enacted in a specific moment-a feat that is unlikely to be accomplished in real time. There have already been attempts to deconstruct the totality of love into its essential constituents. Haller suggests that love is composed of four distinct components, namely a cognitive, ethical, interactional, and emotional category. Sternberg also identifies three distinct components, with "intimacy" evoking an emotional experience, "passion" calling to the afflicted body and sexual consummation, and "commitment" referring to a cognitive experience and volition. Finally, Baumeister and Bratslavsky discuss "passion" and "intimacy" as indicative of love. These examples may suffice to show that there is no broad consensus about what love actually consists of. Much more decisive for me, however, is the question of the nature of the individual components, beyond their descriptive specifications. What functions do these constituents entail, in themselves and in relation to other aspects of love? It seems obvious that the sense data constituting, for example, a communication and a sexual experience appresent very different meanings, which in turn are idiosyncratic for the quality of the sense data in which the objectification is experienced. There are some distinct consequences to this question. I have already argued that objectifications appresenting meaning are an externalization of the lived experience of actors. But the experience is not specific only to its appresented "content" or the universe of meaning it pertains to. The experience is also dependent on the type of sense data actors perceive. When lovers become visually aware of the partner, the experience is bound to the perceived quality of sense data: the "content of this awareness of one's experience's being visual can be specified only in terms of the sensory phenomenology associated with visual experiences, i.e., what it's like to see colors, shapes, etc. So the phenomenological content essential to having an experience is part of an experience's intentional content, whatever else may be part of its intentional content" . Beyond a phenomenological approach, it is feasible to ask whether different forms of experience are also idiosyncratic. With Luhmann I will call these qualities "media." As with visual sense data being exactly this-visual-and visually shaping the perception of the object, it can be assumed that different "media of experience" also appresent content idiosyncratically. Even more, it is feasible to assume that some appresentations are more likely to be conveyed in one medium rather than others. They likewise fulfill a distinct function in this medium, and this one alone. I suggest that there are four distinct "media" of experience in a romantic relationship. First, love is a flow of self-referential communicative acts or communications. Second, love is a series of institutionalized interactions that range from a broad spectrum to specific, everyday life micro-interactions. Third, love is an emotional response to a specific social relation that leads to a mutual sense of closeness with other actors. Fourth and finally, love is a contractual agreement that enables close physical contact. All four media of experiences are self-referential, i.e., they are founded in and at the same time maintain love as a universe of meaning. There are a few points that still need to be addressed before elaborating the different media. First, neither interaction nor communication, emotionality nor sexual activity is specific to love. They become distinct experiences in their own right and pertain to a romantic relationship because they are grounded in a private universe of meaning, an order of existence that compels the actors to perceive sense data as "romantic," and at the same time constitute idiosyncratic appresentations that are discernible to the fullest only by the lovers. Second, content, purpose, and media intertwine. Actors use different media to express themselves in ways that depend on the purpose of their actions, but at the same time are intertwined with the form of the medium used and its ability to appresent meaning more effectively or appropriately than others. I already elaborated on the internalization of objective facticity, becoming the actor's subjective reality. This process not only applies to the meaning of an objectification, but also to the medium the object is experienced in. A hug or a kiss convey the feeling of closeness and the transformation from friendship to love differently than the statement "I am experiencing an emotionally affected condition of the body and the mind, let us begin a monogamous dyadic relationship"-which is why one is used more frequently than the other to indicate romantic intentions . The same is true vice versa. A relationship can be assessed as an intentional emotional engagement on a bodily and affective level of one actor toward the other . Love, then, is a judgment about a social relationship as is hate, despair, or disgust. As actors, we tend to believe in such emotional propositions, even if they are much more blunt and undefined than actual speech and its reflexive abilities. On the contrary, it is difficult to dissuade an actor from certain courses of action, if he or she has already engaged in emotional judgment. It is important to reiterate that the way lovers experience different media is highly dependent on the cultural and social repositories of a society. I will now discuss the four media in which lovers relate to the universe of meaning called love. How do they make possible the totality of the experience called love? --- A typology of four media of romantic experience The four different media of experience in a romantic relationship have yet to be typified to complete my attempt to conceptualize love. --- Communication Love's communication amounts to intimacy. On the surface, it is the most banal statement that there is communication in romantic relationships. Lovers talk, write, and text all the time, perhaps even resort to more exotic means of communication. In this process, a "relationship" develops . Yet, communication in romantic relationships is intimate as it is directed exclusively at one specific actor and can, at least in theory, touch on any subject. This distinguishes love from all other forms of communication. Communication in romantic relationships should be considered from three interrelated perspectives: content, genre, and function. "Content" refers to the meaning that is appresented in the lovers' communication. Leading up to a relationship lovers seem to "talk," disclosing information that should be considered inappropriate for the public sphere in such a concentrated form. This trust-building sharing of one's history as well as a general view on life continues with the beginning of a relationship, excluding others . Then, communication often becomes sexually charged and emotionally saturated , increasing the degree of privacy and separation from the public sphere. It should be noted that communication at the beginning of a relationship is interwoven with all other media of experience, reflecting on them, thus, reinforcing their effects. However, even with the almost inevitable decline in sexual activity in the later stages of a romantic relationship , this change in frequency may be addressed in ways that are only possible with the intimate partner. "Genre" pertains to a lesser degree to meaning, but to "lover's talk" as a generic type of communication . Here, thought objects are modeled as a specificly formed narrative structures that becomes an objectivation itself. This refers to the fact that conversations are socially constructed in terms of their sequential nature, process, para-linguisticity, and purpose/function . Some very specific genres come to mind, that are easily distinguishable: debate, confession, or teaching class. As participants in everyday life we are able to easily identify these genres without the need to know their actual content. The communication between lovers is often constructed around privacy, opting for possibilities to retreat from the public eye, being attentive, or making physical contact, and embedding their personal experiences as lovers into broader themes. Again, there is a dialectical relation between culturally validated knowledge about lover's talk and the genre itself relying on a distinction from the public sphere. Intimate communication's "function" is to increase stability. It does not aim at clarity or inclusion but at the communicative reification of the status as lovers. This eschews all normative or dogmatic implications; even more so, it does not distinguish between "good" and "bad" relationships. Dispute, conflict, and chaos, even abusive or toxic attachment are entailed in the concept of intimate communication, founded in exclusivity, and part of the cultural patterns called love. It is the exclusivity of romantic communication in all three aspects that facilitates the lovers' relationship, as they share an intimacy that would be inappropriate in other contexts . Thus, I depart from any attempts to understand intimacy as a particular form of trust , a willingness to share deeply emotional and "private" experiences with one's partner , or vulnerability . Rather, intimacy is an effect of and the foundation of the separation from the public sphere as well as being based on it. It is the use of those communicative properties that are deemed private by the public. In other words, the particular qualities of intimate communication are mostly public knowledge, and actors are usually well-versed in the art of private communication. This only comes to pass if love as a communicative process starts, thus drawing a distinction to the public sphere ). --- Interaction As far as interaction is concerned, love first and foremost provides security. At first glance, this may seem like a misconception or even a contradiction given the number of separations and divorces. However, looking at the empirical phenomenon in a more theoretical perspective the meaning of the security love offers in terms of interaction becomes clearer. From an anthropological perspective man is endowed only with a relatively weak morphology. Man is not bound to a specific habitat. While able to and often obliged to make almost any stretch of land habitable, this endeavor can only be accomplished through cooperation, collaboration, and specialization. In other words, man needs to compensate for a weak morphology by establishing strong, dependable and predictable social relationships . It is well established that humans have created institutions to deal with this fact and to compensate for the energy they constantly would have to expend were social relations always to be established from scratch, situationally and on the spot. But this security comes at a price-the actor sometimes becomes alienated from the man-made world that is originally intended to provide a functioning and livable environment . This is of course one of the most essential laments about modernity. Modernity's public sphere is highly structured, tightly institutionalized, based on the division of labor, and characterized by technological production and bureaucracyand practically beyond the control of the individual actor. Moreover, as Weber notes: There is a process of selection embedded in every institution. Either actors adapt to the predetermined, institutionalized ways, or they are excluded from the community of those adhering to them eventually. The cognitive logic of the public sphere is virtually impossible to change or modify by an individual actor. It simply takes effect. As institutions persist for centuries, their existence is endowed with an ontological status seemingly independent of actors' will and volition ). While seemingly the antithesis of "freedom," this process rather refers to symbolically legitimizing perceiving the world in specific ways. Only if social relations lose their ontological status by virtue of a weakened plausibility structure, social order becomes "unnatural." This comes with the price of being an alienated stranger in one's own world . From Berger et al. to Giddens to Beck , the finding of a loss of an overarching nomos in contemporary western modernity persists. As a result modern actors are alienated from the very world they built for their survival. Giddens very clearly assumes that "ontological security," i.e. actor's premise of being in control despite the complex structure and sometimes contradictory universes of meaning, is established through routines. While public institutionalizations at their core are supposed to do just this, their lack of controllability for some thwarts the purpose of providing certainty. I propose that romantic relationships counteract this effect-at least to some extent. For Berger and Kellner , committed love constitutes one of the few social relationship in which nomos, an overarching sense of meaning for all experiences, still does take hold. For Beck and Beck-Gernsheim, love in particular becomes the "central pivot giving meaning" to actors. More than that, precisely because interaction in romantic relationships can be institutionalized at least to some degree by the actors themselves. They convey a sense of empowerment to act according to their own volition. Routines that fit one another are enacted, controlled and monitored only by the lovers themselves. The simple fact that most romantic relationships end in separation does not preclude this thought. On the contrary, since there is always the possibility of a "doover," there is also a chance to change things. However, change requires agency and the notion of control. Partners can break up, but a new relationship can be built on the foundation of these experiences. The security of being able to shape relationships as one sees fit is clearly rooted in the privacy of love. The lover's experience translates to a notion of agency . But the opposite is also true. Because the actions of private relationships are perceived as being based on the volition of the actor, their experience is separate from the public realm. --- Sexual activity Sexual activity in romantic relationships is characterized by the notion of property, a term coined by Collins . As with all media of experience, sexual activity amounts to a strengthening of the romantic universe of meaning. The universe of meaning contextualizing all experiences in a romantic relationship, and particularly its exclusivity, can nowhere be found more explicitly than in the property of the other's body. The aspect of property refers to an interaction that is distinct from all others and unique to romantic relationships. This is neither due to the quality of the sexual encounter nor the sex act facilitating pleasure, reciprocity, and intense solidarity . Rather, romantic relationships provide institutionalized access to another actor's body that is far removed from any other form of social interaction. I define "sexual activity" broadly as all physical and erotic acts between two actors that involve transgressions and violations of personal space or "territories of the Self" . There are institutionalized transgressions between adults in public life, for example, in care, nursing, and medical procedures. But these are based on necessity. In romantic relationships, these transgressions are based on volition and the cognitive logic of love. It is after all the notion of exclusivity that characterizes sexual activity. Property, then, does not refer to the sexual actor. Rather, it defines the relationship between two partners, the way one acts toward the other and understands his or her actions in terms of sexual desire . Love implies a contractual agreement between lovers. At the end of courtship, negotiations to enter into a relationship are complete. This "contract" is grounded in a cultural repository and in most cases leads to the institutionalization of exclusivity and monogamy. The agreement to enter into monogamy , and therefore the increased probability of rather risk-free sexual intercourse between partners can be considered the basis of a committed relationship. To clarify this basic principle, some explanations are in order. Property of the other's body simplifies sexual intercourse, even turns it casual and routine in some instances. But while access to the lover's body dispenses with complicated formal discussions, it comes at the price of limiting the number of sexual partners-at least in monogamous relationships. What could be seen as a distinct disadvantage in regard to the variety of sexual encounters and even being at the mercy of a continuous attraction to the lover, again strengthens the universe of meaning by exclusivity. Experiences, even in their unpleasantness or absence, further the distinction between the lovers and the public. Limitation and possibility form a reciprocal distinction . This holds true even if the actual quality and quantity of sexual encounters is unsatisfying. This principle is valid not only for sexual intercourse, but any physical activities that must be considered inappropriate in public life. Exactly how this demarcation is enforced must be negotiated between the partners. The importance attached to an embrace may vary from one couple to the next and depends on the cultural backdrop. Polyamorous and asexual relationships ) further complicate the apparent unambiguity of linking sex and exclusive romantic relationships. Adding to this ambivalence is the fact that the frequency of sexual intercourse and overall sexual satisfaction varies vastly regarding time and form. Generally speaking and disregarding the age of the lovers we know that cohabitating couples have sex more often than married ones. While marriage is commonly known as the only iteration in which sexual intercourse is universally approved of, its frequency decreases the longer the marriage lasts and if lovers become parents . Finally, there are of course sexual interactions outside of romantic relationships, e.g., casual dating, prostitution, and masturbation. Collins goes to great lengths to distinguish individual pleasure seeking from sexual activity as a ritual of interaction that furthers solidarity among participants. That being said, the agreement of access to the lover's body can be postponed by either partner for an undisclosed period of time or reason. Although the term "property" may falsely raise concerns that it is an extension of the patriarchal rights men exercise over women , this implication is not intended. Contractual access does not preclude the situational refusal to have sex. On the contrary, actors are involved in a variety of other institutionalizations that constantly require acts of perception, sense making, and fulfillment of socially recognized requirements, whether in the domestic sphere or in paid employment that will interfere with the couples sexual activity. Elliott and Umberson report a variety of situational breaks in the access to the partner's body that do not necessarily amount to termination of the arrangement itself. Because frequency of sexual activity is correlated with marital satisfaction and marital satisfaction is correlated with the likelihood of separation and divorce, there are institutionalized ways to deal with these irregularities. These are primarily referred to as couples counseling and therapy. Communication and emotionality regarding denial of access to the body are therefore routinely integrated into romantic relationships. The expectation of possession shapes communication, interaction, and emotionality as much as the denial of sexual activity. Perhaps no issue is more central to a romantic relationship than property, regardless if it is in its fulfillment or in its denial. --- Emotion Finally, regarding the one medium most commonly associated with romantic love, its emotional aspects or emotionality are defined by labor. To break this down, it is necessary to examine Hochschild's concept of "emotional labor" in more detail. She proposes that the social world is not only structured in terms of interaction and communication but that actors also are compelled by conventions to feel in accordance with the situation they perceive. As a society, our feelings are ordered-or, institutionalized. Because we know that a particular situation requires an appropriate emotional response, we feel accordingly. Like any emotion, "love" is the appropriate bodily response and expression of a situation within the confines of a historically pre-negotiated and validated framework, and to certain communicative, interactional, and carnal engagements . The actual emotional response of actors in love is the result of the work invested in the ability to feel the right emotions in discernible situations . In other words, reality is structured not only in terms of experience, interaction, and communication but also in terms of emotions. The emotional landscape is highly ordered, and in everyday life we perceive this as normality. There is sadness and grief in death, enchantment in religious experiences, joy in birth and success, and passion and devotion in love. Actors adhere to objectively valid regulative principles or "emotional rules" . The social construction of the body and its emotional aspects are crucial , yet Berger and Luckmann essentially opted out and chose not to discuss them in any meaningful way, shape or form in favor of language. Feeling rules fill this gap. Emotional reality is as socially institutionalized as interactional and communicative reality. Not only do habitualized actions typify the actor-and vice versa -, feelings also typify the afflicted self-and again vice versa. In this understanding, lovers become lovers because they are emotionally affected by an institution called love, which in turn takes place only because a certain typical emotion occurs. It could therefore be argued that love, grief, or desire are not emotions at all. Rather, they are affective institutionalizations that place the actors involved in a particular relationship with each other. It is important to mention that this does not preclude the notion of authenticity any more than actors talking within socially tried and tested communicational defaults. "I love you" is not an novel sentiment, and still it often genuinely expresses an emotional response. There are essential consequences to this. First, I would like to point out once again the processual nature of love. The proper emotional state is not endogenous; it is an effect of the individual actor's socialization and must be constantly maintained by "being felt" in regard to structure . By adopting a particular world perspective, ideally all of the individual's experiences become meaningful within the confines of society's socially validated knowledge. While at a general level the structure of external facticity and inner reality align and correspond , "emotional labor" is constantly required to bring the actor in tune with situational requirements . The "depth," intensity, or "authenticity" of an actor's passion in and commitment to a love relationship cannot be considered on its own. It depends on the partner's actions, the willingness to be "involved," the circumstances of the relationship, and the general culture in which the relationship takes place . For example, in some regions and cultural contexts, homosexual love is still somewhat frowned upon and socially shunned. Therefore, homosexual lovers might not allow themselves to become as emotionally involved as they would if their relationship was perceived in a more positive and favorable manner. That being said, an actor will adapt his or her own emotions to the structure and situation, i.e., how other actors respond to expressions of emotionality. We dress up nicely for dinner together and "get in the mood." And we find it natural to work on our partner's emotionality as well. We are attentive to our partner, maybe give presents or create a special mood during a romantic evening by arranging for appropriate music to be played. The continuity and exclusivity of an emotional bond with the partner cannot be overestimated. As with sexual activity, the feeling of being in love, even more so the feeling of being loved, is unique to the actors in a romantic relationship. It is an experience that cannot be replicated or replaced by anyone else. It is founded in a universe of meaning and essential for its upholding. Therefore, emotional labor is a quintessential part of any romantic relationship. --- Conclusion In this paper I have tried to conceptualize love as an effect of experiences that are based on the constitution and maintenance of a private universe of meaning. Here, all perception is structured in such a way that of all the possible meanings the object could reasonably appresent "love" predominates. Love emerges from four different media of experience: Self-referential communication, idiosyncratic interaction, sexual access to the partner's body, and an emotional response to these experiences. All this derives from set pieces from the cultural repository pertaining to love. In this sense, love is and must be public. But the more the lovers share, i.e. externalize mutual experience and objectivate these into objectifications that are appresenting elements of their love, the more the universe of meaning becomes private and stable. Then, over time and through a multiplicity of four different media of experience, the universe of meaning evolves and expands into a totality. Understood this way, love is not an emotion in the first place, but an affective response to the decision to constitute and participate in a private universe of meaning. It is an emotional proposition about a social relationship with another actor. As such, it affirms attachment to the partner and excludes all other ways of experiencing the world. A love exists outside the public sphere. It is personal and up close. As a love is exclusive it is not accessible to anyone except the actual participants of the respective universe of meaning. Those affected by and at the same time maintaining the universe of meaning are identical to its custodians of the knowledge. In conclusion, I would like to discuss the implications of this concept. The ability of actors to distinguish and differentiate themselves from their environment is the defining feature of love. Partners draw a line in the sand between themselves and everything else. They become exclusive. By this, love overcomes modernity's tightly institutionalized forms, its division of labor, its adherence to technical production and bureaucracy, and its anonymity . The public sphere is as impersonal as it is dispassionate-at least in an ideal world in which public institutions simply fall into place, ensuring everyday life to proceed without major disruption . Love is not. Love is an iteration of the hope to be "somebody" in modernity, to be authentic, to be experienced as an individual in the truest sense by one's partner, and to have agency. Exactly the exclusivity of love is also its greatest weakness. Opportunity and risk go hand in hand: while the extremely limited plausibility structure provided by only two actors allows objectifications to emerge and supersede social appresentations, this also carries a tangible danger: a falling out between the partner cannot be mitigated. It leads to the de-legitimization and destruction of the universe of meaning. Even below the threshold of a relationship's breakdown, personal development and outside interference can be assumed to pose a significant risk to the ability to continue maintaining mutually shared objectifications. That being said, this argument seems to be supported by empirical data. About half of all marriages end in separation and divorce. From the model presented here the main reasons for separation can be understood in a specific way. Conflict, lack of commitment, infidelity, and lack of emotional or physical intimacy all point to the underlying process of having communicational, interactional, sexual, and emotional experiences within the private universe of meaning being disrupted, and finally stopped altogether. What partners then describe as "reasons" for separation and divorce are actually effects. However disparate these reasons to end a relationship may seem, they converge in a lack of mutual objectivation. They are manifestations of a failed agreement between partners to incite a self-referential process that in the end is subject to one function only: maintain the universe of meaning itself by keeping objectivations exclusive and private. But it would be wrong to attribute all this to the shortcomings or unwillingness of the actors. Lovers enact their relationship in a highly complex environmentmuch more so than compared to pre-modern societies. Romantic relationships have changed significantly over the last 250 years. Current modernity incorporates the idea of individuality, of willingly complying even with engrossing institutionalizations. In the wake of this development, relationships are transformed into "confluent love." Confluent love basically constitutes a rational choice agreement in which each partner consents until further notice and is based on the individual benefits provided by the relationship . There are some obvious improvements that fall in line with these transformations, such as attention to sexual satisfaction and the dissolution of a gender-based power structure between lovers. But at the same time, love becomes an institution under the auspices of being deselection at any point. In the framework presented here, the inherent paradox of modern love can be understood differently. It seems entirely plausible that love, because it is private, represents a refuge from the anonymity of modernity, and thus satisfies a need for holistic relationships. At the same time, lovers are also children of their times and cannot fall behind the modern principles of individualization and risk assessment. Romance is based on the idea of surrendering personal freedom and the search for fulfillment in togetherness. Individuality must then be brought in line with the decision to submit to a relationship. In love contradictory forces work against each other, a process which results in establishing a universe of meaning but at the same time allows for it to be discontinued at any time. The fact that with the termination of a relationship, the relevance of anonymous institutions rises is an obvious and yet accepted paradox. All this is not to ontologize romantic relationships. But with the framework presented here, it might be possible to better understand the strains and burdens lovers experience not just on a purely empirical, but on a conceptual level. While this conceptualization of love is theoretical first and foremost, empirical applications are close at hand. Thus, this paper is an invitation to discourse as it tries to understand one of society's key social interactions in a sociological manner. you give appropriate credit to the original author and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Jochem Kotthaus is a professor in the Department of Applied Social Sciences at University of Applied Sciences and Arts Dortmund. He specializes in institutionalization and subjectivation, particularly sociology of knowledge; sociology of professional sports; and theory of social work. For the past three years, he has been working on a transcultural research project on love as a social relationship.

In this paper, I will explore love as a universe of meaning constituted at the crossroads of cultural patterns and actors' biographical experiences. Universes of meaning provide a structure of cognitive pre-selections. While the social in general is composed of a multitude of universes of meaning, they belong to the public. Romantic relationships are private and enable privacy. I will (1) propose a definition of love and a framework that serves to ensure its theoretical validity. I will then (2) analytically deconstruct the unity of communication, interaction, eroticism, and emotionality as love's different media of experience and explore their self-referential functionality.Institutionalization of emotions • Sociology of knowledge • Meaningmaking • Universe of meaning * Jochem Kotthaus

Introduction The incidence of infants born at less than 36 weeks gestation is steadily increasing. Advances in medical intervention have resulted in improved survival rates for preterm infants 1 . Australian data suggests that women residing rurally have poorer obstetric outcomes compared with their metropolitan counterparts, and are more at risk of having infants born prematurely 2 . Infants born prematurely require extensive medical intervention and long-term follow up from a number of health disciplines. Health professionals are required to not only meet the needs of the child, but also to meet the needs of the parents and other family members during this difficult and stressful time 3 . In rural areas, these needs may be unique because families often have to travel great distances to access specialist care for their child, and are likely to have reduced access to support and services in their home town. Parents from rural areas are also less likely to be familiar with large metropolitan hospitals where specialist neonatal, paediatric and neonatal services are located. This study aimed to explore the experience of parents of premature infants living in a rural area during the first year of their child's life. Existing research has been mainly nursing based and focussed on parents' experiences in the neonatal intensive care unit in metropolitan areas. Not surprisingly, parents have reported being in a state of shock when they initially arrive at the NICU 4 . Wigert interviewed such parents and they reported the need for time to inform and educate themselves regarding issues relating to their child 4 . Parents also reported wanting to be treated as an individual with recognition of their own circumstances. This may be especially true for families from rural areas because metropolitan staff may be less familiar with the support and services available in rural areas. Heerman et al described the experiences of parents' from non-rural areas in NICUs 5 . The study revealed that parents shift from being a passive bystander to a more active advocate for their child's needs over time. Parents described being initially focussed on the NICU environment but gradually becoming more comfortable and being able to focus on parenting their child with greater ease 5 . Brinchmann et al described how parents wanted to be consulted about clinical decisions in the NICU; however, they did not want the final decision to be left entirely up to them, feeling too emotionally involved to make a rational choice 6 . The parents described a power struggle between themselves and health professionals regarding the day-to-day care of their child in the NICU. The experiences of fathers in the NICU was explored by Pohlman who reported that fathers have different needs and perceptions in the NICU environment 7 . Fathers often worked with renewed vigour due to the ongoing need to provide for their family. These men seemed to feel in control at work but they often felt helpless in the NICU. Fathers also reported being stressed by having to manage the demands of work and family outside the NICU, their partner and the premature infant. There is a paucity of research examining the rural family's experiences. One study of families from a non-rural area reported that the geographical distance between home and hospital had a negative impact on the stress responses of fathers of premature infants 8 . Difficulty accessing rural paediatric health services, in particular allied health and early intervention services, has been described 9 ; however, the impact on rural families has not been explored. The present study used a hermeneutic phenomological approach to examine parents' lived experiences of caring for a premature infant during the first 12 months of the child's life while living in a rural area. Providing access to families' lived experiences offers health professionals a deeper understanding of the needs of such families, and the ability then to ensure that services effectively reflect these needs. --- Methods --- --- Data collection The research team consisted of a physiotherapist working in paediatrics , a postdoctoral academic with experience researching premature infants and a paediatric occupational therapy academic . Semi-structured interviews were conducted with participants. Two sets of married couples were included; however, participants were interviewed separately in order to gain their individual perspectives. One author conducted all the interviews and the majority of the data analysis. Regular meeting with the other supervising members of the research team were conducted to discuss the thematic analysis in order to gain different perspectives on emerging themes. Participants were asked a series of open-ended questions about their experience of being a parent of a premature infant during the first 12 months of their child's life. Questions --- Ethics approval --- Results and Discussion The participants described a 'journey' which they revealed was emotionally traumatic, stressful but 'character building'. They frequently reported gaining a different perspective on life, and that the hardship made relationships strongerespecially with their partner. --- ...it bought us closer together because it was the most stressful thing we've ever been through and we really relied on each other. Participant 1 Mother When asked about how the experience had affected their personal relationships, without exception all stated that it had brought them closer to their partner. All parents also reported that others, such as friends and relatives, often had little understanding of what they were experiencing which made it difficult to ask them for support. All parents were still together at the time of the interviews. All parents reported initial shock and confusion over what was happening with their infant. This led to a gradual understanding and acceptance as their child made developmental and medical gains. This is consistent with Heerman et al's findings that parents struggled initially with acceptance of their situation 5 . Five of the seven parents had infants who were transferred to a metropolitan hospital. These parents reported feeling burdened by having to travel and relocate to a metropolitan hospital, often in an unfamiliar city. There Three of the participants lived on rural properties on which they worked as well as having other employment. The travelling was particularly difficult for these families because they were unable to take leave and needed to tend to the livestock and perform properties duties. ...especially when you're running your own property as well and working. It was lucky it happened in winter time so there wasn't any hay to make. Participant 2 Father These three parents described being unable to leave their properties for extended periods. They reported that one or both parents were working on the property at night after returning home from hospital during the day. As parents became more familiar with the NICU they came to feel more comfortable in that environment, with the care of familiar health professionals and with a high level of specialist care. Parents described the subsequent anxiety of returning to their local hospital where they perceived the level of care to be less specialised and not individualised. --- That was the biggest shock coming from that high level hospital ...in hindsight I probably wouldn't come back as early if I could do it just 'cause you don't have that connection…and they were really busy. Participant 3 Mother Five parents felt that more preparation for the return home would be beneficial, allowing them to anticipate issues. Further orientation to the rural hospital or to available local services was described as essential for adjusting to the change. Two parents, including one father, felt that they had adequate orientation to local services. All parents expressed relief at getting home from hospital, yet trepidation at having to be the primary carers for their child without medical staff on hand if something went wrong. --- Yeah we were just walking around on eggshells for the first couple of weeks. Yeah you don't know what you're doing, you're just flying by the seat of your pants with them, you haven't got the midwives there to show you what to do or if there's a problem you just can't spin around and ask them. . Participant 2 Father Surprisingly, six of the seven parents reported that living in a rural area may have given them improved access to existing services. They felt they could easily contact clinicians to discuss concerns, and said that living in a smaller community may have allowed clinicians to get to know them on a more personal level. Conversely, four parents expressed concern that the outcome for their child and the support they received may not have been as good had they resided in a more remote setting. --- Major themes Three major themes were identified from the interviews: coping through optimism, stoic survival and striving for normal. Coping through optimism: 'Coping through optimism' was a particularly strong theme across all interviews. Participants often related to very negative parts of their experience in a positive way. For example seeing their child being intubated, a father driving to a metropolitan hospital knowing he was going to miss the birth or having to temporarily leave the child in hospital. --- It just really, really shook us. But in saying that it was also an experience I look back on with fondness too because it did open my eyes and just show me what it was all about and what could happen and what we were up against. Participant 7 Mother Parents generally looked back at the experience in an optimistic way as being difficult but 'character building' and life changing. The parents for whom the premature infant was their first child often said they came through the experience because they didn't know any different. --- Well I guess being my first kids I don't know any different than having a premature [infant]… I've only ever had a premmie baby um yeah so I can't really I don't know, I don't know what a full term would be like.Participant 3 Mother The following extract of a mother meeting her infant for the first time and seeing him being intubated still emphasises the positive aspect of being able to see and spend time with her infant, rather than dwelling on the distress of witnessing the intubation. ….what's happened is you know he's stopped breathing, he's had respiratory failure and all that kind of stuff and all I could see is his little arms kind of getting thrown everywhere cause the doctors are rolling him over... so I think that was a bit full-on but still nice at the same time cause you could just, like once he's settled down and everything I could just sit there and just stare at him and that's all I did for hours, I couldn't touch him cause he was so fragile but um yeah like we could sit and stare. Participant 4 Mother Parents indicated that thinking optimistically was a way of coping with the situation because focussing on the negative aspects would make the experience more difficult or impossible. --- Stoic survival: The theme 'stoic survival' that emerged from the data described families not thinking about the experience in depth at the time or afterwards. Participants stated that they had never discussed the impact of caring for a premature infant with other people. --- ….there was no need to talk about it cause it just brings it up again doesn't it, it brings emotions up and yeah 'cause it was just being normal and not worrying about it, get on with it. Participant 5 Father All parents reported that while the experience was emotionally difficult and stressful, they felt they simply had to do what was needed for the sake of the child or their partner. Five of the seven parents reported having to ignore their own emotional needs in order to be strong for the premature child and the rest of their family. I was eventually wheeled into where he was, and I just broke down and broke down in tears. I just couldn't believe that was my baby, he was just a pathetic sight, I just thought he was pathetic. He just had tubes, machines beeping and going off everywhere and just the tiniest baby that I've ever seen in my entire life and I just, it was too much for me. But I rose above that and overcame the emotion and got on with it. I thought well here I am and they told me what I had to do. Participant 7 Mother This was a strong theme for the fathers who were interviewed. They tended to see their role as being carer for the mother first and the infant second. When asked what he saw as his role in caring for the family one father stated: Oh just for [his wife], to look after her and tell her everything's fine. Participant 5 Father The fathers stated that they were keen to be involved in the care of their infant; however, this was often done by the nursing staff and the mother. The fathers also stated that it was difficult to continue their role as breadwinner, being a husband, spending time at the hospital, attending appointments at later stages, and caring for any siblings, but that it was just something that needed to be done. . Parents reported receiving information from a variety of sources regarding their child's progress. They reported valuing the opinion of an expert who was not emotionally involved with the family the way a grandparent or family friend might be, saying that this assisted them to set realistic goals and expectations for their child. --- Limitations of the study While the parents interviewed provided a vivid account of a wide range of experiences, there were only seven participants. The majority of participant felt satisfied with follow up; however, these responses may vary if participant families had lived in more remote areas with fewer local paediatric services. Parents who were separated or divorced may have different accounts of the way a premature child affected their relationships. Only one of the families had a child with significant long-term health issues. The information provided by this family was surprisingly consistent with those parents whose children had had better outcomes. However, this sample is too small to offer reliable conclusions. Finally, participants' positive responses may have been influenced by the fact that the interviewer was a clinician in the local child development service clinic at the time of the study. --- Conclusions Participants described an extremely emotional, stressful and life-changing experience. Satisfaction was expressed with the services they received in the rural health service, claiming good access to services and reporting that they were able to come to know their clinicians personally. However, the rural town of the study has follow-up services for premature infants with medical and allied health services, and parental experiences may be different where such services are not available, as is the case in many rural areas. Parents reported that the travelling and living away from home while the child was in hospital was an additional stressor; however, they felt that their child had been transferred to get the best possible care in a tertiary facility. Most participants reported that living in a rural area created a financial burden on families at a time of illness, and that families were thankful for services that assisted with this, such as the Isolated Patients Travel and Accommodation Assistance Scheme and Ronald McDonald House. Clinicians working with premature infants and their families should recognise that parents may describe their experience in an overly positive manner and may be avoiding dealing with the experience emotionally as a coping mechanism. This is likely to be the case not only in the initial period in the NICU, but also for many months afterwards. Indeed Holditch-Davis found symptoms of post-traumatic stress disorder in mothers when their child was at a 6 month adjusted age 13 . Health workers should also recognise the unique experiences of the fathers of premature infants and ensure that their needs are met. Educating fathers about how to interact with their child, and discussing how to manage the time demands of work, leisure and family may assist fathers to settle into their new life role. This is consistent with the reported experiences of Swedish fathers of premature infants described by Lundquist and Jakobsson 8 and American fathers, described by Pohlman 7 . Clinicians can assist families to set realistic goals and expectations for their child. This is particularly important for rural families, whose unique situation may need to be managed differently from that of urban families. The bonding process may be assisted by reducing parental frustration from being unsure how to help their child or cope with a child who does not seem to be responding to their efforts. For instance, clinicians trained in child development may notice subtle developmental progress that is not evident to the parents. Further research is needed into the experiences of families where the parents are separated or divorced, and also into the experience of Indigenous families of premature infants, who are likely to represent a largely underserviced group, particularly in rural areas. This research has highlighted the unique experience of ruraldwelling families of premature infants. It begins to illuminate the way families adjust to the shock of the arrival of a premature infant and offers some insight into their coping mechanisms. By being aware of the unique issues confronting families from rural areas, clinicians can be better prepared to meet the needs of rural families in the future.

The birth of a premature infant is a stressful and emotionally challenging time for parents. Families living in rural areas often have reduced access to the specialist services premature infants require, and different concerns from their metropolitan counterparts. Rural families may have to spend extended periods of time at a metropolitan centre while their child is in a neonatal intensive care unit (NICU). Regular return trips to hospital may be required for follow up. Parental experience of this unexpected event and the resultant intensive health service involvement requires further examination; most studies focus on the NICU experience. This exploratory study was undertaken from the rural paediatric allied health perspective. Because there are few accounts of rural families' experiences in the literature, this hermeneutic phenomenological qualitative study focused on the rural parent's experience of having a premature infant in a rural area during the first 12 months of the child's life. Method: Participants were selectively recruited from the case lists of paediatricians in a rural/regional hospital. Seven parents (5 mothers and 2 fathers) of premature infants consented to participate in semi-structured interviews. Interviews were of 60-90 min duration per parent, and were audiotaped and transcribed verbatim. The transcripts were analysed thematically using a phenomenological approach. Process and reflection diaries were maintained in order to provide an audit trail.The participant group reflected a range of experiences. Parents described being initially shocked and then gradually adjusting to being the parent of a premature child. Three major themes emerged: (1) 'Coping through optimism' when parents consistently spoke positively about even stressful and difficult aspects of their experience; (2) 'Stoic survival' where parents did © LT Wakely, K Rae, R Cooper, 2010. A licence to publish this material has been given to ARHEN http://www.rrh.org.au 2 not emotionally deal with their experience, often citing others' needs as having higher priority; and (3) 'Striving for normal' when parents focussed on the aspects of their child's medical care or development that was closer to that of a full term child. Parents described feeling devastated if they felt their infant regressed or was progressing too slowly, and elated whenever a new milestone was achieved. Conclusion: Gaining insight into the experience of parents of premature infants can help health professionals ensure services more effectively meet the needs of these families. Rural families were pleased with their local services although they indicated that travelling to a metropolitan centre was extremely burdensome. While the families interviewed had access to some local specialist services, they expressed concern that if they resided in a more remote area there would be reduced access to services and greater personal strain. Further research is required to determine the experiences of Indigenous families, separated or divorced parents and families living in remote areas.

Introduction --- Background Approximately one-third of people living with HIV worldwide meet the criteria for a common mental disorder , such as depression and anxiety [1][2][3][4]. Among men who have sex with men specifically, the prevalence of CMDs may be magnified, with an estimated 35% reporting a history of suicidal ideation worldwide [5], 43% living with HIV reporting depression [6], and at least one-third in the United States meeting criteria for past-year alcohol use disorder [7]. The pronounced stigma, vulnerability to intimate partner violence, discrimination, and criminalization of same-sex sexual activities that this group confronts worldwide may have multiplicative effects on the development and maintenance of CMDs [8][9][10]. In the context of HIV, CMDs are the important and frequent drivers of suboptimal prevention and treatment outcomes. Depression has been found to be associated with a reduced likelihood of initiating antiretroviral therapy [11], lower ART adherence [12][13][14], increased mortality risk [15], and accelerated HIV progression [16]. Although anxiety disorders, including posttraumatic stress disorder , have been less frequently investigated in the context of HIV, growing evidence suggests that they can reduce the likelihood of achieving at least 80% adherence [17,18]. Anxiety may also increase sexual transmission risk behaviors, suicidal ideation, and cognitive impairments [2]. Among those HIV-, CMDs are linked to hazardous drinking [19], inconsistent condom use [20][21][22], and lower pre-exposure prophylaxis adherence [23], all of which increase the risk of HIV transmission. Smartphones and other information and communication technologies have become increasingly attractive options for optimizing prevention and treatment strategies. Given their accessibility, affordability, and availability, these digital technologies promise to widen access to health care generally and specifically among vulnerable groups by lowering costs and the traditional barriers hindering participation. Mobile health , or the delivery of interventions and other health services via digital technologies, can take the form of SMS text messages, multimedia messages, mobile apps, and social media campaigns, among other options. mHealth interventions have shown promising results in improving medication adherence [24,25], health care engagement [24,26,27], and other health-promoting behaviors such as engagement in physical activity [28][29][30] in diverse populations. Furthermore, mHealth may be an acceptable and feasible approach for delivering care, including mental health care, to MSM both in high-income countries and low-and middle-income countries , including Malaysia, China, India, Thailand, and Vietnam [31][32][33][34][35][36][37][38][39][40]. This is because MSM are often early adopters of technology and may already use such platforms to access health information and search for sexual partners [34,41]. A recent survey of young MSM between the ages of 18 and 24 years found that approximately 70% were willing to participate in a web-based or mobile HIV prevention program, whereas only 1% would attend in-person programs exclusively [33]. As the access to and use of digital technologies do not appear to differ by mental health burden or HIV risk [42,43], mHealth may be a particularly attractive option for engaging underrepresented groups in settings where services are scarce. Given the heterogeneity of risk among MSM [44][45][46], mHealth may further enable the tailoring of mental health and psychosocial interventions to the members' individually varying needs. However, one of the key limitations of the literature on mental health and HIV acquisition risks and treatment is its reliance on cross-sectional data, whereupon participants are tasked to recall symptoms or behaviors unfolding weeks or months prior. Recall can bias individuals to particular events, for instance, by increasing the salience of negative memories when depressed [47], modifying the memory of past judgments when providing new information [48], and skewing estimates of symptom severity based on its peak or most recent occurrences [49]. Similarly, although these data allow for estimations of an event frequency , they provide little information about the contextual factors leading up to, during, or after the event, that is, on the triggers and HIV-related sequelae of depressive-anxious symptom experience. Mobile ecological momentary assessments , or the repeated sampling of individuals in real time during the flow of their everyday lives using mobile phones, can reduce the impact of these biases and provide the high-resolution data necessary to model these relationships longitudinally within an individual. Comparisons of ecological momentary assessment to retrospective approaches using calendars or audio and computer-assisted self-interviews have suggested that the retrospective approaches may contribute to the underestimation of sexual and substance use behaviors by as much as 50% and lead to distortions in specific details, including partner characteristics [50,51]. Comparing EMA reports of patients' experience of side effects associated with antidepressant use with their reports to their general practitioner, 1 study found that although 43% of the patients reported experiencing dizziness in the moment, <20% of them reported this to their general practitioner [52]. By clearly delineating the relationships between mental health and HIV risk-taking behaviors, mEMA approaches may thus contribute to a scientific understanding of what drives differentials in temporary and long-term treatment compliance, outcome, and cessation. This can inform the design of ecological momentary interventions that cater to the dynamically changing risks of individuals across time and context. --- Objective Although many mHealth studies have been implemented in LMICs [53], to our knowledge, only 1 mEMA study has been conducted in an LMIC. Specifically, Soong et al [54] evaluated the feasibility and acceptability of monitoring tobacco use in urban India and found lower compliance than is typical in substance use research conducted in high-income countries . This suggests the need to evaluate how EMA protocols can be adapted to low-resource settings and particularly vulnerable populations to maximize their feasibility, acceptability, and usability for future intervention design. Although there is a lack of consensus on how such protocols should be culturally adapted, qualitative methods can be particularly beneficial in understanding the target population's needs and contexts of use. Integration of the target population's input into the design process can increase the perceived ownership of the technology and optimize the relevancy, uptake, and eventual utility of the platform. Thus, this study sought to determine the feasibility and acceptability of an mEMA app focused on the behavioral and psychosocial linkages between mental health and HIV-related behaviors among high-risk MSM in Hanoi, Vietnam. In Vietnam, MSM continue to experience significant HIV and mental health burdens amidst decline among other key populations [55][56][57][58]. The findings from this study can optimize the design of future mEMA protocols and lay the foundation for EMIs tailored to this population. --- Methods --- Study Design This study recruited participants from an existing pool of MSM who had previously participated in a study examining the association between HIV and PTSD in Hanoi, Vietnam [59]. All participants in the original sample were recruited from sexual health clinics and community-based organizations in Hanoi, were between the ages of 18 and 29, had reported having engaged in anal intercourse with a same-sex partner within the past 6 months, and had a smartphone. From the existing pool of participants who had agreed to be recontacted , a subsample was derived based on the HIV serostatus and probable PTSD diagnosis of individuals, which was indicated by their scores on the Modified Posttraumatic Stress Scale [60]. The criteria for PTSD are dependent on the presence of a criterion A trauma alongside at least one intrusive symptom, three avoidance or numbing symptoms, and two hyperarousal symptoms that have been present for at least one month, as indicated on the Modified Posttraumatic Stress Scale. A total of 4 PTSD-HIV strata were created, and a list of individuals belonging to each group was randomly generated. Individuals were recontacted in that order. Upon enrollment, participants completed assessments of their mental health and sexual and drug use history. The study staff then instructed the participants on how to install a secure web app enabling real-time data collection, KoBoToolbox [61], on their smartphones. Briefly, the participants were first added to the KoBo system and then sent a link to download the web app onto their phone, where it could then be accessed. The web app was powered by Enketo and specifically designed for low-resource settings. Other countries where KoBo has previously been implemented include Vietnam, Indonesia, Ghana, and Malawi [62,63]. This platform was selected as it was free and open source, adaptative in a range of challenging contexts with varying Wi-Fi availability, and had accessible features that could be further expanded and customized depending on user feedback during the exit interview. Data collected through KoBo were stored on a secure server hosted by Amazon Web Services, which provided network and infrastructure security and monitored host and end point security. Only the study staff had access to the data. Screenshots of the app are provided in Figure 1. After KoBo was installed, the participants were asked to complete a sample survey. The study staff were available to answer any questions or concerns. During the next 7 days, the participants were prompted to complete 6 to 8 surveys daily: 1 survey every morning upon waking , 4 to 6 surveys at random time points throughout the day , and 1 survey before sleeping . Most prompts occurred during midday. This decision was motivated by our scientific interest in understanding the dynamic interactions between PTSD and HIV-related risk-taking behaviors. Specifically, one of the features of PTSD is emotional lability or intense, unpredictable, and frequent shifts in emotional experience [64]. Emotional lability can hinder social relationships and functioning and contribute toward risk-taking behaviors. Repeated sampling of emotions, symptom experiences, and social experiences throughout the day would enable us to characterize emotional lability and the trajectory of symptoms and affective experiences leading up to a risk-taking episode. Through this study, we thus aim to test the acceptability and feasibility of such a granular measure of daily experience among a high-risk population. Prompts were structured around participant-provided schedules and delivered randomly in 1-to 3-hour windows; the windows were approximately 30 minutes apart . With each prompt, participants received an SMS text message, reminding them to complete their survey on KoBo within the next 30 minutes. If the participants were unable to respond within that time frame, they were instructed to wait until the next prompt. Each morning and night survey could be completed in approximately 3 to 7 minutes, whereas midday surveys were completed in approximately 3 to 5 minutes. Prompts were uploaded automatically upon completion if internet access was available; otherwise, they were stored locally on the participants' smartphones. At the exit session, participants were interviewed about their experiences using the app, including their motivations for participating, perceived benefits or drawbacks of using the app, any worries or concerns they experienced, and their recommendations for future research. Interviews were semistructured and conducted in Vietnamese by trained research staff. Questions were open ended and intended to elicit feedback on the content, appearance, and functionality of the app. Participants were compensated 150,000 for completing the exit interview, which averaged 20.4 minutes. For each EMA survey they completed, the participants were additionally compensated 7000 . The maximum amount individuals could be compensated for participation in the EMA component of the study was 350,000 . --- Ethical Considerations The study was approved by the institutional review boards of Emory University and Hanoi Medical University . All participants were provided detailed information about the study procedures and expectations, risks, and benefits before their providing of written consent. The participants were told that they could choose to discontinue the mobile surveys at any point. --- Measures In all surveys , participants answered a standard set of questions. At each survey, the participants reported their current location, including perceived safety, and number of social interactions since the last prompt. For their most recent interaction, participants were asked to identify their relationship to that individual and rate the ease of the interaction, the quality of the interaction, and their perception of what the other person had thought of them on a 7-point Likert scale. In addition, participants were asked to rate the degree to which they felt safe on a 6-point Likert scale and report whether they had experienced a range of PTSD and somatic symptoms within the past hour. Emotional states were assessed by asking the participants to rate the intensity of their emotions using the Positive and Negative Affect Schedule [65]. In only the morning survey, the participants were asked about their sleep and ART or PrEP use within the past 24 hours. For the night survey, sexual and drug use behaviors and desires within the past 24 hours were additionally assessed. For each reported sexual activity, the participants answered questions about the sexual act , the sex of the partner, the nature of their relationship , condom use, and drug use during sex. Finally, in the night survey, the participants were asked to recollect, in an audio recording, an event within the past 24 hours that made them tired, anxious, or stressed about the future. They were instructed to freely associate, in detail, the potential causes, consequences, or scenarios associated with this event, including their feelings about what has occurred or could have occurred. Table 1 shows how these different modules were distributed across the morning, midday, and night surveys. Table 1. Modules in the morning, midday, and night surveys administered to young men who have sex with men in Hanoi, Vietnam, between January and April 2019. --- End-of-day prompt Midday prompt Morning prompt Variables ✓ ✓ ✓ Time ✓ ✓ ✓ Location ✓ ✓ ✓ Whether someone else is present and their identity ✓ ✓ ✓ In their most recent interaction, the quality of interaction ✓ ✓ ✓ In their most recent interaction, the ease of interaction Measures of PTSD and somatic symptoms, as well as Positive and Negative Affect Schedule, had previously been culturally adapted using the 5-step translation process by van Ommeren et al [66]: translation, review by mental health professionals, focus group discussions, back translation, and pilot of the measures among MSM in Hanoi, Vietnam. This 5-step approach has been used in several LMICs to maximize conceptual, semantic, technical, and content equivalence of measures in a new population or context [67]. Other items on the surveys were forward and backward translated by 3 bilingual translators. --- Qualitative Data Analysis Interviews from the exit session were transcribed verbatim and anonymized by 2 research assistants and then imported into MaxQDA , a qualitative data analysis software. Using content analysis [68], 2 authors , who were bilingual, then independently reviewed the transcripts and generated a preliminary codebook before meeting to refine primary and secondary code definitions. The codebook contained both inductive and deductive codes. Coding was guided by the Unified Theory of Acceptance and Use of Technology model, which posits that adoption of technology is dependent on performance expectancy or perceived benefit of using the technology; perceived effort or ease of use; social influences, including norms around the social acceptability of the technology; and facilitating conditions or availability of technological, cultural, or organizational resources supporting technology use [69]. Of particular interest to coding was feedback on existing study design and questions, including the perceived relevance of each in the daily lives of the participants, and recommendations for future research design. Once the codebook was established, 2 analysts coded each transcript. The two met regularly to refine the codebook, discuss emergent themes, and resolve any disagreements about coding. After all transcripts were coded, the authors reviewed the excerpts to identify salient themes and select quotations that illustrated the major themes and subthemes. --- Results --- Participant Characteristics Of the 198 participants in the original sample, 50 returned to complete this study between January and April 2019. The four groups differed in their participation rate: 95% of those in the PTSD-/HIV-group who were invited to participate did, 67% among the PTSD+/HIV-group, 54% among the PTSD-/HIV+ group, and 57% among the PTSD+/HIV+ group. Among those who did not participate, 40% had moved and were no longer in Hanoi, 30% could not be reached, and the remaining 30% refused. Among those who agreed to participate, later, 4 participants withdrew from the study after the initial interview because of personal reasons , leaving a final analytical sample size of 46. The sociodemographic characteristics of the participants are summarized in Table 2. On average, the participants were aged 23.5 years. Approximately two-thirds had some university education. Most of the participants earned > 35 million or more annually, which was considered the minimum needed to survive in Hanoi, where the average annual salary was approximately 53 million to 7,600,000 [70], which approximately 59% of our participants either met or exceeded. In the sample, 41% had neither PTSD nor HIV, 13% were HIV+ and without PTSD, 15% were HIV-and with PTSD, and 30% had both HIV and PTSD. Among those with PTSD, the mean score on the PTSD Symptom Scale was 30.3 , whereas, among those without PTSD, the mean was 12.4 . --- Compliance Participants completed an average of 6.9 days of EMA activity reporting. During this time, they each received approximately 38.3 prompts and completed an average of 21.7 prompts. The average response rate was 56.5% but ranged from 0% to 97.4%. Excluding individuals who did not complete any surveys , the average compliance rate was 61.8% . Among these individuals, the response rates were 64.5% on the first day, 67.8% on the second day, 66.9% on the third day, 71.6% on the fourth day, 64.7% on the fifth day, 61.8% on the sixth day, and 37% on the seventh day. Demographics, group membership , and the number of prompts the participants received were not significantly associated with the response rate. Table 3 depicts representative quotes of each qualitative theme identified across groups of varying levels of compliance . --- Qualitative Results --- Relevance Overall, participants considered questions about their social interactions , mood, levels of stress, and health, including their experience of PTSD symptoms and medication use, most relevant to their daily experiences. Participants favored these items as they thought these items provided a dedicated space for daily reflections and made several participants feel as though someone cared for them. For these reasons, the audio recording was often singled out as the most meaningful survey component as it allowed participants to elaborate, in their own words, what had transpired that day: [After the recording], I felt a lot more comfortable, as though there was nothing more I wanted to share. --- Whenever I can share everything I want, I will feel lighter. [My mind] still feels heavy, of course, but my mental state will feel better than if I were to just dwell on it and keep it inside. [20 years, PTSD+/HIV-] In some cases, the audio recording was seen as even therapeutic as it allowed participants to share distressing thoughts or feelings that they otherwise would have concealed. One of the participants contrasted the freedom he exercised in sharing details of his life in the app with the caution he felt when disclosing to friends: Others thought that despite their infrequent use of substances, the questions were still relevant as they were still exposed to drug-related activities in their everyday lives. Several participants felt that they should be able to document those events as those exposures can make them feel uncomfortable or increase their desire to use. [I] feel As many questions focused on PTSD and negative affect, some participants felt that the prompts were more appropriate for someone in greater psychological distress than them. For this reason, some suggested the addition of more positive questions . Furthermore, as multiple questions measured the same construct , items were sometimes seen as repetitive and requiring a more careful read by participants. Some participants believed that the emotion measures, for instance, required them to parse their affect too finely. Instead of having items such as hopeless or happy, several participants favored using a Likert scale ranging from negative to positive: --- Benefits of Using the App Nearly half of those who used the app said that they believed it increased their self-understanding. In particular, many believed that using the app enabled them to better understand their emotions and the individuals or situations that elicited those feelings. In doing so, some participants believed that the app made them more truthful about their emotions and what specifically they had accomplished that day: When I do --- Worries and Concerns Participants reported a few worries or concerns associated with using the app as they perceived the survey as similar to other questionnaires they had completed before and because of the detailed briefing they received before starting the study: Actually Others expressed privacy concerns related to the accessibility of their surveys. In particular, some participants worried that others might see their screen while they were answering sensitive questions about their sexual behavior or medication use, although only one of the participants experienced a situation in which someone had seen his responses: On --- Usability Participants found completing the initial EMA survey with a research staff helpful. Most participants did not experience any difficulties in navigating the EMA form. Participants struggled the most when uploading an audio file. Depending on the model of their phone, some were not able to record and upload directly onto KoBo but had to use a third-party platform to save the files. Others were also initially confused about the kinds of files they could upload and submitted photos instead. When asked whether they would use the app outside of the study, most participants said that they would but that their use would depend on how busy they were. Some emphasized that their use of the app was largely motivated by their current interest in mental well-being: When asked what features would encourage their long-term use of the app, a few said that they would feel more encouraged to use the app over a longer period if they could see daily or weekly trends in their responses or were given personalized recommendations to improve their health or general well-being . Most participants emphasized that their continued use depended less on new features than on changes to the existing study design, including the number and scheduling of the prompts: It --- When I am available, I will do it. When I am not, then I won't do it because it'll be too much. But if I'm free, then what else am I doing but playing [Candycrush]? [23 years, PTSD-/HIV-] Others thought that their willingness to answer the prompts was variable throughout the day, depending on their mood: When I --- Recommendations When asked to provide recommendations to increase future compliance and better tailor content to their everyday experiences, 5 main themes emerged. First, there was an emphasis on supplementing the initial training with detailed reminders on how to use the app, as the participants reported initial confusion when completing the first few surveys on their own and having to contact the study staff for assistance. Second, the wording of the daily prompts and study questions should be made as personable as possible. Although the current phrasing was deemed easily comprehensible, it was also seen as too clinical. Where possible, the participants thought the text should be more MSM friendly, prompting people to share their everyday experiences rather than strictly completing a survey. Having interactive features was also found to be critical in increasing engagement. Third, although open-ended questions took more time to answer, they were considered more experientially relevant than the Likert scale responses. Integrating optional open-ended questions throughout the day was favored over having only the audio recording before sleep. The option to type rather than narrate the responses was considered very important, as some participants thought that the audio recording was inconvenient when others were nearby. Although generally preferring open-ended questions, participants also felt that all survey items, including the audio recording, should be optional. Reflecting on his experiences, one of the participants shared his belief that many Vietnamese people needed to be eased into doing the audio recording as it did not come naturally: --- It's because people feel embarrassed. Why? The audio recording is like a diary, except you write in a diary; here you have to record, retell. It's [related to] personality type, like, you might feel embarrassed because you're more introverted. It makes you feel autistic . Personally, I feel too embarrassed to say aloud my inner thoughts. Writing is okay, but I fear sitting alone and talking aloud like that because when I do I feel as though I'm being autistic ( ). Singing is okay, but talking aloud like that is similar to splitting yourself in half [to sit there, look at yourself, and judge yourself]. [26 years, PTSD+/HIV+] Fourth, most participants found the number of assessments acceptable but took issue with their scheduling. Although participants were told that they did not have to answer the surveys when busy, many reported feeling obliged to do so and actively monitoring their phones for prompts. Participants recommended sampling only once or twice during prime hours and potentially lengthening the time frame they had to respond. This was seen as particularly useful when participants' schedules deviated from those they provided at enrollment, which often happened because of the types of work young MSM were engaged in. Finally, participants recommended reducing the number of questions to prevent future participants from answering the prompts haphazardly . This could be done by combining similar questions, occasionally changing the order of questions to make the surveys more interesting, and introducing new questions. Having access to aggregate data, particularly those that compared participants with the group average, was seen as potentially motivating, alongside receiving custom health information: [The app] keeps asking these questions, everyday these same questions. I think it needs to be a little different, so there's enthusiasm to answer. [28 years, PTSD-/HIV-] --- Discussion --- Principal Findings This study found that mEMA is a feasible and acceptable way of monitoring dynamic interactions between mental distress and HIV risk among young MSM in Vietnam. Overall, participants found the questions pertinent to their everyday experiences of psychosocial stress and were willing to use the app in the future. Participants identified items related to PTSD symptoms, emotions, and social interactions as most relevant to their everyday experiences. Most preferred having open-ended questions to elaborate on their stressful encounters, as they found doing so therapeutic or personally informative. Some expressed concerns about data privacy. Although the number and length of EMA surveys were considered potential hindrances to long-term participation, the impersonality of the app and the scheduling of the prompts rather than their frequency were identified as primary barriers to participation. Participants recommended that the language be made more MSM friendly and that future iterations of the app include personalized end-of-the-week reports to encourage long-term use. Given that our sample was highly exposed to trauma, identification of barriers to feasibility and acceptability was important. Our compliance rate was noticeably lower than those reported previously in similar studies with MSM [71][72][73]. Although our assessment period was relatively shorter , we sampled participants more frequently and included more questions because of our interest in acquiring a thorough understanding of how PTSD interacts with HIV risk. This may have reduced the compliance rate. In addition, young MSM may introduce unique challenges. A recent meta-analysis of EMA studies found that age was the most significant predictor of compliance rate [74]. In particular, although the response generally declined over time, this decline was steeper among younger participants compared with that in older participants. Smiley [75] reported a response rate similar to ours among young gay and bisexual men (aged 21-25 years, which they attributed to the participants' work schedules. In the context of Vietnam, these issues may be further exacerbated, as several participants had highly variable work schedules. For instance, one of the participants did not answer any of the EMA prompts as shortly after enrolling in the study, he was assigned to a mining post in rural Northern Vietnam for the study duration and had limited access to his phone. Cases such as these lowered our overall response rate but may accurately reflect the challenges typical in implementing EMA protocols in LMIC contexts. Incentivization structures can motivate compliance. Doherty et al [76] highlighted the use of extrinsic and intrinsic incentives to motivate participants' EMA engagement. Extrinsic motivation includes monetary remuneration per survey, compliance monitoring and feedback, and emphasis on participants' contributions to science during enrollment. In particular, providing feedback to participants during an EMA study, whether through reminder messages or weekly reports, has been found to increase their perception that someone cared and thereby motivate compliance; however, this may also contribute to reactivity [77]. In addition, although financial incentives are useful, they can also contribute to selection bias [76]. This was a concern that our participants also voiced, as some thought that the length of the survey, alongside financial incentivization, might encourage participants to answer haphazardly . Coupling extrinsic incentives with features that cultivate intrinsic motivation may reduce these unwanted effects. These features enhance or support participants' inner desire to engage with the app, independent of momentary awards. For instance, Hsieh et al [78] found that participants who had access to visualizations of their data had a 23% higher compliance rate than those who did not. Similarly, the attitudes of research staff during enrollment and participants' ability to access their own response rate have been shown to motivate participation [79]. In our study, interest in mental health was a particularly strong motivator. Participants also perceived having an MSM-friendly language as highly important. A recent meta-analysis of mHealth interventions for young sexual and gender minorities suggests that these linguistic issues are often not well attended to; across studies, the most common concerns centered on language, specifically on prompts that were seen as too text-heavy, patronizing, or superficially targeted toward LGBTIQ+ people [80]. Among our study participants, an intrinsic motivation evidenced throughout multiple interviews was the desire to better understand and regulate their mental and emotional states. Our concern that mental health stigma may lead participants to underreport experiences of mental health and psychosocial distress was largely unfounded [81][82][83]. For some participants, mental health stigma may have even enhanced the perceived attractiveness of the app, as it provided them an outlet to share details they otherwise could not. These findings underscore the value of testing assumptions regarding acceptability in the wild. Most of our participants readily looked to the app for mental health and psychosocial support information, which is encouraging for digital mental health interventions. However, this inadvertent reliance on the app for mental health and psychosocial support contributed toward a second set of concerns about reactivity or when the frequency, intensity, or quality of a target measure change as a result of monitoring [84]. Specifically, several participants reported that they intentionally used the app to track and regulate their emotional states; some reported becoming more aware of everyday triggers. Whether this altered the experience of mental distress or the relationship between mental distress and HIV risk behaviors is unclear. Variable effects have been documented in the literature. Although some studies report an improvement in mood or change in risk-taking behavior among participants using EMA [85,86], others have found no significant reactivity effect [84,87,88]. Given the scope of this study, we were not able to determine whether the frequency, intensity, and correlations between measures varied significantly over time. Responses to the survey further highlighted some of the cultural specificities that might contribute to differences in the perceived acceptability of an app in different populations. Notably, our sample strongly preferred open-ended questions about their life but disliked the option to self-record as they found the behavior-when done alone-to be indicative of , which is highly stigmatizing within the country. As mHealth and telehealth approaches are becoming more widely adopted worldwide [89], the question of how to culturally adapt and enhance the acceptability of these applications has become imperative. However, caution must be exercised in interpreting differences as necessarily cultural in nature. The feedback we received on the assessment of emotional states is illustrative: studies of emotion worldwide have demonstrated cross-cultural differences in emotional granularity or the degree to which individuals differentiate emotional states and identify them with precision and specificity [90]. Anthropological investigations of emotional experience worldwide have further demonstrated how conceptualizations and management of emotions differ worldwide [91,92]. These differences point to the need not only to investigate the cultural appropriateness of EMA measurement approaches during the app design and evaluation process but also to evaluate alongside this the degree to which experiences of the app reflect differences based on mental health status. Specifically, emotional granularity has been shown to be reduced in a number of mental health conditions, including PTSD [93]. It is possible that those with higher PTSD symptom severity may experience greater difficulty in recognizing, differentiating, and labeling the different emotions they experience. Building in the means to tease apart the potential causes of difference is critical to tailor EMA protocols appropriately to local contexts and populations. --- Study Strengths and Limitations This study is innovative in several ways. To our knowledge, this is the first study to examine the feasibility and acceptability of using mobile technology to model the dynamic relationships between mental distress and HIV risk in a low-resource setting, and it is the second EMA implemented in an LMIC [54]. The study demonstrates the potential of leveraging such technologies not only to complement existing prevention and intervention strategies but also to advance EMIs that cater to the dynamically changing risk of individuals across time and context. Furthermore, compared with similar feasibility and acceptable studies, the sample size for this study was larger than those that have been conducted among high-risk populations such as MSM, people living with HIV, and people with psychiatric disorders [71,73,94], allowing for a more granular analysis of how adherence to, tolerance of, and experience with EMA protocols differ in this population. Recruitment of individuals with both PTSD and HIV proved particularly challenging not only because many refused to participate but also because many either had physically relocated or been unreachable at follow-up. Of the 4 individuals who withdrew from the study, 3 were also from this group. Although it is not possible to discern the reasons that motivated this decision, future research should investigate how psychiatric comorbidities of HIV may affect the receptiveness of MSM to varying demands of mEMA research and explicitly assess the reasons for dropout across those different study designs. The primary limitations of this study include the generalizability of the findings and potential reactivity. First, although the sample size was larger than comparable studies, participants were recruited from community-based organizations and clinics and, thus, present a subpopulation in Hanoi that accesses sexual health services and are engaged in care. Participants were additionally drawn from a previous study among those who had agreed to be recontacted; therefore, these individuals may be more inclined to volunteer in research. Compared with the national average, our sample was also more highly educated and had a higher income [95,96]. Concomitantly, all our participants owned a smartphone and, therefore, may represent a subpopulation that is more comfortable using mobile apps than their counterparts. This is unlikely to be representative of older Vietnamese MSM in urban settings and those in the country more generally among whom HIV risk and mental health burden may be both more stigmatized and pronounced [83,97,98]. Thus, the generalizability of these findings to other MSM should be taken with caution. Second, although participants noted that they had become more aware of their emotions and were using self-reports to monitor and regulate aversive triggers, it is inconclusive whether using the app actually changed the frequency with which they engaged in risk-taking behaviors or experienced particular PTSD symptoms or emotions. Previous EMA research has found minimal reactivity to EMA [84,99]; however, given the stigma associated with both HIV and mental health issues within Vietnam, sharing otherwise concealed details of their lives to the app may have heightened therapeutic effects. --- Conclusions Our study illustrates the feasibility and acceptability of mEMA studies among high-risk Vietnamese MSM. Participants were receptive to and invested in future studies and interventions. The findings of this study can inform the design of future EMA studies to optimize relevance, usability, and acceptability. Achievement of the Joint United Nations Program on HIV/AIDS [100] 95-95-95 targets by 2025 requires the integration of such novel technologies with existing prevention and intervention efforts. Given their ubiquity, mEMA studies hold tremendous promise for furthering our understanding of the proximal mechanisms potentiating HIV risk, including the contextual and psychological conditions under which they occur, which are conditions difficult to recall later on. Such information is essential for the design and delivery of EMIs that can efficiently and effectively provide resources to when, where, and whom they are most needed. --- Conflicts of Interest None declared. --- Abbreviations

Background: Men who have sex with men (MSM) are at a disproportionate risk for HIV infection and common mental disorders worldwide. In the context of HIV, common mental disorders are important and are frequent drivers of suboptimal prevention and treatment outcomes. Mobile ecological momentary assessments (EMAs), or the repeated sampling of people's behaviors and psychological states in their daily lives using mobile phones, can clarify the triggers and HIV-related sequelae of depressive-anxious symptoms and contribute toward the design of ecological momentary interventions (EMIs) that cater to the contextually varying needs of individuals to optimize prevention and treatment outcomes. Objective: This study aims to characterize the feasibility and acceptability of mobile EMA among high-risk MSM in Hanoi, Vietnam. It aims to evaluate the perceived relevance, usability, and concerns of this group with regard to the content and delivery of mobile EMA and the potential of leveraging such platforms in the future to deliver EMIs. Methods: Between January and April 2018, a total of 46 participants were recruited. The participants completed 6 to 8 mobile EMA surveys daily for 7 days. Surveys occurred once upon waking, 4 to 6 times throughout the day, and once before sleeping. All surveys queried participants' perceived safety, social interactions, psychological state, and mental health symptoms. The morning survey further queried on sleep and medication use within the past 24 hours, whereas the night survey queried on sexual activity and substance use and allowed participants to share an audio recording of a stressful experience they had that day. At the end of the week, participants were interviewed about their experiences with using the app. Results: Participants completed an average of 21.7 (SD 12.7) prompts over the 7-day period. Excluding nonresponders, the average compliance rate was 61.8% (SD 26.6%). A thematic analysis of qualitative interviews suggested an overall positive reception of the app and 5 recurring themes, which were centered on the relevance of psychological and behavioral items to daily experiences (eg, mental health symptoms and audio recording), benefits of using the app (eg, increased self-understanding), worries and concerns (eg, privacy), usability (eg, confusion about the interface), and recommendations for future design (eg, integrating more open-ended questions).

Sometimes you just feel like you need to escape; to get away from the busyness of the city, to a place where nature has greater presence. And so it was that I found myself owning a boat, a means of escape. A substantial body of literature has developed over the last two decades, across multiple disciplinary fields, asserting that contact with nature is good for us. The implications are clear: in an increasingly urbanised world, there is value in maintaining green and blue space in our cities. Access to such space benefits our mental health and wellbeing 1 , restoring us from stress and mental fatigue 2 , increasing our cognitive abilities 3 , improving our emotions and moods, and increasing life satisfaction 4 . It benefits our physical health, boosting cardiovascular and endocrine systems, and immune function 5 . Healing times are reduced, as are incidences of respiratory illness, allergies and obesity 6 . Access to green space provides social benefits, facilitating social contact and social cohesion, and reducing health inequalities 7 . Among children, access to natural spaces facilitates learning and development 8 : through exposure to actual and perceived risks 9 ; through physical and emotional challenge 10 ; and through engagement of the senses to encourage movement, developing motor skills in the young 11 . Such findings have begun to inform policy 12 , and to appear in the popular press 13 . Nature is good for us, and that fact is becoming common knowledge. For some extolling the virtues of green/blue space, at core is EO Wilson's notion of biophilia 14 ; the idea that humans are inevitably and necessarily attracted to life. However, the majority of this literature refers only to access to green/blue space, 'natural environments' and 'contact with nature'. Nature unexamined; an unproblematic, all-encompassing, non-human Other 15 . Yet the concept of 'nature', in its binary relation with the category 'human', is culturally specific, characteristic of Anglo-European thought 16 , and perhaps uniquely so 17 . A cultural myth. The literature on nature and human wellbeing is biased towards the northern hemisphere, dominated by research in Europe and the United States 18 , such that the 'us' in the paragraph above is also culturally specific. Perception of both separation from and the need for more contact with nature are cultural constructions, functions of a shared idea of 'nature'. And yet separation from, and contact with, nature are also experienced by individuals, perceived as real, illustrated by the opening lines of this paper. There is potential for cultural geographers to interrogate the ways this culturally specific 'separation from' nature is experienced, thereby contributing to a more nuanced understanding of urban green/blue space and the benefits of nature. In this paper I seek to offer an alternative depiction of the experience of 'green/blue' space, via a phenomenological account of small boat sailing. I take inspiration from a growing body of work exploring embodiment, emotion, affect and the materialities of the world; geographies that emphasise more-than-cognitive knowing in a more-thanhuman world. A key impulse in this work is a sense that representation can never adequately convey a world that is so much more; that disjuncture, an unbridgeable gap, exists "between the language of the world and the language of the word." 19 The fullness of human experience in the world cannot be captured in words alone. Geographers have drawn on the relational ontologies of theorists such as Deleuze and Latour to better achieve a sense of humans-in-the-world; alongside, combined with, and shaped by non-human others of all kinds. 20 Non-representational 21 and more-thanrepresentational 22 geographies have been at the forefront of this drive to rethink the constitution and experience of place, space, politics, environments and events: the doings of the world in their multifaceted, tangible and intangible wholeness. These theoretical developments lend themselves particularly well to accounts of outdoor places and spaces, and encounters with/in/through non-human nature. Geographers have addressed an increasing array of such situations, from walking , mountainbiking 25 , rock climbing 26 or rescuing 27 in mountains and uplands to kayaking ecotours 28 , bodyboarding 29 , surfing 30 and other geographies of the sea 31 . Social scientists working in sport and physical culture have concurrently developed similar interests 32 , seeking to access the 'unrepresentable' aspects of triathlon 33 and scuba-diving 34 experiences, and to understand experiences of 'nature' through physical activity programmes 35 or through river rafting 36 , canoeing, sea kayaking 37 and windsurfing 38 . I thus intersect with this literature on outdoor activities. My approach is autoethnographic or 'autophenomenographic' 39 , attending to my own experience through Merleau-Ponty's phenomenology 40 . Some would say geography has moved beyond phenomenology; the embodied, affective, non-representational geographies referred to above having been badged 'postphenomenological' 41 . Despite the phenomenologists' attempt to dismantle subject/object distinction, phenomenology retains a perceiving subject at its core, the existence of others granted only by this perceiving subject 42 . Yet I wish to state a claim for the value of phenomenology. Merleau-Ponty specifically draws attention to nature's excess over the perceiving subject, always inexhaustible 43 . Following this through, as I aim to demonstrate, the water environment itself simply will not allow its agency to be disregarded. For some this may not decentre the human subject enough. But if we are to understand how human subjects experience the nature/culture binary, human perception is of central concern. Sundberg 44 questions the logic of trying to challenge the Anglo-European dualistic nature/culture ontology by continuing to draw on Anglo-European traditions of thought, of which Merleau-Ponty's phenomenology is one example. However, Toadvine's reading of Merleau-Ponty emphasises that the sedimented habits of perception always incorporate the historical, cultural, economic, gendered, ethnic situation of the phenomenologist. A phenomenological account of nature cannot step outside of this: "if there is any access to nature, even as alterity, it will only be through our cultural mediations rather than apart from them" 45 . The account of nature produced in this paper is very much situated within Anglo-European, and particularly British, culture. I thus try to make minimal use of the universalising terms 'we' and 'our', unless highlighting that they should be read as situated within this cultural context. If sedimented habits of perception are culturally specific, this opens up the possibility that there is still scope for Anglo-Eurocentric theories such as Merleau-Ponty's to further reveal the ways in which this culturally specific nature/culture ontology becomes manifest. Rather than presenting a direct challenge to the nature/culture binary, then, I seek here to shine new light on this binary, highlighting the embodied experiences through which 'nature' is encountered. The next section explains the methodological approach of the paper, and provides some contextual background. A phenomenological account of being on a boat follows, focusing on themes of embodied knowledge, embodied spatial awareness, and encounters with nature as other-thanhuman. Finally, I return to the nature/culture binary and consideration of urban green/blue space. --- Doing boat things The Rivers Tamar and Plym, in South West England, meet the sea at Plymouth Sound . Geomorphologically this is not a sound but a ria, a river valley drowned by rising sea level after Britain's last glacial phase. Between the two rivers lies the city of Plymouth, a city with enviable access to the water. The submerged river channel offers deep water for naval and commercial vessels within the natural harbour of the Sound. A nineteenth century breakwater provides additional protection from southerly winds. With the rivers either side, the city has something like 16km of waterfront. Alongside military and commercial uses, Plymouth Sound and its rivers provide a setting for a multitude of water-based recreational activities, on , in , under and around the water: a big, 'blue' playground on the city's doorstep. This research did not start out as research. In July 2010, having lived in the city for some years, I acquired a boat 47 , with minimal prior experience of sailing. The boat in question is a 27-foot motor sailer. With small sail area and an in-board 36 hp engine, optimum performance is achieved through motor and sail power combined. I rapidly discovered that boat-owning entailed participation in an unfamiliar lifeworld, one that I would adapt to in a multitude of ways. This included conscious and cognitive adaptation: accessing training in a variety of aspects of boating life to help ensure safety; learning technical terms, yacht parts and actions associated with them. But there was also a less conscious, more-than-cognitive dimension to this lifeworld, and it was this that was the most profoundly different, in the sense of challenging the norms of my terrestrial urban dwelling. This sense of 'newness' led me to want to record the experience, with no particular end goal in mind. I did this through a series of diary entries, written on computer as soon after events as was practicable; usually either once I returned home that day or the next. I wrote whatever came to mind about places, events, feelings and doings, trying to capture as much as I could. 'Taking the plunge' and acquiring a boat with little prior experience of sailing offers some justification for the empirical and theoretical approach. A cornerstone of phenomenology lies in epochē and bracketing 48 , attempting to set aside prior assumptions and knowledge about a phenomenon, to approach it anew. Being new to sailing, I had little in the way of preconceptions to set aside. Vannini asserts that "it is precisely during the first one or two years of dwelling in a new place that a certain acuteness of observation and introspection is at its best, before things become too taken for granted" 49 . By extension, it is when new to an activity or a practice that its peculiarities are most evident. My sailing is restricted to April through October each year 50 , a seasonality that places further constraints on my capacity to become familiar, and for this reason my account draws on my first five summers of sailing, 2010-15. The focus here on embodied experiences of sailing is undoubtedly a product of developing familiarity with embodied, affective and non-representational geographies through that time period. Analysis involved initial reading of the entire file of diary entries, identifying emerging themes. Paying considered attention to embodied experiences of sailing, identifying a sense of contrast with my urban everyday life, then led me towards literature on urban green and blue space. This 'experience first' approach raises the charge of autoethnography being no more than introspective indulgence, an 'academic selfie' produced by those privileged to have the time, inclination and education to afford it 51 . Such privilege is important to acknowledge. I may not have had many preconceptions about sailing, but my academic interests have always centred around nonhuman 'nature' in one way or another. Still, this should not entirely negate the value of autoethnographic research. Academics are embedded in, and products of, culture as much as anyone else. My desire to 'get out of the city' reflects a milieu of influences: a mostly rural childhood and early adulthood; social context, working among colleagues who value outdoor experience; intellectual interests; but also a dominant cultural conception of nature as 'out there', away from the urban, free from human interference, sublime. And the sense of any narrative is formed in the telling 52 ; in others' responses to, and self-recognition in, the material 53 . Triangulation thus occurs gradually, through the responses of others 54 , including readers of this paper. Ultimately, I aim here to demonstrate that understanding experiences of nature requires attention to the character of the space dwelt in, as much as to the presence of non-humans. Now, to the boat: --- Being on a boat Small boats require a particular way of being; cultivating and practicing a certain kind of body 55 . --- May 2011: I remember well how useless I felt in those first weeks of owning the boat. How clumsy. I couldn't do anything right -I couldn't do anything at all. I kept banging my head, every time we were on the boat, every time I tried to do something inside the cabin or move from one part of the boat to another. I still bang my head sometimes, but now I am reasonably capable on the boat. It's like I have learnt, but it's not just a cognitive how-to, it's just as much an embodied how-to. May 2013: Saturday morning, we were on our way down the River Plym towards the Sound. The boat had just gone back into the water after some extended winter maintenance, and we had tied the halyards up to prevent them clanking on the masts while we worked in the cabin. I went on deck to rearrange them. One had become looped around the radar reflector, way up the mast. It looked problematic. I detached the two ends of the halyard from the mast, stepping back with one in each hand. Arms spread, holding the ends apart to make sense of what was where, my eyes followed the lines upwards. I raised my left hand slightly and flicked it, sending a ripple up the line. That unsnagged it. It struck me almost immediately that I hadn't thought at all during that process. I looked up the line, saw, and moved. That was it. Somewhere my brain must have processed what I saw and turned it into action 56 , but I wasn't conscious of that. There were certainly no words, no "If I…". See: Move. Merleau-Ponty's phenomenology foregrounds embodiment. Our being-in-the-world is an embodied being. The body is the means by and through which the world is perceived and, equally, the world is our means of understanding our bodies. Perception and understanding thus inhere in the bodyworld intersection, our knowledge of the world predicated upon our bodily capacity to act in it. "Consciousness is in the first place not a matter of 'I think that' but of 'I can'." 57 I did not 'think that' the halyard could be unsnagged from the radar reflector if the line were moved in the right way; I understood that I could do it, experienced the "harmony...between the intention and the performance" 58 . Understanding is embodied, and Merleau-Ponty thus challenges the Cartesian dualist ontology of body/mind, which situates the body as 'object' against a cognitive 'subject'. He similarly challenges dualist notions of body/environment separation. The body-subject experiences the world and changes in response, primed to experience the world again -and so we develop 'habit', an understanding of how to, through the sedimentation of experience in our bodies. Not being used to sailing, my normal state of being was as a mobile entity in static spaces. Being in/on any form of transport changes that, but living 59 on the boat is a bit different. On the boat I am doing everything one might do in a caravan -cooking, eating, drinking, washing dishes, putting things in cupboards -as well as boat-specific functions such as getting fenders out of lockers or putting sails up or down. Being a mobile body in a restricted and moving space requires different corporeal knowledge. I initially did not understand how to be on a boat, and so my capacity for 'I can' was limited. I was clumsy, unable to function. That changed over time, as the constant motion of the boat and the movements through which I could function in/on it became sedimented into my body, became habit. This is not to deny the role of cognitive, 'scientific' knowledge. Such knowledge helps to make sense of, and anticipate, the water environment: August 2012: We head upstream in the afternoon and drop anchor in one of our favourite places, a big meander. We sit at the back of the boat, looking around, always slightly unsettled when first anchored. A successful anchorage is a coming-together of so many things: anchor placement; chain length; water depth; bedforms; tide changes; current; wind. If we want a good night's sleep we need to be confident that it is right. I look up-and down-river, taking in the meander, then comment: "We're too close to the edge". When the tide drops we will bottom out on the steeply sloped edge of the currently submerged channel. We need to be nearer the middle. We were starting to drag the anchor any way. So we move further into the river, set anchor, drag, set anchor again -got it this time. When the tide fell we could see that we had been right to move. A depth-sounder is an essential piece of technology, revealing changing depths of water beneath the boat. But it shows changes in depth only in the direction in which the boat is moving. If we happen to be following a contour, the slope across that contour is invisible. Interpretation is thus necessary and, in the vignette above, it was my prior cognitive knowledge of fluvial geomorphology that gave me some sense of where the thalweg would be. But the implications return to embodiment: bottoming out on a steep slope would mean an uncomfortable night, at best. At worst the boat could end up on its side. Similarly, scientific knowledge in the form of weather and tide forecasts are central in sailing. Knowing that wind and tide are in opposing directions means anticipating choppy waves, a much bumpier journey than when the two are aligned. Understanding scientific knowledge entails understanding corporeal implications. Meaning and materiality are inseparable 60 . --- Boat spaces Being on the boat is to be in a confined space that is almost always in motion 61 . Coming to understand how to be on the boat, developing the habit of being on the boat, the body must take up both the motion and the space -or, rather, the space-in-motion. It is this that becomes sedimented into conscious and unconscious actions. Rather than being in space, our body inhabits space 62 . But at the same time as inhabiting this space-in-motion, through that very motion the body responds to space beyond the boat: May 2011: A brilliantly sunny weekend in Salcombe, we are on a visitor mooring in the area known as 'The Bag', out of sight of the town. The place is busy, visitor moorings full. --- It's evening and I am in the cockpit cooking dinner. Sitting on the motor housing, which offers a convenient seat from which to use the hob, I have two saucepans on, rice bubbling in one and a chilli underway in the other. Another boat approaches. The man at the helm clearly realised that I was cooking: he slowed down to pass us, minimising the wash from his boat. I had seen him coming and so already had my hands on the pans to steady them, but I smile and wave thanks, appreciating his consideration. We rock gently from side to side -boat, people, saucepans -with the familiar clanking of the open cabin door rattling on its hook. Just as the body/environment binary is problematic, the boat/environment binary breaks down. The lived space-in-motion of the boat is never bounded from its surroundings. The example above acknowledges human agency, but the motion experienced derives from a combination of the ways these two particular boats move in, disturb, and respond to, the water, along with the state of the water itself. Always the water. The constant movement of the boat ensures that non-human agency is ever-present, dominant. Sea: Move. This extension of body to boat is also significant in the boat's mobility through space, its functioning as a mode of travel. Here the boat demands a different spatial awareness than I was used to. I was familiar with the spatially restricted territory of car or bus, constrained to lanes, a narrow ribbon of permitted travel rendering the space of passing encounters with other vehicles largely predictable. The space of boats -the water space -is different. Other boats approach from all angles, varying in speed and intention. While there are rules, not everyone is familiar them. Every encounter requires interpretation. Constant checks through 360 o quickly become the norm as the possibilities for encounters are less bounded than in grounded motorised travel. A different lifeworld makes different spatial demands. And then there is the 'invisible beneath'. The attention required of the car driver is predominantly two-dimensional: in front and behind; left and right at junctions. On a boat, the third dimension of 'below' becomes much more important, yet is largely hidden from view 63 . In the immediate vicinity, the space beneath is rendered visible by the digital display of the echo sounder, and when this space is limited, the instrument becomes an obsession: 0.9m, 0.8m, turn one way, 0.6m -turn the other way quickly…The possibilities of movement across the visible two dimensions are constrained by the invisible beneath. Beyond the immediate location of the boat the chart encodes the terrain, revealing routes that may be possible, places accessible, at certain times only, dependent upon the tide. But whatever is below the boat can never be fully be conveyed by such technologies. Just occasionally the water yields glimpses, hints of the unknown: May 2013: The river was becoming a slow upstream flow, the tide having turned just an hour or so before. We were now travelling upstream with it. We passed the first of the huge naval ships, docked on the city side of the river to our right, alongside the enormous sheds of the dockyard. Cruised between the ferries that cross the river on their clanking chains, shuttling people from Devon to Cornwall, Cornwall to Devon, day and night. More naval vessels, more dockyard sheds and cranes to our right , the river to our left littered with small boats on their moorings. Suddenly I hear "Woah!" and our engine is dropped out of gear. "A dolphin! It was dead in front of us!" We let the boat drift as we scan the water, looking, looking, looking. Sure enough, it appeared off to our port side, maybe 15-20 metres away. The curved back, the dorsal fin, its tip ragged, damaged. Slowly, so slowly, it arched through the water and disappeared, then reappeared in another slow arch, its long nose just breaking the surface momentarily. Another arch, and another. I resisted the urge to jump in the water 64 . It seemed to turn back towards us and I hoped, hoped it was going to come closer again. But it disappeared. Gone. It seems odd that such a large animal is in the river here. This busy river, here at its most industrial, always full of boats, full of noise, full of people. And that slow arch of the dolphin, slipping silently out of view. Incongruous. We were still smiling half an hour later. June 2013: Suddenly there was a sickening 'thunk' and a lurch as we hit something beneath the water. Knock the engine out of gear, check over the stern: nothing there. In gear again, and the propeller was still working, nothing tangled around it. We circled, scanning the water, searching for any sign of what we might have hit, any explanation. We saw nothing. As we continued on our way, I checked inside the cabins, inside all the lockers, for water ingress, but all were dry. Moments like these -momentary sightings of dolphins, seals or jellyfish, or the scrape of anchor chain shifting on riverbed, reverberating through the hull in the silence of night -reinforce a sense of the water environment as a "different world" 65 . This is nature as 'Other' in terms of unknown, and that which we can never know; the "impenetrability of non-human interiority" 66 . For Merleau-Ponty, space is central to being, embedded in existence. Perception of space is shaped by this existence, by our corporeal means of inhabiting space 67 . Here lies the connection between the space-in-motion of the boat, the spatial awareness demanded by mobility of the boat, and the sense of otherness of what may lie beneath. As a means of travel, the boat extends the body's potentiality of volume, the "demand for a certain amount of space" 68 , just as a car or other vehicle does 69 . But the restrictive potentialities, delimiting the space available, are very different in/on the water from on the road. On the surface, there is less restriction -and hence less predictability of interactions -on the water than on the linear space of a road. But the restrictive potentialities of the water environment are resolutely three-dimensional, the presence of that additional dimension being reinforced by the relative unknowability of the underwater environment. And this difference in spatial awareness demanded by the water is compounded by the vagueness of motility on/through water. The boat does not respond with the linear determinism of a motor vehicle, which you can point in the direction you want to travel. The boat crabs sideways with the tide, slews suddenly with a change in current, lurches in the waves. Steering and stopping are more vague, less precise: negotiation-with, rather than control-over. This renders the potentiality of volume, the corporeal demand for space, uncertain -again reinforcing the demand for a different kind of spatial awareness. The space-in-motion of the boat, the spatial awareness demanded for mobility on the water, and the sense of otherness of what may lie beneath, all contrast with my 'normal' means of inhabiting the world, and so give a sense of Otherness. --- Changing space June 2011: A sunny day, sitting at the back of the boat, the endless, lazy sea stretches to the horizon. The waves can be traced from a distance, coming closer, closer, closer and then the rise and fall of the boat over this one, while my eyes begin to trace another. I'm aware of other boats on the water, all some distance off. I'm aware of the cliffs in the distance. I know where the navigation buoys are in relation to them, if I think about it. But the dominant feeling is of big, endless space. Time to raise or drop a sail, and I'm up on deck. There is only me, the task at hand, and the moving surface of the boat, rising, dropping, rocking one way and then the other. Loosen the halyard from the mast ; lurch to the front and drop the sail, guiding it to the deck ; stagger back to the mast to tie the halyard off ; stumble forwards and onto my knees to undo the knots attaching the sheets to the sail ... For these moments, nothing exists beyond the edge of the boat. The horizon, other boats, the cliffs, all gone. The physicality of the job to be done reigns in my vision, my attention, my awareness of the space around me. July 2012, Salcombe to Plymouth: the swell was coming up the estuary, so we knew it would be bouncy once past the sand bar. And how bouncy! But the little boat was fine, heading straight into those rollers. Around Bolt Head, heading westwards, the waves were not quite as high but still a 'moderate' sea. It rained. Wipers on. This was our first time at sea in the rain. No other boats were in sight. Three miles from land. Just the two of us and the boat. Our boat. I felt safe, insulated, despite the motion. Up, down, up, down, with the whir of the wipers filling the wheelhouse. Then a wiper blade fell off. It was the one directly in front of the wheel. We knew we had a spare in the cabin. Neither of us spoke for a few seconds. Like we were avoiding it. Then, "Are you OK to go up on deck?" I nodded. We agreed that I should wear a safety harness this time, go up through the front hatch, and hook on before I climbed out. As always in this kind of sea, once I am on deck nothing else exists. Up, down, side to side, I slowly lurch along the wet, moving surface of the boat. On my knees on deck, directly in front of the windscreen, wiper blade in hand. How many attempts to line it up to the wiper arm? Everything is moving. One, two, threenearly that time -eventually I get it there. It takes some effort to click it into place, to get the 'push' in the right direction, given the motion of the boat. Then retreat back along the boat -still on my knees, it's easier -towards the hatch. It is not until I am back in the safety of the wheelhouse that I realise how wet I am. Psychologists have a concept of 'attentional narrowing' 70 , whereby emotional arousal -often interpreted as stress, fear or anxiety 71 -impacts on cognitive performance. This can have opposing effects 72 : arousal may slow down performance of intellectual tasks; or may result in reduced processing of task-irrelevant information, the person becoming more focused on what they need to do. The latter seems to correspond to my experiences here. However, two points are of note. Firstly, the psychological literature emphasises emotion 73 . Phenomenologically, what dominates these experiences for me is not emotion but the corporeal challenge of negotiating the task in a space-in-motion. Perhaps the psychological processes of my brain's functioning are associated with emotions of which I am less conscious, but if that is the case, they are rooted in the materiality of the situation. Secondly, this shift in my attention is inherently spatial 74 . Perception of space, as a function of our embodied inherence in the world 75 , changes with bodily circumstances. The water affords a particular perception of space at any one time, just as much as being on the water demands it. Small boat sailing, then, in coastal and estuarine waters, renders present a sense of nature as 'Other' in terms of agency and affordance. Echoing Hollenberg's 76 analysis of Hemingway's The Old Man and the Sea 77 , the sea irrupts into consciousness to point out human limitations. It does not allow human freedom to 'master' nature. --- The Other of green and blue spaces Taking to the water certainly did offer a means of getting out of the city, to places where 'nature' has greater presence. Nature is ever-present in many different ways, including some I would never have anticipated. The mysterious Other of the 'invisible beneath' is the most obvious and predictable; the presence of a lifeworld, a mode of being that, as human body-subjects, we cannot know. The invisible beneath always exceeds perception and yet must be attended to. But being on the water, coming to understand the space-in-motion of the boat-in-environment, means the presence of an Other can always be felt. The agency of water and weather is always there, in the unending motion, communicated through the boat and thus through the body. And these embodied spatialities of being, the lived perceptions of space, are central to this sense of Otherness. This encompasses visual, and haptic, somatic sensations of space 78 with cognitive; the perceiving subject's combined, multidimensional understanding of being in space. The lived space-in-motion, the vagueness of motility through the water , the different spatial awareness demanded by mobility on the water, and shifts in spatial awareness resulting from the water/boat motion; all are different from terrestrial urban living. To encounter nature on the water is to experience space differently. As Merleau-Ponty illustrated, the extraordinary can reveal something of the everyday. Returning to the homogenising-yet-culturally-specific 'we': it can be argued that normal daily experience, perhaps particularly for urban residents , is of linear space; mobility constrained by the material and social structures of roads and streets, and rules associated with them 79 . Even our field of vision is often constrained into linear pathways by buildings, walls and other boundaries. In the midst of a city, Euclidean geometric space is largely invisible. In leisure, the places we encounter and enjoy nature most powerfully are often open spaces, in the form of parks, National Parks, lakes; places where vision and motion is mostly not constrained in such a linear way 80 . Our civilised lives have been linearised, contrasting with the unlinearised spaces of recreational and therapeutic places. If our separation from nature is a cultural construct, it is not just an idea. It is materially constructed in the fabric of our towns and cities, not only in the artefacts of bricks and mortar, but in the way we order space. Separation from nature is embedded in how we experience space. Returning to the benefits of nature, the past decade has seen a move towards relational perspectives to understand therapeutic landscapes and spaces. This entails recognition that therapeutic space is an emergent outcome of "a complex set of transactions between a person and their broader socio-environmental setting" 81 . Some of this literature 82 offers echoes of the arguments presented here, research participants describing their experiences of coastal spaces as contrasting with enclosed, linear, urban and artificial spaces. Within this therapeutic spaces literature, Gorman 83 highlights a need to consider what constitutes the 'wild' of the wild spaces. His focus is non-human participants, and I wish to extend this further: we also need to pay attention to space. Not just the two-dimensional, mappable space often assumed, but to the lived experiences of space, the embodied spatialities of being. We encounter 'nature' through our bodies, and embodiment is inherently spatial. Foley and Kistemann 84 argue that the incorporation of nature into health policy is hampered by "limited understanding of which components of nature deliver which health benefits". Such reductionist thinking seems to dominate the 'benefits of nature' literature, but the encounters I describe here suggest that experiences of 'nature' are complex, diffuse and holistic. The key point is that the Western cultural idea of human separation from nature has become sedimented into our habits of perception, part of our embodied experience of the world. To understand green or blue space as 'natural', we may need to understand dwelt perception of the space as much as of the 'green' or 'blue'. Cultural geographers surely have much to offer. Gabriel. Big thanks for Ben Garlick for incisive and insightful feedback during the revision. I am also grateful for the constructive criticism and suggestions from the reviewers and editor. This paper would not have been possible without my partner-in-crime, Kevin, or without the boat. --- Author biography Pauline R Couper is Subject Director: Geography at York St John University. With a longstanding interest in philosophy of geography, her research has spanned multiple forms of knowing nature, from the scientific to the embodied .

2018) The embodied spatialities of being in nature: encountering the nature/culture binary in green/blue space. Cultural geographies, 25 (2). pp. 285-299.

Introduction In this article, I outline findings from a research project on young people's involvement in organised crime. The project was commissioned by the UK Home Office to respond to the Serious and Organised Crime Strategy's commitment to develop educational resources about young people and organised crime . My role in the project was to facilitate consultations with young people and practitioners, support the development of the educational resources and conduct an evaluation of the associated toolkit. i This article does not represent the official view of the Home Office, but draws on my own analysis of the data collected, and is not focusing on the toolkit itself but on some of the key themes that emerged from the research with practitioners and young people and their implications for policy. These consultation and evaluation activities offer a contribution towards some much needed research on the topic of young people and organised crime, particularly in a UK context. The consultations employed qualitative methods and the evaluation involved mixed methods . The themes outlined here are primarily from the consultation activities with reference to the evaluation where it builds on these themes. The article is divided into two main sections. In the first, I explore international academic and policy literature that is relevant to young people, organised crime and youth crime prevention. This includes the challenges in defining organised crime, what research tells us about the scope of young people's involvement and their motivations, and a critical review of policy approaches to tackling the problem of organised crime. In the second half, I outline the methodology and discuss key themes arising from the research with practitioners and young people. I argue that young people's involvement in organised crime is complex and not as simple as making a conscious choice to become involved. In particular, I demonstrate that young people are vulnerable to exploitation by organised criminal groups and that factors such as their sense of responsibility towards their families and lack of alternative employment, may add to the risk factors that make them vulnerable. The paper concludes with a review of the policy implications. --- Part 1: Definitions, Research and Policy Contexts --- Definitional challenges One of the main challenges in understanding the causes of organised crime and evolving policies for prevention, lies in the very definition of organised crime itself. A review of the literature highlights major variations within and between policy-maker and academic usage of the term, and a degree of conceptual blurring between 'organised crime' and 'gangs'. Organised crime is defined in the Home Office's Serious and Organised Crime Strategy as 'serious crime planned, coordinated and conducted by people working together on a continuing basis. Their motivation is often, but not always, financial gain' . However, in the same document, serious crime is defined as 'crime, which may not always be "organised" but requires a national response, notably many aspects of fraud and child sexual exploitation' . Consequently there is some overlap: organised crime is also serious crime, but serious crime, is not always organised. Child sexual exploitation, for example is understood as serious but not necessarily organised. The interpretation in the Strategy is that organised crime involves larger networks than something that operates at just the local level, and that it involves a number of layers around the organised criminal group at the centre and is usually national and often international in reach. Therefore, organised crime as defined by the strategy is distinct from 'gangs' as we know them at the local level . A gang is not, in itself, an organised criminal group as per the Home Office definition. This definition does, however, recognise that existing gangs may be utilised by organised criminal groups, for example, to distribute drugs within a local area. Academic research, has challenged the policy definition of organised crime as highly organised and carefully structured by a 'mafia' type group at the centre. Hobbs , for example, suggests that the networks are 'messier' with less of a clear hierarchy than popular definitions suggest and that the perpetrators have been over-glamorised. Based on ethnographic research in East London, Hobbs suggests that what we typically view as organised crime is usually more opportunistic than the term suggests. Similarly, Hopkins et al suggest the term is ambiguous and has generally been used to create categories around structures of association and to a lesser extent structures of activity . Arguably, it is difficult to establish what constitutes organised crime and how it presents itself, due to the 'moral panics' and false assumptions that exist and are communicated by the media, politicians and other commentators . Von Lampe recognises that there is no consensual definition or understanding even among academics. As such, there are varying forms of crime included in academic discussion of organised crime but with some overlapping themes and interpretations. He suggests that definitions of organised crime have drawn on three main categories: the type of criminal activity; the particular criminal structure or organisation; or on organised crime as a form of illegal governance . Within these categories there are different interpretations and understandings. In terms of activity, there are different conceptions of what forms of activity are, in fact, organised crime and the level of continuity or sophistication of these activities. The understandings of organised crime that centre on the group or structure differ according to how large or organised a group needs to be to constitute organised crime. The forms of illegal governance that have been understood as organised crime vary according to their motivations; for example, whether they are financial or political. Von Lampe proposes a conceptual framework for understanding organised crime that incorporates all three categories and the various interpretations that exist within these, rather than attempting to find consensus on one specific definition. The definitional challenges outlined above become further complicated when reviewing the literature on gangs. As with organised crime, it is difficult to find a consensual definition of 'the gang' and some academics suggest that the 'gang problem' has been over-stated with numbers overinflated . Similarly, academic research suggests that there is also no definitive answer as to how far gangs overlap with organised crime groups. Pitts suggests that different forms of gang exist in the UK, some of which do involve organised crime. However, there is a lack of conclusive research determining whether youth gangs might evolve into adult organised crime groups. Muncie argues there is little evidence to suggest that they do -with youth gangs being fairly loose, unstable and temporary overall. Indeed, he claims that in both the UK and the US, those gangs which might be more aligned with organised crime and those that engage in 'serious violence and threats to assert control locally', are adult gangs rather than juvenile ones . Densley , however, argues, that gang crime is increasingly morphing into organised crime, becoming more structured and developing illegal business-like enterprises. He suggests this is happening less so in London than other areas of the UK. It has been recognised that gangs appear more structured and business-like in the US than in the UK . However, there are questions over whether even these American gangs are structured and continuous enough to constitute organised crime . Hallsworth and Young outline a model of 'street collectives' that distinguishes the 'gang' from organised criminal groups. They divide street collectives into three main groups, these being; peer groups, gangs, and organised crime groups. In doing so, they emphasise the differences between these collectives demonstrating that groups may be mislabelled as gangs when they are, in fact, either a non-criminal gathering of peers or more organised criminal groups. They distinguish particularly 'between the more organized and professional elite of criminals who by and large have an "off-road" presence, from the far more volatile "on road" presence that gangs which occupy the periphery of this underworld typically exhibit' . This emphasises that gangs may be on the peripheries of organised crime groups but are distinguishable from them. --- Who becomes involved in organised crime and why Whilst there is scarce literature on organised crime and young people in the UK, a research study commissioned by the Home Office on criminal careers in organised crime does offer some useful statistics on a range of demographic factors including age . As with Muncie's research outlined above, this study showed that young people were not highly engaged with organised crime. In fact, organised crime offenders were found to be older than general offenders with the average age at which the 'inclusion offence' was committed for the specific category of organised crime being 32 years old . Only 1% of organised crime offenders were under 18 years of age . However, it should be noted here that the methodology for Francis et al's study applied strict rules in terms of which offences needed to be committed for the person to be considered an organised crime offender. The research focused on 'criminal careers' in organised crime and was thus concerned with those for whom engagement is long-term and not temporary. Given the varying definitions and understandings of organised crime, there are also a wide range of arguments about why people become involved and the extent to which the structure and purposes of particular organisations may have a role to play in understanding the reasons for, and nature of, belonging. For Von Lampe these might include whether factors such as profit-making or violence are essential features of the organised criminal activity, as well as the extent to which, organised criminal groups engage people through fear and manipulation or through the promise of social, financial or other benefits. Hopkins et al suggest that the links between organised crime and fatal violence in the UK have largely been over-stated. They argue that fatal violence is more a feature of organised crime in other countries such as Mexico and the USA, than it is in the UK. Corsino found that for Italian organised crime groups in Chicago, the opportunities provided by social networks and social capital were more efficient than violence in keeping people engaged. The literature on gangs also highlights the importance of social networks as key motivating factors. For example, a study by Joseph , emphasises the role of relationships in young people becoming involved in gangs, with 'close friendships' as significant in young people's exposure to gangs and their social proximity to gang activity a key factor. He also found that young people's involvement with gangs was motivated more by the social capital that the gang provides rather than simply economic benefits, especially where they face social or educational exclusion more widely. This is supported by White and Cunneen who also recognise that young people facing institutional exclusion may seek identity and belonging in gangs and that a social motivation to become involved exists alongside an economic one. In the Netherlands, however, Kleemans found that a combination of social relations with violence and manipulation supported people's involvement in organised crime. Within the UK, a recent report by the National Crime Agency has highlighted that 'in some cases, individuals will have been groomed, exploited or coerced into [organised crime] as a means of re-paying a debt or supporting friends and family' . And research by Sturrock and Holmes , into the links between missing children and gang involvement emphasises the vulnerability of these young people and that, as such, they become key targets for exploitation by gangs and criminal groups. In sum, reasons for young people's involvement in organised crime are complex and contested and cannot be understood simply as a conscious choice to become engaged. Within the policy relating to organised crime there is no specific strategy relating to young people though there are some particular commitments within the overall strategy such as protecting young women from FGM and developing educational tools to prevent young people's involvement with organised crime . Serious and organised crime policy does acknowledge that vulnerability to exploitation and grooming is a factor in people's involvement with organised crime. --- Policy approaches to tackling youth involvement in organised crime: a critique --- Some individuals are knowingly drawn into criminality; others, such as victims of Modern Slavery, can be unwittingly seduced and groomed and then exploited. These factors, left unabated, enable the 'stock' of serious and organised criminals to be replenished or backfilled despite police and law enforcement disruptions. The UK's National Crime Agency also recognises the role of grooming, exploitation and coercion in individuals being drawn into organised crime . Whilst this vulnerability is recognised and Prevent is a key area of the Serious and Organised Crime Strategy, the priorities and interventions that follow have largely focused on pursuance with the key priorities being the disruption of organised crime activities, the reduction of organised crime offenders with a focus on prosecution, and ensuring the legal powers are in place for such . Home Office research has indicated that organised crime costs the UK £24 billion per year and reducing this cost is identified as a clear priority . Whilst organised crime policy does not have a key focus on young people, there is plethora of policy documentation on gang involvement and youth violence. The government's current key priorities for 'ending gang violence and exploitation' include: • Tackle county lines -the exploitation of vulnerable people by a hard core of gang members to sell drugs • Protect vulnerable locations -places where vulnerable young people can be targeted, including pupil referral units and residential children's care homes --- • Reduce violence and knife crime -including improving the way national and local partners use tools and powers • Safeguard gang-associated women and girls -including strengthening local practices --- • Promote early intervention -using evidence from the Early Intervention Foundation to identify and support vulnerable children and young people • Promote meaningful alternatives to gangs such as education, training and employment Whilst these priorities clearly recognise young people's vulnerability to exploitation, many of the significant interventions remain largely punitive. Even within these supportive-seeming priorities responses include, for example, under the final priority above, reductions in the threshold for 'persistence absence' from school . Similarly, the significant government intervention of 'gang injunctions' is acknowledged as being for protection from as well as prevention of gang-related violence. However, they are a predominantly punitive tool that places prohibitions on the young person, with custodial sentences the likely outcome of breaking these orders . The research literature on young people and organised crime, highlighted above, would suggest that policy-makers need to engage with the complexities that underpin involvement in such crime, and take a critical approach to the notion of 'choice' in the context of structural disadvantage and individual vulnerability. As Sturrock and Holmes in their research on the links between missing children and gang involvement, have stated: A]dolescent-specific risks ….need new approaches in terms of safeguarding, interventions and social care. Children and young people who experience these risks are too often criminalised rather than receiving a safeguarding response, and the risk-taking behaviour that is an inherent part of adolescent development is interpreted as a 'choice' with young people 'putting themselves in harm's way'. It is essential that criminal justice, gangs and children's social care services do not work in silos but work in partnership to provide the most effective response. However, critical policy research has shown that moral panics, myths and lack of consistent agreement around organised crime and gangs specifically, has affected policy and practice interventions with young people . A key theme in this literature is that policy-makers in the UK have tended to over-simplify and hence distort the problem of young people and crime, with negative consequences for the types of intervention that have been promulgated and resourced. This is seen in the policy responses discussed above which, whilst recognising vulnerability to coercion, focus dominantly on punitive interventions. Joseph , for example critiques what he perceives as a 'one size fits all' approach to prevention work around gangs in London. He argues that prevention work needs to take account of individual and cultural factors, rather than assuming a pre-packaged intervention can be created. He argues that attempts to create an official definition of the 'gang' by academics and policy-makers in the UK have exacerbated the problems of a homogenous approach to practice that has led to misplaced interventions . Pitt's work adds important insights to this debate. He argues that both popular and policy discourse on youth crime has been framed by a 'culture of entitlement' which has served to distort understanding of young people's involvement in offending and been used to justify more punitive responses. According to Pitts, this 'culture of entitlement' discourse is levelled only at the working classes and is contrasted with a 'culture of merit'. The discourse of entitlement ignores inequalities and suggests any differences in levels of success between different groups of young people is due to moral rather than structural differences. Therefore, young people's involvement in crime is seen as stemming from an immoral sense of greed and an entitlement to have what they want. Pitts highlights how this discourse shaped the dominant response to the English riots of summer 2011. For example, David Cameron, the then Prime Minister, stated in a speech immediately following the events: Irresponsibility. Selfishness. Behaving as if your choices have no consequences. Children without fathers. Schools without discipline. Reward without effort. Crime without punishment. Rights without responsibilities. Communities without control. Some of the worst aspects of human nature tolerated, indulged -sometimes even incentivised -by a state and its agencies that in parts have become literally de-moralised. Lightowlers , similarly, suggests that the response to the riots took the view of the young perpetrators as simply morally depraved whilst ignoring the inequalities and deprivation they faced during a time of austerity following the fiscal crisis. Fundamentally, approaches to youth involvement in organised crime and its prevention need to recognise the complex needs of young offenders rather than viewing them as simply criminal. There has been a dominant focus on the threat of penal consequences in youth crime prevention both in the UK and more widely. However, the effectiveness of this approach is being increasingly questioned and this critique offers some insight into the problems with the policy responses to youth gangs and organised crime in the UK. Darke , for example, argues that a 'prevent as enforcement' approach has been over-used in youth justice and is ineffective. The 'law and order' discourse that focuses on the crime and not the vulnerabilities of the perpetrator is also critiqued by Hughes . In research comparing the approaches to youth crime and anti-social behaviour prevention in England with that in Victoria, Australia, Hughes found that the two localities have very different models. The 'law and order' discourse dominates in English policy and practice, focusing on the crime and its penal consequences, whereas in Victoria, the young person and their needs is the focus . Hughes concluded that the approach in Victoria is more effective at preventing crime. Arthur has also found that approaches to Youth Justice in Ireland and England disregard young people's needs whereas the Scottish approach is more focused on the welfare needs of the young people. Case and Haines advocate for a 'Children First, Offenders Second' approach to Youth Justice work. Whilst their discussion is more related to response than prevention, working with young people after an offence is committed, their Positive Youth Justice model is worthy of note here and has clear implications for a more progressive approach to the prevention of young people's engagement with organised crime. Case and Haines argue that young offenders need to be recognised as children and not treated as 'mini adults'. They refer to this tendency in youth justice to view the response to youth crime through an adult lens as the 'neoliberal responsibilisation' of young people. These approaches that focus on young people's needs and vulnerabilities could offer effective solutions to dealing with young people's involvement with gangs and preventing their engagement with organised crime by offering more supportive interventions that current responses such as gang injunctions as discussed above. As part of a more progressive approach to young people's engagement with organised crime, a focus on using positive relationships may be essential, especially as policies such as gang injunctions cut young people off from their social circles and even their families. The literature suggests that positive relationships are important in both the prevention and response to youth crime. Creaney advocates for a 'relationship-based practice' in Youth Justice that uses positive relationships between the professional and the young person as a tool for change. This is supported by Drake et al who also argue that effective practitioner-young person relationships are crucial to the effectiveness of interventions, as well as by Lemmon's research in Australia, which suggested that young people at risk of engaging in crime need long-term support rather than short-term interventions for prevention to be effective. --- Summary Overall, the literature explored suggests that an approach to crime prevention through enforcement and a focus on penal consequences is ineffective. It reveals that current interventions, particularly in the UK, have been based on some problematic and distorted assumptions about, for example, why young people engage with crime and how easily terms such as 'organised crime' and 'gangs' can be subject to universal and over-simplistic definition. It has also found that relationships and support are considered essential to Youth Justice and crime prevention if it is to be effective. As the literature reveals that the focus of interventions has been misplaced, it is arguable that young people need to be consulted as to what interventions would have impact. This reinforces the need for more research around young people and organised crime. Case argues that the qualitative accounts of young people should be sought in defining the risk factors for their engagement with crime. This justifies the importance of the study outlined in this paper which goes some way to gather the qualitative accounts of young people and practitioners who work with them so that preventative interventions can be authentic and effective. --- Part 2: The Home Office Project and key findings --- Aims and Methods The purpose of the research carried out with practitioners and young people in England was to support the development of the educational resources that the Home Office had committed to create in the Serious and Organised Crime Strategy. It was conducted in three main stages; consultative, developmental and evaluative. The consultative stage involved both practitioners and young people. Two semi-structured day-long workshops were conducted with practitioners who work with young people in a range of sectors including youth work, youth offending, education, secure institutions, probation, police and social work. One workshop took place in London and the other in Liverpool. Forty-two practitioners attended these workshops. They involved activities designed to gain the practitioners' qualitative feedback on their knowledge and experience of young people's involvement with organised crime. These were run as workshops rather than focus groups due to the high numbers of participants in the sessions and the activities allowed for smaller sub-group discussions. Three unstructured focus groups were conducted with young people in London. One group was approached via an open access youth club session, another via their youth offending worker and the other through a targeted youth club session for young people considered to be 'at risk' of educational or social exclusion. Through these groups, 19 young people were consulted. The sample of young people therefore included both young people in the primary target group for the preventative resource and those who had already offended. The developmental stage of the research drew primarily on an advisory group of seven practitioners who commented on various edits of the resource, as well as consulting with the young people to whom they had access. The evaluation of the final toolkit involved: two surveys sent to practitioners to look at and/or trial the toolkit; four focus groups with practitioners in London and Greater Manchester; and four observed sessions of youth practitioners using the resource with young people in Greater Manchester and London. One hundred and forty three practitioners took part in the evaluation and the toolkit was trialled with two hundred and sixty young people. . This research was conducted in accordance the ethical protocols of the British Society of Criminology. In particular, it was made clear to all young people that there was no obligation to participate and that, should they take part, their confidentiality would be respected. The discussion in this article draws on the qualitative themes from the consultative stage of the research to explore young people's and practitioners' perceptions of young people's involvement in organised crime. It also draws on the qualitative elements of the evaluative phase where there is further evidence of the themes. There are some challenges in drawing academic conclusions from an applied project. However, due to the lack of empirical research on the topic of young people and organised crime in the UK, this study goes some way to illuminating the key factors and impacts, to be built on with further research. --- Research Themes As stated earlier, the Serious and Organised Crime Strategy is divided into four sections; Protect, Prevent, Pursue, and Prepare. The commitment to develop educational resources is located under the 'Prevent' section of the Strategy. Therefore, the project to develop educational resources and the associated research began from this starting-point. During the consultations, 'drug running' was identified as the most prevalent link to serious and organised crime for the young people that practitioners have contact with. Child Sexual Exploitation was also discussed but was considered to be less prevalent. As the commitment to develop educational resources had been framed as a 'Prevent' issue rather than a Protect issue, it was decided that 'drug running' was the most appropriate focus for the toolkit. However, it emerged throughout the research with practitioners and young people, that the distinction made in the Strategy between 'Prevent' and 'Protect' was too simplistic and that young people's involvement with organised crime was far more complex than such crude distinctions allowed for. Four key themes emerged from the research: 1. The role of grooming/exploitation 2. Opportunity vs. austerity 3. Protecting family 4. Supportive interventions from trusted professionals I explore each of these in turn below. --- Theme 1: The role of grooming/exploitation A key message from young people and practitioners involved with the project was that when young people become engaged with activities that are on the fringe of organised crime, they are often unaware of the full extent of what they are becoming involved with. In particular, young people who become involved with 'drug running' do not tend to know whether or not it is linked to an organised criminal group . Practitioners, particularly police, and young people said that the young person would not have met or be aware of the organised crime offenders at the centre of the OCG but would be recruited by someone local, often who they see as friend. This 'friend' would have spent a substantial amount of time grooming the young person so that by the time they are urged to become involved with the organised criminal activity they would see it as helping them out or often 'owing them a favour'. One youth worker stated 'those above you don't want to put themselves as risk so they pass it to runner. The runner may not always see it this way'. Practitioners illustrated this blurred line between victim and perpetrator in the consultation workshops: --- Figure 1 -'victim/perpetrator' flipchart extract from practitioner consultation workshop in London During the consultations with young people, a thirteen year old female described a situation in which an older male known to her within her community had asked that she 'walk with me like you're my sister for a minute'. This was not a request she felt she could refuse and it was only afterwards that she realised she was acting as a form of protection, to minimise him looking suspicious to the police who were further down the road. For this young woman there was no big choice or dilemma to respond to and the example demonstrates the subtle way in which a young person can first become involved with drug running. Another example emerged during the evaluation phase of the project when a youth worker working in an area of Manchester known to have high levels of organised crime explained how in their local community, a struggling single mother had been given a new sofa by someone whom she saw as a family friend. Following receipt of the gift, this man had then told her he needed her son to do some work for him. Young people from the youth offending group in the consultative phase also spoke about how, by the time they became involved in drug running, the person who recruited them had often given them free gifts over a period of time such as takeaways and second hand smartphones. During a focus group as part of the evaluation phase, teachers at a school for young people with special educational needs explained that young people with learning difficulties have problems in discerning what a real friendship is and were particularly vulnerable to exploitation. The teachers relayed many concrete situations where this had happened, suggesting that young people with particular vulnerabilities are deliberately targeted. One teacher gave the example of a how young person with special educational needs was groomed into gang involvement then punished when he tried to disengage: He got drawn into a dead end by his so called mates. The gang was waiting for him and beat him badly because he wouldn't do what they asked. The young people in the youth offending group who had been involved with drug running linked to a wider network stated that getting out was not easy., with a majority of the group saying that they would need to move away from the area. One young person suggested that even this would not be enough especially if you still had 'debt' and that safety payments and 'looking over your shoulder' would be essential. The group discussed how getting into debt was something they were potentially set up for by those higher up the criminal chain so that they were trapped. They recognised the futility of being trapped in this way. One young man in the youth offenders group explained how being in debt to the people above him had stopped him from being able to disengage from drug running after it began to have a negative impact on his family. A young person during the evaluation sessions stated 'It's a no-win scenario, either the police or the gang will get you'. This was reinforced by a police officer in the consultations with practitioners who was aware of situations in which 'runners' were robbed by people within the wider OCG networks in order to keep them indebted to the group The practitioners who were supporting young people recognised the power of peer influence and grooming but felt that older young people might be role models as well as negative influences. They were able to use this in their preventative interventions by facilitating activities and discussions where older, respected young people from the same area spoke to the younger people. --- You've got a lot of them who want to impress [young person X] because he's got a reputation. But he's a lot more mature now, he knows he doesn't want that. If he says something they listen. The practitioners identified this relatability and real-life experience of such role models as significant. --- Theme 2: Opportunity vs. Austerity Whilst recognising that young people were vulnerable to exploitation as outlined above, the practitioners in the consultation workshops suggested the following motivating factors for young people's involvement in organised crime: money, social status, sense of belonging, power, dignity and protection. They also recognised that in the current economic context there is a lack of alternative opportunities for some young people. --- Figure 2 'why young people become involved' flipchart extract from practitioner consultation workshop in Liverpool Similarly, the young people from the youth offenders group cited peer pressure, boredom, desire for independence and lack of opportunity, aspirations and self-belief as their reasons for becoming involved with crime. Status in the local area was cited by the young people's consultation groups as a lure into becoming involved. As well as money, other material incentives such as access to the latest smartphones were seen as linked to involvement. It was felt across all the consultation groups that the idea of quick, easy money was attractive -especially in a context where the possibility of employment feels elusive. This also emerged in the evaluation where one practitioner stated: The biggest practice challenge is how to offer young people in deprived areas with few support networks real alternatives to 'making a bit of money from running' and then supporting young people to walk away from criminal activity when they have so little to lose, especially if they already have or are likely to gain a criminal record. The young people from the youth offenders group also expressed frustration at the government and other institutions suggesting that getting involved was, at least to a small extent, a way of 'kicking back at the system'. --- Theme 3: Protecting family Both young people and practitioners emphasised that young people who become engaged with activities such as 'drug running' often do so, not just to make money for themselves, but also to provide for their families. Practitioners in the consultation workshops said that young people knowing the potential impact of engaging with organised crime on their family would act as a more powerful deterrent than the threat of penal consequences. --- Figure 3 -'what young people need to know' flipchart extract from practitioner consultation workshop in London A police officer suggested having younger siblings taken into care was often a reason for young people to want to change. This was reinforced by the evaluation phase of the project. For example, one practitioner fed back via the targeted evaluation survey that 'Young people found the film quite realistic and said it made them think about the impact of serious crime and mentioned especially the impact on their families'. It was also recognised by the practitioners that the young people have to continue living and surviving in their home context between practitioner meetings and interventions which may not immediately address material or other needs: You go to a meeting where all the agencies get together, people say things, you meet a month later and nothing has changed, a lot can happen in a month, a child has got a lot to live with and put up with in that time. Most of the young men in the young offenders group stated that their own involvement with drug running was, at least in part, motivated by a desire to protect and provide for their families, particularly their mother and younger siblings, and that they felt a responsibility to bring money into the family home. The impact on family was cited across the consultation groups as a key motivating factor to want to become uninvolved with organised crime. For example, when discussing what would enable preventative resources to have a real impact, young people suggested seeing family members injured or killed would be a deterrent. This reflected a discussion they had of other videos and resources relating to crime that had affected them, with the most 'shocking' being deemed the most 'impactful'. As with the practitioners, they emphasised that a focus on the consequences of involvement with crime on their families was more powerful than the vast number of preventative resources and interventions that focus on penal consequences. One young man in the young offenders group said that seeing his younger brother become involved in drug running was a critical moment for him wanting to disengage. One young woman explained that having a baby had been the key factor in her own desire to change. --- Theme 4: Supportive interventions from trusted professionals The most concrete suggestions from young people for what someone might be able to leave their involvement with organised crime for were employment or education opportunities, although they recognised a need for support to achieve this outcome. One young ex offender was keen to emphasise that drug running required sales skills which could be transferable and that selling on a street-market was a good next step. The young people in one consultation group discussed being supported to access NVQ or other vocational qualifications at college as reducing the risk of them engaging with crime. The young people from the young offenders group also highlighted the benefits of becoming involved in other pro-social activities and particularly highlighted a music project their youth offending worker had set up. The young people were very enthusiastic about this project as something to do, a way to express themselves and as a means of building confidence and aspirations. The practitioners recognised a need for interventions that support young people to reflect and think critically about their futures. One youth worker stated: That young man has got a lot of experience of all that, there's a bit of reflection for him there. His cousins have a got a reputation, and he's got to decide whether to live up to it or walk away from it. Practitioners in the consultation workshops appeared to view themselves as positive role models for young people and youth workers and teachers were the professionals they particularly identified as role models. They also recognised the role that less formal role models play citing family in particular but even, in one case, the local barber. --- Figure 4 -'role models' flipchart extract from practitioner consultation workshop in London The young people identified the support of a professional that appears to be 'on their side' as an important factor to getting out of involvement in crime. The young people in the young offenders group unanimously praised the impact of their youth offending worker as helping them to change. A young woman with a baby cited her relationship with her youth worker as a key support in changing her life. However, it was becoming a mother that was her key motivation to change. Overall, it appeared that the combination of a 'critical moment', the appropriate professional support and the presence of a positive alternative was the most likely situation in which a young person would pursue a pathway out of organised crime. --- Implications of the findings As outlined earlier, only 1% of the UK's organised crime offenders were found to be under 18 by Francis et al . Therefore, it could be questioned why there is a policy focus on them in relation to organised crime at all; and why preventative resources were deemed necessary. However, as discussed earlier, the methodology of Francis et al's study was such that young people engaging with fringe activities would not classify as organised criminals. Lower level activities such as 'drug running', identified in our consultations as the main method of engagement by young people would not fit the definition. It appears, thus, that young people are unlikely to be 'career criminals' within organised criminal networks. However, both the international literature explored, and the research findings from England presented in this paper, support the argument that they may be vulnerable to exploitation by such groups. As such, a key implication of the consultations was the need to recognise the blurred line between the young person as criminal or as exploited, and for interventions to view prevention of crime and protection or safeguarding of young people as overlapping. This also supports the findings from the literature of young people as vulnerable to exploitation . The research in this project also supports the view of the literature that positive, longterm relationships with professionals are key to young people's disengagement from crime . The consultation themes also emphasise that young people's involvement in crime is more complex than simply out of a sense of entitlement and have implications for crime prevention more widely, beyond simply organised crime and beyond the English context. The young people in the study informing this paper did not cite selfish reasons for engagement with crime but a desire, even a need, to help out their families by providing extra money. This was couched in the sense that there were few alternative opportunities available with legitimate employment seeming elusive. The findings support the critique of 'prevent as enforcement' interventions that has emerged in the UK and Australia , as they justify a focus not on penal consequences but on the impact on the young person's family. This shift from a dominant emphasis on punishment to one on family consequences might have a powerful impact on the effectiveness of youth crime interventions in various contexts. Indeed, the evaluation of the educational toolkit that stemmed from this project praised the impact of a key focus on family as highly effective. As such, a question remains as to whether interventions based on restorative justice or mediation with victims are as powerful as drawing on young people's empathy for their own family. New interventions based on family will also have limitations for use with some young people, particularly those who are in the care of the authorities. New preventative interventions need to involve a range of agencies working in partnership with an overall focus on safeguarding, as identified by Sturrock and Holmes . They should focus on relationship-based practice with the opportunity for long-term support from a key worker for young people who are vulnerable to involvement with crime . Most crucially, they need to be more supportive than punitive and to recognise young people's vulnerabilities. The Positive Youth Justice approach offers a model that draws on these elements and that could be translated from a post-offending to a preventative intervention. Young people's voices need to inform and guide these interventions in order for them to be relevant and effective . In the research outlined in this paper, this positive preventative approach was seen in the opportunity for some young people to receive ongoing key worker support from their YOT or youth worker, through both individual mentoring and positive group work activities . These supportive activities allowed young people to discuss and express their feelings as well as to develop a positive vision for their futures. Case has argued that young people need to be consulted in defining the risk factors for their involvement with crime. This study goes some way towards this and its key themes represent people's authentic experiences in relation to young people and organised crime. The study emphasises that young people are vulnerable rather than simply criminal. Their involvement in organised crime is often preceded by grooming or exploitation rather than being a clear-cut choice to engage. It also found that communicating the impact of young people's involvement on their family is likely to be a more powerful deterrent than focusing on penal consequences. Lastly, the findings highlight that young people need positive long-term support as part of preventative interventions. These findings have a number of implications for the development of youth crime prevention practice that are relevant to the English context and beyond. In particular, preventative interventions should not add to the 'enforcement as prevention' discourse through highlighting penal consequences. Instead they should recognise the blurred line between young people as perpetrators and as exploited and highlight the impacts on their family who they often seek to protect and provide for. Interventions need to be based on positive relationships with professionals and provide support for young people's needs and vulnerabilities. There is, however, a clear need for further empirical research on young people and organised crime in the UK and more widely as there is a dearth. --- Conclusion

The Serious and Organised Crime Strategy (2013) for England and Wales made a commitment to develop preventative educational resources for use with young people on the topic of organised crime. This paper presents findings from a UK Home Office funded project which was aimed at developing and subsequently evaluating these resources, and explores their wider implications for youth crime prevention policy within the UK and internationally. Based on interviews with youth practitioners and young people, the project found that many young people with vulnerabilities (such as learning difficulties) were in turn vulnerable to exploitation by criminal groups; that the reasons for young people becoming involved in organised crime were complex including a desire to provide for their families in a climate of austerity and unemployment; and that positive relationships with professionals and long-term support were significant for youth crime prevention.

en e ´tudiant la fac ¸on dont les perceptions subjectives de la femme relatives a `son contro ˆle du travail ou a `son autonomie, le stress lie ´au travail et le conflit travail-famille influencent les intentions de fe ´condite ´. Des mesures au niveau national des taux d'accueil d'enfants de moins de trois ans en structure collective et des possibilite ´s de travail a `temps partiel sont e ´galement prises en compte afin de comprendre a `la fois leur impact et leur relation avec les perceptions subjectives du travail salarie ´et intentions de fe ´condite ´. A partir des donne ´es de l'Enque ˆte sociale europe ´enne de 2004/5 provenant de 23 pays, des mode `les de re ´gression logistique multi-niveaux, avec pour variable de ´pendante les intentions de fe ´condite ´, sont estime ´s pour les femmes sans enfant et celles avec un enfant. La probabilite ´de vouloir un deuxie `me enfant est significativement plus e ´leve ´e chez les femmes qui ont un plus haut niveau de contro ˆle sur leur travail. De plus hauts niveaux de stress lie ´au travail sont associe ´s significativement a `des plus faibles intentions de fe ´condite ´dans des contextes de faible taux d'accueil des enfants en bas a ˆge. La pre ´valence du travail a `temps partiel parmi les femmes actives est associe ´e significativement a `l'intention de devenir me `re, mais avec des effets diffe ´rents selon que les femmes aient elles-me ˆmes un travail a `temps partiel ou a `temps plein. The difficulty of combining paid work with family responsibilities has been a prominent focus in fertility research and public policy debates. The massive entry of women into the labour market in the 1960s, and women's gains in educational attainment, is seen as central factors driving fertility postponement . In fact, around 80% of women between the ages of 25-44 in the European Union are now in paid employment compared to around 50% 30 years ago . This is coupled with recent public debates and policy mandates of the European Commission ) to attempt to simultaneously raise both fertility levels and female employment. To strengthen the labour force and increase European productivity, the Lisbon Strategy called for a rise in women's employment . This was coupled with the strategic goal to increase European fertility ). These mandates raise the potential dilemma of how to combine an increase in female employment with an increase in fertility and thus make paid employment more compatible with family responsibilities . It also corresponds to a shift in policy from the male breadwinner/female carer model towards an adult worker model of the family . --- Mots-cle ´s Although the link between participation in paid employment and fertility has been widely studied, employment is often examined by focusing on women's labour market status or number of hours in terms of part-versus full-time employment . Although we have an increasingly adequate picture of the association between labour market participation and number of hours with fertility intentions and outcomes, we lack an understanding of how subjective perceptions of autonomy and control over work, time pressure and levels of workfamily conflict impact fertility decisions. The central argument of this study is that it is not merely employment versus non-employment that is pivotal, but rather certain job characteristics that enable employment to become more conducive to parenthood. Previous research on work-family conflict has identified various characteristics of 'good' jobs, that allow paid employment to become more compatible with family responsibilities, such as flexibility in timing and organization of work and a higher degree of autonomy . The study provides several contributions to fertility research by introducing the examination of working conditions and subjective perceptions of work in addition to empirical measures of institutional circumstances that might impact fertility intentions across Europe. The first contribution of this study is that it demonstrates the usefulness of women's subjective experiences such as their perceived control or autonomy over work, the impact of job strain and work-family conflict to explain fertility desires. A second extension of the literature is the acknowledgement and empirical measurement of national contextual factors that might impact fertility intentions. Women's employment and subjective perceptions do not exist in a vacuum, but are shaped by a wider national context where certain policies enhance or constrain the compatibility of paid work and care. In this study, we consider the opportunity to work part-time and the availability of childcare for young children . The ability to combine paid work with parenthood differs considerably across Europe. In the Nordic countries, labour force participation rates are generally high throughout the entire family cycle, whereas in Southern, Eastern and to some extent Western Europe , female and maternal paid employment remain highly dependent on the number of children and the educational attainment of women . The aim and scope of national family policies such as parental leave and cash benefits for families are often positioned as the underlying reason for the differences found between countries in female labour force participation and fertility. How far these policies are effective in increasing fertility or labour force participation has, however, been the subject of many studies, most of which yielded ambiguous results . This study includes national-level measures of childcare enrolment below the age of three and the proportion of parttime female employment to understand how these macro characteristics are related to fertility intentions and examine their interplay with perceptions of work control and strain on fertility intentions. Our focus is on fertility intentions, and more specifically, the time-dependent intention to have a first or second child within the next 3 years. Some researchers have argued that fertility intentions are more reliable concerning the total desired number of children for aggregate rather than for individual predictions or that the link between intentions and behaviour is weak . A growing number of studies, however, have demonstrated the high predictive power of fertility intentions. In studies that compare fertility intentions and their subsequent realization, it appears that timedependent fertility intentions, which refer to the intention to have a child within a specified time interval, are good predictors of fertility, even after controlling for background and life-course variables . We acknowledge that fertility intentions may be revised because of changing constraints , but also maintain that a measure of time-dependent fertility intentions is appropriate to understand fertility decision-making. We analyse the intention to have a first and second child in separate models, because the intention to become a parent is commonly viewed as a process guided by different considerations than the decision to have an additional child . We focus on the intention to have a second child because considering that the total fertility rate is below the replacement level of 2.1 children in most European countries and the widespread two-child norm, we assume that women who intend to make the transition to a third-or higher-order birth are a special and distinct group with different motivations and fertility ideals . Also time-dependent fertility intentions have been found to be the most reliable for individuals with no child or one child . --- Theoretical Background As stated previously, the aim of this study is to introduce new employment-related factors that are associated with the plan to have a child and test how the effect of these characteristics varies across different institutional contexts. Previous studies relating female employment to fertility have shown that women who work continuously throughout their adult lives have fewer children than women who are not in paid work. In other words, there is a negative relationship between employment and fertility at the individual level. One explanation for this effect is that paid work delays the transition to parenthood mainly by raising the age at first birth . A competing argument is that the negative impact of paid work on employment only manifests itself after the first child is born. This is because women only become aware of the incompatibility of the mother and worker role when they are confronted with the conflicting demands of childrearing and work and thus delay or forego the birth of a second or third child . In order to theoretically and empirically evaluate the two modes in which employment presumably affects fertility, we formulate separate hypotheses for women with and without children and thus also analyse these groups separately. We first discuss the importance of work characteristics followed by an examination of institutional factors. --- Subjective Work Control, Job Strain and Work-family Conflict in Relation to Fertility Intentions Jobs with certain characteristics might be more compatible with family life than others since it is not only the quantity or number of hours, but also the quality of work that matters . Characteristics that have been identified as potentially reducing the conflict between work and family life are control over work, such as flexible working times or arrangements , whilst stressful and irregular jobs tend to increase the conflict experienced between paid work and family . To understand how these aspects might impact fertility intentions, we draw upon a broader body of literature outside of demographical fertility research on perceived work control, job strain and work-family conflict. --- Subjective Work Control The underlying requirements for 'good jobs' that facilitate lower friction between work and family are those with higher autonomy and variety . Both job autonomy and variety have been frequently linked to higher worker well-being and, more recently, to the reduction of work-family conflict . In general, there appears to be an attenuating effect of autonomy and variety on work-family conflict, and a positive effect on parenting style and other personality features . Autonomy and variety afford employees the opportunity to learn new things and maintain a feeling of control over their work, thereby enhancing the feeling of responsibility and meaning . The ensuing motivation, energy and attitudes derived from work can likewise be mobilized to facilitate functioning in other life domains such as the family and result in a better 'synergy' between multiple roles . Another important employment feature is the level of time flexibility provided by the employer. Flexible work schedules have been shown to have a positive impact on work-family conflict by increasing the time available for family responsibilities and the perception of control . In a meta-analysis which reviewed 60 studies, Byron reported that schedule flexibility leads to lower levels of work-family conflict. In a study of the impact of job characteristics on work-family facilitation in the United States, Grzywacz and Butler found that autonomy and variety were associated with lower reported work-family conflict and higher work-family facilitation. In the Netherlands, Mills and Ta ¨ht demonstrated that nonstandard and flexible hours resulted in lower relationship conflict and more time spent with children, particularly for men. In a study conducted in three Finnish organizations, Mauno et al. found that job control protected employees from experiencing high levels of time and strain-based work-family conflict. Other studies, however, have presented mixed results. In the Netherlands, Van Echtelt et al. examined the effect of job designs with a high degree of worker autonomy on what they term 'over-employment' and came to the conclusion that this type of job design is particularly time greedy and causes employees to work longer hours than they desire. Autonomy and flexibility are also job characteristics associated with high-status jobs and a high degree of responsibility for organizational outcomes . High levels of autonomy might therefore also cause the borders between home and work to blur, thereby increasing work-family conflict, especially in high-status jobs. Employees with a high degree of flexibility tend to bring work home more often and have more work-related contacts whilst being at home . Since no previous studies have linked work control to fertility intentions, we enter into relatively uncharted theoretical terrain. It is possible, however, to draw upon the previous findings related to work-family conflict to formulate a hypothesis related to fertility intentions. In this study, we subsume several aspects under the broader theoretical construct of work control, which includes autonomy , variety and time flexibility . Based on previous research, our first hypothesis is that: higher levels of perceived work control will result in more positive fertility intentions to have both a first and a second child. We assume that women who have jobs that are characterized by a high degree of work control are more able to combine the demands of childrearing and paid work and will also evaluate the possibility of having a second child more positively than women with lower levels of work control. --- Job Strain Not only work control, but also the related concept of job strain, has been shown to serve as an important factor for work and family relationships. Job strain may operate via two different mechanisms. The first mechanism emphasizes stress at the workplace. According to this view, job-related strains are reactions or outcomes that result from the experience of stress . Building on this approach, we would expect to find a direct negative effect of job strain on various outcomes in private as well as working life, including lower fertility intentions. A second more nuanced mechanism of job strain has been proposed by Karasek , who argues that mental strain is dependent on the job demands placed on the worker in combination with the discretion permitted to the worker in deciding on how to meet these demands. This definition maintains that the decisive factor in subjective experiences of high job demands is the amount of control granted to the worker. Four types of jobs are distinguished in a 'Job Demand-Control' model, which operates according to the combination of demands and control they offer: passive jobs , low strain jobs , active jobs and high strain jobs . This means that aside from the direct effect of job strain and perceived work control, the combination of high job demands and low perceived work control might result in higher strain. The Job Demand-Control model has been tested on various outcome variables such as well-being, depression, and different aspects of physical health . Several studies also examined whether the model also holds when work-family conflict is considered , showing that increased job demands are associated with higher work-family conflict, whilst control has an alleviating effect. In Sweden, for example, Gro ¨nlund found that high job demands increased work-family conflict whereas high job control had the opposite influence. There was, however, no significant interaction between job demands and control, leading the author to conclude that high job demands are associated with higher levels of work-family conflict with the effect of control being rather marginal. Very similar results were obtained in a study of the effect of job control, job demands and social support on work-family conflict and depression in Canadian lawyers . Job strain and work control independently influenced work-family conflict, but no multiplicative effects in the form of significant interactions were found. Since the focus of this study is on fertility intentions and not job strain, we do not empirically test the multiplicative model here, but rather assume that the effects of job strain and work control operate independently on fertility intentions. Previous research has shown that job strain and perceived work control are relevant and important concepts to predict work-family conflict, yet to date, there has been no application of this theoretical model to fertility intentions. Based on these previous explanations and findings, we hypothesize that a high degree of job strain will lead to lower fertility intentions in women who already have one child. We anticipate that women who do not have children will be less influenced by job strain since they have not as of yet experienced the tension between paid work and family responsibilities that children bring . --- Work-family Conflict A final interrelated concept regarding work control and job-strain is work-family conflict. Work-family conflict has been defined as 'a form of inter-role conflict in which the role pressures from the work and family domains are mutually incompatible in some respect. That is, participation in the work role is made more difficult by virtue of participation in the family role' . Three forms of work-family conflict are generally distinguished: time-, strain-and behaviour-based conflicts. Time-based conflict is regarded as the most common and occurs when work and family life compete for the individuals' time in such a way that the individual is unable to perform at the preferred levels in both domains. Strain-based conflict arises when stress or tension experienced in one life domain spill over into the other domain. This includes worries about work, which often leads to impatient or irritable behaviour at home . Behaviour-based conflict refers to the situation where behaviour required in one role makes it difficult to fulfil requirements of another role. One example is the potential role conflict between the stereotypical behaviour of managers like aggressiveness and objectivity. Behaviorbased conflict occurs when the same individual is expected to express emotions in the family such as being warm, and vulnerable . The different types of work-family conflict often overlap and can be difficult to distinguish empirically. As described in more detail shortly, we use a measure of work-family conflict that refers to both time and strainbased conflict since these are the most relevant for the majority of employees . Our central hypothesis in relation to work-family conflict is that high levels of conflict between work and private life will lead to lower fertility intentions in women who already have one child. Women without children are expected to experience less conflict between their paid work and private life since children put a large claim on parents' time, especially when they are young. We also expect that some women without children will be sensitive to these issues and anticipate that women without children who already place a higher importance on the compatibility of a job with their private lives will be more likely to intend to have a child. --- Institutional Context In addition to employment characteristics and individual subjective perceptions, national level policies operate to enhance or constrain the compatibility of work and care. In this study, we focus on policies designed to maintain or promote the labour force participation of women rather than policies aimed at stay-at-home mothers or the provision of long unpaid leave that might elicit a negative effect on mothers' participation in paid work. Policies that enhance the compatibility between work and care include affordable childcare facilities and the opportunity to work parttime since both reduce the opportunity costs of having children . --- Childcare Availability Adopting the assumption that a reduction in the opportunity costs of childbearing will result in higher fertility, childcare availability has frequently been hypothesized as having a positive influence on fertility . The empirical evidence generally shows that a higher availability of childcare has a positive impact on fertility, but there are some mixed findings . Brewster and Rindfuss found a positive effect of childcare availability on the combination of childrearing and paid employment, showing a higher return to the labour market after childbirth. Others have likewise demonstrated that fertility is positively influenced by reducing childcare costs and increasing childcare availability . Hank and Kreyenfeld found that access to informal childcare arrangements significantly increased the transition to first birth in Germany, concluding that availability and not affordability of childcare was central. Rindfuss et al. ) also found that increased childcare availability in Norway clearly and consistently had a positive effect on fertility. In this study, we focus on childcare for the youngest group of children below 3 years old, which has been demonstrated as a crucial period for the labour market re-entry of women . We consider it as a close empirical proxy to capture childcare as a policy designed to maintain a mother's continuous labour force participation throughout her childbearing years. In this context, our hypothesis is that a higher availability of childcare for young children will have a positive influence on the intention to have a first and second child for women in paid employment. Furthermore, as an extension of our previous argumentation regarding the importance of work control for fertility decision-making, it is expected that a heightened level of work control is more important in countries with less institutional support in the form of childcare availability. This is attributed to the fact that women in these countries are more dependent on their individual resources if they want to combine childrearing and work. We likewise expect that higher levels of job strain and work-family conflict are associated with lower intentions to have a child particularly in countries where the institutional support of childcare availability is lower. --- Part-Time Work Part-time work, usually defined as working less than 30 h per week, is one of the most widely used work arrangements amongst women of childbearing age to manage work and family responsibilities. In fact, roughly one third of women aged 25-49 employed in Europe work part-time . The prevalence of part-time work varies considerably across countries, ranging from over 70% in the Netherlands, to less than 5% of women in Bulgaria and Slovakia . When part-time work is considered as a way to reduce the incompatibility between work and family life, the assumption is that women work these hours deliberately and voluntarily. This assumption, however, is potentially problematic as one aspect of part-time work is that it is often associated with 'bad' jobs, which are jobs with unfavourable working conditions that offer limited perspectives for advancement or have a stigmatizing effect, signalling to the employer that commitment to work is low . Besides being concentrated in certain sectors and lower-paying occupations, there is little empirical evidence that parttime workers in Europe experience their jobs as being worse in quality than a comparable full-time position. A recent examination of working conditions and work-family reconciliation in Europe showed that more than 60% of women in parttime work cited family responsibilities as a reason to not work full-time, whereas only one fifth indicated that they could not find a full-time job . In addition, in terms of work-life balance, mothers in particular report positive effects of reduced working hours . In addition to controlling for whether women work part-or full-time at the individual level, we also include the opportunity to work part-time in this study as a characteristic of the institutional context. The prevalence of part-time work amongst women in a country is associated with institutional aspects and overall working-time regimes, such as the rigidity of the labour market, and thus affects the ability of individuals to opt for reduced hours. In this respect, we hypothesize that a higher prevalence of part-time work amongst women within a country is associated with positive fertility intentions in the case of women who already have one child. This is because these women are more likely to have already experienced time scarcity in full-time employment because of their family responsibilities. Since women who do not have children are expected to be more concerned about establishing themselves in the labour market and building a career, the prevalence of part-time work is not anticipated to influence the intention to have the first child. Furthermore, we also expect an additive effect of work characteristics when the institutional context offers less opportunities to work reduced hours and expect that heightened levels of work control are more important for intending a second child in countries with a lower availability of part-time work. Because women in these countries are expected to experience a stronger time-squeeze when combining paid work and family responsibilities, we also anticipate that higher levels of job strain and work-family conflict are associated with lower intentions to have a second child in countries with a lower availability of part-time work. --- Method --- Data and Sample Data are used from the second wave of the European Social Survey , a largescale quantitative survey administered in 2004/5 across Europe. We excluded Turkey and Ukraine from the analyses, leaving 23 countries .1 In each country, a representative random probability sample was drawn with strict quality controls employed to ensure that all national samples met the requirements. Each wave of the ESS consists of a core questionnaire on attitudes and values and rotating modules. The 2004/5 wave contained a module on family, paid work and well-being which included information on family life and fertility intentions. The total sample consisted of 49,066 respondents, which was subsequently reduced further to examine fertility intentions as realistically as possible. We also opted to examine women only because of the very different mechanisms involved in paid employment and fertility for women versus men. This meant that the sub-sample used in these analyses included women only who were not older than 45 years at the time of the interview and lived together with their partner or husband. The sample also only included respondents who were engaged in paid labour as a dependent worker2 for at least 1 h in the week before the interview. The final restricted sample used here consists of 1,533 female respondents. For details of the descriptive statistics, refer to Table 2. --- Measures The dependent variable is fertility intentions, which was measured by the question of whether the respondent intended to have a child within the next 3 years, answered on a four-point scale . This variable was subsequently recoded as a dichotomous measure of fertility intentions in which the answers probably not and definitely not were coded as no, and the answers probably yes and definitely yes was coded as yes. In a series of preliminary analyses, we estimated ordered logit models, but since this did not produce considerably different results, we opted for a more straightforward binary measure of intentions. Explanatory micro-level variables. Perceived work control is measured by an index of six items that asked whether the respondent is allowed to influence the pace of work, daily organization of work, policy decisions, as well as whether the work done requires learning new skills, offers variety and challenge; and, is not closely supervised. These six items were averaged into a scale that ranged from 0.5 to 7.5 with higher values indicating more perceived work control. Internal reliability of the scale is sufficient . In order to facilitate interpretation, the scale was centred around the mean. Job strain is measured by the extent to which the respondent feels that there is never enough time to get everything done at work . Work-family conflict is measured by a scale that is constructed using four items asking how often the respondents keep worrying about work problems when they are not working, feel too tired after work to enjoy the things they would like to do at home, find that their job prevents them from giving the time they want to their partner or family; and, find that their partner or family gets fed up with the pressures of their job. The four items were averaged into a scale running from 1 to 5, with higher values indicating more conflict . Once again, to facilitate interpretation, the scale was centred around its mean. The importance placed on being able to combine family and work when choosing a job was also included as a continuous measure. Control variables. The educational attainment of respondents was measured by the total number of years each respondent was enroled in full-time education. Also the number of years the respondent was employed is included to account for individual differences in labour market attachment. To facilitate interpretation, both variables are centred around the grand mean. Furthermore, respondents' age and, if applicable, the age of the first child is included in the analysis. The number of weekly working hours was also included. As described shortly, models are estimated for women with and without children. In the model for women without children, a cross-level interaction term of the proportion of women working part-time on the country level by a dummy indicating whether the respondent herself works in a part-time job is included. The cut-off point of 30 h is in line with the definition applied by the OECD in comparative studies . Furthermore, the educational attainment of the partner is included . Owing to high levels of non-response, it was not possible to include additional information about the partner, such as weekly working hours or access to flexible working times. Explanatory macro level variables. The availability of child care facilities is measured at the country level, with a variable that indicates the percentage of children of ages 0 to 3 enroled in formal childcare . Opportunities for part-time work are measured by the share of women working part-time . In order to facilitate interpretation of the coefficients, both variables are divided by 10 and centred around their mean . --- Analysis A multilevel binary logistic model was run in Stata 10.1 with separate models estimated for women without children and those with one child. The multilevel binary logistic model is a two-level random coefficient model with respondents nested in the country cluster j, which includes a random intercept u j for clusters in the latent response model . The amount of variance attributable to the country level ) can be calculated as r 2 u0 / in logistic multilevel models. Although we do acknowledge that the estimates should be interpreted with some caution since this approach of calculating the VPC is influenced by the explanatory variables in the model. In other words, the inclusion of certain level-one variables might increase the estimated VPC compared to the empty model . In our analysis, in the empty models, the betweencountry variances in fertility intentions are estimated as 7.3% and 10.2% . After estimating the full model, the between-country variance in the sample of women without child is fully explained by introducing the contextual variables and the cross-level interactions. In the sample consisting of mothers with one child, the between-country variance in the full model is reduced to roughly 5%. In order to test whether the effect of the main explanatory variables varies between countries, random slopes for these variables were introduced in the model, but there was no evidence that the relationship between fertility intentions and work characteristics varies across countries. Therefore, the final model only contains a random intercept to account for differences in fertility intentions in the various countries. --- Results The results of the regression analysis are presented in Table 1. The first finding is that the results are markedly different for women with and without children. This supports previous research, which has also found that issues such as work-family conflict, job strain and control only become salient after the birth of a first child . As stated earlier, our first expectation was that a heightened level of perceived work control would positively influence the intention to have a first child and a second child . The results support this expectation for women who already had one child. In other words, women who experience higher levels of work control are more likely to intend to have their second child. We find no support for our second hypothesis, where we anticipated that a high level of job strain would lead to lower fertility intentions for women who already have children. As mentioned earlier, job strain taps the subjective perception that one is unable to get everything done at work, which is a job characteristic that is intrinsically difficult to combine with having another child. One explanation is that these are challenging jobs that offer career perspectives perceived as being incompatible with having additional children. However, as indicated by the significant interaction coefficient between job strain and childcare availability, the direction of the effect of job strain on fertility intentions differs according to the level of childcare availability. To facilitate interpretation of these interaction effects, we have graphed the predicted probabilities in Fig. 1, which were calculated holding the group-level residual at its mean of zero . Turning to the right panel of Fig. 1, we see that women who experience higher levels of job strain , are more likely to intend to have a second child when childcare availability is high whilst the relationship goes in the opposite direction when childcare availability is low. This is in line with our cross-level interaction hypothesis where we expected that work characteristics would have stronger effects in settings where the institutional context is less supportive. Our results suggest that higher childcare availability is related to higher second birth intentions for women who experience high levels of time pressure at work. Our third central hypothesis predicted that higher levels of work-family conflict would translate into lower fertility intentions for women who already have children. The results show a strong and positive effect of work-family conflict for mothers with one child, contradicting our expectations. A plausible explanation is that the women who are experiencing the highest levels of work-family conflict are also those placing the highest value on their job and family. Therefore, these women are most likely to stay in paid work, whilst at the same time not compromising their fertility plans. Turning to the institutional effects, our initial expectation as stated earlier was that the higher availability of childcare would have a positive impact on the intention to have both a first or second child. We do not, find any linear positive effect of childcare, neither for women without children or for mothers of one child. There is however, as described previously, evidence of a diverging effect of higher enrolment in formal childcare on fertility intentions for mothers with one child who experience high levels of job strain . The quadratic term of childcare enrolment is significant and positive in the analysis of women without children, suggesting that the relationship between childcare enrolment and fertility intentions is u-shaped with positive effects at the lower and higher ends of the scale. The reason for the absence of a linear effect might be that the enrolment of children below the age of three in formal care might not sufficiently tap into the availability of care of individual parents, which has been shown to have a positive effect on fertility ). Nationwide availability, we believe, indicates the general policy climate concerning the compatibility of paid work and childrearing for women. Previous research has shown that particularly in countries in Central and Eastern Europe, characterized by a low availability of formal childcare and high female labour force participation, informal care provided by grandparents and other relatives plays a key role in attenuating conflicts between work and family responsibilities , which can be one explanation for the u-shaped effect found for women without children. Our final expectation was that a higher prevalence of part-time work amongst women within a country would be associated with positive fertility intentions in the case of women who already have one child. The results do not, however, support this hypothesis. We do find, however, a consistent and negative effect of part-time work on fertility intentions for women without children. To explore this unexpected finding further, we added a quadratic term of this variable into the analysis to see whether the negative effect is linear or u-shaped. The quadratic term proved to be significant which leads us to conclude that women without children in countries with either a very low or very high proportion of women working part-time have higher first birth intentions than women from countries with moderate part-time employment possibilities. To explore this effect further, we then added a dummy variable measuring whether the respondent works part-time herself and allowed this variable to interact with the institutional indicator of the proportion of part-time work amongst the female workforce. This interaction was significant and positive, indicating that the relationship between fertility intentions and part-time work on the country level is u-shaped for women who work more than 30 h per week, but J-shaped for women working part-time themselves . As Fig. 1 illustrates, the two lines intersect slightly below the mean of female part-time work, indicating that women who work full-time have higher fertility intentions compared to those who work part-time in countries where the proportion of female part-time employment is lower than the overall sample average. One explanation for this finding might be the type of jobs where part-time work is located in different institutional settings. In countries where part-time work is less prevalent, part-time jobs tend to be of a lower quality in aspects such as wages. In these contexts, it would be particularly the women without children who might view part-time employment as more of a constraint than an opportunity to combine family and work responsibilities. Conversely, in countries where a large proportion of women are employed in reduced working hours, part-time work is less likely to be related to the quality or level of the job. Finally, it is interesting to note the varied effects of the control variables that we included in the models for women with and without children. For women without children, higher labour force attachment, indicated by more years in paid employment, is associated with lower fertility intentions whilst women's own education, their partners' education and a higher number of weekly working hours predict the intention to become a mother. For women with one child, the results show a different pattern of the control variables, with only age and the age of the first child serving as significant predictors of fertility intentions. In order to explore the idea that women without children are less influenced by the characteristics of their present job because they are not yet aware of difficulties that might arise once they have to combine employment with parenting, we introduced a variable in the model to measure the extent to which the respondent thinks that when choosing a job it is important that the job allows them to combine work and family. Placing a higher importance on this hypothetical job characteristic significantly predicts the intention to become a mother but is not associated with the intention to have a second child. --- Conclusions This study contributes to the existing fertility literature by introducing working conditions, subjective perceptions of work and empirical measures of institutional circumstances to understand fertility intentions across Europe. To theoretically and empirically evaluate the different ways in which employment affects fertility intentions, separate hypotheses for women with and without children were formulated and tested. Previous research on perceived work control, job strain and work-family conflict outside of demographical fertility research was used to develop theory and hypotheses about how these factors might influence fertility intentions. Our results show that whilst the 'objective' indicators of labour market position and conditions are strong predictors of the intention to become a mother, the intention to have a second child is more strongly associated with work characteristics. Perceived work control in the form of autonomy and variety in work and time flexibility was previously shown to have a positive effect on reducing work-family conflict. Extending these previous assumptions to fertility intentions, we found support for our expectation that those with higher levels of work control are significantly more likely to intend to have a second child. Previous research in the domain of work-family conflict has also shown that job strain is an important factor, yet to date, there is a lack of research linking this type of conflict to fertility intentions. The current study found that. for women who already have a child, higher levels of negative job strain have an effect on the intention to have a second child only in combination with the availability of formal childcare for young children. We believe that this is because if one already experiences an inability to get everything done at work, the perception appears to be that it would be difficult if not impossible to combine this job with having an additional child when the level of institutional support is low . Whilst the effect of work control operated in the expected direction, we found an unexpected positive effect of a higher amount work-family conflict on the intention to have a second child. We believe that this effect is attributed to the fact that mothers of one child who remained in paid work but place high importance on family life and on having another child might be more likely to experience higher levels of conflict between paid work and family responsibilities. These women might not want to compromise on family size but still place high importance on participating in the labour force. Unfortunately, we have no measure of labour market attachment in our data that would permit us to test whether this is the case. That the effect of workfamily conflict on fertility intentions is not as straightforward as expected is highlighted by the fact that the only other study, to our knowledge, which used workfamily conflict to predict fertility intentions by Shreffler et al. , failed to find a significant direct effect of the perceived conflict between work and family of men and women on the intention to have a child within the next 3 years in a sample of dual-earner couples in the United States. The results of this study demonstrate clear differences for women with and without children, once again underlining the finding that issues such as work-family conflict, job strain and control only become salient after the birth of a first child, underlined by previous studies . This does of course not mean that women without children are oblivious to these issues, which was demonstrated by the effect of the measure of the importance of work-family compatibility when choosing a job. Placing a higher importance on being able to combine a job with family life strongly predicts the intention to become a parent, indicating that women without children are aware of potential friction between these domains of life but relate these potential frictions less to their present job and work characteristics than mothers of one child for whom no effect of this measure was found. It is plausible that many researchers would argue that a woman's employment career is endogenous with her fertility decisions and ultimate fertility outcomes. This endogeneity can be controlled for statistically by implementing structural equation models or adopting a fixed-effects modelling strategy . It was not possible, however, to overcome this issue within our current study design because only employed women were asked to evaluate their subjective employment perceptions and work characteristics within this cross-sectional data source. It was also not possible to, for example, estimate additional sensitivity analyses that did not include variables related to women's employment to see if results changed. It would be desirable to attempt to tackle these issues in the future either using longitudinal data or by applying the above mentioned techniques. A strength of this study was to go beyond previous theoretical discussions that underline the importance of institutional factors in shaping fertility, to empirically examine how national level policies in combination with subjective perceptions of paid work can enhance or constrain the compatibility of work and care. We empirically examined how childcare availability for children under 3 years of age and the prevalence of part-time work amongst women within a country impacted the intention to have a second child. We obtained some mixed and very interesting findings in this respect. Contrary to our expectation, only the second-order terms of both institutional indicators predicted the intention to have a first child, whilst no significant main effects were found for mothers of one child. We also found two interesting cross-level interactions indicating that childcare availability is positively associated with fertility intentions for mothers in stressful jobs and that part-time work is only associated with a lower intention to become a mother in contexts where only few other women work part-time. In conclusion, it appears that to enable women across Europe to fulfil their desires and ambitions in the work as well as the family domain, the institutional as well as the workplace characteristics and the subjective experiences of employment can play a crucial role. The option of staying at home for extended periods of time during the most productive years of life is no longer a viable option for the large majority of European women, neither economically nor in terms of self-actualization and societal participation. Foregoing the birth of children or having a smaller family than preferred, on the other hand, may seem to be a very high price to pay in exchange for employment. Creating and implementing policies that are effective in promoting work-life balance certainly remains a challenge for many policy-makers. It is the hope that this study provides some insights into the more nuanced mechanisms and impact of workplace characteristics and subjective experiences of employment and work-family balance that afford individuals with the ability to fulfil multiple roles in both the workplace and family. Open Access This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author and source are credited. --- Appendix Tables 2 and3.

The link between employment and fertility is often only examined by focussing on women's labour market status or the impact of part-versus full-time employment. This study introduces a new explanation by extending research to examine how women's subjective perceptions of control or autonomy over work, job strain and work-family conflict influence fertility intentions. National-level measures of childcare enrolment under the age of three and the occurrence of parttime work are also included to examine their relation to fertility intentions and their interplay with perceptions of work. Using data from 23 countries from the 2004/5 European Social Survey (ESS), multilevel logistic regression models of fertility intentions are estimated separately for women without children and women with one child. Women with higher levels of work control are significantly more likely to intend to have a second child. Higher levels of job strain (time pressure) significantly lower fertility intentions for mothers in contexts where childcare availability is low. The prevalence of part-time work amongst the female work force significantly predicts the intention to become a mother but has different effects for women who work part-time themselves compared with full-time employees.

Introduction On the 14 th anniversary of the exogenously focused scenario that constituted the painful events of 2 July 1993, we don't want to see outsiders in Sivas. … We shall lay flowers in memory of the dead. There's no need for outsiders to participate . Thus read the press statement released by a number of NGOs in Sivas, central-eastern Turkey, ahead of the 2007 episode of an annual commemoration the city has hosted since 1994. The commemoration concerns an arson attack that took place on 2 July 1993, when a hotel hosting the participants of a culture festival organized by an association representing followers of the Alevi faith-one of Turkey's demographically minor social groups-was set alight by tens of assailants before an inactive law enforcement, thousands of onlookers, and live TV cameras, resulting in the death of thirty-three festival participants and two hotel workers. Delivered nearly a decade and a half after the arson attack, the statement epitomizes the conviction that the atrocity has not been fully investigated; only some 30 assailants caught on camera have received sentences, although a crime of such scale most likely involved the backing of a much larger network that has yet to be revealed. While this conviction is widespread across different social groups in Turkey, not as commonly shared is the answer to the question of exactly whom the network in question comprises. For those upholding the legacy of the festival participants killed in the blaze, the answer lies in the inaction of the governing authorities and that of the thousands of onlookers: it is they who encouraged the assailants . For mainstream politicians, journalists and various other influential figures in Sivas, however, the arson was but one part of a larger plot orchestrated by foreign powers . Hence the opening statement's reference to the atrocity as "exogenously focused" : an event whose origins remain outside the field of vision. As the statement indicates, in the years following the arson attack, allegations regarding "exogenously focused scenarios" turned out to affect new developments in and around the site of arson. First, they engulfed the commemorative procession held annually in Sivas by those upholding the legacy of the festival participants targeted in the arson, an initiative that pioneered various calls by intellectuals in the late 1990s and early 2000s for Turkey to "reckon with" its violent past . More recently, an important implication of such allegations marked a state-sponsored initiative to commemorate the atrocity on-site. The implication is that, insofar as the arson attack can be blamed on the outside, those inside are all victims, which therefore include not only the festival participants targeted in the arson but also the assailants who were in effect naive pawns. In 2011, the state's transformation of the site of atrocity introduced a victims' list that is inclusive of two individuals who were among the crowd outside the hotel during the blaze and were killed as it was belatedly dispersed by law enforcement. The transformation gave the site unprecedented visibility, evident in high-profile government representatives' visits and mainstream networks' live broadcasts of the annual commemoration held by those upholding the legacy of the festival participants killed in the arson. But it also introduced ever tighter security measures comprising checkpoints and barricades that sought to isolate the procession from Sivas and its inhabitants. The notions of inside versus outside and visual clarity versus obscurity, which characterize the above-mentioned references to exogenously focused scenarios, are among the most prominent tropes employed in both conspiracy theory and its scholarly discussions . Yet, their significance here transcends mere figuration. The conspiracy theory, which implies that those who started the fire and those killed in it are both victims in their own right, inspires a state-sponsored on-site memorial. Conjointly, a change of attitude among mainstream politicians and media gives the arson attack greater visibility through live broadcasts and high-profile visits. But, on the ground, various law enforcement apparatuses are introduced to isolate the commemoration, physically reinforcing the inside-outside delineation that characterizes related conspiracy theories. If these developments establish conspiracy theory's role in negotiating violent events, they also demonstrate that it hinges on a constant interplay between the figurative and the physical. Understanding this role, therefore, requires that the physical underpinnings and consequences of the visual and spatial tropes involved in conspiracy theory be considered seriously. How might this consideration contribute to analyses of the work conspiracy theory does with respect to violent events? I explore this question in light of my fieldwork on commemorative practices in Sivas, which I carried out at intervals over the two years following the transformation of the site of atrocity. In rethinking conspiracy theory and spatial memorialization through one another, I draw on Wagner-Pacifici's recent criticism of predominant approaches to analyzing commemoration for their tendency to cast the event being commemorated as complete rather than as continuing to unfold through commemorative practices. I argue that the environments in which the arson attack took place and is commemorated are significant to understanding the atrocity not because they serve as evidence refuting or validating conspiratorial thinking but because they act as catalysts of two of conspiracy theory's deadliest characteristics: indeterminable authorship and non-linear causality. --- Conspiracy theory, physical space, and the commemoration of violent histories Among the most prominent tropes employed in conspiratorial thinking and its scholarly analyses are those that engage the medium of visuality. This is due largely to the preoccupation that both analysts and conspiracy theorists have had with transparency. Whether understood as remedy to conspiracy theory or as its symbiotic other, transparency has loomed large in relevant literature primarily because it captures the sense of sight that conditions much of conspiratorial thinking. A closer look into this literature, however, demonstrates that transparency may have spatial as well as visual implications. The first way transparency has featured in relevant literature concerns critiques of conspiracy theory's commonplace associations with irrationality. Jamer Hunt suggests that conspiratorial thinking is "a style of interpretation" rather than "wild irrationality or psychosis" . If conspiracy "relies on secrecy and invisibility," conspiracy theorists aim at "making visible the clandestine causal agents" . Mark Fenster posits conspiratorial thinking as a "practice that longs for a perfectly transparent, accessible democracy" . If such Habermasian approaches have treated transparency as an antidote to conspiracy and thus as the ultimate goal of conspiracy theorists, they have increasingly been subjected to criticism in the aftermath of the Cold War , a historical period marked by not only the collapse of a physical wall preventing visual and spatial access to the political Other but also transparency's becoming the mainstay of liberal democracies . Consider Harry West and Todd Sanders' understanding of the relationship between transparency and conspiracy in this period. While acknowledging that conspiratorial thinking is preoccupied "with the operation of secret, mysterious, and/or unseen powers," they question the assumption that this indicates a quest for an abstract idea of transparency . Rather than take transparency at face value "in a world where varied institutions claim to give structure to the 'rational' and 'transparent' operation of power," conspiracy theorists maintain that power continues to operate "in realms normally concealed from view" . Michael and Kim Fortun similarly contend that conspiratorial thinking continues to proliferate despite liberal projects of transparency, because it "religiously set[s] up two separate worlds, this one of appearances and that other one of secrets" . For Kathleen Stewart, this indicates that conspiracy theories are driven by "a desire for an Other order of a true US and THEM coming from someplace outside our control" . At stake in the dynamics between transparency and conspiracy theory, then, is not just visuality but also spatiality: a delineation separating the space of power's formulations from that of its operations. Moreover, this delineation is physically charged, as the delineator is taken to determine the degree to which the space of formulations is visible from that of operations. Noteworthy in this respect is West and Sanders' discussion of transparency as the purported characteristic of "a surface to power" that conceals the space from which power operates . If their discussion understands this "surface" to function in a more figurative than physical register, it might operate in the latter, too, especially in contexts characterized by technology's impact on conspiratorial thinking. This is demonstrated by Jean and John Comaroff's account of how developments in optic technologies between the 17 th and 19 th centuries shaped epistemological notions of not only transparency but also its "obverse: a concern with refraction, distortion, concealment, collusion," accompanied by metaphors such as "the camera obscura" and "the hidden hand" . Comaroffs' historical account is brought up to date by Nicholas Holm, who discusses the relation between conspiratorial thinking and surveillance technologies in post-9/11 US and UK. For him, this has implications for the surveillant as well as the surveilled: the latter suffers from a "fear of constant observation of the self," or from the "paranoia-of-the watched," while the surveillant feels obliged "to be constantly privy to all information in order to address potential threats," which Holm calls the "paranoia-of-the watcher" . These two modalities of paranoia interact to produce results diametrically opposed to surveillance's purported disciplinary motivations, engendering "excessive and aberrant behavior" rather than establish "bodily and social discipline" . A case in point is presented by Misty Bastian's discussion of the impact of live television broadcasting on a mid-1990s' violent episode in the southeastern Nigerian city of Owerri, where conspiracy theories regarding ritual killings triggered a public riot. Rather than result in despair and inertia, broadcast images of a murdered child's body parts prompted ordinary citizens to take to the streets and perform "a sense of local Owerri values and a local Owerri place at the heart of the globalized, surveilled, and modernized city" . If these discussions reiterate that sight and space may feature physically as well as figuratively in discourses and practices of transparency and/or conspiracy theory, they also indicate exactly what sort of a role physicality may play in this respect. Consider how optic improvements have engendered a concern with not only a visually enhanced sense of epistemic transparency but also its obverse, how surveillance technologies have in certain cases produced the opposite of their purported aim of social discipline, and how the dissemination of visual information on socially infuriating events has at times fueled rather than quell popular reaction. These instances all show that, far from serving as passive backdrop or conduit to discourses and practices of transparency and/or conspiracy theory, physical spaces and objects may well be shaping these actively. Frederic Jameson has proposed to understand conspiracy theory as "the poor person's cognitive mapping" in the late capitalist age . Leaving aside momentarily the question of whether conspiracy theories are necessarily the preserve of "the poor person," I would like to reconsider Jameson's "mapping" in light of the discussion above: as not just a cognitive but also a physical spatial practice. This understanding recalls West and Sanders' framing of transparency within Arjun Appadurai's notion of "ideoscapes" , as a keyword that condenses a larger set of ideas and facilitates their export to and superimposition on "geographical landscapes" . But it differs from the way they place heavy emphasis on the first half of Appadurai's notion while neglecting its -scapeness. My methodology here prioritizes the latter. In so doing, I hope to contribute to analyses of conspiracy theory as a "search for the missing plot" rather than "a rigid, all too clear plot"-as "practice" rather than just "theory" or "prefabricated ideology" . This requires attending not just to the actors involved in conspiracy theory but also to the settings in which it unfolds. As Mathijs Pelkmans and Rhys Machold have demonstrated, anthropological work on the topic has often focused sharply on actors that produce, promulgate and/or subscribe to conspiracy theories. Yet, "[w]hat distinguishes conspiracy theories from other theories is the precise manner in which such theories are embedded in sociopolitical fields" and the various and often unforeseeable "use values" they acquire as a result . Pelkmans and Machold's call to shift focus from actors to fields is significant for two reasons. First, it complicates the notion that conspiratorial thinking is the preserve of one set of actors or another à la Jameson's "poor person." Secondly, as any sociopolitical field is constituted not only by humans and their faculties but also by physical environments, it demonstrates the need to take seriously the role of objects and spaces in this constitution. This combination of an interest in physical environments and an attention to the productive force conspiracy theory becomes-virtually independent of its authors-when embedded in particular fields resonates with the way rumor features in Veena Das' work on civilian unrest in post-Partition India. In a way that recalls George Marcus and Michael Powell's argument that post-Cold War conspiracy theorizing is characterized by the reversal of "cause and effect" as it seeks to reckon with "the invisible, unpredictable and incalculable risks of our contemporary world" , Das understands rumor as a practice that is less about representing a preexisting reality than about precipitating a new one. As such, rumor does not merely communicate but produces events "in the very act of telling" by authorizing them . The authorization, suggests Das, involves reanimating particular "regions" of the past, which are the stuff of collective consciousness, and the "regionality" of which is defined not by chronological proximity but by affinity to the "affective qualities of the present moment" . Rumor produces new events-in this case, violent events-by sociohistorically anchoring them in these regions . But rumor's productive capabilities concern space as well as time; its materializations are not uniform across geography. Certain physical environments across which rumor circulates significantly shape its consequences in degree if not in kind . Rumor, therefore, is a means of spatial production as well as a sociopolitical one. There are glimpses of an understanding of conspiratorial thinking as just such a production in Carol Delaney's ethnography of a village in central Turkey. She discerns the spatial demarcation of an inside from its outside as a trope employed frequently by her interlocutors vis-à-vis adverse events or the likelihood thereof. "Bad and threatening things," they believe, "come from outside … the body, the family, the village, or the nation" . Tangible implications of this trope are discernable in the ways villagers experience the village and their houses. The village's topographical positioning enables a clear view of those approaching it, giving "villagers the sense that they have some control over who enters" . New roads connecting the village to the world outside are met with skepticism by the villagers: if roads mean better logistics, they also expose the village "to polluting influences from town" . On an architectural level, villagers' perception of their houses as clean and streets as dirty has physical underpinnings: "Houses are swept several times a day," whereas rubbish "is dumped outside the house onto the street" and remains untended except by passing animals that eat the organic bits . Unlike Das, however, Delaney's account does not feature a particular event or set of events, and is therefore somewhat historically unspecific. In this article, I aim to further explore the glimpses Delaney provides into the physical spatial implications of conspiratorial thinking in Turkey. In so doing I understand conspiracy theory in light of the discussion above: as a practice whose authors may not always be singularly determinable and whose causality is not always necessarily linear. This, moreover, is a practice that mediates events by operating across specific histories and anchoring the present in parts of the past which are qualitatively linked to it and to one another. In my case these histories center on an arson attack that took place in 1990s' Turkey, and the ways it has recently been commemorated on-site. Thinking conspiracy theory and spatial commemoration together in such a way has implications for memory studies, whose rapid growth over the past three decades has rendered it something of a natural home for scholarship on commemorative practices. Throughout these decades, memory studies arguably witnessed a gradual but significant shift; collective remembrance has come to be increasingly understood as not "a thing" but rather "a process" with "varieties, contradictions, and dynamism" , and as productive of not just "meanings" about the past but also "affects and effects" in the present . While this resonates with my proposal to rethink commemoration and conspiracy theory as interrelated practices, it has left intact a longstanding methodological assumption I aim to overcome-the assumption that the past and present in question each constitutes a discrete and singular event. In questioning this assumption, I take my cue from Robin Wagner-Pacifici . For Wagner-Pacifici, memory studies scholars overwhelmingly assume commemorative practices to pertain only to "the aftermath, after-effects, or afterlife of . . . actual events that have, essentially, ended" . For instance, representations of an event deemed complete, such as those produced through memorialization , are misunderstood as, or mistaken for, "memory" , whereas they are in effect "part and parcel" of its continuing development . The assumption becomes especially problematic in the case of violent events; while the violence that marks the "past" being commemorated is acknowledged, the violence inherent in the "cultural" work of that past's being named, appropriated, and displaced in its apparent "aftermath" is often overlooked . The problem is perhaps most evident in the specific body of literature on spatial memorialization. Monuments and memorials are explored in terms of their various processual and relational aspects, including the contestations and controversies surrounding their design and construction , their relationships to the urban settings in which they are located , and the myriad ways publics interact with them . This multifarious exploration has influentially demonstrated the significance of memorialization as pertaining not only to its representation of the past event but also to the various present-day events involved in its cultural production and reception. But it has left intact the methodological assumption that the event being memorialized, i.e., "violence," is complete and therefore ontologically distinct from the events surrounding its memorialization, i.e., "culture." My focus on conspiracy theory as thus far developed requires that this assumption be dispensed with. For, due to its non-linear causality and violently productive capabilities, conspiracy theory refuses to perceive events as complete and to settle with only explaining them. Rather, it contributes to their continuing development. What role, then, if any, might the physical environments where events as such take place play in this contribution? And how might an understanding of this role contribute to analyses of the work conspiracy theory does with respect to violent histories? --- Conspiracy theorizing in Turkey Conspiracy theorizing has become increasingly influential over the past three decades in Turkey such that it now has across-the-board resonance. The roots of this influence lie in the early 20 th century, which witnessed the transition from the Ottoman Empire to the Republic of Turkey, and more specifically in two events that marked this transition and over time became the stuff of conspiracy theories. The first is the 1908 constitutional revolution led by the Young Turks, which climaxed in the toppling of Abdülhamid II-the last Ottoman monarch to reign not just on paper but also in practice over the crumbling Empire-and laid the groundwork for the founding of the secular republic in the early 1920s . That the Young Turk movement was based in Salonika , then still a predominantly Jewish-inhabited city and antecedently the headquarters of a 17 th -century wave of religious conversions from Judaism to Islam, was made by anti-secularists into the stuff of the conspiracy theory that its members were crypto Jews acting in the interests of the global Jewry-a theory later extended to the modernday republic's Salonika-born founder Mustafa Kemal Atatürk . The second relevant event is the post-World War I Treaty of Sèvres, which partitioned much of today's Turkey into Allied-controlled territories. Although the Turkish National Campaign of 1919-22, which enabled the founding of the modern-day republic, was largely successful in overturning Sèvres, over time the treaty became the stuff of what is now known as the "Sèvres paranoia" : "the conviction that the external world is conspiring to weaken and carve up Turkey" . Often working in tandem with the above-mentioned anti-Semitic conspiracy theories , the Sèvres paranoia became especially prevalent in the post-Cold War era when the preceding decades' symmetrical political alignments among states began to complexify . It has continued to thrive in the new millennium, as previously peripheral conspiracy theories permeated the mainstream. This was possible first because a political movement rooted in religious conservatism, which has historically been among the proponents of the above-mentioned conspiracy theories as it was largely excluded from the secularist republican project for the better part of the 20 th century , rose to power in the shape of the AKP and consolidated its grip through consecutive majority governments -an ongoing phenomenon that has just entered its fifteenth year at the time of writing. Secondly, having now found themselves in the margins of the new political mainstream and facing ideological bottleneck, secular nationalists sought to regain influence by resorting to populism, the prime pillars of which in Turkey comprise conspiracy theories of the above-mentioned sort . Therefore, such theories now find appeal in Turkey among intellectuals, ordinary citizens, and politicians of various and otherwise highly different persuasions . This background is relevant to the Sivas case in at least two respects. The first concerns the historical context of the arson attack: the immediate aftermath of the Cold War when conspiracy theories began to gain wider popularity especially in geographies like Turkey "where cold-war [sic] disciplines and interventions shaped the experience of civil society" . It was in this context that conspiracy theory began to function increasingly as a practice grounded in "actual experience" rather than as a merely discursive representation of affairs . The second reason concerns the inaugural decade of the 21 st century when, simultaneously with the proliferation of conspiratorial thinking, long-fought campaigns for truth and justice regarding the Sivas arson attack and various other political atrocities in Turkey's 20 th -century history gained unprecedented visibility. The decade culminated in a state-sponsored project of what senior figures like the then Prime Minister Tayyip Erdoğan called "democratization" and "transparentization" . In 2011 these developments yielded results relevant to the site of the Sivas arson attack, whose conversion into a memorial museum had long been the subject of a campaign by those upholding the legacy of the festival participants killed in the blaze. In response to this campaign, and following a series of meetings with the city's "notables" and representatives of its "civil society" , the state transformed the site into a commemorative-cumeducational institution . --- Conspiracy theories regarding the Sivas arson attack and their spatial underpinnings It is against this background that the mainstream press in Sivas and the city's "notables" have espoused the sort of conspiratorial thinking that marks this article's opening quote, implying that the culture festival in 1993 whose guests were attacked in the arson, the arson attack itself, and subsequent initiatives of commemorating it on site are all the work of outside forces. This raises the question: outside of what or of where? To explore this question, consider, alongside the opening quote, the below excerpt also from the local press in Sivas, which is especially significant in this respect as its coverage of the culture festival in late June and early July 1993 is known to have aggravated the atmosphere that culminated in the arson attack . Written by the long-time chair of the Sivas journalists association, who also authored a monograph on the arson attack , the excerpt was published as the editorial of one of the city's best-selling newspapers a couple of days before the atrocity's tenth anniversary when the commemorations had begun to draw thousands of participants instead of the hundreds they had until then managed to attract. Neither the murderers of ASALA [the Beirut-based Armenian militant organization active between the late 1970s and the early 1990s], who were behind the events [the arson attack], nor the festival participants were invited to Sivas by its people. … These plotters are the same hitmen of ASALA who insist on keeping the events on the agenda. Let's all be level-headed and vigilant tomorrow [during the commemoration], and avoid falling into ASALA's trap. … Tomorrow, the perpetrators will likely mourn alongside them [those who claim the legacy of the festival participants]. … They first kill, and then weep at graves . As epitomized by both the editorial and the quote that opened this article, conspiratorial readings of the arson attack blame it on actors from outside Turkey. While in the above quote these actors are said to consist of ASALA, this has proven interchangeable with any other group considered "exogenous" at a given time, as evident in the blaming of the pro-Kurdish armed organization PKK in this respect by indeed the same writer cited above . Such readings of the affair therefore imagine Turkey as an immaculate and peaceful inside threatened by a hostile and polluted outside. It is this sociospatial imagination that has been mapped onto the physical space of central Sivas, as indicated in the above quote's emphases on the city and its people. These emphases originally implicated two sets of actors. The first were the alleged ringleaders of the arsonist crowd, on the basis that they were entirely unbeknownst to the people of Sivas and have not reappeared ever since, whether as suspects in court or as inhabitants of the city . Secondly, the organizers of the culture festival whose participants died in the blaze were labeled as outsiders, accused of both conducting an event many of whose aspects were allegedly foreign to the city's way of life and refusing to cancel it despite being reputedly given the opportunity to do so vis-à-vis the likelihood of a violent backlash which became increasingly apparent in the run-up to the arson . But, over time, such accusations began to also engulf the organizers and participants of the commemorations held annually in Sivas, arguably to the effect of amplifying, by association, the earlier insinuations leveled at the organizers of the culture festival in July 1993. Unpacking this effect requires first revisiting certain organizational and historical aspects of the 1993 culture festival. Organized in central Sivas by an Alevi association headquartered in Ankara, the festival was not the first but fourth of its kind. The inaugural festival had taken place in summer 1978 in Banaz in provincial Sivas, which is famously the native village of a 16 thcentury minstrel revered in Alevism . Although the festival had been intended as an annual event, it was interrupted due to sociopolitical unrest across late 1970s' Turkey, which partly also targeted Alevis in the country's central and eastern cities such as Malatya, Çorum, Sivas and Maraş and led many to flee for metropolises like Istanbul, Ankara, and Izmir, or even beyond, for Europe . Then followed the 1980 military coup, whose drastic restrictions on social rights precluded events like this festival. Therefore, the organizers had to wait another decade to repeat the event. Like the inaugural festival, its second and third episodes in 1991 and 1992, respectively, were held in Banaz and appealed mainly to Alevis. But in 1993, the festival board took the unprecedented decision of partly relocating to central Sivas and composing a two-day program that was not only Alevism-related. They thus facilitated what was for many an organizer and participant a return to the center of the geography that they or their parents had been forced to flee in the 1970s and 1980s. Part of a wider phenomenon called the "Alevi Revival" , it is this aspect of an outreach or return to the city that has led to the perception that the festival was the work of outsiders infiltrating Sivas . A similar perception began in the 2000s to color the way the commemorations held in Sivas were featured in the media, as the decade saw the annual event attract increasing numbers of participants each year. Evidence to back this perception has been derived from the significant role of the Cologne-based European Alevi Unions Confederation in organizing these events and drawing large crowds to them-an example of the financial and ideological support Alevis in Europe have provided to those in Turkey . To be sure, the main reason why such an organization exists in Germany in the first place is that, insofar as 1970s ', 1980s' and 1990s' migration from Turkey by members of marginalized social groups like Alevis is concerned, fleeing political violence was as significant a motivation as were better economic prospects . Indeed, several individuals with such a migration background were among the Sivas arson attack victims . That Europe-based Alevis, many of whom are indeed originally from Sivas, may continue to engage with Turkey is implicated by conspiracy theorists as an intervention by the EU and specifically by Germany, where they were given in the 1980s the opportunity for the first time to legally organize under their ethnic and religious identities and where many of them continue to reside . So mainstream has this perception become as to repeatedly feature in statements made by the long-time Prime Minister and current President of Republic Tayyip Erdoğan . In sum, conspiratorial readings of the arson attack have made two interrelated insideoutside delineations. The first delineation has featured the sociospatial imaginary of Turkey as inside, and has blamed the arson attack on its outside. The second has treated central Sivas as inside and, in so doing, has helped concretize the first delineation. In the next section, I unpack the ways these delineations have come to shape and be shaped by on-site commemorations of the atrocity, which assumed new significance in the early 2010s when discourses of "reckoning with the past" culminated in tangible developments. --- Conspiracy theory and spatial commemoration in Sivas My fieldwork in Sivas consisted of two components. The first was an ethnography of the site of atrocity within the daily life of Sivas. This involved spending entire workdays inside the building, whose recent state-sponsored transformation has opened it to free-of-charge visits between 8.30am and 5pm during the workweek, and in-depth conversations with a group of Sivas residents. The second component was to participant observe the site's significance within the annual commemoration, which I attended twice with groups coming from outside central Sivas. The site's recent transformation has introduced a "Memory Corner," a 70-squaremeter room whose centerpiece is a 3-by-4.5-meter stainless-steel structure complete with a victims' name list, an electrically operated set of thirty-seven fountains, and a couple of statements. There are as many fountains as there are names commemorated: thirty-seven. As previously mentioned, the number includes not only the thirty-three participants of the culture festival and the two hotel workers killed inside the hotel but also the two members of the crowd outside, a decision state authorities have defended as "a human-centric" refusal "to discriminate between the dead" . Of the two statements, one is more relevant to this article than the other and so will be explored at greater length. This statement is unsigned and reads "In the painful incident that took place on 2 July 1993, thirty-seven of our people have lost their lives. With the wish that such pains do not recur…" It is in effect a synopsis of the following speech that Minister of State Faruk Çelik delivered in 2010 when he became the first government representative to visit the site: 2 July 1993 is one of the painful days in our history … On this day, insidious foci sought to stage their dark scenarios. … [This] is the pain of the whole of Turkey. There can be no sides in this incident; to take a side in this incident means to not extinguish the fire. … The screen of fog surrounding this incident has not yet been lifted. … It is no other person than yet again us, the people of Sivas, who have the remedy for our problems. … I remember with grace and respect our thirty-seven citizens who lost their lives on 2 July 1993 . These emphases on "the people of Sivas" and/or "the whole of Turkey" as sufferers, on "our thirty-seven citizens" as victims, on the "dark scenarios" that constituted the arson, and on the "screen of fog" surrounding it echo conspiracy theories that have placed the blame on "exogenous foci" . The echo is discernible not only textually, i.e., in the statements, but also numerically, i.e., in the thirty-seven names and fountains. This has led those upholding the legacy of the thirty-three festival participants to refuse to enter the building that now hosts the Memory Corner, as they believe doing so would be to legitimize it. But there are also various others who pass by the building without ever setting foot inside, and do so due to certain assumptions they have about it. As some of these assumptions concern the building's exterior, it merits further exploration. The recent transformation preserved the façade's structural composition but reclad it entirely. The fenestration now consists of oneway mirror windows that render interiors invisible from the outside. The rest of the façade is clad in composite panels in the pastel shades of dark red and beige characteristic of state buildings. There are no signs outside the building regarding the services provided or the working hours. The only sign, except that which bears the institution's name, is a plaque that reads, "this building is monitored twenty-four hours by CCTV." Made of glossy brass, the plaque looks much more ostentatious than the usual CCTV disclaimer; I observed that it misled potential visitors who were confusedly drawn to it in the absence of any other sign. The confusion was most evident during lunch breaks when the institution closes its doors for an hour. Potential visitors would come, see the closed doors, search for a sign, only to notice the CCTV disclaimer, turn away perplexedly and leave. All these features led the building to exude an appearance that potential visitors found inaccessible at best and intimidating at worst. In 2011 and 2012 these issues were compounded due to the way the institution was staffed. In fall 2011, four people-a director, a kindergarten teacher and two primary school teachers-were employed here by the Ministry of National Education. When I returned next summer, the time of year when the building's visitor numbers are at their highest due to visits to the city by people with migration background who originally hail from Sivas, the director was still in place but the teachers had left-each visiting class was now asked to bring along its own teacher, I was told. In their stead, there were four new staff members: two men in their late twenties, a man in his mid-thirties, and a woman in her early twenties. They were all unskilled workers hired through the state's employment agency on contracts ranging from six to twelve months. There was no training scheme in place regarding the site's historical significance or anything else for that matter. Not much was expected of the employees except, in the words of the man in his mid-thirties, "to keep this place clean, tidy and orderly, as the director likes it that way." The director's occasional admonitions also included, according to the man in his late twenties, reminders about how "we should never speak with any visitors about the incident." If in its institutional inconspicuousness the building sought to downplay the contested legacy of the arson attack, this in effect exacerbated speculations regarding on whose side its occupants might be vis-à-vis the atrocity. Consider the case of a couple of women , who in summer 2012 passed by the entrance back and forth several times without entering inside. The female employee noticed them and went out to welcome them in. The younger woman explained that "we were here last year, too, but hesitated to enter. You see; my mom is covered, and although I'm not, we thought her headscarf could cause eyebrows to raise." 1 The employee replied that they welcome all members of the public indiscriminately. Standing at the doorstep, the women explained why they had second thoughts about entering the building. The mother said: I chair the AKP women's branch in my hometown. Plus, I am covered. Hence my belonging to the opposite camp. Nevertheless, I condemn this terrible atrocity. Being human is enough to condemn it, although you and I might not be of the same opinion. The staff member interrupted the woman's comments: "What do you mean by the opposite camp? To whom did you think this place belongs? The Alevis?" The women both nodded in affirmation. The employee corrected them: "No, no, no; not at all. This place belongs to the state." Only after having received this clarification did the women go inside. Many residents of Sivas have also yet to visit the revamped building. I observed this not only during my research inside the building, when I found out that only about one in ten visitors was a resident of Sivas, but also during the regular and lengthy evening conversations I had with men in their late twenties and early thirties at what is one of the oldest and most popular coffeehouses in the city. 2 Described by my interlocutors as "the local intellectual hangout," this is an alcohol-free establishment whose clientele consists virtually only of men. My interlocutors here included several outspoken supporters of the BBP , whose ideology synthesizes Sunni Islamism and Turkish nationalism . Sivas is famously the BBP stronghold and, at the time of my fieldwork, was the only municipality governed by one of its members. The party is immensely pertinent to the arson attack, owing to how some thirty survivors fled the blaze: through the air well to an adjacent building then occupied by the BBP. BBP supporters take pride in this story and present it as proof that they have done their utmost to mitigate the consequences of what was a dark plot orchestrated by external powers . The survivors, however, have recounted that they had to force their way into the building; at first, they were cursed at and chased away by those inside the party's offices . BBP supporters believe that such claims have unjustly stigmatized them as perpetrators, and caused innocent onlookers to receive heavy court sentences . Hence the interest my research stimulated among the members of my coffeehouse circle, whose numbers ranged from five to twelve per night. Many of them not only openly condemned the arson attack as the work of "exogenous foci" but also maintained that the AKP "government do not want 2 July to be elucidated," as evinced by the continuing lack of a comprehensive investigation that could "reveal the actual plotters." Regarding the recent transformation of the site of the blaze, my interlocutors were unanimous that it is not a project that speaks, or would be of interest, to "the people of Sivas:" The ordinary resident of Sivas will never go there. He/she would say "what's going on behind those mirrored windows?" and think that they are sacralizing that place and that the Culture Center is the penultimate stage before it becomes a cemevi [place of Alevi worship; literally: house of gathering]. Yes; many think that the place will soon become a house of worship where they will start performing semah [an Alevi ritual]. I personally have not been there to this day, and neither have any of the people I know. This range of reactions demonstrates that the architectural banality and institutional ambiguity resulting from the site's recent transformation has led to its being disowned by not just those upholding the legacy of the thirty-three festival participants but also various others. This banality and ambiguity, therefore, are not merely a consequence of conspiracy theory-the theory that the arson attack was an "exogenously focused" event-but also its instigator. If the project's implicit subscription to the conspiratorial reading of the arson attack as the work of "exogenous foci" is packaged in a refusal "to discriminate between the dead," the architectural banality and ambiguity embodying it have caused conspiracy theories to ramify rather than subside. The extent of this ramification is such that the project itself has triggered new conspiratorial rumors regarding how certain Alevis are collaborating with the authorities to hijack the building. The physical implications of conspiracy theories involving the building extend beyond its walls. Consider the commemorative procession held in Sivas annually by those upholding the legacy of the festival participants killed in the arson. Over the past decade this event has drawn increasingly larger crowds to Sivas from across Turkey and beyond. Following the hotel's commemorative-cum-educational transformation in 2011 amid state-endorsed discourses of reckoning with the past, the commemoration was unprecedentedly broadcast live by a popular nationwide TV channel. The event proceeds along a two-kilometer route starting in Alibaba, the reputed "Alevi neighborhood" of central Sivas and one of the scenes of the late 1970s' wave of violence against Alevis and leftists. It ends at the site of the blaze with the laying of flowers. Most commemoration participants then return to Alibaba in groups albeit not in the shape of a single collective, which they take during the procession. There are numerous ways this annual event shapes and is shaped by the insinuation that it is the work of outside forces. In both 2011 and 2012 the police conducted checkpoints at the border of central Sivas to stop and search those travelling to the city for the commemoration. Police officers invariably greeted people here by saying, "good morning our guests, welcome to Sivas." One of my interlocutors, a civil servant in his mid-fifties who indeed originally hails from Sivas but moved to Ankara in the early 1980s for sociopolitical as well as economic reasons, challenged the police: "Guests? Who says we're guests? I'm from Sivas. You may now be stationed here but who knows where you're originally from!" He later told me that, while these checkpoints have always been conducted, they used to be more technical than social affairs of the sort the officers' remarks sought to render them. My interlocutor therefore exposed the function implicit in what otherwise seemed a benign gesture of hospitality: to demarcate the space of the commemorations as the outside of Sivas. Such attempts at delineation continued throughout the day and became much more overtly physical, as streets leading to the route of the procession were double-barricaded. An interlocutor in his early thirties, who is a non-Alevi resident of central Sivas and lives in a building situated about halfway through the route, remarked that in 2011 he left home wanting to participate in the commemoration only to run right into barricades. Police officers told him that if he wants to join the procession he ought to go up to Alibaba, the reputed "Alevi neighborhood" where the procession starts. My interlocutor "didn't feel like doing so, as I don't know anyone there," and gave up the idea. Nearer the site of the blaze, the distance between the double barricades expanded to constitute a buffer zone across which commemorations participants and residents of Sivas gazed at one another. At times this spatial separation took forms less peaceful than the exchange of gazes. Disapproving gestures coming from some balconies along the route were met with a group of young commemoration participants who chanted "Sivas' arsonists-watching from balconies." 3 Finally, the procession itself became subject to an obstacle in what was perhaps the least technically justifiable "security" measure by the police. Whereas commemoration participants would formerly be able to walk right up to the doorstep of the hotel and lay flowers, from 2011 onwards-following the site's transformation-the authorities began to try and obstruct the procession nearer the building. In 2011 they mounted barricades about fifteen meters ahead of the building, blocking off the street where it is located. When the participants insisted on laying flowers at the building's doorstep as per tradition, the scuffle between them and the police resulted in the latter's use of teargas. The following year, the police moved the barricades about 400 meters further up along the route, only to pull it back to where they were in 2011 after an hour-long sit-in by commemoration participants. Even then, once the procession reached the street of the building targeted in the blaze, it emerged that the barricades had morphed from simple police shields into a steel wall, which has since then become a standard architectural feature of the site on commemoration day. Rather than serve their purported purpose of providing safety and security for the commemoration, the barricades therefore functioned as an instigator of aggression. They did so not only by isolating the procession and thus rendering it the outside of downtown Sivas, but also by enabling the conditions for the portrayal of commemoration participants as troublemakers aiming to provoke the city's residents. The portrayal continued to reverberate after the procession, as teargassed scenes captured by the press and especially by the mainstream TV channel that live broadcast the event circulated in news reports and national dailies the following day under such headlines as "18 years on, and yet another provocation" . --- Concluding remarks I have argued that conspiracy theories dwelling on "exogenous foci" did not only continue unabated but also ramified in Sivas throughout the early 2010s when the state-sponsored project of "democratization" and "transparentization" was in full swing. Contrary to what has been rightly suggested with respect to various other cases , however, this was not due to an adverse reaction to the project. Indeed, many of those who subscribed to conspiracy theories did not so much categorically disbelieve the promises of "democratization" and "transparentization" as they expected these to translate into tangible, judicial results in the shape of a comprehensive investigation into the full range of culprits. Conspiracy theories dwelling on "exogenous foci" ramified in early 2010s' Sivas because their central trope found physical expression in practices of spatial commemoration concomitant with the state's project of "democratization" and "transparentization." These practices included the annual procession, where checkpoints and barricades blurred the lines separating security from provocation, hospitality from hostility, and aggressee from aggressor. They also included the site of the blaze, whose architectural transformation resulted in an aura of bureaucratic anonymity and a sweeping approach to remembrance that echoed understandings of the arson attack as outsiders infiltrating the inside. If these developments may appear to have aimed at curbing the contestation surrounding the atrocity, they produced the exact opposite effect. And they did so not only among those upholding the legacy of the thirty-three festival participants but also among the adherents of conspiratorial thinking, who for instance began to embroil the architectural transformation of the site of atrocity itself, and those behind it, in the ways they theorized the arson attack as conspiracy. Conspiracy theory, then, seeps through the "state" as well "society," reiterating that these categories are empirically much less distinct from each other than often assumed . That is, it serves less as a challenge posed by "society" to the "state" than as a prominent mechanism through which to forge the former as distinct from, and at times even antagonistic to, the latter. This, moreover, is a spatial mechanism, and one which becomes especially pernicious in the case of violent events. By conceptualizing "society" as "inside," conspiracy theory flattens various other, empirically grounded distinctions produced by violence . By seeking physical evidence for this conceptualization in environments that have in effect been shaped-partly, if not wholly-by violent events, it naturalizes violence's spatial legacy. It is this flattening and this naturalization that a close and nuanced attention to conspiracy theory's spatiality might help unravel. --- Funding --- The fieldwork on which this article builds was partly funded by the Graduate School, UCL, the German Academic Exchange Service DAAD and the Istanbul-based art institution SALT's Research Fund. --- Works

This article discusses the relationship between conspiratorial thinking and physical space by focusing on the ways conspiracy theories regarding political violence shape and are shaped by the environments in which it is commemorated. Conspiratorial thinking features space as a significant element, but is taken to do so mainly figuratively. In blaming external powers and foreign actors for social ills, conspiracy theorists employ the spatial metaphor of inside versus outside. In perceiving discourses of transparency as the concealment rather than revelation of mechanisms of governance, conspiracy theorists engage the trope of a façade separating the space of power's formulations from that of its operations. Studying the case of an arson attack dating from 1990s' Turkey and its recent commemorations, this article argues that space mediates conspiracy theory not just figuratively but also physically, and as such serves to catalyze two of its deadliest characteristics: anonymity and non-linear causality. Attending to this mediation requires a shift of focus from what conspiracy theory is to what it does as a spatial practice.

Introduction Impacts related to natural hazards are increasing and threatening globally. Tailings dam failures used to be considered rare events . However, in the last 20 years, the number of accidents has practically doubled around the world . In the last six years, Brazil had two major environmental disasters caused by the collapse of dams . The impacts caused by the collapse of the Córrego do Feijão tailings dam in Brumadinho in 2019 demonstrate that although the areas affected by these disasters, or susceptible to them, are the center of actions and strategies that minimize their impacts, preparation, response, and mitigation efforts must also involve, at different stages, the local communities . The Brumadinho disaster highlighted that not only the environmental impacts are important, but also the socioeconomic inequality of populations is a determining factor. Therefore, including the social dimension in environmental disasters studies should be considered as an important step in the development of strategies based on a socially fair approach for different stakeholders, enabling to achieve more comprehensive and lasting results . Social vulnerability research emerges as a central issue to describe, through pre-existing conditions of the populations involved, the inherent capacity of systems to prepare, absorb, and respond to impacts caused by extreme events . The different degrees of vulnerability are associated both with the proximity and impact level of the events and the previous forms of social and economic organization of the communities, which are expressed by a set of assets, resources, and structures in which its access or insu ciency indicate a situation of socioeconomic inequality of the population involved . Further, socioeconomic inequality is re ected in the capacity or incapacity of communities to predict and respond to these events as well as to overcome the impacts on the environment, infrastructure, economy, and sociability. Understanding the factors that drive social vulnerability, caused by different levels of inequality, is essential in studies of disaster management programs, projects, and actions, as it demonstrates different capacities for involvement and participation of affected communities, by providing a more comprehensive and integrated portrait of its reality world . There is a relevant conceptual debate on the physical, and social components of disaster and about the use and development of indexes to measure social vulnerability . As in the United States, the use of social vulnerability indexes has been recently replicated and adapted to different regions, mainly in studies at the census sector level , rather than at the municipality level, such Social Vulnerability Index , developed by , and in studies carried out to assess the greatest environmental disaster in the country, at the Rio Doce watershed, in Mariana . However, empirical studies are still lacking and major challenges permeate the development of vulnerability indexes . Some challenges are: to recognize the importance of measuring social aspects in disaster management to attribute weights and to form a consensus in the development of variables in scales that allow the knowledge of the speci cities of the populations involved since the available methods tend to map at scales that are not relevant for implementation. Thus, given the lack of social indicators in disaster management, the main goal of this article is to ll this gap by proposing the construction of an index at the census sector level in Brazil. This article also shows the use of a set of internationally validated and acknowledged variables that were adapted to our context and scale of study. The index proposed here is more robust, as it incorporates, in addition to variables already approved in the literature and freely available for the entire country, some methodological adjustments, such as weights attributed by method of paired comparison by various experts, and census tract level scales, designed to provide more detail than municipal levels. This study aims to outline a set of metrics nationally and internationally validated to assess social vulnerability in environmental disasters to repair, mitigate, and prevent responses in municipalities exposed or susceptible to them. For this purpose, two study areas are evaluated, one where the disaster has already happened and another where it could happen. The rst is a major environmental disaster in Brazil, also considered one of the biggest in the world: the collapse of the Córrego do Feijão tailings dam in Brumadinho in 2019 . The second is the Casa de Pedra tailings dam in Congonhas, currently considered the largest open-pit mine in urban areas in Latin America . The unit of analysis will be formed by the census sectors inserted in the watershed of the Paraopeba River in Brazil . The paper is organized as follows: this Introduction; Section 2 reviews the literature on the concept and index of social vulnerability; Section 3 details the case studies, the choice of the database and set of variables, the weights, and methods for building the index and indicators; Section 4 explains the empirical results; Finally, we discuss and conclude in sections 5 and 6, respectively. --- Social Vulnerability Vulnerability can be understood as conditions determined by physical, environmental, social, and economic factors or processes that increase the susceptibility of a community or system to extreme events drivers . The Programa das Nações Unidas para o Desenvolvimento PNUD , completes this de nition by emphasizing that vulnerability is a condition or a process of human nature resulting from physical, environmental, social, and economic factors that determine the scale of damage from the impacts of these events, whether of natural or anthropogenic origin . Hence, vulnerability can be de ned as the susceptibility of social groups to potential losses due to damage caused by extreme events considering that the existing differences in the living conditions of individuals in uence their capacity to anticipate and deal with these events Although multiple de nitions of vulnerability have been proposed , here we understand it as pre-existing conditions among different groups of people in a community that tends to increase and/or reduce the susceptibility to losses as a result of a real or potential risk event, as well as her capacity to recover from these events . It is important to highlight that not all communities that live close to a disaster are under the same risk. Socially vulnerable communities will have more di culty dealing with the impacts of that event. Thus, social vulnerability provides a broader view of pre-existing conditions that in uence the different experiences of disasters. Social factors are related to different exposure and impact and slow or insu cient recovery . Social vulnerability is in uenced by many factors that, according to Singh et al. , are associated . Variables such as employment, and diversi cation of income ; quality of housing and access to treated water, electricity, garbage collection, adequate sanitation , and essential services such as schools and health facilities ; family structure ; and social structures, such as racial minorities and ethnic groups, are factors whose access or absence result in different levels of social inequality and are directly or indirectly linked to historical patterns that act as obstacles to change, to combat, and to recover from disaster contexts . --- Methodology --- Case studies in the Paraopeba Watershed, Brazil The Paraopeba River Watershed is a sub-watershed of the São Francisco River with an area of 12,054.25 km² that corresponds to 5.14% of the territory of the main watershed. The vegetation is a transition between the Cerrado and the Atlantic Forest biomes, both considered focal points for the conservation of global biodiversity . It covers 48 municipalities and 3,732 census sectors. Its population density is 93.24 hab/ km² and a total of 1.3 million inhabitants . To deal with changes in vulnerability in this watershed, we operate at the census sector level in two distinct perspectives: where the disaster has already happened and where it could happen . The rst perspective refers to a major environmental disaster in Brazil, also considered as one of the largest in the world, the collapse of the Córrego do Feijão tailings dam, which collapsed on January 25th, 2019. About 12 million cubic meters of tailings and mud were released from the dam and traveled 8.5 km to the Paraopeba River, extending for more than 300 km along the river towards the São Francisco River . The event resulted in the death of 262 people and the disappearance of another 8. Besides the environmental impacts, social problems are still in their beginning, as families, jobs, housing, among other factors were affected, resulting in many challenges for the society, the environment, and the mining company . The second perspective refers to the Casa de Pedra tailings dam. Created in 1946, in the municipality of Congonhas, it is the oldest operating mining area in Brazil and also the largest open-pit mine in urban areas in Latin America , hence, justifying its notoriety and the importance of studies, plans, and action policies to increase the preparation to deal with extreme events in susceptible areas. The Córrego do Feijão tailings dam and the Casa de Pedra tailings dam are in the Quadrilátero Ferrífero, in the state of Minas Gerais, Southeastern Brazil. Quadrilátero Ferrífero is one of the largest mineral provinces in the world . The failure of dams usually causes serious accidents . According to the National Mining Agency and the Minas Gerais Institute for Water Management , 108 Brazilian dams have a medium to high risk of failure ; from these, 65 have a high potential for associated damage. Currently, 264 dams fall into the high damage potential category . The Dam Safety Report classi ed the dams under study, including the one collpsed, as having a low risk of failure and a high potential for pollution, that is, high potential for human and ecosystem life loss, in addition to other impacts of social and environmental origin . --- Database and variable selection The statistical data for the indicators and indexes used in this study were obtained directly from the Brazilian Demographic Census , available on the website of the Brazilian Institute of Geography and Statistics . This data is the only o cial data source in the country that best represents the speci cities of the territory, due to its scale of action. Tasnuva et al. , when portraying social vulnerability in Bangladesh, also used data from the 2011 census as they are the most recent. All data are available in the form of electronic spreadsheets at the census sector level and will covered the period 2010. The data referring to the layers of the census sectors were obtained from IBGE, and those referring to the hydrographic limit, rivers, and watercourses were obtained from the National Agency of Water and Basic Sanitation in shape le format . To ensure that the indicators and the index were empirically valid and not redundant measures of the concepts, a review was made on important and validated variables that operationalize the indicators in the contexts in which they were studied. Subsequently, at the level of census sectors, the variables that are commonly used and endorsed in the national and international literature were selected. Besides the variables, our methodology was also different in terms of the weights to obtain the index by considering the understanding and analysis of specialists and the speci cities in the 3,732 census sectors that make up the Paraopeba River Watershed. We used the statistical environment R to select all variables, enter the weights and carry out the calculations. The ArcGis® 10.5.1 was used to spatialize and ll the result of sectors without information. 4 Results --- Variables and criteria and weights for obtaining indicators Table 2 shows the distribution of the weights of each variable to obtain the social, economic, and infrastructure indicators. We used AHP to assign the weights of the indicators considering a scale from 0 to 1. The criterion weights were assessed from data obtained from important literature on the subject, the analysis of experts, and the speci cities of the communities involved. Note: Weights were assigned to the variables to obtain the indicators. To calculate IVS, no justi cation from the specialists or literature that could increase and/or decrease the importance of one indicator over another was found. IV Si = ∑ n i=1 x i p i --- Variables, calculation of indicators, and social vulnerability index The results are grouped and color-coded to show the different vulnerability classes: very low , low , medium , high , and very high . The analyzes, at the census sector level, are discussed considering the two case studies: the susceptible dam, Casa de Pedra tailings dam; the one affected by the rupture, Córrego do Feijão tailings dam; and the municipality divisions and regional divisions of the watershed . In Table 2, we have an overview of the description of the chosen variables and the literature that validated them in different geographic contexts. The variables are positively related to social vulnerability, so the higher the values of the variables, the higher the social vulnerability. For variable II01 this method was not applied. Figure 3 shows the social, economic, and infrastructure indicators used to calculate the IVS across the watershed. The rst indicator, social, ranged from 0 to 1 with an average of 0.49 and a standard deviation of 0.12 . The second, the economic indicator, ranged from 0 to 1 with an average of 0.35 and a standard deviation of 0.17 . The third indicator, infrastructure, ranged from 0 to 0.1 with an average of 0.20 and a standard deviation of 0.19 . In the Paraopeba Watershed, the IVS ranged from 0.00 to 0.78 with an average of 0.34 and a standard deviation of 0.12 . Figure 5 detail the social, economic, infrastructure indicators and o IVS of area affected by the collapse and susceptible area the watershed. The results show considerable variability across the watershed. In rural areas, it ranged from 0 to 0.74, and in urban areas from 0 to 0.78. We noted that about 69.8% of the sectors are classi ed with very low and/or low vulnerability. About 26.8% of them are classi ed with medium variability and 3.3% as high and/or very high. Among the main, the municipalities with the greatest quantity of the most vulnerable sectors in the watershed are Curvelo and Felixlândia, and the least vulnerable ones are Betim, Contagem, Pará de Minas, and Sete Lagoas. We found that 47.1% of the sectors close to Brumadinho, the municipality affected by the disaster, are classi ed with very low and low vulnerability and around 47.1% with medium, and 5.7% with high. In the susceptible area , the results indicated the inexistence of very low vulnerability. It was found that 63.6, 30.3, 6.1% of the sectors are classi ed with low, medium, and high vulnerability, respectively . In Fig. 4, we noted that the IVS medium and high vulnerability classes are more present in the Lower Paraopeba and less in the high Paraopeba. In the High Paraopeba region, there were municipalities with highly vulnerable clusters . It was found that 3.9, 65.6, 28.0, 2.5% of the sectors from the High Paraopeba are classi ed with very low, low, medium, and high vulnerability, respectively; In the Middle, they correspond to 4.3, 73.4, 20.9, 1.3%; and in the Lower, at 1.6, 78.5, 46.4, 7.9%, respectively. --- Discussion Although the discussions and results suggest that social factors can shape and in uence populations to prepare, respond, and/or recover from disasters, the inclusion of these factors in research, planning, management, and mitigation is still scarce . The vulnerability mapping in the Paraopeba Watershed at a detailed scale offered a unique opportunity to identify characteristics in these communities that can interfere, positively and/or negatively, in the possible impacts from a disaster, whether in the context of preparation and response in areas susceptible to disasters or the correction, and mitigation in areas affected by the collapse of the tailings dam in the watershed under study. Our results corroborate the ndings by Kuhlicke et al. who stated that, so far, there is no consensus on the indicators of social vulnerability. In this study, special attention was given to the selection of variables. Another important issue that is different from other indexes and makes IVS an adjustable index for other contexts is the weight, which was assigned to each variable using process AHP, what is method of paired comparison by experts. The weights 0.06, 0.20, and 0.08 assigned to the variables IS01, IS02, and IS04 were based the greatest importance it IS02 have about other variables . So on the historical characteristics of the watershed population, which tend to act as obstacles to change, and the literature which shows that communities with a higher percentage of children, elderly, and female heads of households are more prospect to have low purchasing power and access to basic resources and assets, in addition to needing more care and assistance . Were estimated as equally important to variables IE01, IE03, and IE05, resulting in the attribution of equal weight. Based in on the fact that the effects of vulnerability to income were felt at all stages of the disaster. Furthermore, there is no scienti c basis in the literature that justi es different weight attribution. When economically disadvantaged populations are exposed to negative externalities from extreme events, it is impossible to expect them to be able to anticipate and respond effectively to external changes and threats without assistance. These families, mainly, reside in regions with precarious housing without infrastructure and without resources to help them in times of crisis . Although IE05 is not directly linked to inequalities like others used here, this variable is added by the age factor . The equal weight to variables II02, II03, II04, and II05 is based on the fact that we did not nd any justi cation that could increase and/or decrease the importance of one factor over another. As the studies developed by the Fundação Renova suggest, access to basic sanitation, infrastructure, and housing services such as treated water, collection via mains sewage or rainwater network, or even septic tank, and garbage collection should, in principle, to be available to everyone, as they are rst necessity. Their deprivation generates signi cant effects on the well-being of individuals, especially when this unequal access is enhanced by the impacts resulting from disasters . . As expected, we found that social vulnerability is inequitably distributed across the watershed with some places more vulnerable than others. The Middle Paraopeba portion, where the Metropolitan Region of the capital of Minas Gerais is located, presented a large discrepancy in the vulnerability levels. The population from Lower Paraopeba, on the other hand, has higher levels of infrastructure vulnerability than the one from the Middle and Upper portions. This inequitable pattern in the watershed was also veri ed within the municipalities , it is recommended that research, programs, and/or action plans for disaster preparation and recovery should not be based on a single strategy for the entire affected and/or susceptible territory, as there are signi cant differences in social inequalities between regions and also within the same municipality. The greater presence of sectors classi ed with medium and high vulnerability close to the disaster-prone area reinforces their need to receive high priority assistance. On the other hand, the lower variability and lower vulnerability of the sectors close to the affected area reinforce the argument that proximity to the disaster cannot be the only factor to be considered. This is an important result, as it shows that greater proximity to the disaster is not always spatially related to the most vulnerable populations . Vulnerability in the susceptible area is driven by social e indicators, however, economic and social indicators have a signi cant impact on improving the vulnerability of areas affected by the collapse of the Córrego do Feijão tailings dam. The vulnerabilities in the area affected by the collapse cannot be associated with the collapse of the Córrego do Feijão tailings dam in the municipality of Brumadinho since the local population already had a history of great social inequality caused by the region's development process . Although the data used in this research are from before the disaster, they can help us understand where the impacts will be more or less severe and what are the main vulnerabilities in each reality. Our results are designed to allow governments, researchers, and a wide range of stakeholders to manage measures and resources more effectively, to respond and mitigate the effects of disasters and eventual disasters more assertively and targeted to local characteristics. As in the studies by Thompson et al. , which portrayed the environmental aspects, our study shows that the impacts caused by the rupture were added to the impacts that already existed in the watershed. Sete Lagoas, Contagem, Pará de Minas, and even Brumadinho are among the municipalities with the highest number of less vulnerable sectors. Besides high population density and good quality of life, these municipalities have a developed economy with better literacy levels , for example. Curvelo, and Felixlândia, are the most vulnerable municipalities. The results obtained in this research suggest that these municipalities have higher concentrations of racial minorities , elderly, and children . Thus, communities in the Lower Paraopeba Watershed have less access to resources to prepare and/or rebuild after disasters. Additionally, the study showed that the highest levels of illiterate people are found in the lower part of the watershed. Therefore, the IVS and the indicators and their components, added to the studies by Cutter et al. , Morrow, , Bolin et al. , and Ainuddin and Routray , indicate the importance of higher levels of education , as low levels of education can create barriers during all phases of the disaster, for example, to understand the measures and procedures to be adopted and even in the resolution of bureaucracies. Ethnicity or racial minority can also affect vulnerability, however, its effects rarely occur on their own. The literature points out that this variable must be carefully analyzed and correlated with other variables, as race or ethnicity may not always be related to areas that are more or less vulnerable . The results of this study indicate a greater proportion of racial minorities in the Lower Paraopeba part, however, in the Upper part, close to the municipalities of Congonhas and Ouro Preto, there is also a large concentration of these minorities. Similarly, the lower part has the highest percentage of mothers with low levels of education, which tends to be associated with informal jobs with low salaries . Thus, our results re ect other studies that emphasized regional differences in economic and social indicators with the conclusion that many of the highest levels of vulnerability, in this study, are located in the lower and some parts of the Middle Paraopeba, while the high remains particularly less vulnerable to threats. This fact shows that, instead of following random spatial standards, the IVS pointed to the existence of regions with vulnerabilities ranging from low to high risk with signi cant implications for emergency planners and for the authorities responsible for identifying and directing resources to areas in need . Given the scenarios described, the results of this study emphasize how the socioeconomic differences of a population must be integrated into physical/environmental strategies for the prevention and mitigation of environmental disasters. The IVS showed high spatial heterogeneity throughout the watershed with signi cant clusters that evidenced the different vulnerability characteristics of the local communities. The social analysis of vulnerability is necessary, as it addresses the issue of how individuals and social groups predict, resist, and deal with a real threat as well as how to recover from them . Thus, the knowledge of previous forms of social, demographic, and economic organization of these communities, expressed by a set of assets, resources, and structures, can be considered an important factor to increase the opportunities to expand the possibilities of emergency management, correction, and mitigation to obtain more satisfactory results . --- Conclusion In this paper, a model for social vulnerability assessment was proposed and applied at the census sector level for the entire Paraopeba Watershed. Based on national and international literature and AHP criteria, we determined the weight of each variable and calculated the social, economic, and infrastructure indicators as well as the social vulnerability index. This study represented the rst approach to obtain and spatially assess social vulnerability to disasters at the census sector level in the Paraopeba Watershed and Brazil. After facing one of the biggest environmental disasters of the world, the results showed the location of the communities most susceptible to negative impacts due to their socioeconomic and demographic characteristics. Our results can help us understand where the consequences will be more or less severe if a disaster happens since vulnerabilities are an inherent condition of society and the impacts of the disaster add to the existing ones. Analyzing the watershed from two different angles: where the disaster has already happened and where it could happen , our ndings indicate that the areas closest to the disaster will not always overlap spatially with the most vulnerable ones. Furthermore, the data showed the existence of more vulnerable areas in the Lower part of the watershed while in the Middle and Upper Paraopeba the results showed high variability and less vulnerability, respectively. In the Upper part, more vulnerable clusters close to the susceptible area were evidenced. This research can serve as a basis for different stakeholders to expand their strategies based on a more socially equitable approach for preparing, correcting, and mitigating the consequences of impacts from extreme events to achieve more inclusive and permanent results. This social vulnerability assessment model showed the advantages of the proposed method since this type of study is still scarce in the Brazil as well as its applicability in the assessment of social vulnerability where the disaster has already occurred and where it may occur. The use of a census data source for the elaboration of the index and indicators aims to allow the higher level of detail replication of the proposed methodology throughout the national territory, as these are o cial data available for the entire country and freely accessible. It is noteworthy that the lack of recent to the set of factors is a limitation of the study which, on the other hand, allowed a greater level of detail and recognition of the speci cities within inside the municipality. So far, the existence of empirical studies on disaster social vulnerability in the Paraopeba Basin is unknown. --- Declarations Figures Figure 1 Geographic location of the Paraopeba River Watershed, detail area affected by the collapse of the Feijão tailings dam in the municipality of Brumadinho and susceptible area, Casa de Pedra tailings dam in the municipality of Congonhas. Figure 2 --- Data availability The datasets generated and/or analysed as part of the current study are available from the corresponding author upon reasonable request. Declaration of competing interest The authors have no relevant nancial or non-nancial interests to disclose. --- Author Contributions All authors contributed to the study conception and design. Material preparation and data collection were performed by Roque, M.P.B and Ferreira Neto, J.A. Vieira, W.C, Roque, M.P.B performed the analysis. All authors read and approved the nal manuscript.

The increase in the number of environmental disasters, in recent years, has led to a reorientation of research and programs. This study proposes a set of 16 theoretically signi cant variables to measure and map vulnerability in Brazil, after about a major socioenvironmental disaster in Brazil, also considered one of the biggest in the world. They were grouped into three indicators, social, economic, and infrastructure aspects of the population, obtained from the Brazilian census database. The Córrego do Feijão tailings dam, in Brumadinho, and the Casa de Pedra tailings dam, in Congonhas were taken as case studies. The vulnerability of 48 municipalities and 3,732 census sectors from the Paraopeba River Watershed were explored from two different perspectives: where the disaster has already happened and where it could happen. The index and indicators used were developed in the open-source software and the maps in a geographic information system (SIG). The results show considerable variability. The most vulnerable municipalities are in the lower part of the watershed while the High Paraopeba region is less vulnerable. This paper presents methodological and empirical contributions of the social dimensions of vulnerability the preparation, response, and recovery of affected communities to environmental disasters.

‫ﺑﺎﻟﻴﻤﻦ‬ ‫اﻟﺜﺎﻧﻮي‬ ‫اﻟﺘﻌﻠﻴﻢ‬ ‫ﺑﻤﺮﺣﻠﺔ‬ ‫اﻻﺟﺘﻤﺎﻋﻴﺔ‬ ‫اﻟﻤﻮاد‬ ‫آﺘﺐ‬ ‫ﻓﻲ‬ ‫اﻟﺪﻳﻤﻘﺮاﻃﻴﺔ‬ ‫ﻣﻔﺎهﻴﻢ‬ ‫ﺗﻮاﻓﺮ‬ ‫ﻣﺪى‬ ‫ﻟﻬﺎ‬ ‫اﻟﻄﻠﺒﺔ‬ ‫اآﺘﺴﺎب‬ ‫وﻣﺪى‬ The --- ‫اﻟﻤﻔﺘﺎﺣﻴﺔ‬ ‫اﻟﻜﻠﻤﺎت‬ : ‫اﻟﻤﻔﺎهﻴﻢ‬ ، ‫اﻟﺪﻳﻤﻘﺮاﻃﻴﺔ‬ ، ‫اﻻﺟﺘﻤﺎﻋﻴﺔ‬ ‫اﻟﻤﻮاد‬ ، ‫اﻟﺜﺎﻧﻮي‬ ‫اﻟﺘﻌﻠﻴﻢ‬ ، ‫اﻟﻴﻤﻦ‬ . --- Abstract This study aimed at identifying the availability of democracy concepts in the social studies books in secondary education in Yemen through the content analysis, and the extent of acquisition of students of these concepts, and whether there are statistically significant differences between the average gain for these concepts due to differences in sex and location of school ?. To check the availability of democracy concepts in social studies books the researchers prepared a list of the concepts of democracy that must be met in the books of social studies, The list has-been finalized to concepts, and the analysis of stability has been confirmed by coefficient Holsti . The sample books under analysis in this research were .The researchers adopted the results of content analysis of samples showing the availability of only concept, and of varying degrees, They also prepared a test consisting of items, and to ensure the stability of the test, they used Pearson's correlation coefficient, which showed reliability coefficient , and by using Cronbach coefficient stability reliability coefficient showed . Discriminatory force was calculated, accepting each item exceeding of discriminatory force and accepting each item ranged between at the level of difficulty. Three items were deleted, to leave the test at items. The results of content analysis showed that the second level books got to the first rank while the books ,of the first and third levels got to the second and third ranks respectively. Among the seven books the history book of the second level achieved the first rank. Among the concepts, that of Shura got to the first rank. As for the results related to the second question the arithmetic mean of students acquisition of these concepts has reached which is less than the acceptable level educationally and specified in this research as . The results --- ‫اﻟﻤﻘﺪﻣﺔ‬ --- ‫ﻳﺄﺗﻲ‬ ‫آﻤﺎ‬ ‫اﻟﻨﺘﺎﺋﺞ‬ ‫وآﺎﻧﺖ‬ :

indicated that there are significant differences between the averages of students due to the variable of sex in favor of females, but did not show any significant differences attributable to the variable location of school. The researchers have recommended the following: social studies books should to include concepts which did not appear adequately. They also recommend drawing up a syllabus for course in civics. And finally it should be taken inconsideration that concepts should be presented through the same significance.

INTRODUCTION Among the factors associated with obesity, significant attention has been given to birth weight. High birth weight , especially greater than or equal to 4 kg, 1 has been correlated with overweight and/or obesity over the long term, in epidemiological studies. [2][3][4][5] Particularly in developed countries, HBW is the main risk factor for obesity among children and adolescents whose mothers were affected by gestational diabetes mellitus, because of the elevated number of adipose cells acquired by the infant. 6 However, other factors may overlap the association between birth weight and overweight/obesity, such as the mother's body mass index , 7 both the mother's and the father's BMI, 8 family income 3,8 and the type of school attended, with regard to developing countries. 9 These factors may reduce the importance of the birth weight variable in multivariate analysis. 3 Systematic reviews have shown that correlations between birth weight and overweight/obesity among children not only have presented contradictory results, but also have differed regarding the methods used. Thus, these reviews have demonstrated that research on this subject needs to be conducted with the proper methodological rigor, in order to identify factors that might reduce the effect of birth weight on overweight or obesity among children and adolescents. 10,11 Because of the contradictions in these studies, it is important to clarify the association between birth weight and overweight/obesity, so as to determine the stage of life at which this relationship first appears. Longitudinal studies have revealed these associations in adulthood. 12,13 A study conducted in the city of Florianópolis in the year 2002 by de Assis et al. 14 , among children aged 7-9 years, found that the prevalence of overweight was 22.1%, in accordance with the criteria of Cole et al. 15 Data from this city's live births registry, obtained in 2005, showed that the prevalence of low birth weight among live births in Florianópolis was 8.3%. 16 These proportions are similar to those found in developed countries, where an increase in HBW has been observed with a simultaneous decrease in LBW. 17,18 Florianópolis was classified as the fourth highest city in Brazil in terms of human development in 2000 and the third city in 2013. 19,20 Thus, the importance of the present study lies in the fact that in Florianópolis, elevated HBW prevalence that could be correlated with high rates of overweight and obesity would be expected. --- OBJECTIVE The purpose of this study was to investigate whether overweight and obese children and adolescents aged 7 to 14 years living in Florianópolis, Santa Catarina, southern Brazil, were born with low or high birth weight, taking into consideration the possible interactions between socioeconomic factors and other biological variables. --- METHOD Design, setting and ethics This study was conducted on a retrospective cohort from the year 2007. An assessment was made on a probabilistic sample of schoolchildren aged 7 to 14 years who were enrolled in public and private elementary schools in Florianópolis, Santa Catarina, southern Brazil. The schoolchildren in this investigation were included after obtaining consent from their parents or legal guardians, who signed a free and informed consent statement. This study had previously been approved by the Ethics Committee for Human Research of the Federal University of Santa Catarina through project number 028/06. --- Sample The sampling was divided into two stages. Firstly, the 221 schools in Florianópolis were listed according to their geographical location in the city and their kind of institution . The number of schools selected in each of the four geographical areas was defined taking into account: the proportion of schools in each area compared to the totality of schools in the city and the proportion of private and public institutions in each geographical area, totalizing 17 schools . The selection of the 17 schools was made randomly, by simple draw, but taking into account the previous list . Subsequently, students in each school were selected taking into consideration the ratios of schoolchildren registered in the 2004 school census in Florianópolis in the following categories: geographical location of the student's home, kind of school, gender and age group. Sample size was calculated taking the prevalence of overweight among children aged 7 to 9.9 years to be 10%, 21 and 17% for adolescents aged 10 to 14 years, 22 with 95% confidence levels and a two-tailed sampling error of 2%. The design effect was estimated to be 1.3 and the power was taken to be 80%. This calculation resulted in a requirement for a sample size of 1,100 children aged 7 to 9.9 years and 700 adolescents aged 10 to 14.99 years. In addition, all the children who had participated in a previous study conducted in 2002, 14 and who in 2007 were adolescents enrolled in these randomly selected schools were also included. It was possible to locate 30% of all the students who had participated in the previous study in these randomly selected schools. Considering the error margins for losses in tests, the total sample was estimated to be 1,200 children and 1,900 adolescents . Data on 2,863 students were collected. Those younger than seven years of age and those who were 15 years or over were excluded, as were those for whom no valid data for weight or birth weight was available. Thus, the final sample investigated comprised 2,696 students , i.e. 94% of the total. --- Data collection Biological data and socioeconomic data were collected. Information relating to birth weight, gestational age, age, parents' weight and height and socioeconomic data were collected by means of a self-administered questionnaire that was sent to the parents and legal guardians of the students. Anthropometric data on the students were collected by previously trained anthropometry technicians, 23 following a protocol based on the recommendations of Lohman. 24,25 Weight was assessed using a Marte electronic scale, model PP 180 . with a capacity for 180 kg and precision of 100 grams. Height was measured using an AlturExata stadiometer with 1.0 millimeter precision. Skinfolds were measured using a Cescorf caliper with 0.1 millimeter precision. The children and adolescents were measured without shoes and wearing light clothes. A pilot study was conducted among subjects who were not included in the sample and, following the recommendations from the World Health Organization, 26 the intra-rater technical measurement error for skinfold measurements was calculated. The intra-rater TME showed a reliability coefficient greater than 0.95, which showed that all of the anthropometry technicians made skinfold measurements properly, thus resulting in low variability in the data. 23 The data were entered into EpiData 3.2 and were fully checked by the duly trained data entry team, and automatic consistency and amplitude checks were made. --- Statistical variables and analysis The outcome variables were overweight and obesity. Overweight was defined as BMI ≥ 85 th percentile, according to age and gender, as proposed by Must et al. 27 and as recommended by the Brazilian Ministry of Health until the year 2008. 28 Obesity was defined as BMI ≥ 85 th percentile, according to age and gender, as proposed by Must et al., 27 along with triceps and subscapular skinfolds in millimeters ≥ 90 th percentile as proposed by Johnson et al. 29 The criteria of Must et al. 27 were chosen because they are widely used in the literature and have been recommended by the World Health Organization. 24 These were also recommended by the Brazilian Ministry of Health 28 for evaluating nutritional status among children and adolescents until 2008, when this study was conducted. Birth weight and birth weight according to gestational age were the independent variables and were classified as follows: a) birth weight was classified as a single piece of data, as LBW , appropriate birth weight and HBW ; 24,30 and b) birth weight was correlated with gestational age in order to classify schoolchildren as small for gestational age , i.e. below the 10 th percentile; appropriate for gestational age , i.e. between the 10 th and the 90 th percentiles; and large for gestational age , i.e. above the 90 th percentile. 30 The World Health Organization classifies children as presenting insufficient birth weight if they are born weighing between 1,500 and 2,999 g and as presenting low birth weight if they are born weighing 2,500 g or less. 24 In our study, these categories were unified because the number of responses was low and this could have hidden an association between the outcome and the independent variables. The BMI of the students and their parents was obtained by dividing the weight measurement by the square of their height . Non-elderly parents were classified as overweight if their BMI was between 25 and 30 kg/m 2 or obese if their BMI was ≥ 30 kg/m 2 , as proposed by the World Health Organization, 24 and elderly parents were classified as overweight if their BMI was ≥ 27 kg/m 2 , as proposed by the American Dietetic Association. 32 Both of these sets of criteria are recommended by the Brazilian Ministry of Health. 28 An analysis on the consistency of the database was made using the Stata version 9.0 statistical package. A descriptive analysis was conducted to show the prevalence ratios of overweight and obesity for each independent variable, and the prevalence of LBW, HBW, SGA and LGA. In an inferential analysis, models divided by age group and gender were used to calculate associations among children aged 7 to 9.9 years and, separately, among adolescents 10 to 14.9 years, because of the possible effect of sexual maturation among the adolescents and the differences between genders regarding the prevalences of overweight and obesity. In order to compare the prevalences of overweight and obesity for the different categories within each variable studied, an analysis was performed using Pearson's modified contingency coefficient, based on chi-square statistics. Poisson univariate analysis was used to investigate association between birth weight and overweight and obesity . Multivariate analysis was performed to ascertain the extent to which the exposure variables influenced the outcomes. Additionally, 95% confidence interval and P-values were estimated. The Poisson analysis model was used because for high-prevalence outcomes in cross-sectional studies , odds ratio estimates are said to either overestimate or underestimate associations with outcomes, in comparison with prevalence ratios. 33,34 All the analysis took into consideration the effect of the sampling design, through the svy command in the Stata software, which is used to analyze data from complex samples. Associations among the variables for which the P-value was ≤ 0.05 were considered to be statistically significant. --- RESULTS Data on 2,863 children and adolescents were collected. The proportion of the data that comprised refusals or exclusions was 5.8%. Students less than 7 years of age and more than 14.9 years of age were excluded because these ages were outside of the study range. Invalid weight data and birth weight were also excluded. Occurrences of no response or discrepant values were considered to be invalid data. Thus, 2,696 students were studied. In relation to the initial number calculated , this study presented a data loss rate of 13%. Table 1 describes the data on the population studied. The LBW prevalences among the children and adolescents were respectively 8.2% and 7.8%. HBW was found in 7.4% and 7.9% of the children and adolescents, respectively. The prevalences of children and adolescents who were born SGA were 14.4% and 13.0%, respectively. Children and adolescents who were born LGA accounted for respectively 10.2% and 12.9%. The prevalence of overweight among the children was 31.5% and of obesity, 10.9%. Among the adolescents, the prevalence of overweight was 21.0% and of obesity, 6.0%. Table 2 shows the prevalences of overweight and obesity among the children according to gender and the variables investigated. In this table, it can seen that the prevalence of overweight among children was significantly higher in the following cases: boys born LGA ; children of both genders whose mothers were obese ; and boys whose fathers were obese . The prevalence of obesity was higher among girls whose mothers were obese ; and among boys whose fathers were obese . The other variables investigated did not demonstrate any significant association with overweight or obesity in children . Table 3 shows the prevalences of overweight and obesity among the adolescents . The prevalence of overweight was higher among the following: boys with HBW and LGA ; boys in the top monthly per capita income quartile ; boys whose mothers and fathers were obese; girls who attended public schools ; and girls whose mothers or fathers were obese. The prevalence of obesity was higher among the following: adolescents with HBW ; LGA ; and those whose mothers or fathers were obese . analysis , it ceased to be correlated after controlling for confounding. The father's and mother's BMI were probably the variables that were most strongly associated with the outcomes. Table 5 shows the prevalence ratios for obesity among adolescents and their adjusted associations with birth weight. A significant association was found between birth weight and overweight among male adolescents, for those with HBW (PR = 1.14; 95% CI = 1.02-1.27; P = 0.03). However, it needs to be noted that this association observed between HBW and overweight among male adolescents was weak, because the prevalence of overweight among the male adolescents born with high birth weight was only 1.14 times greater than the prevalence of overweight found among adolescents born with appropriate weight. Hence, birth weight does not seem to be the principal factor that determines overweight in adolescence. Birth weight and birth weight according to gestational age were not significantly associated with These data show that a higher number of variables were associated with overweight and obesity among adolescents than among children . Among children, the biological variables were those that established a positive, proportional and significant association with the prevalences of overweight and obesity. Among adolescents, socioeconomic variables were also associated in addition to the biological variables. The main variable of interest seemed to be associated only among males. Table 4 shows the prevalence ratios in adjusted analysis on the outcomes of overweight and obesity among children and their associations with the exposure variables. No association between birth weight and overweight or obesity was found among the children in multivariate analysis. Thus, although birth weight according to gestational age was correlated with these outcomes in univariate * P = 0.01; † P = 0.001, ‡ P = 0.001, § P = 0.001, || P = 0.01, ¶ P = 0.001, ** P < 0.001, † † P = 0.003, ‡ ‡ P < 0.001, § § P < 0.001, |||| P = 0.002, ¶ ¶ P < 0.001, *** P = 0.003, † † † P = 0.03. --- DISCUSSION Investigating the influence of birth weight on overweight and obesity among children and adolescents is important, according to public health studies. This study in Florianópolis was justified by the city's high Human Development Index, which reached the fourth position in Brazil in 2000 and this index is continually increasing. Currently, Florianópolis is the third city in the country in terms of human development. 19,20 Moreover, the prevalence of newborns who were LGA , was similar, in 2000, to that found in developed countries. 17,18 The LBW prevalence found in this study is similar to that recorded in the Ministry of Health's Live Births Registry for the city of Florianópolis, which was 8.3% in 2005, 16 and the same has been found in developing countries such as Argentina and Mexico. 35 However, it must be pointed out that intrauterine growth restriction, which is an indication of poor nutrition during pregnancy, reached 27.4% in this study. The prevalences of overweight and obesity found in this survey were similar to those found in other studies on schoolchildren that used the same diagnostic criteria. For example, an assessment made by Dutra et al. 5 36,37 These findings may indicate that overweight and obese infants also tend to have overweight and obesity during childhood and adolescence. Regarding the influence of birth weight on the prevalence of overweight in the sample investigated, a significant association was found in the univariate Poisson regression analysis for male adolescents who were born LGA. No association between birth weight and overweight was found among children of both genders, or among female adolescents. In an analysis adjusted for the parental socioeconomic and biological variables, the most significant association found was between HBW and overweight among male adolescents. Additionally, associations with obesity were found in univariate analysis for male adolescents who were born LGA and for those with HBW. However, after inclusion of parental socioeconomic variables and parental BMI in the multivariate analysis, the associations with obesity ceased to be significant. Gilman et al. 4 conducted a cross-sectional study on a cohort 13 years and found that birth weight between 2,890 and 3,110 g and birth weight ≥ 3,110 g were associated with overweight according to the cutoff points set by Cole et al. 15 , in a multivariate analysis. Monteiro et al. 3 obtained similar results to those of the present study, in a cross-sectional investigation embedded in a cohort of children born in the city of Pelotas in 1982. Using the same diagnostic criteria for overweight and obesity as used in this study, they also found through a univariate analysis that there was an association between birth weight according to gestational age and obesity among adolescents aged 14 to 16 years. However, the association did not remain significant after adjusting for family income and maternal BMI. The present study showed that being born with high birth weight is a factor associated with overweight among male adolescents. However, this association seemed to be weakened by the variables of obesity in the mother and obesity in the father, in the multivariate analysis. In childhood, this association between birth weight and overweight/obesity was not observed. This seems biologically plausible, since there are reports in the literature showing a strong association between these variables in adulthood. 39 So, even though it was not so strong, the relationship between overweight in male adolescents and being born with high birth weight indicates that health services should promote nutritional monitoring among adolescents with a focus on lifestyle, in order to reduce the chances of overweight and obesity in adulthood. In addition, it is recommended that epidemiologists and other researchers should investigate whether adolescents with elevated birth weight were born from mothers who presented gestational obesity or gestational diabetes, because the origins of overweight may be found in intrauterine development. In relation to birth weight and parental weight and height, it should be stressed that this information was reported by the children's legal guardians through a self-administered questionnaire. This method was chosen in order to make it easier to collect data and administer the data collection instrument. Araújo et al. conducted a validation study on the degree of agreement between reported birth weight information and birth weight measurements made immediately after birth, among eleven-year-old adolescents who were part of a cohort in Pelotas and found a high level of agreement regarding information on low birth weight , with disagreements of the order of -20.0 g . 40 However, when stratifying birth weight information, the authors found that reported information tended to be overestimated in the case of LBW children and underestimated in the case of HBW children. They pointed out that the linear relationship between birth weight and BMI was more consistent when birth weight was measured immediately after birth and not so precise when birth weight was reported. Therefore, validation for birth weight measurements in a subsample is highly recommended, so as to compare the reported variable with data gathered by hospital registries or on health record cards. In spite of these limitations, the present study had external validity, was probabilistic and had a complex sampling plan. Additionally, the researchers took care to train the anthropometry technicians, which resulted in reliable data. In addition to appropriate data gathering, care was taken in compiling the research data, which were fully entered and checked by a trained team, using software that enables data entry control. Therefore, the results from this study seem not to have been influenced by selection or measurement bias. --- CONCLUSION No significant association was found between high birth weight/being born large for gestational age and obesity after adjusting for the control variables, either among children or among adolescents. The same results were found for low birth weight and being born small for gestational age. Even though the association was not very strong, a relationship between high birth weight and overweight among male adolescents was observed. between the body mass index of schoolchildren aged between 7 and 14 years and that of their parents in the city of Florianópolis, in the State of Santa Catarina, Brazil]. Rev Bras Saúde Matern Infant.

CONTEXT AND OBJECTIVE: Being born heavier than 4 kg is associated with current overweight and obesity over the long term. The objective here was to ascertain whether birth weight was related to overweight or obese status, among 7 to 14-year-old schoolchildren, taking into consideration the possible interactions between socioeconomic factors and other biological variables. DESIGN AND SETTING: Retrospective cohort study on a probabilistic sample of 2,696 children and adolescents living in Florianópolis, Santa Catarina, Brazil.The following data were collected: anthropometric (student's weight, height and age; and parents' weight and height), socioeconomic (family income, number of people in house and parental schooling level), birth weight and gestational age. Overweight and obesity were classified using percentiles of body mass index and triceps and subscapular skinfolds. The outcome variables were overweight and obesity and the main explanatory variables were birth weight and birth weight according to gestational age. The control variables were the parents' nutritional status, their schooling level and the per capita family income. Poisson multivariate regressions were carried out. RESULTS: Higher prevalence of high birth weight was observed among overweight male adolescents (PR = 1.14; 95% CI = 1.02-1.27; P = 0.03), but this was not observed among obese male adolescents. Low birth weight and being born small for gestational age were also not associated with the outcomes. Among overweight and obese children, birth weight was not significantly different from that of normal-weight children. CONCLUSION: No significant association between birth weight and obesity was observed. However, there was a weak but significant association between high birth weight and overweight, among male adolescents. RESUMO CONTEXTO E OBJETIVO: Nascer com peso > 4 kg se associa a sobrepeso e/ou obesidade em longo prazo. O objetivo foi verificar se o peso ao nascer foi um fator relacionado ao estado de sobrepeso ou obesidade atuais, em escolares de 7-14 anos, levando em consideração interações possíveis com fatores socioeconômicos e outras variáveis biológicas. TIPO DE ESTUDO E LOCAL: Estudo coorte retrospectivo numa amostra probabilística de 2.696 crianças e adolescentes residentes em Florianópolis, Santa Catarina, Brasil. MÉTODOS: Coletaram-se dados antropométricos (peso, estatura e idade dos escolares, peso e estatura dos pais), socioeconômicos (renda familiar, número de pessoas na casa, escolaridade dos pais) e de peso ao nascer e idade gestacional. Sobrepeso e obesidade foram definidos utilizando-se os percentis de índice de massa corporal e as dobras cutâneas tricipital e subescapular. Sobrepeso e obesidade foram desfechos, e peso ao nascer e peso ao nascer/por idade gestacional foram variáveis explanatórias principais. As variáveis de controle foram: estado nutricional dos pais, seu nível de escolaridade e renda familiar per capita. Foram realizadas regressões de Poisson. RESULTADOS: Nos adolescentes do sexo masculino com sobrepeso, houve maior prevalência de elevado peso ao nascer (RP = 1,14; 95% IC = 1,02-1,27; P = 0,03). Naqueles com obesidade, não foi observado o mesmo. Baixo peso ao nascer e nascer pequeno para a idade gestacional também não se associaram aos desfechos. Nas crianças com sobrepeso ou obesidade, o peso ao nascer não diferiu daquele entre as eutróficas. CONCLUSÃO: Não se observou associação significativa entre peso ao nascer e obesidade. Porém, houve associação significativa, ainda que fraca, entre elevado peso ao nascer e sobrepeso, nos adolescentes do sexo masculino.

Introduction Studies have consistently linked adolescents' high academic expectations or aspirations with higher academic achievement [1][2][3], and more positive mental health [4][5][6]. A better understanding of the potentially modifiable factors that influence adolescents' academic expectations may be important for the development of potential educational and mental health trajectories. Although research on the determinants of academic expectations is relatively scant, several previous studies have determined some important potentially modifiable antecedents of academic expectations or aspirations. Parental attitudes towards education may be of particular importance for adolescents' development of academic expectations. In fact, parental expectations for their children appear to be one of the most consistent predictors of adolescents' own academic expectations [7][8][9]. Parental involvement in their children's education, however, is less clearly understood. While this has been described as one of the most important predictors of academic expectations [10], others have reported null findings [11,12]. According to the Expectancy-Value theory, academic motivation and achievement are predicted by the combination of ability self-concept and the subjective value put on certain tasks [13]. Previous academic achievement is likely to inform ability self-concept, and academic engagement may be an important aspect in the valuing of education and schoolwork. Previous studies have found intrinsic motivation towards schoolwork [7] and own academic ability or achievement to be important contributors to adolescent academic expectations [7,8]. Some studies, however, cast doubt on the extent to which adolescents actually adjust their academic expectations according to their own ability or achievement [14,15]. Mental health may also be an important aspect in relation to the development of academic expectations and goals and has previously been found to predict academic expectations [16]. Identity development is another factor which is closely linked to educational and vocational choices and future planning in general [17]. According to Erikson's psychological stage theory, adolescence is the developmental stage where young people primarily face conflicts between identity synthesis and confusion. It is important at this stage for them to establish a sense of who they are and what their goals are [18]. Though identity formation is a very individual process, at least one school-based intervention has succeeded in modifying identity development [19]. Few quantitative studies have investigated aspects of identity formation in relation to academic expectations. During adolescence, academic expectations from external sources often become more demanding, while intrinsic motivation in school [20] as well as academic self-concept [21] tend to deteriorate. At the same time, specific goals and future planning are likely to become more definite, as decisions become closer and more central in the lives of adolescents [8]. Along with increasing rational thinking, some adolescents may adjust their expectations to more realistic or achievable goals over time. Previous studies have come to conflicting conclusions about whether own expectations on average tend to increase [22], decrease [23] or stay the same [10,14] during early adolescence. Our understanding of potentially malleable factors that influence the development of adolescent academic expectations, however, remains poor. For instance, few studies have addressed the question of which factors explain positive changes in academic expectations during the critical time of adolescence. One particularly interesting change at this age is the transition between uncertain expectations to more definite visions about own future educational development, a topic often neglected in this type of study. On one hand, uncertainty in early adolescence may be a natural response related to healthy identity formation and adaptiveness [24]. On the other hand, prolonged uncertainty may lead to delayed goal setting and subsequent undesirable outcomes such as poor academic performance and attainments later on [7]. A better understanding of which factors influence an improvement in academic expectations before the transition to high school, including resolving uncertainty would be important in order to design consistent strategies to promote positive adolescent educational goals. In Sweden, 9 th grade marks the end of compulsory school. At this point the students choose whether or not they want to attend upper-secondary education and if so, whether they would like to follow a vocational track, or a theoretical track, this latter often in preparation for university studies [25]. The time between 7 th and 9 th grade is therefore particularly critical for the development of educational plans in the Swedish context. This study aims to explore potentially modifiable factors in the family environment and at the individual adolescent's level that may promote positive change in academic expectations during the final block of the Swedish compulsory school, i.e. between 7 th and 9 th grade. Specifically, we explored whether parental involvement with school, parental expectations, student engagement, academic grades, sense of identity, and mental health are associated with an improvement in academic expectations or with a resolution of uncertainty in these expectations. --- Methods --- Study population and design This study is part of the KUPOL cohort study . Details of the study design and the data collection methods have previously been published [26] and will be summarized here. One hundred and one schools were recruited from southern and central Sweden. Within these schools, the parents of 3,959 students gave informed consent for their children's participation in the study during the 2013-2014 or 2014-2015 academic years. Of the students allowed to participate, 3,671 answered the baseline questionnaire in the 7 th grade . Yearly follow-up surveys were conducted in the 8 th and 9 th grade using similar questionnaires. There were 3,203 adolescents who participated both in the baseline and in the second follow up survey. Parents of the index adolescents also answered a similar questionnaire each year with 3,645 answering the baseline questionnaire and 2,977 answering at both the baseline and 2 nd follow-up. The KUPOL study was approved by the Stockholm Ethics Review Board . --- Measures Outcome. We measured academic expectations by asking adolescents both in 7 th grade and in 9 th grade how far they thought they would go in their education with the possible responses; "I don't know", "high school-vocational track", "high school-theoretical track", and "university". From these answers we derived two outcomes representing positive changes in academic expectations in two sub-populations. We first considered those with uncertain expectations in the 7 th grade and compared those who remained uncertain with those who endorsed a specific expectation in the 9 th grade. Next, we considered those who reported some definite expectation in the 7 th grade and compared those who raised their expectations between the 7 th and the 9 th grade with those who did not. We excluded those participants who reported university expectations at baseline from the main analysis, as it was not possible for them to experience any upward development in expectations. However, in secondary analyses we compared those who lowered their expectations between 7 th and 9 th grade with those who sustained or raised their expectations . Predictors. Parental involvement in their child's education was measured using the childreported Family Support for Learning subscale of the validated Student Engagement Instrument [27]. This subscale measures how engaged and supportive children feel that their families are concerning school life using four items with five response alternatives ranging from "strongly agree" to "strongly disagree". The total score is calculated as a mean of the four items [28] where a higher score indicates more family engagement in school. The Cronbach's alpha was .79 in this sample. Parental academic expectations for their children were measured by asking how far parents expected their children to go in school. We categorized this variable as "university expectations" vs. "less than university expectations". Adolescent engagement with school was measured using the Future aspirations and Goals subscale of the Student Engagement Instrument. This subscale includes five items related to future academic plans, general future prospects, and the importance put on school. A sample item is "School is important for achieving my future goals". Items are answered on a five-point scale ranging from "strongly disagree" to "strongly agree". The scale score is an average of the five items with a higher score indicating higher engagement. The Cronbach's alpha was .79 in this population. Academic grades obtained during the 7 th grade were retrospectively reported by the adolescent in either 8 th or 9 th grade for the subjects Swedish, English, and mathematics. Each grade was represented by a numeric value that was then added across subjects to derive a total score . Identity synthesis was assessed using five items from the Identity subscale of the Erikson Psychological Stage Inventory. This subscale indicates how secure the child feels with his or her own identity and has previously been validated [29]. The score was calculated as a mean of the five items, with a higher score indicating stronger identity synthesis. The Cronbach's alpha for this scale was .71. Mental health was self-reported by the adolescent using two measures. The well validated Center for Epidemiological Studies Depression scale for Children [30] was used to measure depressive symptoms. This scale assesses how often depressive symptoms have occurred during the last week. We used the cutoff point of 30 recommended for Swedish adolescents to separate normal and high scores [31]. The Strengths and Difficulties Questionnaire was used to measure multiple dimensions of mental health [32]. This scale assesses symptoms related to emotional problems, peer problems, conduct problems, and hyperactivity during the last six months. The recommended cutoff point of 18 was used in order to separate normal and high scores [33]. All predictors were measured at baseline when adolescents were in the 7 th grade except for academic grades which were reported retrospectively. Covariates. Some additional variables were considered as potential confounders because they could theoretically determine both academic expectations and their predictors. These were: adolescents' gender; family living arrangement , derived from the child's questionnaire; parents' education and country of birth, obtained from the parents' questionnaire and categorized as "at least one parent with university education vs. neither parent with university education" and "both parents born in Sweden vs. one parent born outside of Sweden" respectively. Being female, living in a two-parent family, having a higher SES, and identifying as an ethnicity other than white have all previously been found to be related to having higher academic expectations [34]. These factors are also likely to be related to parental and own engagement in school, parental expectations, academic achievement, sense of identity and mental health. --- Statistical analysis We analyzed baseline characteristics of the study population separately according to the two main outcome variables using chi-square tests for categorical variables and Kruskall-Wallis tests for continuous variables. We used logistic regression models to estimate: A. the likelihood of endorsing specific expectations in 9 th grade rather than remaining undecided among those reporting undecided expectations at baseline. B. the likelihood of raising expectations between 7 th and 9 th among those who reported a definite expectation in 7 th grade. Models were adjusted for sex, living arrangement, parents' education, and parents' country of birth. As a secondary analysis, we explored the additional outcome of lowering expectations from 7 th to 9 th grade compared to those who maintained or raised their expectations. We used multi-level models to accommodate for clustering at the school level. Because estimates and model test statistics were nearly identical in the two models, results are presented ignoring the school-level cluster. All analyses were conducted using SAS enterprise guide 7.1. --- Results About two thirds of adolescents who did not endorse a definite academic expectation at baseline resolved their uncertainty by the 9 th grade. Among those who reported definite academic expectations at baseline, 47% raised their expectations. Table 1 shows the distribution of 9 th grade expectations according to 7 th grade expectations. Adolescents living with both parents were more likely to resolve their uncertainty. Median student engagement, parental support and grades were higher among those who resolved their uncertainty between 7 th and 9 th grade. Girls, adolescents with parents with university education, those whose parents expected them to go to university, and those with at least one parent born outside of Sweden were more likely to raise their expectations between 7 th and 9 th grade. Median grades in 7 th grade and student engagement were higher among those who raised their expectations between 7 th and 9 th grade. . Higher parental support was associated with a greater odds of transition to defined expectations , while high parental expectations were not. Student engagement in school was also associated with a significant increase in the odds of resolving uncertainty in expectations . Higher academic grades were associated with greater odds of resolving uncertainty . Identity synthesis and the absence of problematic depressive or mental health symptoms did not appear to be significant predictors of resolving uncertainty . Parental university expectations were associated with greater odds of their children raising their own academic expectations , as was adolescents engagement in school and having higher academic grades . Parental involvement in their child's education, identity synthesis, and absence of depressive or emotional and behavioral symptoms did not appear to be important predictors of raising academic expectations . Secondary analyses of lowered expectations revealed a similar pattern, with higher parental expectations, student engagement, and academic grades all being significant predictors of a lower odds of adolescents lowering their expectations. Parental engagement, identity synthesis, and a normal score on the Strengths and Difficulties Questionnaire did not significantly predict lowered expectations. However, having a normal score on the CES-DC depression scale was associated with a decreased odds of lowering expectations . --- Discussion In this large longitudinal study, we found that both parental attitudes and adolescents' engagement and achievements were associated with an improvement in the future expectations about own education. Specifically, higher parental involvement in education, higher student engagement, and academic grades predicted whether students who in the 7 th grade could not endorse any definite expectations would resolve this uncertainty two years later. Likewise, higher parental academic expectations, higher student engagement and higher academic grades predicted raised academic expectations between 7 th and 9 th grade. However, psychological factors at a typically individual level such as identity synthesis and positive mental health were not longitudinally associated with positive change in academic expectations in this cohort. Previous studies have consistently found parental expectations to predict adolescents' own expectations [7][8][9]. However, studies investigating the association between parental involvement in or support for learning and their children's academic expectations tend to come to different conclusions. Several found parental involvement in education to be an important predictor of adolescent academic expectations [9,10,35] while others reported null associations [11,12]. However, measures of parental engagement vary between studies. For instance, some of these studies measured parental involvement in terms of how often parents met with teachers or attended meetings. This behavior may not be an uncontroversial measure of engagement, since it is likely to also reflect children's difficulties in school, thus hampering a direct interpretation of this association. We measured parental involvement through the children's perception of their parents' engagement and communication related to schoolwork. Interestingly, in this study parental involvement in the child's education appeared especially important for resolving uncertainty in expectations, while parental academic expectations appeared especially important for predicting a raise in adolescents' academic expectations. It is possible that parental non-directive engagement in the adolescents' education promotes the process of autonomous decision making and goal setting rather than the absolute level of these goals [36]. On the other hand, parents communicating high academic expectations may prompt their children to strive for higher specific goals. Higher own engagement in school has been found to be related to academic expectations in qualitative [37] and quantitative studies [38], to the point that expectations and engagement are sometimes measured as a single construct of "engagement with educational goals" [39]. Our findings support that engagement in school predicts a positive shift in academic expectations during this phase of adolescents. A bi-directional relationship between academic achievement and academic expectations has previously been hypothesized. Some studies have found that adolescents adjust their expectations based on their own achievement. For instance, one study found that adolescents' high achievement predicted higher expectations [8], but other studies did not find such a relationship. In one study, only very large changes in average grades predicted a change in expectations [14]. In another study the majority of adolescents did not adjust their expectations to reflect their current achievements [15]. Our findings present some support for the hypothesis that own achievement is important in inspiring higher expectations. This is in line with Expectancy-Value theory, where achievement and motivation are predicted by ability self-concept and subjective task value [13]. It is likely that academic achievement and expectations reinforce each other over time indicating that adolescents adjust their expectations according to their own achievements, and that improving expectations is likely to also improve future achievements. Sense of identity has sometimes been described as closely related to the formation of academic and career goals [40], but a long period of exploration takes place before identity commitments are made [41]. In our study, a greater sense of identity synthesis was not related to resolving uncertainty of expectations or to raised expectations. It may be that both identity and academic expectations develop simultaneously rather than early adolescent identity development predicting later changes in expectations. Though identity formation is supposed to take place during adolescence, it is unclear at exactly what age identity synthesis should be achieved [42]. The adolescents in our sample may have been too young at baseline to convey useful information about their identity development. It is also possible that identity formation and academic expectations actually represent two different and independent processes. Adolescents may be able to separate their specific academic goals from who they feel they are as a person overall. Previous studies have found that measures of mental health or well-being tend to be associated with higher educational expectations or aspirations [6,12,16,43], though some of these studies were cross-sectional and offered different explanations of the causal direction of these associations [6,43]. In our study, mental health measures did not predict a positive change in academic expectations. It may be that the relationship between mental well-being and positive expectations is already established at an early age. Inequalities in Swedish education are growing in terms of an increasing gap between the highest and lowest achievers in general, as well as a growing achievement gap between those with high and low socioeconomic status [44]. Those who are thriving in the current academic system may be particularly adaptive individuals with supportive families and more optimistic perceptions of their potential opportunities for the future. Several limitations of our study need to be considered. The original cohort had a low response rate at the individual level and was rather selected with an over-representation of highly educated parents and an under-representations of parents born outside of Sweden [26]. For this reason, the generalizability of these results may be limited. Additionally, we lacked finer insights in the question measuring academic expectations, as university was the highest level of education possible for the student to report. We also only had a select number of items from the Erikson Psychological Stage Inventory, and thus may not have captured all aspects of identity development. Relying on retrospectively reported grades also led to poorer quality data in this variable. Finally, unmeasured confounding cannot be ruled out, for instance parents' mental health, the child's cognitive ability, or personality traits such as optimism. --- Conclusion This study found that potentially modifiable factors at the family and at the individual level such as parental engagement in education, parental academic expectations, and adolescent engagement and achievement in school are predictors of a positive development in academic expectations during the final years of Swedish compulsory school. These findings point toward the high value of school-family communication, which is likely to impact on parental engagement in children's education and on how parental expectations are formed and communicated. In turn these processes would facilitate adolescents' identification of their academic goals and ultimately their social, psychological, and educational trajectories. --- Supporting information --- S1 Table. Odds ratios and 95% confidence intervals for lowering own academic expectations among adolescents reporting a definite academic expectation at baseline according to family and individual predictors . ---

Adolescents' high academic expectations predict future health and successful societal integration. Yet, little is known about which factors may promote adolescents' expectations of their future education and academic achievement.To explore whether potentially modifiable factors such as parents' engagement and expectations regarding their child's education; or student individual factors such as school engagement, academic achievement, sense of identity, and positive mental health predict positive development of academic expectations in early adolescence.A longitudinal study of 3,203 adolescents and their parents was conducted with information collected between 7 th grade (13 years of age) and 9 th grade (16 years of age). Parental and adolescents' own academic expectations and engagement in school, academic achievement, identity synthesis, and mental health were self-reported in annual questionnaires. We used logistic regression to analyze the associations between the aforementioned factors and two binary outcomes related to changes in expectations from 7 th to 9 th grade: A. resolved uncertainty regarding own academic expectations; B. raised academic expectations.Student engagement, and higher academic grades predicted both resolved uncertainty in expectations and raised academic expectations. Higher parental involvement in education was related to resolved uncertainty, while high parental expectations were related to raised student expectations. Identity synthesis and mental health did not appear to predict either outcome.

In many locations around the world, war and intergroup conflict have destroyed healthcare infrastructure and have rendered large segments of the population vulnerable to disease and injury. A shortage of medical services and healthcare professionals in these regions often means that community members without medical training are the default first responders. The World Health Organization has advocated for an expansion of prehospital training courses, such as short courses in first aid, to empower lay people all over the world with easy to learn lifesaving knowledge and skills to help themselves and others when emergencies arise and healthcare professionals are not readily available [1]. Although the medical value of these training programs is clear, such trainings also have valuable and untapped peacebuilding potential. Our perspective on this issue comes from conducting a small research project in South Sudan. We were part of a group, the Maine-African Partnership for Social Justice , that investigated whether training in wilderness first aid techniques could increase medical knowledge in South Sudanese villagers with minimal access to healthcare resources. Although we provide a comprehensive account of our research elsewhere [2], in brief, 50 villagers participated in a multi-day training course held in the Kit Region of South Sudan. The course was a mixture of lectures and hands-on sessions that taught basic first aid principles and how to use everyday objects in one's environment to dress wounds, treat dislocated joints and fractured bones, and transport people with injuries. Villagers from across the region were brought together to take the course. Participants were from different tribes, spoke different languages, and practiced different religions, but they all had the same desire to learn basic medical information and techniques. Our work focused on assessing the effectiveness of the wilderness first aid training, and it did not occur to us during planning that our project might have peacebuilding benefits. However, at the conclusion of the training, several participants told us that the opportunity to interact with people from other villages in a supportive setting allowed them to forge new relationships that they thought would strengthen community bonds. The collaborative energy generated by the training was also evident in spontaneous discussions among participants about how the villages could pool resources and work together to lobby the local government to improve healthcare for their region. Robert Owot, one of the South Sudanese men who was instrumental in organizing the first aid training, a parallel labor and delivery training, and another effort to bring a regional school to the area, wrote a letter to our group in which he said, "This region inhabits four tribes that have been living indifferently for decades, but the health trainings initiated by MAPSJ, which have been introduced to these people without discrimination, have improved relations and coexistence greatly amongst the people." In hindsight, we wish we had collected empirical data to support the anecdotal indications that our training facilitated positive interactions that helped people transcend group boundaries. Despite the absence of data, this experience inspired us to reflect on the features of our program that might have had a positive social impact on these villagers. In what follows, we highlight several observations and attempt to connect them to foundational social psychological theory on improving relations between groups and also lessons learned from the WHO Health as a Bridge for Peace program and similar peace through health initiatives. We believe that our observations add to the existing peace through health literature because the majority of this work has focused on healthcare professionals working in conflict areas and not healthcare trainings for lay people who primarily have other professional roles. To be clear, our goal is not to provide a comprehensive analysis or make strong claims about the peacebuilding effectiveness of our particular program. Such assertions would require further research and long-term assessment. Instead, we hope that sharing our experience motivates other organizers of healthcare trainings for lay people in regions of conflict to consider how their programs can improve medical knowledge and simultaneously strengthen community relations. There were several features of our training program that we believe facilitated positive interactions among members of neighboring villages. The first feature was the decision to bring members of multiple villages together to the same location for a single training, instead of the more typical approach of conducting separate trainings for people of different groups. Our rationale for conducting a single training was practical. It was more cost effective, allowed the training to reach more participants, there was relative stability in the area, and our South Sudanese partners encouraged the approach. However, bringing people together to the same location proved more valuable than we anticipated because it served as a necessary condition for members of various villages to interact and have positive social contact. It was also critical that our training filled a medical need. Because the training offered skills and knowledge that were novel to participants and valued by the local community, the village leaders sanctioned our training program and encouraged people to participate. This buy-in from respected members of the local community was crucial because it provided trainees with reassurance that the training was worthwhile and also reduced the perceived risk of interacting with previously unknown participants from neighboring villages. Several features of our program helped create a psychologically safe environment and allowed participants to feel personally invested in the experience. First, we did our best to treat members of all villages equally, which signaled respect and equal status. Additionally, we provided participants with a voice to shape their educational experience. At the end of each day, we asked the trainees what they liked and did not like about the training sessions and adjusted the lessons for the next day accordingly. We did this to maximize the relevance of the training to the participants. However, we think that this hierarchy flattening dialogue also increased the trainees' investment in the program because it communicated that we were interested in what they wanted to get out of the training and were not just talking at them. Participants also seemed to give special value to the name badges and other materials we gave them as part of the training. The purpose of the name badges was to help everyone learn each other's names, but it is possible that these badges provided a visual cue of their new shared identity "first aid trainees," which helped trainees see themselves and others as a group with a common purpose. Finally, hands-on training exercises put people in close physical contact with members of other villages. In these exercises, some trainees role-played various injuries while others worked together to provide the recommended treatment. Working together provided moments of laughter, physical touch, exchange of ideas, and helping each otherhumanizing experiences that create social connections and challenge stereotypes. These features of our training program emerged organically, but it is notable that they resemble social psychological principles that have long been thought to promote positive relations between groups. For instance, classic theorizing by the psychologist Gordon Allport highlighted how intergroup interactions benefit from the support of authority figures, equal status between groups, common goals and superordinate identities, and intergroup cooperation [3]. The importance of these factors for facilitating positive intergroup contact has been empirically supported by a recent meta-analysis [4]. Other influential social psychological work has demonstrated the value of giving people a voice in their experiences. This work indicates that when everyone is treated equally and given a voice to express their opinions and concerns, people are motivated to identify with the group that is affording interpersonal respect and also engage in cooperative behavior [5]. The public health and medical literature on peacebuilding through health, which we will collectively refer to by the WHO moniker "Health as a Bridge for Peace ," provides further clues about how basic first aid trainings might be used to promote peace. The critical insight of the HBP approach is that healthcare and peace are often intrinsically linked. This is the case because discord between groups can cause disparities in healthcare, injuries from violence, mental health problems, and conditions for infectious disease to spread. At the same time, health professionals can promote peace in a number of ways, including reframing conflicts in public health terms, leveraging medical care and resources to promote ceasefires, and providing mental health treatment to heal the psychological trauma of war [6]. As with the social psychological literature, the importance of building relationships through respect, equal status, and giving a voice are emphasized by HBP theorists as critical for bringing people together [7]. Medical trainings figure prominently in the HBP literature, but the focus is primarily on trainings of healthcare professionals. Evidence from the Balkans to the Middle East has shown that bringing together healthcare professionals from differing sides of a conflict for trainings and discussions can improve both regional healthcare services and cross-group cooperation [8,9]. To our knowledge, very little HBP work has examined how healthcare trainings for people without medical backgrounds can serve as a bridge for peace. This likely reflects the reality that many of the practitioners involved in basic first aid trainings around the globe are not part of the HBP community. Yet, principles from HBP trainings for medical professionals are very translatable to first aid trainings for non-healthcare professionals and provide a useful guide for best practices. It is also possible that healthcare trainings for lay people may have peacebuilding advantages over conventional conflict resolution interventions, particularly when groups participating in the trainings are not actively seeking intergroup dialogue. The reason for this is that unlike conflict resolution programs that emphasize direct discussion about conflicts, participants in healthcare trainings are not required to discuss these difficult issues. Talking about disagreements and injustices can be aversive, especially when people anticipate and experience confrontation. This uneasiness might cause people to avoid conflict resolution programs that involve such discourse. In contrast, people in healthcare deprived regions are highly motivated to participate in healthcare trainings for the medical knowledge. Because these trainings focus attention on the medical skills and information that can be acquired, the social interactions are in the background. These social interactions, however, provide an important opportunity for conversation to occur among people of different groups. A common interest in healthcare creates a built-in topic for trainees to discuss that can make the conversation pleasant and fluent. As the conversation turns to small talk, people might uncover similarities that they did not know about each other. The rapport established in these nonconfrontational interactions can increase feelings of social connection and mutual respect, which can be invaluable if future discussions involve direct dialogue about intergroup tensions. Healthcare trainings that bring groups of people together might not always be feasible and could backfire in certain circumstances, such as when groups are engaged in intense conflict. It is notable that the region of South Sudan where we conducted our training was not embroiled in violence. In areas of active fighting, buy-in from community leaders might be difficult to obtain and potential trainees might be fearful of participating. As a result, healthcare trainings might be most effective at promoting peace when used preventively to increase social connections between groups that are at risk for conflict but currently coexist without trouble, or reparatively in post-conflict regions and refugee resettlement areas, particularly when the raw emotions of war have begun to dissipate. It is also important to recognize that well meaning interventions can sometimes have a net negative effect on the communities they are designed to help [10]. It is currently unknown whether incorporating peacebuilding features into healthcare trainings could have detrimental and unintended consequences. As a result, people conducting such efforts should pay particular attention to any concerns expressed by trainees and other community members. Our hunch is that many global health professionals who are designing healthcare training programs for lay people in low and middle-income communities are focused on the public health objectives and have not contemplated how their programs can also build crossgroup trust and friendships. However, we urge these individuals to carefully consider whether their programs can be used to promote peace. We also hope that HBP practitioners and researchers will further develop the ideas we have proposed and rigorously assess their utility. Given that healthcare trainings are regularly conducted all over the world, structuring them to promote peace could have large and multiplicative benefits on the stability of communities around the globe. --- Availability of data and material Not applicable. --- Authors' contributions KR wrote the letter and LK provided critical feedback and revisions. Both authors read and approved the final letter. --- Competing interests The authors declare that they have no competing interests. --- Consent for publication Not applicable. Ethics approval and consent to participate Not applicable.

Global health professionals regularly conduct healthcare trainings, such as first aid courses, in disadvantaged communities across the world. Many of these communities lack healthcare infrastructure because of war and political conflict. The authors draw on their experience conducting a first aid course in South Sudan to provide a perspective on how healthcare trainings for people with no medical background can be used to bridge ethnic, political, and religious differences. They argue that a necessary step for turning a healthcare training into a vehicle for peacebuilding is to bring people from different communities to the same physical space to learn the course material together. Importantly, simply encouraging contact between communities is unlikely to improve intergroup relations and could be detrimental if the following features are not incorporated. Buy-in from respected community leaders is essential to ensure that training participants trust that their safety during the training sessions is not at risk. Trainers should also create a supportive environment by conferring equal status and respect on all trainees. Finally, hands-on training exercises allow for positive interactions between trainees from different groups, which in turn can challenge stereotypes and facilitate cross-group friendships. These features map onto social psychological principles that have been shown to improve intergroup relations and are consistent with lessons learned from peace through health initiatives in public health and medicine. By adopting peacebuilding features, healthcare trainings can serve their primary goal of medical education and provide the added benefit of strengthening social relations.

Background Gobalisation, poor living conditions, war and conflict are major factors contributing to forced migration. In 2016, the number of people displaced by conflict and persecution worldwide was estimated to be 65.6 million. Of these 2.8 million were estimated to be asylum seekers and 22.5 million refugees, which the United Nations High Commissioner for Refugees suggests is the highest level ever recorded [1]. Among this population, 49% of refugees were women, a similar proportion as reported annually since 2003 [1]. The impact of migration on health is far-reaching, making migrant populations particularly vulnerable, fuelling health inequalities and resulting in serious implications for global health. Research on migrant populations is challenged by the diverse terminology and definitions used. For the purposes of this systematic review, we use the following UNHCR definitions [1]: Asylum seekers are individuals who have sought international protection and whose claims for refugee status have not yet been determined, irrespective of when they may have been lodged. An asylum seeker has applied for asylum on the grounds of persecution in their home country relating to their race, religion, nationality, political belief or membership of a particular social group. This population remains classified as asylum seeker for as long as the application is pending. Refugees have been forced to leave their country in order to escape war, persecution or natural disaster. The 1951 Convention relating to the Status of Refugees describes a refugee as "a person who owing to a well-founded fear of being persecuted for reasons of race, religion, nationality, membership of a particular social group, or political opinion, is outside the country of this nationality and is unable to or, owing to such fear, is unwilling to avail himself of the protection of that country". A refugee is an asylum seeker whose application has been successful. Migrants include those who move, either temporarily or permanently from one place, area or country of residence to another for reasons such as work or seeking a better life , for family reasons or to study. People also migrate to flee conflict or persecution, which is where the definition converges with the terms refugee and asylum seeker. Timely access to perinatal healthcare is an effective method to optimise pregnancy outcomes and the lifelong health of women and their offspring. Late access to maternity care can result in adverse perinatal outcomes. Vulnerable pregnant women, including women with asylum seeker and refugee status, face barriers to accessing healthcare [2] including maternity care [3]. A recent report of vulnerable women in social crisis in Europe included pregnant women seeking or having been refused asylum and found that 65% had no access to antenatal care, 42% accessed care after 12 weeks of pregnancy and two thirds were classified as being 'at risk' requiring urgent or semi-urgent care [4]. This disparity in access to, and use of, perinatal healthcare can lead to significant health inequalities. Failure to effectively reach and provide optimal perinatal care for women with asylum seeker and refugee status will result in failure to reduce health inequalities for this vulnerable group of women and their babies. There has been a recent escalation of systematic reviews investigating different aspects of perinatal health in women who have migrated, which includes asylum seeker and refugee populations. For example, multiple systematic reviews were published in 2016 and 2017 on topics including perinatal health outcomes [5][6][7][8] and experiences of antenatal care [9][10][11]. However, there is a lack of published systematic reviews that explicitly address pregnancy among asylum seeker and refugee populations, and there is a tendency to group all migrant populations together in syntheses. Given this, we have chosen to undertake a systematic review of systematic reviews to assess the research gaps and provide direction to future research specifically relating to women with asylum seeker and refugee status. The aim of this systematic review was to provide an overview of the existing evidence base drawn from systematic reviews that have examined perinatal healthcare and outcomes among women with asylum seeker or refugee status. --- Methods The Joanna Briggs Institute methodology for umbrella reviews was used to guide this systematic review of systematic reviews [12]. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses reporting guidelines and checklist have been used to report each stage of the systematic review methods and findings [13]. The protocol for this systematic review has been registered in the PROSPERO database . --- Identification of studies Electronic bibliographic databases were searched using PICOS criteria: Population ; Intervention ; Comparator ; Outcome ; Study design . A search strategy for database-specific search terms and subject headings was developed with the support of an information scientist for the databases MEDLINE, Embase, Scopus, Cumulative Index to Nursing and Allied Health Literature, JBI database, PROSPERO, Cochrane Database of Systematic Reviews, Google Scholar, Science Direct, Web of Science, PubMed and ProQuest . Database searches were supplemented with hand searching of the reference lists of all included systematic reviews to identify any further relevant reviews. All included systematic reviews were also subjected to citation searches using all citations produced by Google Scholar. Any systematic reviews identified by the supplementary searches which met the inclusion criteria were also subject to reference list and citation searches until no further eligible reviews were identified. The detailed search strategy was carried out between June and July 2017 and restricted to systematic reviews published within the past 10 years as per the JBI recommendation [12]. No restrictions were placed on country or region of study or on low-, middle-or high-income status of the host countries. Inclusion criteria were as follows: Systematic reviews with a quantitative, qualitative or mixed methods evidence synthesis Published in the English language Included any perinatal health outcomes or perinatal care during the preconception, antenatal and postnatal periods Clearly stated that women with asylum seeker or refugee status were populations within the included studies. This included reviews of migrant women where asylum seekers and refugees were part of the included population Reviews were excluded if they were: Scoping reviews which aimed to identify the extent and nature of the evidence base without a formal evidence synthesis Published abstracts without full texts and protocols of systematic reviews. We searched for any subsequent full text publications of these works Reviews that focussed on refugees living in camps Two authors independently screened titles, abstracts and full texts for inclusion in the review. Disagreements regarding eligibility for inclusion were resolved through discussion; a third independent reviewer was available where no agreement could be reached . References were managed and recorded in EndNote version X7. The flow of reviews through each stage of the searches and screening and the reasons for exclusions are presented using a PRISMA diagram . Data extraction and quality assessments were carried out in duplicate for all included systematic reviews. Independent data extractions and quality assessments were combined by two authors and agreed with recourse to a third reviewer if no agreement could be reached . --- Quality assessment The JBI Critical Appraisal Checklist for Systematic Reviews and Research Syntheses [12] was used for quality assessment. The checklist comprises 11 questions relating to methodological rigor, transparency of reporting and appropriateness of conclusions and recommendations, with options of 'yes' if the review clearly meets the checklist criteria and 'no' , 'unclear' or 'not applicable' if the review does not clearly meet the criteria . The reviews were awarded a score of 1 for each checklist criterion clearly met, with a maximum possible score of 11. The reviews were considered to be of high quality if they scored 8-11, moderate quality for scores of 4-7 and low quality for scores of 0-3. No reviews were excluded based on quality score. The percentage of included reviews meeting the criteria was calculated for each of the 11 checklist questions. --- Data extraction The JBI umbrella review data extraction form was adapted to meet the needs of this mixed methods systematic review of systematic reviews . The following data were extracted for each included systematic review: aim, objectives and focus of the review including review type, aims, objectives, type/definition of included population, inclusion and exclusion criteria and outcomes included in search strategy; search details including date range of the search, search strategy and restrictions to the search; appraisal rating including whether the quality appraisal was reported, what method/tool was used and summary of quality of included studies; key results including the number of included studies, publication date range, sample size, host countries, description of included population, summary results and conclusions for the overall population and also detailed results and conclusions explicitly relating to asylum seekers and refugee populations. We implemented a process of systematically extracting data which was directly relevant to women with asylum seeker and refugee status for all of the systematic reviews which combined data from multiple populations in their syntheses . This involved two stages of searching for relevant data in the tables, figures and narrative in the results, discussion and conclusions sections of the included systematic reviews. First, the relevant sections of the reviews were searched for data that the authors had explicitly described as being relevant to asylum seeker or refugee populations, and these data were extracted. The second stage involved identifying whether any of the included studies in the systematic reviews were exclusively among asylum seeker or refugee populations. When studies which were exclusively among these populations were identified, data were extracted for any results which had cited these studies as part of the evidence base which informed that specific result. This second stage was only carried out if we were confident that the population of the included study were exclusively asylum seekers and/or refugees; for example, data were not extracted for studies which included migrants and refugees, as we could not be confident that the data that had informed the result originated from women who were migrants or refugees. --- Evidence synthesis Evidence synthesis in systematic reviews of systematic reviews should provide a summary of existing research syntheses in tabular format with a more detailed narrative description of the systematic review characteristics and relevant quantitative and qualitative results [12]. The results have therefore been summarised in tables to describe the characteristics of the included systematic reviews, results for overall populations included in the systematic reviews and also results explicitly relevant to women with asylum seeker and refugee status. Tables are supplemented with a narrative discussion of the included systematic reviews grouped by the review themes of perinatal health outcomes and perinatal healthcare access and experiences for women who are migrants and for asylum seekers and refugees. Each theme has several data-driven sub-themes. As per reporting recommendations [12], any overlap in original research studies in the included systematic reviews is reported in the results Fig. 1 PRISMA flowchart of searches, screening, and inclusion and exclusion of studies table and narrative for the asylum seeker-and refugeespecific data. Evidence synthesis was carried out in duplicate. The first stage of synthesis was performed by JR, HB and AP to group the systematic review data into the review themes and to provide a descriptive summary of the topic-specific data. The second stage involved NH validating the review themes and performing a detailed synthesis of sub-theme data reported in the systematic reviews. --- Results --- Included systematic reviews A total of 3415 records were identified by searches, of which 29 were systematic reviews which met the inclusion criteria . Twenty-one systematic reviews were identified through the database searches, three from reference list and five from citation searching. Fourteen of the included reviews were quantitative, eight of which included a meta-analysis, nine were qualitative and six involved mixed methods . The systematic reviews were published between 2009 and 2017; the fewest reviews were published in 2011 and 2012 , and most were published in 2017 . The number of studies included in the reviews ranged from eight to 133, and the publication years of the included studies were from 1956 to 2016. Only one systematic review identified was exclusively focussed on women with asylum seeker status [14]. The populations included in the remaining systematic reviews were migrant populations including women with asylum seeker and/or refugee status and marginalised women , which included those with asylum seeker and/or refugee status along with women who had experienced domestic violence, minority ethnic groups, travelling communities, women of low income, those with substance abuse problems, teenagers and women who were homeless. There was no consistent definition in the use of terminology to describe women, and many reviews did not adequately define their populations. We use the terms asylum seeker, refugee and migrant as previously defined, and the term 'women in the host country' to collectively describe comparison groups that have been reported in the systematic reviews which include a range of definitions including non-migrant, native-born, etc. . --- Quality of evidence The quality scores ranged from six to 11 . Of the 11 questions in the JBI Critical Appraisal Checklist for Systematic Reviews and Research Syntheses, all of the included reviews scored 'yes' in four questions , while only six reviews scored 'yes' for the question 'was the likelihood of publication bias assessed?' . Additionally, only 14 reviews used methods to minimise data extraction errors , 18 carried out quality appraisal and only 21 used adequate sources and resources to search for studies as recommended in guidelines for systematic reviews of observational studies [15]. --- Perinatal health outcomes amongst women who are migrants Nineteen systematic reviews reported perinatal health outcomes including perinatal mental health, mortality , mode of delivery, birth weight, preterm birth, congenital anomalies and additional morbidities. The results are summarised in Table 2, and a narrative summary is presented for each outcome. --- Perinatal mental health Mental health, which included postnatal depression, antenatal depression, anxiety and post-traumatic stress disorder, was the most frequently reported outcome and was included in eleven systematic reviews; six were quantitative [5,6,[16][17][18][19], three were qualitative [8,20,21] and two mixed methods [22,23]. The reviews reported prevalence and risk factors for mental health disorders. Prevalence of perinatal mental health disorders All systematic reviews reporting prevalence data concluded that perinatal mental health disorders were more frequent in migrant women than in women from the host countries [5,6,[16][17][18][19]22]. Postnatal depression was the most frequently reported perinatal mental health outcome in the systematic reviews. Prevalence of postnatal depression amongst migrant women was reported as 11. 2-60% [16], < 1-59% [5], 24-42% [18], 2.9-52% [22] and 20% 17-23%) [17]. Prevalence of antenatal depression amongst migrant women was reported to be 12-45% [5], and prevalence of any depressive disorder was 31% [6]. There were also significantly increased associations with mental health disorders amongst migrant women compared with women from the host countries. Anderson et al. [5] reported that anxiety was increased in migrant women with non-English-speaking backgrounds, and post-traumatic stress disorder was 15% compared with 0% amongst non-migrant women. Nilaweera et al. [22] reported that odds ratios for postnatal depression in their included studies ranged Risk factors for the development of perinatal mental health disorders Seven systematic reviews reported quantitative and qualitative evidence of factors associated with increased risk, or having a protective effect on perinatal mental health disorders [5,6,[16][17][18][19]22]. There were similarities between the systematic reviews, and results are reported under the themes of stress and support, adjustments to host country, pregnancy care and infant feeding, health status and history and sociodemographics. Stress and support. This was the most frequently and consistently reported risk factor for the development of mental health disorders amongst migrant women. Examples provided included emotional stress, a history of violence or abuse, having witnessed or experienced stressful life events and their premigration experience such as having migrated for political reasons or problems with the police or army in their home country [5,6,[16][17][18]. Lack of social support and lack of family support were also reported to be important risk factors. There was a consistent pattern of low social support increasing the risk and good social support being protective against perinatal mental health disorders [5,6,[17][18][19]22]. Having no relatives or friends, a lack of emotional support from their spouse, being unmarried, having no partner, having migrated for marriage, marital adjustment problems and a lack of domestic decisionmaking power in relation to the child were all risk factors for perinatal mental health disorders amongst migrant women, whereas having a close relationship with their partner was reported to be protective [5,6,16,17,22]. Adjustment to host country. The most commonly reported risk factors for perinatal mental health disorders were difficulties with the host country language [5,6,17,19,22] and being unfamiliar with local life [19]. Anderson et al. [5] reported inconsistent evidence in their included studies relating to the length of time resident in the host country, whereas other reviews reported that shorter duration of residence was a risk factor for perinatal mental health disorders [6,17]. Fellmeth et al. [6] reported that adherence to traditional postpartum practices was protective against postnatal depression in migrant populations. Pregnancy care and infant feeding. Experience of perinatal healthcare including operative caesarean and instrumental delivery and poor satisfaction with support [6,18] and also infant feeding experience including formula feeding and feeding problems [6,17,22] were risk factors for the development of perinatal mental health disorders reported by four systematic reviews [7,17,18,22]. Health status and history. The risk of perinatal mental health disorders was increased when migrant women perceived their overall health to be low [17,19] or had a history of mental health disorders [6,22]. Fellmeth et al. [6] reported ORs for postnatal depression to be between 24.9 and 29.7 when there was a personal or family history of depression. Socio-demographics. Risk factors included low income or socio-economic status, unemployment [5,16,17], low education [17], having a visible minority status [19] and primiparity [6]. Fellmeth et al. [6] also reported that maternal age > 30 years and < 25 years were risk factors for increased postnatal depression. --- Mortality Two systematic reviews reported data on maternal mortality [24,25]. Pedersen et al. [24] reported the relative risk to be twofold [27]. When the meta-analyses were restricted to migrants from non-European countries, the risk increased for stillbirths and slightly increased for perinatal, neonatal and infant mortality . Conversely, migrant women in the USA had better outcomes than USA-born ethnic minorities , demonstrating a healthy migrant effect. Adjustments for risk factors in the meta-analyses only accounted for a small proportion of the excess mortality risk [27]. Gagnon et al. [25] reported meta-analyses for feto-infant mortality . They found that Asian and North African migrant women had a significantly increased association with feto-infant mortality than women in the host country . There was no significant difference between majorityreceiving-country women and European-born migrants or Latin American-born migrants [25]. The metaanalysis for African women showed the largest effect size, but this was not significant [25]. Note that these meta-analyses only included two or three studies for each country of origin and had high levels of heterogeneity. Bollini et al. [26] found an increased association between offspring mortality and migrant women compared to women from the European host countries . The authors hypothesised that pregnancy outcomes amongst migrant women were influenced by the degree of implementation of integration policies in the host countries, where a strong integration policy would be demonstrated by countries which had entrenched equality and social cohesion in their societies [26]. They carried out further meta-analyses adjusting for maternal age, parity and national level of implementation of integration policies and found the associations to be attenuated when there were strong implementation policies compared with weak implementation policies ; although the implementation of strong integration policies attenuated the association with offspring mortality, the difference in effect did not reach significance [26]. --- Mode of delivery Three quantitative systematic reviews investigated mode of delivery amongst migrant women compared to women from host countries [7,25,28]. Gagnon et al. [25] reported that 40% of the 25 studies included in their review found operative modes of delivery to be higher amongst migrant women; the remaining studies reported reduced operative mode of delivery outcomes for migrant women , mixed results or no difference between migrant women and women from the host country . Merry et al. also reported mixed results for caesarean delivery in their 2013 review [28]; associations between migrant women and caesarean varied by country of origin and by receiving country. The authors reported a significantly increased odds of caesarean amongst women migrating from former colonised Caribbean states , South Asia , the Philippines and Somalia . Women migrating from Africa had increased odds of caesarean which differed according to receiving country: France , Australia , Canada and North/West Europe . However, these increased odds were not observed amongst women migrating from North Africa to Canada or France . Similarly, women migrating from Latin America had significantly increased odds for caesarean in Norway and Canada , but not in Southern Europe . Odds for caesarean were significantly reduced or no different from those of women from receiving countries when women migrated from Vietnam , Kosovo , Russia/Baltic States and East Asia , USA , and Australia, UK, Canada or Finland ) [28]. The 2016 review of Merry et al. [7] was an update of the 2013 review and identified that migrant women from sub-Saharan Africa had higher caesarean section rates, whereas migrant women from Eastern Europe had lower rates than women in the host countries. Higher emergency caesarean deliveries were also reported for women migrating from Latin America, North Africa and the Middle East compared with women in the host countries [7]. --- Birth weight Low birth weight or small for gestational age outcomes were reported by four reviews [19,25,26,29] with contradictory results. A meta-analysis of LBW amongst migrant women residing in European countries showed significantly increased association compared with women in the European host countries [26]. There was a significant attenuation of LBW when analyses adjusted for age, parity and level of implementation of integration policies ; weak implementation resulted in an increased association and strong implementation reduced the association , although the association remained significantly increased compared with results for non-migrant women [26]. Conversely, a metaanalysis of international data not restricted to women residing in Europe showed a reduced aOR for LBW and SGA amongst migrant women with borderline significance [25]. Meta-analysis by migrant origin showed increased odds amongst women born in African and Asian countries and reduced odds amongst European, Latin American and North African-born women, although no sub-group meta-analysis reached statistical significance [25]. The paradoxical healthy migrant effect in relation to reduced risk of LBW and SGA was discussed by De Maio [19] and Villalonga-Olives et al. [29], although there were some reported inconsistent findings. The systematic review by De Maio [19] discusses how the patterns of reduced risk amongst migrants compared with women in the host countries are influenced by maternal socio-economic status, country of origin and maternal education, where migrant women with low levels of education have better outcomes and there is an increased risk of SGA and LBW amongst migrant women with higher education [19]. Villalonga-Olives et al. [29] discuss how the apparent healthy migrant effect in the USA is contrasted by the health inequalities in Europe, where the associations are reversed. The US studies show a reduced risk of LBW and SGA amongst Latina migrants, although this does not extend to Black and Puerto Rican migrants-who have increased risks-and Asian women show no difference in risk compared to women from the host country [29]. In contrast, there is a lack of a healthy migrant effect in Europe with the exception of studies from two countries, Spain and Belgium. However, data from these countries are also conflicting, showing that outcomes differ by migrant origins and also by the severity of outcome [29]. --- Preterm birth Three reviews reported preterm birth outcomes [19,25,26]. A meta-analysis by Bollini et al. [26] identified a higher odds of preterm birth amongst migrant women in Europe . There was a significant attenuation when analyses adjusted for age, parity and level of implementation of integration policies ; weak implementation resulted in increased odds of preterm birth and strong implementation policy decreased the odds [26]. A meta-analysis reported by Gagnon et al. [25] found differences in risk of preterm birth by migrant origin. Compared with women in the host countries, there was a higher odds for migrant women from Asia and Africa ; a lower odds for Latina migrant women ; and no difference for migrant women from Europe and North Africa [25]. De Maio [19] discusses the healthy migrant effect for preterm birth outcomes being influenced by maternal education and length of residence in the receiving country. Migrants with < 5 years residence had a lower prevalence of preterm birth compared with women in the host countries , and those residing > 15 years had the highest prevalence [19]. Further, a 5-year increase in length of residence significantly increased the odds of preterm birth amongst migrant women , which was potentially influenced by maternal stress and discrimination [19]. --- Congenital anomaly Two reviews reported on congenital anomalies [25,26]. Migrant women had a significantly increased risk of a pregnancy affected by a congenital anomaly compared to women in the host countries . There was a significant attenuation of congenital anomalies when analyses adjusted for age, parity and level of implementation of integration policies ; when there was weak implementation, a significant increased association remained , whereas having a strong implement policy resulted in a significantly lower odds of congenital anomalies amongst migrant women [26]. Gagnon et al. [25] combined congenital anomalies with other infant morbidity admission and low Apgar score) and found that 62.5% of the 16 studies included in their review reported worse outcomes for migrant women compared with women from the host country; no studies found this outcome to be better for migrant women. --- Additional morbidities Three reviews reported additional maternal or infant morbidities [19,26]. Bollini et al. [26] investigated maternal postpartum haemorrhage, but no summary data were reported. De Maio [19] identified a healthy migrant effect for risk of placental dysfunction amongst women residing in Ontario, Canada, for < 5 years which was influenced by length of residence: the lowest odds were for women residing < 3 months , which increased the longer the duration of residence ; however, the OR remained lower than that for women in the host country for all durations of residency. Gagnon et al. [25] had a composite outcome for maternal health and reported that 50% of their 32 included studies showed worse outcomes for migrant women, 21.9% showed better outcomes and the remaining studies were mixed or reported no difference. The authors also reported maternal and infant infections to be worse amongst migrant women in 63.6% of included studies and better in 9.1%; the remaining studies showed mixed results [25]. Bollini et al. [26] and Gagnon et al. [25] reported that admission to a NICU or special care was higher amongst offspring of migrant women . --- Perinatal healthcare access and experiences amongst women who are migrants Twenty systematic reviews reported access to, and experience of, perinatal healthcare amongst migrant women; 11 were qualitative [8,9,11,16,20,21,[30][31][32][33][34], 5 were quantitative [6,18,19,25,35] and 4 were mixed methods systematic reviews [10,22,23,36]. Results are summarised in Table 2, and a narrative summary is presented of the themes and sub-themes relating to: access to and utilisation of perinatal healthcare and experience of perinatal healthcare. --- Access to and utilisation of perinatal healthcare Sixteen systematic reviews reported data relating to access or utilisation of perinatal healthcare [8-11, 16, 19-23, 25, 30, 32, 33, 35, 36]. All systematic reviews reported that access to perinatal care, including routine care and specialist care such as mental health support for postnatal depression, was worse amongst migrant women. Heaman et al. [35] reported that 86% of the 29 studies included in their review showed inadequate prenatal care for migrant women compared with women in the host countries, with 15 studies reporting large effect sizes . Gagnon et al. [25] reported that prenatal care was worse amongst migrant women compared with women in the host countries in 58.3% of their 12 included studies, and no studies reported care to be better amongst migrant women. Barriers to accessing care were consistent across all systematic reviews and are summarised here under the themes of structural and organisational barriers, social barriers and personal and cultural barriers. a. Structural and organisational barriers Ten systematic reviews reported unfamiliarity with local healthcare provision, culture and systems as a barrier [8, 10, 19-22, 30, 32, 33, 36]. Issues included a lack of knowledge and awareness of services and support on offer, a lack of information provision about how to get support, difficulties with navigating healthcare systems, managing bureaucracy and a lack of information about regular appointments and check-ups which resulted in missed appointments. Ten systematic reviews reported language barriers to accessing perinatal healthcare [10, 11, 19-23, 32, 35, 36] including proficiency in being able to verbally communicate with health professionals, access to translators and understanding written communication. Physician availability, long waiting lists for services, especially those specialising in migrant care, a lack of postnatal follow-up and perceptions that health services did not want to take migrant women were additional structural and organisational barriers reported in three systematic reviews [20,23,36]. b. Social barriers Nine systematic reviews reported social barriers to accessing care which centred on the competing priorities of real life worries [21] that migrant women faced such as poverty, safe housing, employment and caring for their other children [8,10,11,20,21,31,32,35,36]. Financial constraints were frequently reported including a lack of health insurance, cost of care and wider poverty issues such as having no phone, childcare or transport [8,10,11,20,21,32,35,36]. Further social barriers included having an unplanned pregnancy, being single and maternal education level [31,35]. c. Personal and cultural barriers Six systematic reviews reported personal and cultural barriers to accessing services [8,9,20,21,23,36]. Three reviews focussed on accessing perinatal mental health services [8,20,21] and reported a reluctance amongst migrant women to talk about mental health, a lack of cultural acceptability to seek help, beliefs about women's strength and self-coping, the fear of labelling, stigma and alienation in some cultures, beliefs that depression was not a real health condition, that health professionals and services were for physical health, a lack of understanding of the condition and associated terminology and fears of having their child removed. Additional barriers reported were a lack of culturally appropriate therapists and services available [20,36] and a preference for female health professionals due to religious reasons and the intimacy of body areas during pregnancy [9,23]. However, the preference for female providers was negated in emergency situations, and the competency of the health professional was considered more important than gender [9]. --- Experience of perinatal healthcare Twelve systematic reviews reported data relating to migrant women's experiences of perinatal healthcare [8, 10, 11, 16, 20-22, 30, 32-34, 36]. There were some consistent experiences reported in the systematic reviews, and these are summarised under the themes of negative communication and discrimination, relationships with health professionals, cultural clashes and the receipt of clinical perinatal healthcare. a. Negative communication and discrimination Language barriers and having to rely on translators had an impact on communication experiences [8,10,11,30,[32][33][34]. Systematic reviews also reported themes of insensitive and hurtful communication, perceptions of racism, cultural stereotyping and discriminatory interactions between migrant women and health professionals [10,33,34,36]. Small et al. [33] reported that the migrant women felt that care was not kind or respectful and that they were less likely to be spoken to with respect, understanding and in a way they could comprehend. --- b. Relationship with health professionals The interpersonal relationship between migrant women and health professionals was reported to be an important influence on experience of perinatal care. A positive experience resulted from health professionals who were kind and friendly and who listened to the woman's concerns [10,32]. However, the majority of data related to struggles with relationships and a lack of connection; migrant women were less likely to describe health professionals positively than women in the host countries [20,30,32,33]. There was a common theme of migrant women feeling rushed during interactions with health professionals [8,10,32,36], misunderstandings with health professionals and a lack of confidence to express concerns or ask questions [11,30]. c. Cultural clashes Ten systematic reviews reported a lack of cultural knowledge and sensitivity in their experiences of perinatal healthcare [8,10,11,20,21,30,[32][33][34]36]. The majority of these clashes stemmed from differences between cultural, religious and traditional beliefs and practices and Western biomedical approaches to perinatal healthcare. The systematic reviews reported that migrant women lacked understanding about Western medicine and care, felt pressure to adapt and were labelled as non-compliant if they resisted Western approaches in favour of traditional practices [11,20,21,30,32,34,36]. d. Clinical perinatal healthcare Nine systematic reviews reported migrant women's experiences of clinical perinatal healthcare including breastfeeding support, decision making about care and Western approaches to medicine and technology [10, 16, 20-22, 30, 32, 33, 36]. Higginbottom et al. [36] reported some positive experiences amongst migrant women relating to provision of breastfeeding support in hospital; however, the remaining data relate to negative experiences of care. The reviews reported that migrant women were less positive about the care they received and reported that health professionals discussed their care with them less frequently than with women in the host countries, especially relating to mental healthcare needs [16,22,33,36]. The reviews also reported that migrant women did not feel involved in decision making about their care or did not feel they had options [10,33,36]. There were reports of poor experiences of care and pain management amongst migrant women who had female genital mutilation [30,33,36], too much focus on technological and procedural approaches to care and childbirth [30,32] and an over-reliance on prescription of medications which were culturally or religiously inappropriate rather than access to supportive care [20,21,30,32]. Results specific to women with asylum seeker or refugee status Twenty-two of the included systematic reviews either explicitly reported results relevant to women with asylum seeker or refugee status, or they cited studies where the participants were exclusively women with asylum seeker or refugee status [5-11, 14, 16, 18-20, 23, 27-34, 36]. In total, the reviews cited 54 studies, although there was some overlap in the original studies relevant to asylum seekers and refugees included in the systematic reviews and used to inform the analyses ; data from 43 unique studies were cited by these 22 systematic reviews. Eight systematic reviews use primary data that view asylum seekers and refugees as separate sub-groups [5,6,16,18,19,23,30,36]. Five systematic reviews [7, 9-11, 34, 36] combine asylum seekers and refugees as a sub-group of the migrant population. Seven systematic reviews [8, 20, 27-29, 32, 33] use primary data solely involving refugees. Two systematic reviews [14,31] use primary data looking at asylum seekers only. Seven systematic reviews [17,21,22,[24][25][26]35] do not distinguish asylum seekers and refugees from other types of immigrants in their analysis. The data reported for asylum seekers and refugees were limited, and the majority of detailed data came from qualitative studies on women's access to and experiences of perinatal healthcare. The results table is summarised in Additional file 6, and a narrative summary is presented for perinatal health outcomes and healthcare access and experiences amongst women with asylum seeker or refugee status. --- Perinatal health outcomes amongst women with asylum seeker and refugee status Fourteen systematic reviews reported perinatal health outcomes for asylum seeker and refugee populations [5-7, 14, 16, 18-20, 23, 27-30, 36]. Perinatal health results are presented for perinatal mental health, offspring mortality, mode of delivery, birth weight, preterm birth and additional morbidities. Perinatal mental health was reported most frequently by the systematic reviews. No data specific to asylum seekers or refugees were reported for maternal mortality or congenital anomaly outcomes. Perinatal mental health amongst women with asylum seeker or refugee status Nine systematic reviews reported data for asylum seekers and refugees [5, 6, 16, 18-20, 23, 30, 36] citing 11 original studies [37][38][39][40][41][42][43][44][45][46][47]. Prevalence of perinatal mental health disorders Five reviews cited data from Stewart et al. [37] which found that rates of postnatal depression were significantly higher amongst women with refugee and asylum seeker status compared with women in the host country of Canada . They also found a significantly increased odds of scoring 10 or more on the Edinburgh Postnatal Depression Scale for refugees and asylum seekers [5,16,18,19,23]. Similar rates were reported in a systematic review by Fellmeth et al. [6]; 37.3% of refugees and 41.8% of asylum seekers living in Canada experienced symptoms of depression, somatisation or anxiety and significantly increased odds for the prevalence of any depressive order . Increased prevalence for posttraumatic stress disorder were reported, where asylumseeking women had the highest prevalence , followed by refugees and migrants [5,6] . Higginbottom et al. [36] reported that of 50 refugee mothers who received a home visit at 4 months postpartum, 26 were found to have symptoms of postpartum depression . Data from original studies exclusively on women with asylum seeker or refugee status [40][41][42][43][44]47] were used in a metasynthesis by Balaam et al. [30] and contributed to the findings that stress and low self-esteem were common, and that women had mental health problems such as depression, feelings of loneliness and isolation and expressed sadness, vulnerability and anxiety together with severe nausea. Risk factors for the development of perinatal mental health disorders Three systematic reviews reported risk factors for the development of perinatal mental health disorders specifically relevant to asylum seekers and refugees [6,18,19]. Fellmeth et al. [6] reported data from Matthey et al. [45] which showed statistically significant associations between anxiety and the number of premigration traumatic events experienced or witnessed, but no association with anxiety or post-traumatic stress disorder and history of living in a refugee camp prior to resettlement. Collins et al. [18] and De Maio [19] presented data from Stewart et al. [37] which found that refugees and asylum seekers had significantly lower social support than women in the host country of Canada , including support from family, friends, groups and systems, as well as personal, emotional and instrumental social support. Tobin et al. [20] also reported that women who were refugees attributed their depression to social factors such as family problems or economic hardship rather than biological factors . Offspring mortality amongst women with asylum seeker and refugee status Two systematic reviews [14,27] reported offspring mortality amongst women who were refugees using data from nine original studies [40,[49][50][51][52][53][54][55][56]. Gissler et al. [27] reported that in European studies women who were registered refugees or originated from refugee source countries at the time of arrival had a significantly increased risk of stillbirth , early neonatal mortality and perinatal mortality compared to women in the host countries of Norway, Sweden, Ireland and the Netherlands. However, women from Vietnamese backgrounds had lower mortality than women in the host country of Norway [27]. Evidence from the former Yugoslavia showed that women who were refugees had increased risk of early neonatal mortality and perinatal mortality but no difference in risk of stillbirth . Deaths attributed to congenital anomalies, pregnancy complications or intrauterine growth restriction were similarly distributed amongst refugees and women in the host country. Live birth and abortion There were additional data relevant to offspring mortality for women with asylum seeker and refugee status that were not reported in the data for migrant women. Hadgkiss and Renzaho [14] reported that asylum seekers had a higher incidence of sexual assault, unwanted pregnancies and induced abortion-to-live birth ratio compared with women in the host countries . Asylum seekers with longer duration of stay had a lower live birth and abortion rate [14] . Mode of Delivery amongst women with asylum seeker and refugee status Three systematic reviews [7,14,28] reported caesarean delivery for refugee and asylum seeker women using data from four original studies [38,40,57,58] with conflicting results. Merry et al. [28] and Hadgkiss and Renzaho [14] reported data from two studies [40,57] which found no significant difference in caesarean delivery rates amongst asylum seekers compared to native-born women . However, Merry et al. [7] reported that refugees and asylum seekers were at a reduced risk of an emergency caesarean compared with economic and student migrants but an increased risk compared with women in the host country of Canada . Birth weight amongst women with asylum seeker and refugee status Two systematic reviews [14,29] reported data for low birth weight and intrauterine growth retardation using data from five original studies [40,49,[59][60][61]. Villalonga-Olives et al. [29] reported no difference in LBW between refugee populations in Ireland or undocumented Latina migrants in the USA and women in the host countries . However, Somali refugees in Belgium, Canada, Finland, Norway and Sweden had lower rates of LBW compared with women in the host countries . Hadgkiss and Renzaho [14] reported prevalence of intrauterine growth restriction to be one of the most prevalent outcomes amongst women who were seeking asylum, 7% of the population . Preterm birth amongst women with asylum seeker and refugee status Two systematic reviews [14,27] reported preterm birth amongst women who were refugees using data from two original studies [40,54]. Hadgkiss and Renzaho [14] reported premature labour to be one of the most prevalent outcomes in women seeking asylum at 15% of the population , and Gissler et al. [27] reported that women who were displaced from the former Yugoslavia had higher preterm rates than women in the host country . Additional morbidities amongst women with asylum seeker and refugee status Two systematic reviews [14,30] reported additional maternal morbidities and data from six original studies [40-42, 55, 56, 62]. The additional morbidity outcomes reported for women with asylum seeker and refugee status are similar to those reported for migrant women . There were additional data reported explicitly for women with asylum seeker and refugee status that were not reported for migrant women showing increased risk of severe acute maternal morbidity , gestational diabetes, anaemia and uterine rupture. There was a lack of data explicitly for women with asylum seeker and refugee status and offspring infection and admission to special care units. Hadgkiss and Renzaho [14] reported that asylum seekers faced a range of complex obstetric issues including bleeding, gestational diabetes, anaemia, 4.5 times higher incidence of SAMM than the general obstetric population , uterine rupture and eclampsia ; but lower incidence of obstetric haemorrhage . Baalam et al. [30] reported poor health amongst women with asylum seeker and refugee status which caused complications for the women and the newborn babies, including infected wounds, HIV and hepatitis . --- Healthcare access and experiences amongst women with asylum seeker and refugee status Twelve systematic reviews reported access to or experience of perinatal healthcare amongst women with asylum seeker and refugee status [8-11, 20, 23, 30-34, 36]. The sub-themes presented are the same as the results for migrant women. However, there are additional results within the sub-themes for asylum seekers and refugees that were not present, or not as detailed, for migrant women. Access to perinatal healthcare amongst women with asylum seeker and refugee status The barriers to accessing care are summarised here under the themes of structural and organisational barriers, social barriers and personal and cultural barriers. --- Structural and organisational barriers Seven systematic reviews reported structural or organisational barriers for women with asylum seeker and refugee status to access perinatal healthcare [10,11,23,30,31,33,36] including data from 15 original studies [39-44, 47, 57, 63-69]. Results relating to limited ability to speak the language of the host country or understand the verbal or written information provided [10,23,30,33] were similar to the results for the overall migrant population, as were challenges navigating, and a lack of familiarity with, the healthcare systems and inadequate information about what support services exist [10,11,30,31,33,36]. Additional data relevant to asylum seekers and refugees included a lack of knowledge about availability of support services which led to feelings of social isolation [36]. There were assumptions amongst asylum seekers and refugees that they would have to pay for perinatal healthcare when they were entitled to free care [10,31] and mistrust of healthcare professionals who were perceived to be a threat to the emotional and physical safety of asylum seekers who did not engage with antenatal care [31]. Higginbottom et al. [23] also reported that learning the host country language was not a priority for women, and that the men in the household attended language classes while the women stayed at home. --- Social barriers Six systematic reviews reported social barriers to accessing perinatal healthcare [8,10,11,30,32,36] including data from 12 original studies [39-44, 64, 66, 70-73]. Many of the social barriers to accessing or continuing with perinatal healthcare were similar to those for migrant populations such as a lack of finances, transport, issues with housing and a lack of family and friend networks [8,10,11,30,32,36]. However, these difficulties were described in the systematic reviews to be particularly challenging for women with asylum seeker or refugee status due to temporary and uncertain status, not being permitted to work in their host countries and the impact of these factors on available resources and having a 'normal life' [8,36]. For example, Higginbottom et al. [36] describes postnatal refugees skipping meals because of a lack of resources, and Balaam et al. [30] reported that some types of accommodation for refugees and asylum seekers are restricted by fixed mealtimes which imposed practical challenges with flexibility to attend appointments. Mengesha et al. [10] reported that home visits by refugee health nurses were positively received, and Balaam et al. [30] reported that childbirth was a critical milestone towards a better social status, and that the baby represented a new beginning and a health resource. --- Personal and cultural barriers Five systematic reviews reported personal and cultural barriers to accessing perinatal healthcare [9,20,23,30,32] including data from 11 original studies [40-44, 67, 71, 74-77]. The systematic reviews reported similar results to those for migrants in relation to a lack of cultural understanding of postnatal depression and a preference for female health professionals. Further context was provided on gender preference for asylum seekers and refugees. Aubrey et al. [9] reported that higher rates of caesarean deliveries amongst Syrian refugee women resulted from avoidance in seeking antenatal care due to the lack of female health professionals and the fact that only 5 out of 18 African refugee women in the USA would accept care from a male health professional. However, these findings were in conflict with other studies in their review which reported that African refugee women accessing obstetric care in Australia, and Somali women in the USA, would accept care from a male health professional in an emergency [9]. Experience of perinatal healthcare amongst asylum seekers and refugees The themes identified in the systematic reviews around experience of care related to negative communication and discrmination, relationship with health professionals, cultural clashes, and clinical perinatal care are summarised below. --- Negative communication and discrimination Seven systematic reviews reported negative communication and discrimination data for asylum seeker and refugee women [10,11,23,30,33,34,36] reporting data from 12 original studies [39-44, 63, 64, 66-68, 73]. There were similar negative communication experiences to the results for migrant women, including reliance on interpreters and experience of discrimination. However, these negative experiences were more widely represented in the data specific to women with asylum seeker and refugee status than for general migrant populations. Balaam et al. [30] reported that refugee and asylum-seeking women were less willing to state their needs and wishes. Data relating to reliance on interpreters represented an inadequacy of service provision leading to delayed care, women's reliance on body language and facial expressions to communicate, their needs not being met and the women being unable to express their concerns. There was a reported need for more consistent professional interpreting support for women with asylum seeker or refugee status including integrated services, continuity of competent interpreters and improving of health professionals' knowledge of when interpreting services are required [10,23,30,33,36]. The systematic reviews reported that the most vulnerable women with asylum-seeking or refugee status had the most difficult situation and negative encounters with health professionals including openly racist and discriminatory care, cultural stigma, disrespect, hostility, stereotyping and being treated as 'primitive people' [10,30,33,34]. These experiences are demonstrated in a quote from an included study reported in the systematic review by Wikberg and Bondas [34]: "An African woman asked for help when she got an infection but was not met with respect: She looked at me like this and said, 'You are OK'.. . She said to another midwife, 'These Africans. .. they come here, they eat nice food, sleep in a nice bed, so now she doesn't want to move from here!' . .. When she said this I didn't say anything, I just cried… she doesn't know me, who I am in my country. And the other midwife said 'What's wrong with them, these Africans?' and some of them they laughed" . Women reported that these interactions were influenced by skin colour, their language ability and communication problems, and that they wanted supportive, non-discriminatory care [30,33]. --- Relationship with health professionals Four systematic reviews reported data on the relationships between health professionals and women with asylum seeker and refugee status [10,11,30,33] from 10 original studies [40-44, 47, 63, 64, 68, 70]. There were similar findings to the results for overall migrant populations in relation to the importance of a supportive relationship with health professionals, negative experiences such as feeling health professionals were too busy and a lack of confidence to discuss their issues with health professionals [10,30,33]. Positive interactions were experienced when health professionals had respect for practices from the country of origin or were of the same ethnicity or religion, and positive support increased confidence in asking questions and acceptance of the new healthcare system and practices [11,30]. --- Cultural clashes Five systematic reviews reported cultural clashes in perinatal healthcare experience amongst women with asylum seeker and refugee status [10,11,30,32,33], reporting data from five original studies [63,64,66,70,71]. All data specific to women with asylum seeker and refugee status duplicate the findings of the overall migrant women, such as tensions between feeling the need to adapt to host country medical practices and women's preferences for traditional cultural or religious practices. No new findings were identified in the data specific to women with asylum seeker and refugee status. --- Clinical perinatal healthcare Six systematic reviews reported issues with the clinical perinatal healthcare amongst women with asylum seeker and refugee status [10,20,30,32,33,36] reporting data from 15 original studies [39-44, 47, 57, 63, 64, 66, 68, 71, 73, 78]. There were some similarities with the results for migrant women relating to negative experiences amongst women with asylum seeker and refugee status, health professionals showing a lack of knowledge and sensitivity relating to FGM, women receiving poor explanations of care and lack of discussion of options, a lack of assessment and referrals for postnatal depression, an over-reliance on technology and Western practices which lacked cultural sensitivity [10,30,32,33,36]. Additional findings in the data for women with asylum seeker and refugee status include the following: outcomes being better amongst women who were able to exhibit resilience and adjust and change their cultural beliefs; disappointment and lack of preparation for the lack of practical postnatal help and support; recommendations for advocacy or link-worker schemes; and the need for culturally appropriate health education materials on labour and delivery and health professional training on Somali refugee women's culture, traditions, values and expectations [30,33]. The systematic review by Tobin et al. [20] reported discrepancies on the topic of support groups for postnatal depression in their included studies; one study reported limited use for refugee women who preferred individual therapy due to privacy, confidentiality and a cultural stigma related to the condition, whereas another study found that social networking and support groups were important in facilitating help seeking and the healing process. --- Discussion This systematic review of systematic reviews aimed to summarise the existing evidence base of perinatal health outcomes and perinatal healthcare amongst women with the status of asylum seeker and refugee. Although all included reviews incorporated data for women with asylum seeker or refugee status in order to be eligible for inclusion, the data reported specific to this population were limited. Only one included systematic review was exclusively focussed on asylum seekers, and the remaining data for asylum seeker and refugee women were grouped with those for heterogeneous migrant populations or other vulnerable women in the evidence syntheses. We found that a number of perinatal health outcomes were worse for migrant women than women in the host country, including mental health disorders, maternal mortality, preterm birth and congenital anomalies. The qualitative and quantitative evidence specifically relevant to women with asylum seeker and refugee status suggests that they have worse outcomes and experiences compared to the evidence from wider migrant populations and to women in the host country, particularly relating to complex obstetric issues , mental health, offspring mortality, sexual assault and unwanted pregnancy, FGM, infectious disease and anaemia. However, similarities in population risk between asylum seekers, refugees and wider migrant populations were observed for some perinatal health outcomes, such as caesarean deliveries. The healthy migrant effect was reported in some of the systematic reviews, particularly relating to LBW where the risk was similar to or better than that for women in the host countries. This was reported by some authors as being an explanation for better outcomes. The evidence suggests that the healthy migrant effect is context-specific and does not translate across all migrants from all countries of origin or receiving countries. Systematic reviews reported a healthy migrant effect amongst specific populations where outcomes tended to be improved compared with women in the host country, either native-born or other migrant groups. However, health inequalities were reported amongst migrant populations from other origin and/or host countries and amongst refugee and asylum seekers who, for certain outcomes, fared worse than either other migrant women or women from the host country. The heterogeneity between migrant, asylum seeker and refugee population leads us to further question the appropriateness of grouping migrant populations in research, practice and policy. Combining populations may mask the true differences in perinatal health outcomes and care requirements, and without these data the development of targeted interventions to prevent adverse outcomes is hindered. Despite the lack of systematic reviews exclusively focussing on women with asylum seeker and refugee status, there were some data on these populations available to explore perinatal health issues amongst these groups of women. The majority of the literature which specifically focussed on women with refugee and asylum seeker status explored access to and experience of perinatal healthcare. These data showed similar barriers to access and use of perinatal healthcare as for wider migrant populations. However, additional depth of data relevant to asylum seeker and refugee women included social isolation resulting from barriers to care, mistrust of health professionals and financial concerns and poverty; the latter barriers were particularly challenging due to the inability to work and temporary and uncertain status of residency. Women's experiences of care also showed similarities to those for wider migrant populations but with apparently increased challenges with language and communication barriers and more widespread experience of racism, discrimination, stigma and stereotyping in encounters with perinatal healthcare services and professionals. This systematic review has several strengths, particularly the comprehensive search strategy. We searched 12 databases, using a search strategy developed with an information scientist with expertise in database searching. The search strategy was developed and pre-tested using MEDLINE, then refined and retested until we were confident that it was both sensitive and specific. We also searched the reference lists of all of the included systematic reviews and implemented citation searching. These supplementary searches identified a further eight systematic reviews, which demonstrates the importance of supplementing rigorous database searches with additional search strategies. This is particularly important when searching for qualitative or observational evidence, which can be limited when using databases alone, and is a recommended approach to search strategies in the Meta-analysis Of Observational Studies in Epidemiology guidelines [15]. The quality of the included reviews was judged to be either moderate or high; no review was considered to be of poor quality. However, supplementing electronic database search strategies was only carried out by authors of 20 included systematic reviews, which suggests that there may be some element of publication bias in the existing evidence syntheses. Only 21% of studies explored publication bias, which may compound bias from combining heterogeneous migrant population definitions in the analysis. A further strength includes screening and data extraction carried out independently by two researchers. We used a validated quality assessment tool to assess the quality of each included review. However, despite our use of a comprehensive search strategy, we cannot be certain that we have retrieved all relevant reviews, as our searches were restricted to English language reviews. The main limitation of this systematic review of systematic reviews relates to data availability in the existing reviews. We had set out to consider selected pregnancy care and perinatal health outcomes specifically for women who were asylum seekers or refugees, but this couldn't be undertaken in depth as the existing evidence identified often did not allow for these sub-groups to be analysed separately. Despite the large number of studies of migration and perinatal health, there was limited evidence available for a number of pregnancy outcomes including pregnancy complications such as obstetric haemorrhage, maternal infections, maternal mortality and congenital anomalies, and although all reviews included data for women with asylum seeker and refugee status, there were limited results specific to this population. This highlights that although studies on migrant health have increased in recent years, certain maternal and offspring health outcomes remain under-researched, which limits the conclusions that can be drawn. There were also limited data exploring the risk factors for developing adverse outcomes between different migrant populations. The majority of risk factor data related to the development of mental health disorders, although these data were not stratified by the specific migrant population in question, which challenges the interpretation and application into routine care. A review of systematic reviews will naturally result in overlapping data from multiple reviews incorporating the same original study data. We have addressed this in our review relating to the analysis of data specific to asylum seekers and refugees, detailing the number of unique studies that contributed to the results and listing these studies in Additional file 6: Asylum and Refugee Data, and by reporting the data explicitly for women with asylum seeker and refugee status separately from the results for migrant women including asylum seekers and refugees. Due to the primary focus of this review being on asylum seekers and refugees, and the volume of data relating to migrant populations , it was not feasible to go into this level of detail for overlapping studies for this population; this is a limitation. However, we do not believe that the identification of overlapping studies included in the systematic reviews that were not explicitly related to asylum seeker and refugee populations would have added to the interpretation of results, given that the major challenge to interpretation was in the grouping of these heterogeneous populations. Our systematic review of systematic reviews suggests a number of areas that warrant further research. There is limited evidence for women with asylum seeker and refugee status on particular perinatal outcomes such as maternal mortality, obstetric complications such as haemorrhage and infections and congenital anomalies. There is also a paucity of research into the potential causal pathways between migrant statuses and adverse health outcomes. Migrants, asylum seekers and refugees are specific populations; investigating health outcomes for these groups when they are combined presents challenges for furthering research as well as for policy and practice. When it was possible to compare migrant populations including asylum seekers and refugees with asylum seekers and refugees in this review, we were able to demonstrate some similar findings for particular health outcomes but also different and worse outcomes which are masked when groups are combined. We were unable to analyse data specific to asylum seekers and specific to refugees. We only identified one systematic review specific to asylum seekers, which suggests that further research is needed. Our systematic review specifically searched for systematic reviews on asylum seekers and refugees, but studies on other groups of vulnerable women, e.g. undocumented and migrant workers, are also needed. The development of effective interventions to support these women will not be possible if heterogeneous groups continue to be combined for research. Our findings on the healthcare experiences of women with asylum seeker and refugee status have implications for practice. Interactions with healthcare professionals were far from optimum, with communication, discrimination and stereotyping reported. Current UK [79] and Australian [80]) guidelines share the common recommendations of health professionals needing to understand the specific needs of these groups of women; that a variety of means should be used to support women; and that there is a need to inform women of antenatal services and how to use them. Given the findings of this review on risks of maternal mental health and obstetric complications, the provision of mental health services and facilitation of timely access to antenatal care is essential for this population of women. Healthcare commissioners should also have a clear understanding of local needs so that appropriate services can be planned [79]. Implementing these recommendations into practice and providing culturally specific training for health professionals have the potential to reduce some of these negative experiences for women and also for health professionals. --- Conclusions This systematic review of systematic reviews demonstrates that women with asylum seeker and refugee status have worse perinatal health outcomes, including mental health, offspring mortality and preterm birth, compared to women from other migrant groups. Further research is warranted on particular perinatal health outcomes, e.g. maternal mortality, as well as on understanding potential causal pathways. Access, use and experience of perinatal healthcare were also reported to be far from optimal. This represents inequalities for migrant women, especially those with asylum seeker or refugee status. Improvements in care are urgently needed to increase access and enhance the experience amongst these vulnerable populations. There is an urgent need for the inclusion of unambiguous definitions of migrant groups to be used in research and for analysis to be stratified by migrant status and other migration indicators, e.g. country of origin and length of time spent in the host country. The results of this review support the need for future research on perinatal health which can make specific recommendations for policy and practice. --- --- Additional files --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Global migration is at an all-time high with implications for perinatal health. Migrant women, especially asylum seekers and refugees, represent a particularly vulnerable group. Understanding the impact on the perinatal health of women and offspring is an important prerequisite to improving care and outcomes. The aim of this systematic review was to summarise the current evidence base on perinatal health outcomes and care among women with asylum seeker or refugee status. Methods: Twelve electronic database, reference list and citation searches (1 January 2007-July 2017) were carried out between June and July 2017. Quantitative and qualitative systematic reviews, published in the English language, were included if they reported perinatal health outcomes or care and clearly stated that they included asylum seekers or refugees. Screening for eligibility, data extraction, quality appraisal and evidence synthesis were carried out in duplicate. The results were summarised narratively. Results: Among 3415 records screened, 29 systematic reviews met the inclusion criteria. Only one exclusively focussed on asylum seekers; the remaining reviews grouped asylum seekers and refugees with wider migrant populations. Perinatal outcomes were predominantly worse among migrant women, particularly mental health, maternal mortality, preterm birth and congenital anomalies. Access and use of care was obstructed by structural, organisational, social, personal and cultural barriers. Migrant women's experiences of care included negative communication, discrimination, poor relationships with health professionals, cultural clashes and negative experiences of clinical intervention. Additional data for asylum seekers and refugees demonstrated complex obstetric issues, sexual assault, offspring mortality, unwanted pregnancy, poverty, social isolation and experiences of racism, prejudice and stereotyping within perinatal healthcare. Conclusions: This review identified adverse pregnancy outcomes among asylum seeker and refugee women, representing a double burden of inequality for one of the most globally vulnerable groups of women. Improvements in the provision of perinatal healthcare could reduce inequalities in adverse outcomes and improve women's experiences of care. Strategies to overcome barriers to accessing care require immediate attention. The systematic review evidence base is limited by combining heterogeneous migrant, asylum seeker and refugee populations, inconsistent use of definitions and limited data on some perinatal outcomes and risk factors. Future research needs to overcome these limitations to improve data quality and address inequalities. Systematic registration: Systematic review registration number: PROSPERO CRD42017073315.

Introduction Rural areas in the United States are ravaged by the opioid epidemic [1,2]. Residents of rural Appalachia are 43% more likely than the rest of the US to die from a drug overdose [3]. The rural opioid overdose epidemic has not spared Oconee County, which is located in the northwestern corner of South Carolina, and is designated as an entirely rural Appalachian County by the US Census Bureau [4]. The overdose death rate in Oconee County in 2020 was 18.9 deaths per 100,000, [5] and the South Carolina state overdose death rate was 34.9 per 100,000, which was ranked as the 17th highest among 50 US states and Washington DC [6]. While the underlying reasons for the high risk of opioid overdose mortality in Oconee County are unknown, excess availability of substances, a risk factor for misuse, [7] was noted by a US Drug Enforcement Administration report, which indicated that a pharmacy in Oconee County receives the 4th highest number of prescription opioids in the state [8]. Additionally, stigma towards persons with opioid use disorder manifests in low rates of willingness to help someone experiencing addiction or overdose [9] and fits the pattern of low social capital seen in rural communities [10][11][12]. A single definition of social capital is not well established [13][14][15][16]. It has been defined as the ability of people to work together for common purposes in groups and organizations, [17][18][19] and as a capability that arises from the prevalence of trust in a society or community, with trust being the expectation of regular, honest, and cooperative behavior from other members of a community [20,21]. Recent literature has continued to emphasize the trust and collaboration that are inherent to a solid definition of social capital [22]. In addition to these less tangible aspects of social capital, a tangible definition of social capital has been undertaken, and includes such metrics as the number of establishments in religious organizations, civic and social associations, business associations, political organizations, non-profit organizations with a local/regional mission, athletic clubs, and the like [12,23]. So while social capital has been rightly described as a collective manifestation of the behaviors, attitudes, and values of individual members of a community, [12] it is also more than these attributes. These attributes should ultimately manifest themselves in support for and participation in community-level institutions. When these different perspectives on social capital are combined, a holistic definition emerges: social capital is a capability arising from both: the intangible attributes within a group or community -including trust and a willingness to collaborate to achieve a common goal; and the presence of tangible social assets -institutions capable of improving the life of the community. Large swaths of the country are negatively affected by low social capital, including many rural areas, the South, and Appalachia [12]. Exploring the reasons for this phenomenon falls outside the scope of this study, but several references are included for interested readers [24][25][26]. A community with low social capital will find itself at cross-purposes as it attempts to solve problems like the opioid overdose epidemic [27]. While the perception that addiction is a failure of morals or of willpower has long plagued America's approach to the opioid epidemic, [28] increased social capital in communities has been linked to improved protection from drug overdose [29]. Still, ideologies and attitudes that penalize cooperation and solidarity result in a lack of community resources that might help neighbors experiencing poverty, addiction, mental health conditions, homelessness, and other societal ills . Low social capital creates a vicious cycle, or Fukuyama's 'distrust tax' , [20] as those most in need of the community's help end up stigmatized, marginalized, and ignored, the problem becomes larger, and the community feels even less safe. Studies have shown that inadequate social capital at a societal level is associated with increased prevalence of OUD [11,29,30]. Since low social capital fuels the OUD epidemic and its worst outcomes, improving both the intangible and tangible aspects of social capital that directly address the opioid overdose epidemic will attack it at one of its roots and break the vicious cycle keeping OUD so prevalent in rural communities. Harm reduction and recovery support services fall into the category of tangible social capital as non-profit organizations with the mission of improving the well-being and health of the community. To combat the local effects of the opioid epidemic, Oconee County Opioid Response Taskforce was formed in 2019 by multiple community stakeholders, including the Prisma Health Addiction Medicine Center. OCORT members include representatives from local law enforcement, emergency medical services, primary care practices, county administration, the school district, and the departments of health and social services, treatment and recovery organizations, and concerned members of the community. To assess community needs and make a strategic plan to address the most pressing opioid unemployed populations, could be a step toward improving community uptake of the harm reduction and recovery service resources critical to individual recovery efforts. Keywords Social capital, Rural, Overdose, Opioid use disorder, Recovery, Harm reduction epidemic issues, OCORT conducted a comprehensive community needs assessment in 2019 by disseminating a 24-item survey to county residents to measure stigma, knowledge, and opportunities for interventions and educational initiatives pertinent to OUD [9]. This survey was updated, revised and expanded in 2022 with newly added demographic items and additional recovery and harm reduction services items to assess support levels for organizations capable of mitigating the opioid overdose epidemic. In this study, using the OCORT 2022 survey responses, we aimed to identify demographic factors and knowledge, beliefs, and attitudes toward harm reduction and recovery support that, if improved, might increase support for community resources to mitigate the opioid overdose epidemic. --- Methods --- Setting and design Oconee County has a population of 78,314 with Whites composing 82.3% of the population according to the 2020 census. With respect to tangible social capital, the county has one hospital, Prisma Health Oconee Memorial Hospital, two community improvement organizations, three family support and assistance organizations, organizations for poverty and hunger, and 34 churches [31]. Although some of these organizations provide support for individuals with OUD, the county is currently lacking in resources intended to reduce substance use disorders and mitigate the opioid overdose epidemic. The 2022 OCORT Community Attitudes Survey had a total of 46 items, inclusive of demographic factors, and elicited beliefs and attitudes toward individuals with OUD and medications for OUD , in addition to SUD knowledge, and support for harm-reduction services. The survey was disseminated via online methods using the REDCap platform and solicitation of paper surveys which were then entered into the REDCap portal by research assistants. The survey was disseminated between May and June 2022 to the general population of Oconee County. To this end, it was distributed through social media marketing on Facebook pages run by OCORT members, via Prisma Health emailing lists targeted to Oconee County zip codes, and through direct solicitation of participants by research assistants in 4 primary care offices throughout the county on a set schedule throughout the month. The survey received a Prisma Health Institutional Review Board exemption. --- Demographic factors The survey items included the following demographic factors, and we dichotomized multiple responses of each item in the following manner for between-group comparisons: Age , Sex , Race , Ethnicity , Marital Status , Employment , Household Income . Item responses with "prefer not to answer" were treated as missing. --- Harm reduction and recovery support score outcome We developed an outcome to measure a level of support for OUD recovery and for building harm reduction service resources. This composite score included 2 items addressing recovery and 7 items addressing harm reduction services. The score was calculated as the number of "Agree" responses to these 9 items. This composite score is herein referred to as "Harm Reduction and Recovery Support Score" . The score ranged from 0 to 9 with the internal consistency estimated by a Cronbach alpha as 0.72 ). The HRRSS items included: 1. Emergency naloxone boxes should be placed in public places for emergency response to overdose . 2. I would support provision of MOUD in the incarcerated population . 3. I would support HIV and Hepatitis C screening in the county . 4. I would support condom distribution in the county . 5. I would support having a syringe service program in my neighborhood . 6. I would provide financial support for a syringe service program . 7. I would support the syringe service program by utilizing the services they provide . 8. I would support the concept of syringe services in the community . 9. I would support safe consumption sites , defined as health services where individuals can inject or consume substances in a hygienic environment under the supervision of trained staff, and have opportunities to engage in other health and social services . We note that there were additional three items on support for MOUD: methadone, buprenorphine, and naltrexone. However, these items were not included in the above HRRSS since a factor analysis revealed that these items belonged to a factor different from that of the above harm reduction and recovery support items, possibly representing a different underlying construct. --- Survey item predictors We considered 22 items related to OUD knowledge, attitudes, and beliefs related to OUD, individuals with OUD, and harm reduction as potential factors associated with the HRSSS outcomes. . Two groups of each item were classified based on the response to each survey questionnaire item; the first and second groups consisted of respondents who agreed/true and those who disagreed/false, respectively. However, "positive" responses between agree and disagree depend on items and are not consistent across all items, and some item responses are neutral between them. --- Statistical analysis Descriptive statistics included mean, standard deviation , frequency and percentages of survey responses. Significance of difference in mean HRRSS between groups was tested by two-sample t-tests. To identify factors associated with HRRSS after adjusting for demographic factors, we tested significance of adjusted mean difference in HRRSS between groups of each candidate factor using a general linear regression model that included all demographic factors as adjusting covariates. We anticipated that at least N = 300 residents would respond to the survey to detect with > 80% statistical power a relatively small effect size of Cohen's d unit scale) = 0.3 between groups categorized by survey responses. All statistical analysis was conducted using R software v.4.0.5 [32] with RStudio [33] and test results with a two-sided p-value < 0.05 were declared statistically significant. --- Results --- Participant demographic compositions A --- Distribution of predictor item responses A vast majority of participants agreed that anyone can become addicted to pain medications , that it is possible to sustain recovery from OUD , that it is important for individuals with an opioid use disorder to be part of supportive community , and that pregnant women with substance use disorders should have access to comprehensive prenatal care, including appropriate counseling and/or MOUD . In contrast, a distinct minority of participants agreed that OUD only affects low-income individuals , and that individuals who receive rehabilitation or treatment will just overdose again . Approximately three quarters of the participants agreed that an opioid use disorder is a real illness like diabetes and heart disease. Approximately one quarter of the respondents agreed that they can easily spot an individual in their community with an opioid use disorder. The rates of "agree/true" responses of all the other items ranged between 25% and 75%. --- Distribution of positive responses of HRRSS items The "Agree" response rates for individual HRRSS recovery support items were: 58.3% for placement of emergency naloxone, and 26.2% for HRRSS support for MOUD among the incarcerated. The 'agree' response rates for harm reduction services support items were: 90.7% for support for HIV and HCV screening, 82.9% for support for condom distribution, 21.0% for support for SSP in neighborhood, 11.5% for financial support for SSP, 10.1% for support for utilization of SSP, 42.3% for support 3) --- Factors associated with HRRSS The strongest factor associated with the HRRSS was the understanding of OUD as disease, "OUD as disease", which has the greatest adjusted mean difference of HRRSS , p < 0.001) after adjusting for demographic factors. The "Effectiveness of MOUD" factor had the second greatest adjusted mean HRRSS difference of 1.11 , p < 0.001). Other factors significantly associated with greater adjusted HRRSS scores are displayed in Table 4 as follows: agree on "Willingness to live in neighborhood of OUD", "Same right to a job", "OUD part of supportive community", and "Naloxone administration to a stranger"; and disagree on "Can stop drug use", "Abstinence therapy only", and "OUD dangerous". --- Discussion This study found a relatively low overall harm reduction and recovery support score in Oconee County, South Carolina, at a mean of 4.1, below half of the maximum HRRSS. This is consistent with the known low county-level scores on metrics of social capital and suggests that HRRSS could be indirectly measuring levels of social capital as it relates to OUD recovery potential, particularly as harm reduction and recovery support service organizations fall into the category of tangible social capital. Furthermore, the rates of positive response were smaller than 50% for 6 of the 9 HRRSS items. Nonetheless, HRRSS of younger and employed respondents are significantly greater than their older and unemployed or retired counterparts, although a further analysis revealed that the effect of the employment status was no longer significant after adjusting for age . Respondents who agree that OUD is a disease and who agree with the effectiveness of MOUD also had greater HRRSS. In addition, respondents with less stigmatizing attitudes in general had significantly greater HRRSS. A majority of survey respondents support placement of naloxone boxes in public places, condom distribution, and HIV and HCV screening for recovery support or harm reduction services. Although respondents are neutral for supporting the concept of SSP and SCSs, they appear to disagree somewhat strongly on locations and utilizations of SSP, financial support for SSP, and MOUD treatment for incarcerated population. The factors associated with support for SSP found in a sub-analysis of the 2019 OCORT data [34] were mostly replicated in the present analysis. While respondents perhaps understand the need for harm reduction resources such as SSPs or SCSs, they might also be somewhat resistant to building or placing physical resources to that end. Such resistance, quantified as lower HRRSS, is greater for those who disagree that OUD is a disease and that MOUD is effective, and is also greater for those who hold stigmatizing attitudes towards individuals with OUD. Thus, there is a clear and significant positive correlation among higher HRRSS, knowledge regarding OUD/SUD, and less stigmatizing attitudes toward individuals with OUD. These results reported in Table 4 were not significantly different when the employment status was adjusted for in the model even after further broken down into three categories of employed, seeking, and retired . In rural communities, the beneficial effects of SSPs and SCSs have not been well studied, although a qualitative analysis and systematic review [35] supported expansion of SSPs and SCSs into non-urban setting as evidence-based interventions to reduce overdose and transmission for infectious diseases. In other settings, however, the effectiveness of SSPs and SCSs has been well-documented with consistent evidence showing that availability of a SSP or SCS does not increase drug use or improper disposal of syringes but reduces overdose mortality and crime. For instance, SSPs reduced drug use and increased drug treatment enrollments in Seattle [36] and reduced infectious disease without increasing improper syringe disposal in high population areas such as California [37]. A study of SSPs in Baltimore indicated that after just 2 years of the program improper syringe disposal had been significantly reduced by > 46% [38]. Unsanctioned SCSs have succeeded in overdose prevention, infectious disease transmission reduction, and have even reduced crime in the United States [39,40]. Similar findings have been reported in urban cities in Canada, Australia, Switzerland, and Spain, where studies focusing on supervised injection services reduced public drug injections and improper disposal of syringes without any increase in drug injecting, drug trafficking, crime or overdose mortality [41][42][43]. Naloxone programs have also substantially contributed to drug overdose reversals. From 2010 to 2014, local naloxone program sites increased from 188 to 644 , distributions of naloxone from 53,032 to 152,283 , overdose reversals from 10,171 to 26,453 , and the participating states including Washington DC from 14 to 30 [44,45]. Community harm reduction resources such as SSPs, SCSs, and naloxone distribution programs are tangible local social capital as it relates to mitigating the opioid overdose epidemic, while the attitudes toward these resources are intangible aspects of social capital in a community. A lack of support for these harm reduction resources in this study's results, and the accompanying correlation between lack of support and stigmatizing attitudes suggests that the HRRSS is likely indirectly measuring social capital as it relates to attitudes and institutions that would improve OUD outcomes. Accordingly, increasing levels of OUD-related knowledge and decreasing stigmatizing attitudes in the community could be expected to increase support for harm reduction and recovery support resources, thus tangibly and intangibly increasing social capital levels. These findings highlight the importance of emphasizing education of the general population regarding the neurobiological model of OUD and other substance use disorders in general, along with communitywide educational efforts to reduce stigma towards individuals with OUD. If interventions can successfully target older and retired/unemployed persons, the improvement may be more marked. At the same time, to leverage an increase in social capital for mitigating the OUD epidemic, future research should strive to develop and evaluate comprehensive community-level multidimensional interventions targeting populations with high levels of stigma or with inadequate knowledge of OUD and its treatments to build resources for recovery services and harm reduction in our rural community. A few limitations should be taken into consideration when the study findings are interpreted. First, the measure developed and reported in this study has not been tested in any other population to date. Second, Oconee County may not be representative of US rural counties in terms of rurality, population composition or extent of OUD burden. This limits the generalizability of the findings to a broader rural or urban population. Third, our survey sampling strategy might have not resulted in a representative random sample of the Oconee County population as they might have been more motivated to respond by interest in the OUD epidemic. Collecting survey results in medical office waiting rooms may have contributed to the number of responses from women, as women are more likely to visit a doctor's office. People with SUD or historically excluded groups are less likely to engage with formal institutions, including healthcare, and so also may have been underrepresented in this sample. Therefore, the implicated potential influence of response bias further limits generalizability of the findings. Furthermore, the dichotomized associations chosen for the analysis may have combined disparate groups within the community and affected results. All but two of the survey items read at an 8th grade level, but the other two read at a 9th grade level, so it is possible that readability of the survey affected results. Lastly, the "prefer not to answer" response was treated as missing, and its rate was approximately 11% for the potentially sensitive income item. Although the mechanism underlying any missing item is unknown, it was assumed to be missing completely at random. --- Conclusions Building tangible or physical resources to improve recovery and harm reduction services and improving the intangible attitudes and knowledge that surround and affect the opioid overdose epidemic may mitigate the community destruction caused were the epidemic to spread unchecked in communities with low social capital. Although the need for these services appeared to be well recognized, community willingness to develop the social capital is inadequate as reflected in the low HRRSS reported in this study. Nonetheless, respondents with adequate knowledge of the disease model of OUD and the effectiveness of MOUD treatment -and those with lower levels of stigmatizing attitudes -have greater support for recovery and harm reduction services. Development and evaluation of comprehensive interventions to improve knowledge and reduce stigma and focusing more on older and unemployed or retired individuals could hold promise as a strategy to mitigate the opioid overdose epidemic in rural communities. --- Data Availability The datasets used and analyzed during the current study are available from the corresponding author on reasonable request. --- Abbreviations --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background Rural areas in the United States (US) are ravaged by the opioid overdose epidemic. Oconee County, an entirely rural county in northwest South Carolina, is likewise severely affected. Lack of harm reduction and recovery resources (e.g., social capital) that could mitigate the worst outcomes may be exacerbating the problem. We aimed to identify demographic and other factors associated with support for harm reduction and recovery services in the community.The Oconee County Opioid Response Taskforce conducted a 46-item survey targeting a general population between May and June in 2022, which was mainly distributed through social media networks. The survey included demographic factors and assessed attitudes and beliefs toward individuals with opioid use disorder (OUD) and medications for OUD, and support for harm reduction and recovery services, such as syringe services programs and safe consumption sites. We developed a Harm Reduction and Recovery Support Score (HRRSS), a composite score of nine items ranging from 0 to 9 to measure level of support for placement of naloxone in public places and harm reduction and recovery service sites. Primary statistical analysis using general linear regression models tested significance of differences in HRRSS between groups defined by item responses adjusting for demographic factors. Results There were 338 survey responses: 67.5% were females, 52.1% were 55 years old or older, 87.3% were Whites, 83.1% were non-Hispanic, 53.0% were employed, and 53.8% had household income greater than US$50,000. The overall HRRSS was relatively low at a mean of 4.1 (SD = 2.3). Younger and employed respondents had significantly greater HRRSS. Among nine significant factors associated with HRRSS after adjusting for demographic factors, agreement that OUD is a disease had the greatest adjusted mean difference in HRSSS (adjusted diff = 1.22, 95% CI=(0.64, 1.80), p < 0.001), followed by effectiveness of medications for OUD (adjusted diff = 1.11, 95%CI=(0.50, 1.71), p < 0.001). Conclusions Low HRRSS indicates low levels of acceptance of harm reduction potentially impacting both intangible and tangible social capital as it relates to mitigation of the opioid overdose epidemic. Increasing community awareness of the disease model of OUD and the effectiveness of medications for OUD, especially among older and

Background Declaration on the elimination of violence against women defines violence against women as "…any act of genderbased violence that results in, or is likely to result in, physical, sexual or psychological harm or suffering to women, including threats of such acts, coercion or arbitrary deprivation of liberty, whether occurring in public or in private life" [1]. By year 2013, one out of three women experienced violence [2]. Among those violence, intimate partner violence on pregnant women is a global health concern [2]. The WHO multi-country study on women's health and domestic violence against women found the prevalence of physical intimate partner violence as 15 % in Japan to 71 % in Ethiopia and the prevalence of physical domestic violence against women in pregnancy 1 % in Japan city to 28 % in Peru Province, with the majority of sites ranging from 4 to 12 % [3]. Prevalence of violence against pregnant women in developing countries is estimated to be 4 to 29 % [4]. Analysis of demographic and health surveys and the international violence against women survey showed prevalence of intimate partner violence during 'pregnancy to be 2 % in Australia, Denmark, Cambodia and Philippines to 13.5 % in Uganda, with the majority 4 and 9 % [5]. In Ethiopia, community based studies indicated that 50 to 76.5 % of women experienced domestic/intimate partner violence in their life time [6][7][8][9][10]. Prevalence of intimate partner physical violence during pregnancy in rural Ethiopia is reported to be 8 % [3]. Intimate partner violence during pregnancy has been associated with fatal and non-fatal adverse health outcomes for the pregnant woman and her unborn fetus either direct physical trauma to her body and physiological stress from current or past abuse or fetal growth and development [11][12][13][14][15]. Different studies showed that intimate partner violence against pregnant women was significantly associated with adverse maternal health outcomes i.e. unintended pregnancies, pregnancy-related symptom distress, inadequate prenatal care, induced abortion, spontaneous abortion, gestational weight gain, intra uterine restriction, hypertension, pre-eclampsia, third trimester bleeding and STIs. Pregnant women were at higher risk of adverse outcomes including maternal death [16][17][18]. In low and middle income countries including Ethiopia, there are several gaps and inadequate evidence on prevalence and factors associated with intimate partner violence during pregnancy [4]. Most studies were limited to intimate partner violence to non pregnant women. However, the consequence of intimate partner violence during pregnancy is worse. Thus, this study assessed magnitude and associated factors of intimate partner violence during pregnancy in Abay Chomen district, western Ethiopia. --- Methods --- Study area The study was conducted in Abay Chomen district, Oromia region, western Ethiopia. Abay Chomen district is located in Horro Gudurru Wollega Zone of Oromia region, 246 Km to the west of Addis Ababa, the capital city of Ethiopia. The district has two administrative towns and 18 rural kebeles, which is the smallest administrative unit in Ethiopia. There were five health centers and 18 health posts in the study area [19]. --- Study design and population Community based cross sectional study was conducted in the district in April, 2014. Study population were randomly selected and pregnant women aged 15-49 years who are living in the study area for at least six months prior to the study were targeted. Sample size was calculated using a single population proportion formula using the following assumptions; 8 % prevalence of intimate partner violence during pregnancy in rural Ethiopia 95 % confidence level, 3 % margin of error, 10 % allowance for non-response rate. Accordingly, the total sample was 299. Out of twenty one kebeles of Abay Chomen district, ten kebeles were randomly selected. Updated ANC registration of health extension workers was checked and household pregnant women census was done for one week by six interviewers and numbering was done in the selected kebeles to fix a sampling frame. A total of 2021 pregnant women who lived at least six months prior to the study period were enumerated. Finally, population proportion allocation was done to identify respondents from the selected households as a study unit. In situations where households had two or more eligible subjects, only one was selected by lottery method. --- Data collection method and tools Data was collected by six high school completed female interviewers using standard WHO multi-country study of VAW questionnaire [20]. The questionnaire was translated to local language Afan Oromo by experts in both languages and was translated back to English by another person to ensure consistency and accuracy. The data collection process was closely supervised by two health officers and the principal investigator. The data collectors and supervisors were recruited based on previous experience on data collection and fluency in the local language. In addition, training was given for two consecutive days on how to interview, handling ethical issues and maintaining confidentiality and privacy. Pre-test study covered 15 pregnant women in a separate kebele, which become out of the main study. Pre-test was conducted to familiarize enumerators with the administration of interview process and for ensuring consistency. Debriefing sessions were held with the pre test field staff and the questionnaires were modified based on lessons drawn from the pre -test. --- Data analysis Data was first checked manually for completeness and then coded and entered in to Epi data version 3.5.1. Then the data was exported to SPSS version 16.00 for data checking, cleaning and logistic regression. Cleaning was done by calculating frequencies and sorting. Bivariate analysis between dependent and independent variables was performed using binary logistic regression. P < 0.25 was used as criteria to select candidate variables for multivariate analysis. Multivariable logistic regression analysis was done to adjust for possible confounding variables. P-value < 0.05 with 95 % confidence interval for OR was used in judging the significance of the associations. Results were presented in text, tables and charts. --- Ethical considerations Ethical approval was obtained from Ethical Review committee of Jimma University. Letter of permission was obtained from Abay Chomen district administration and health offices. The purpose of the study was explained to the study participants and written consent was secured before data collection was started and confidentiality of the information was ensured by coding. For study participants aged less than 18, we received written informed consent from the participants themselves because they were married and mature minor. The consent procedure was approved by the ethics committee for all including those aged less than 18 years [21]. The interview with the pregnant women was undertaken privately in separate area within the household compound and fieldworkers were trained to refer women requesting assistance to available sources of support according to WHO Ethical and safety recommendation [22]. However, no pregnant women requested support. --- Result --- Socio demographic characteristics of pregnant women A total of 282 out of 299 study subjects were interviewed making a response rate of 94.3 %. Majority 135 of respondents were in the age range of 25-34 years. The mean age of the respondents' was 27 years . About two thirds of the respondents' wealth quantile was medium. Nearly half 145 of the respondents had no formal education. About two thirds 184 were housewives, and half 143 of them were grew up in the same community as they born or nearby community/ Kebele . About half reported that they had seen women being battered in their childhood and 32 of them reported that they support pregnant women to be battered by her intimate partner. More than eight in ten 242 had a person to whom they might talk when they encounter problem. Forty eight were advised by the person to whom they talk to take violence as it is normal, where as 6 and 4 were advised to go to elders and to ask for divorce respectively. Nearly nine out of ten 50 of respondents marriage had dowry payment, which they consider has positive impact on how they are treated by their husband and his family and 226 of them reported that the entire dowry was paid. Majority 205 of them reported that dowry payment had positive impact on their rest of life. One in three 98 of the conflict between the husband and the pregnant women was solved by the elders. --- Reproductive characteristics of pregnant women One hundred forty two of the respondents married in the age between 20-24 years with the mean age of first marriage 19.6 years . Nineteen of the couples the initiation of marriage was not based on their own choices and one in five 58 were not volunteer to marry their current partner. Similarly, 95 of them had never conducted marriage ceremony. Thirty seven of them married to partners who had other wife. The mean age at first sex was 17.9 years . --- Prevalence of violence and its forms Intimate partner violence during recent pregnancy was 44.5 %. More than half 157 women experienced all the three forms of intimate partner violence during recent pregnancy. The simultaneous occurrence of intimate partner physical and psychological violence during pregnancy as well as joint occurrence of intimate partner physical and sexual violence was 160 . --- Physical violence The prevalence of intimate partner physical violence during recent pregnancy was 29.2 %. The commonly reported type of physical violence was batter 34 followed by hit with fist/something else that could hurt them 18 and 25 had reported scar or wound . The abdominal beat during recent pregnancy was 38 , and 72 were beaten before pregnancy. --- Sexual violence The prevalence of intimate partner sexual violence during recent pregnancy was 85 . About 56 of the respondents reported that their partners had forced them to have sexual intercourse without their interest or consent during recent pregnancy. In addition, 84 of respondents experienced sexual intercourse during their pregnancy due to fear of their partners. --- Psychological violence The prevalence of psychological/emotional intimate partner violence during recent pregnancy was 46 . Nearly four in ten 18 of the participating women were verbally insulted and made feel bad about themselves at least once during recent pregnancy. The proportion of both pregnant women humiliated in front of other persons and insisted on them knowing where they were all times was 7 . The percentage of pregnant women who reported stress and depression as a result of violence were 24 and 13 respectively . --- Factors associated with IPVDP Result of binary logistic regression showed that childhood growth areas, partner's education, partner's age, age of marriage, age at first sex, ever lived with partner's family, type of marriage ceremony, dowry payment, dowry impact in future life, having discussant to talk their problem, battering seen during childhood, support of pregnant women battering, money provision and marriage level were identified as significant predictors of experiences of IPVDP while respondents' educational status and occupation were not associated. In multivariable logistic regression three variables i.e. dowry payment impact, partner education and undergoing marriage ceremony were associated. When compared with literates, illiterate partners were 50 % less likely to use violence against their intimate partner during recent pregnancy . Pregnant women who reported that dowry payment has positive impact on how they are treated by their husband and his family were 8.7 times more likely to experience IPVDP than those who reported no positive impact on how they are treated by their husband and his family . Pregnant women who didn't undergo marriage ceremony during their marriage were 4.1 times more likely to experience IPVDP . --- Discussion The study showed that intimate partner violence during recent pregnancy was 44.5 % which was consistent with study from Zimbabwe [23] and higher than studies conducted in USA [24], Japan, and South Africa. This could be because this study included psychological violence as an addition for measuring intimate partner violence during pregnancy [3,25,26] unlike the other studies. The higher prevalence found in this study also might indicate that women were disadvantaged segment of the population and the country's patriarchal society norms. About three in ten 82 respondents experienced physical violence during recent pregnancy by their intimate partner. It is in agreement with study conducted in Namibia [3] and Tanzania. However, this finding is higher than finding from previous studies in Ethiopia and Serbia and Montenegro [3]. The possible reason might be the presence of traditional norms that support beating pregnant women in the study area because 32 of the respondents reported that they support pregnant women to be battered by her intimate partner. Likewise, in this study 13.4 % of pregnant women were beaten on their abdomen during pregnancy in which 86.2 % of the assailants were the biological father of the baby they were carrying, which is slightly higher than WHO finding which was 8 % [3]. This can be explained by inter cultural difference among the two study setting as they are south and west Ethiopia settings. Furthermore, it might be due to the fact that WHO study covered a large section of the population where as this study is in one district. The perpetrator, the biological father of the baby she was carrying was lower than previous Ethiopian study [3]. This might be due to presence of polygamy in the study area because 37 of respondents married to partners who had other wife. One in three, 85 of pregnant women had sexual violence during recent pregnancy. This is similar with study conducted in Tanzania [27] which might be due to shared similar socio economic status. It is higher than findings from Namibia [3]. This might be due to sexual autonomy imbalance among the two study areas. Overall, the prevalence of psychological intimate partner violence during recent pregnancy was 46 which is lower than study conducted in South Africa which was 49 % [28]. This could be due to difference in the perception of violence in the respective communities [29]. Women who ever lived with their intimate partner's family were 45 % less likely to experience intimate partner violence during pregnancy when compared with women who had not ever lived with their partner's family. This might be explained if pregnant women were attached to their husband's family, their partner's respect might increase and so lower violence. Unlike other studies, compared with literates, illiterate partners were 50 % less likely to experience violence by their intimate partner during pregnancy. Pregnant women who reported that dowry payment has positive impact in the rest of their lives were 8.7 times more likely to experience IPVDP than those who reported no positive impact. Pregnant women who didn't undergo marriage ceremony during their marriage were 4.1 times more likely to experience IPVDP. Dowry impact on how they are treated by their husband and his family was also a risk factor for IPVDP as dowry is the payment to be made to the groom's family to marry a daughter, and it takes different forms in different cultures. However, the size of the dowry is a common reason for disputes between the families, with the groom's family demanding more than the bride's family can offer, resulting in violence of brides even to death in India and southern Asian countries [30]. --- Conclusions Intimate partner violence during recent pregnancy is observed in nearly half of the pregnant women in the study area. One in three pregnant women experienced both physical and sexual violence. More than one in ten experienced intimate partner psychological/emotional violence. The implication of the study findings regarding the practice of intimate partner violence during recent pregnancy is common in the study site. To that end, increasing community awareness about the consequences and its adverse reproductive health outcomes of the practice could be an important. Moreover, the health extension workers should be engaged in education, screening and referral of IPVDP victims as they are near to the community. In addition the district education office, office of women's affair should strengthen education on prevention of intimate partner violence in schools. The unique finding from this study is about the association between education and intimate partner violence among pregnant women where it associated positively with less education. To make the associations of demographic variables more clear, we recommend further longitudinal studies to further strengthen the magnitude and associated factors and its effect on pregnant women and the fetus locally. Moreover, the longitudinal study will help in checking the repeatability of the findings in different areas and with different study design possible to scale it up to incorporate in the government policy. --- Competing interests We, the authors declare that we didn't have competing interests. ---

Background: Intimate partner violence during pregnancy is the most common form of violence that harms the health of women and the fetus but practiced commonly in developing countries. There is scarcity of information regarding intimate partner violence during pregnancy in Ethiopia. Thus, this study aimed to assess the prevalence and associated factors of intimate partner violence during recent pregnancy in Abay Chomen district, Western Ethiopia. Methods: Community based cross sectional study was conducted among married pregnant women in Abay Chomen district in April, 2014 using a standard WHO multi-country study questionnaire. Two hundred eighty two randomly selected pregnant women aged 15-49 years participated in the study. Logistic regression and multivariate analysis were employed. Results: The prevalence of intimate partner violence during recent pregnancy was 44.5 % (95 % CI, 32.6, 56.4). More than half 157 (55.5 %) experienced all three forms of intimate partner violence during recent pregnancy. The joint occurrence of intimate partner physical and psychological violence during recent pregnancy as well as joint occurrence of intimate partner physical and sexual violence was 160 (56.5 %). Pregnant women who were ever lived with their partner's family were 46 % less likely to experience recent intimate partner violence. Dowry payment decreases intimate partner violence during recent pregnancy (AOR 0.09, 95 % CI 0.04, 0.2) and pregnant women who didn't undergo marriage ceremony during their marriage were 79 % are less likely to experience violence (AOR 0.21, 95 % CI 0.1, 0.44). Conclusion: Nearly half of interviewed pregnant women experienced intimate partner violence during pregnancy implying the prevalence of such practice in the study site. To that end, increasing community awareness about the consequences of the practice could be important. Moreover, as health extension workers works closely with households, they could be crucial players to increase community awareness about intimate partner violence on pregnant mothers and halt it or its risk factors.

Introduction In Andalusia, the fishing industry is predominantly maledominated, but there is growing recognition of women's increasing involvement in onboard activities. AGAPA highlights the crucial roles women play in various aspects of fisheries, such as transformation, handling, sale, administration, and management. However, unlike other regions in Spain, women's participation in activities like net mending, shellfishing, and fish marketing is not prominently visible in Andalusia. Studies conducted in different parts of Spain have emphasized women's involvement in shellfish gathering , fish processing , and family business management . Nevertheless, in Andalusia, these roles have often been overlooked and undervalued, with women's contributions considered as unpaid household work . Furthermore, various studies highlight the significant contributions of women in the fishing sector, encompassing a range of roles such as net mending, preparing fishing gear, managing vessel administration and documentation, handling labor paperwork, cooking for the crew, assisting with product unloading and sales, among other tasks. These contributions, although significant, have often been unnoticed . Moreover, in many cases, women's tasks in the fishing sector were considered part of women's contributions to the household economy, resulting in them not receiving payment or being formally employed for these duties . A study conducted by the regional government of Andalusia in 2021 reveals that women's involvement in the fishing sector has increased by 30% over a span of 9 years, rising from 15.1% in 2011 to 19.7% in 2020. However, the presence of women in extractive activities remains relatively low . Various studies point out different factors that have influenced the limited presence of women in extractive activities, such as the lack of structural adaptation of fishing vessels and the challenging working conditions inherent in this sector . Additionally, the way in which the reproduction of the household economy was articulated with industrial and fishing capital fostered the instrumentalization of the tasks carried out by men and women, and subordinated and made women invisible . Furthermore, the study by AGAPA highlights that the inclusion of women in the fishing sector is not an easy task, as they face challenges not only in accessing various job opportunities but also in advancing professionally once they enter the sector. The ratio of female vessel captains to male captains is approximately one per every 1500 male captains. Additionally, it is worth noting that women have been actively engaging in associations in the past few decades to enhance their visibility within the fishing sector and ensure equitable compensation for their contributions. Moreover, women are assuming an increasingly pivotal position in various initiatives related to the blue economy, including marine tourism, environmental conservation, sustainability projects, and emerging technologies . Within this framework, this research aims to analyze the labor reality of women in the Andalusian fishing sector, identifying the ideological factors and socio-economic conditions that are hindering the access of women to this sector. Implicitly, we also ask ourselves whether the progressive incorporation of women could help address various challenges and contribute to solving different issues related to the social sustainability of fisheries in the Andalusian fishing sector. We refer to challenges such as lack of generational replacement, aging of the working population and a weak innovation in the artisanal fishing sector; gender imbalances and inequalities, loss of local knowledge, and certain occupational skills in the fishing world. Moreover, the study analyzes the contributions that women are bringing to the fishing sector through projects, associations, among other initiatives. --- Methodology For this research, data has been collected through in-depth 70 questionnaires conducted via personalized interviews. These interviews were recorded with the consent of the participants and were conducted both in person and online. This approach provided an opportunity for the informants to offer unexpected narrative insights, and the collected discursive material was analyzed as part of the research process. Participant observation was carried out in the ports where the interviews were conducted. Additionally, the team participated in activities organized by the Andalusian Association of Women in Fishing , obtaining information about women's organizations, the relationship among women's associations in different ports, their main concerns regarding the fishing sector, and their role in management and extractive activities. The sample has a dual territorial and social dimension. Territorially, it is organized by ports , and socially, it consists of a total of 65 interviewed women. However, a series of interviews were also conducted with men to provide even a glimpse of differentiation in their perspectives . The sample data includes information about the family situation , which is significant in examining the perceptions and narratives collected, especially in relation to whether the female informants have or have had family responsibilities, including caring for children. This allows us to assess the perspectives of women who have successfully balanced their family duties with their work in the fishing sector or other unpaid endeavors. This is despite the fact that the gender distribution of tasks in the family group was clearly unequal to the detriment of women. The sample also reflects the diversity of socio-professional roles performed by women associated with the fishing sector today. Service-related responsibilities exceed those associated with extraction tasks , placing a distinct emphasis on management, especially administrative duties. However, roles in associations, spanning political management positions and the extractive sector, are experiencing gradual growth. The sample also reflects the recent emergence of women in technical advisory roles and entrepreneurial tourism. Three profiles that have a long-standing historical presence of women were not included in the sample. These profiles include women workers in canning factories, fish vendors in marketplaces, and employees of aquaculture companies . These particular groups have been extensively studied and have ample statistical and academic information available . The research focused on analyzing fishing ports as the primary unit of analysis, considering them as socially cohesive environments with a high degree of redundant social relationships . Due to the method of contacting informants through personal networks, certain profiles, such as women workers in the aquaculture sector, were not represented in the study since they typically do not belong to the specific social context being examined. --- Results and discussion --- Challenges to social reproduction in artisanal fishing: gender dynamics and socio-economic factors --- Labor conditions and economic insecurity The labor market in the fishing sector is characterized by segmentation and redundancy, as it relies on a limited pool of individuals. According to AGAPA´s study it can be observed that the fishing sector does not have a high representation of young people. The age group of 16-29 constitutes 20.4% of the workforce. In contrast, the majority of the workforce falls within the 30-64 age range, with 39.2% belonging to the 30-44 age group and 38.9% to the 45-64 age group. Furthermore, the crews belong to a well-defined social sphere, with no incorporation of people from other social backgrounds, including young people. This feature has emerged repeatedly in the surveys carried out. Moreover, the fishing industry in Spain has faced challenges with depleted fish populations and unfavorable market conditions, resulting in decreased profitability. The reform of the Common Fisheries Policy aims to promote sustainability but involves restrictive measures impacting catches, fishing time, and the overall socio-economic framework. Economic factors contributing to declining earnings include stagnant market prices, reduced fishing yields, and increased operating costs . There is widespread recognition that pay conditions are not sufficiently attractive and do not provide a sustainable income to support a family economy. The causes pointed out by informants are the perception of uncertainty in fishing trips; the tendency for prices received by fishermen to stagnate; and the so called share system as a mechanism of remuneration in small-scale fleets . In addition, there is a perception that work at sea is hard, with working hours that are hardly compatible with what is considered a normal life. During the last decades, the generalized recruitment and hiring system was the "part-time and permanent seasonal contracts." Shipowners could terminate crews in case of any contingency. This created a vulnerable and insecure position among sailors, which made recruitment unattractive. The Spanish government implemented a Labour Reform in 2022 on urgent measures to guarantee employment stability and the transformation of the labour market.1 In all the ports investigated, hiring has been adjusted to the new legal framework, but, so far, no response of greater interest on the part of seafarers to be recruited in the new context has been detected. One effect of this crisis in the fisheries labor market is the loss of strategic occupations, such as fisheries technicians or, in particular, net menders. Additionally, the values and expectations of young people have shifted, with a greater emphasis on pursuing higher education, limited availability due to incompatible working hours, and a lack of commitment to fishing companies. As a result, they are prone to abandoning the fishing activity at the slightest difficulty or when alternative opportunities arise. This situation poses a significant challenge for ensuring generational replacement in the small-scale fishing sector. In any case, in ports where fishing activity is socially well rooted, in both Atlantic and Mediterranean ports, it has been noted that the crisis in other sectors such as construction caused some of the fishing workers to return to the sector: "back to mother fishing -Madre pesca -," as an informant from the port of Motril told us. Another limiting factor for new enrollments is the scheduling of the mandatory training courses for sailors. Sometimes courses are organized at times when the fleet is operational; i.e., their timing is not adjusted to the seasonality of the fishing operations. In addition, these courses are often announced well in advance, leading to a situation where young individuals who express interest in participating may ultimately not attend when the course is finally held, which could be several months later. The main response of shipowners to find crews in the absence of a local population is to recruit migrants. This trend is more established in Mediterranean ports, although it is also widespread in Atlantic ports. These are people of Moroccan, Senegalese, or Mauritanian origin, especially from the coastal areas of the north-west coast of Africa. The first concern is to regularize the situation of these --- AdministraƟon AdministraƟon Fig. 4 Role of women that were interviewed . Source: Own elaboration based on field work workers and to ensure that they complete training courses. The incorporation of Latin American crews is also known in some fishing ports. In certain ports like Barbate, Chipiona, Adra, and Almeria, where various fishing modalities exist, shipowners engage in competition by offering incentives to attract the most skilled and reputable fishermen. Even in some ports, such as Almeria and Punta Umbria , we have identified that some shipowners from the most intensive fleets are offering contracts with a salary, guaranteed the so-called inter-professional minimum wage legally defined. --- And women? Within the framework of active policies to encourage women's participation in the workforce, there is an observed increase in hiring, labor activity, and social recognition of women in various areas of the fishing industry. Women are taking on roles such as fish buyers at auctions, administrative and political positions of responsibility, technical and research positions, and in maritime-fishing tourism activities . Based on the data gathered during field research, an approximate estimation of the number of women employed in the fishing industry in Andalusia was obtained. In defining profiles, it is important to acknowledge the unique social and cultural conditions of fishing communities, as well as the challenge of representing dynamic social roles in statistical analysis. Our approach involved simplification for statistical purposes, focusing on individuals' predominant roles to better understand their discourse and ideological positioning. The obtained results align closely with the data provided by the AGAPA administration regarding the analyzed job sectors in our field survey. However, the participation of women in the extractive sector remains limited and progresses at a slow pace. Along the entire South Atlantic coast, Punta Umbria stands out as a port where women are employed in all types of vessels, and in other ports in Cadiz, their presence is noted in different fishing modalities. In the Mediterranean, Motril and various ports in Almeria are notable examples. In all cases, this labor participation is closely linked to pre-existing family relationships, such as sisters/daughters or as wives. It can therefore be argued that, simultaneously, the family network acts as both a facilitating and limiting factor, as beyond these familial ties, there is no possibility of being hired as a female fisher. However, in certain instances, women from fishing families, possessing the required qualifications are incorporated through fictitious enrollment to meet bureaucratic obligations instead of being genuinely hired. The survey has shown that the role of associations can be a factor that promotes greater female labor participation, as seen in the demonstrated effect in Punta Umbria or Almeria. Newly established associations in recent years are already showing their capacity to drive change, while other more traditional entities, established in the province of Almeria or women net menders association in Barbate, still need to validate their transformative capacity. In contrast, in ports where there is no such associative activity, the registered perception is one of a radical incompatibility between women's visible social action and the fishing industry. This makes more difficult to change the inertia of invisibilization of the different roles of women in fishing industry, and to promote the presence of female fishers on board, directly participating in harvesting activities. --- Values, ideas, and discourses on the difficulties of women's access to the extractive and fishing sector and the recruitment crisis --- Women's contribution to the fishing sector The women who were interviewed emphasized that women entering the fishing sector can make diverse contributions. Firstly, they play a crucial role in managing and administering fishing operations. Often, women handle administrative tasks and maintain the necessary documentation for the vessels. Secondly, they bring a distinctive perspective to fishing activities, characterized by their conciliatory nature, rationality, organizational skills, and orderliness. Additionally, the women interviewed exhibit a particular sensitivity towards addressing environmental concerns, promoting education, and supporting the development of sustainable tourism. Based on the conducted interviews, women indicate the following areas and how they can contribute to the fishing sector if they are included in offshore activities: • Work-life balance and rational decision-making: According to the informants view, women can bring a unique perspective to the table, enabling them to analyze situations calmly and contribute to a management style that helps prevent or minimize sanctions in the fishing sector. • Efficiency and organization: Women are recognized for their ability to introduce order and organization into various aspects of the sector, including administrative tasks and fishing operations. • Leadership and administration: Women have played a significant historical role in the management of fishing companies, especially in small-scale fleets In the last decade, in addition, women are achieving managerial and administrative positions in fishing organizations. They could continue to offer valuable contributions in this area. • Environmental awareness and education: According to informants perceptions, women possess a special sensitivity to environmental issues and can implement projects that contribute to the preservation of marine ecosystems. They also state that they have the potential to drive environmental education initiatives to promote awareness of sustainable practices, highlight the importance of the fishing sector, and emphasize the need for marine species conservation for economic and food security reasons. • Gender equality: Women and men are equally capable of working in the fishing industry. The focus should be on knowledge and skills rather than gender differences, promoting a fair and inclusive work environment. By recognizing and harnessing these contributions, the sector can benefit from the valuable perspectives and skills that women bring, leading to a more inclusive and successful fishing industry. According to the survey, 40% of men hold the view that women do not have a distinct impact on the sector. However, the remaining 60% acknowledge that women can contribute in multiple ways, including enhancing fisheries management, promoting order and cleanliness, and introducing fresh perspectives on organization and work approaches. Women bring drive, optimism. They also have a different point of view and have improved fisheries management. . Regarding the involvement of women in addressing the issue of generational succession in the artisanal fishing sector, there were different perspectives. Some women believed that they can contribute across all areas of fishing and thus fill the gaps left by the shortage of sailors, skippers, netmenders, and engine operators in many ports. However, it was acknowledged that there is a barrier preventing women from joining fishing vessels. Most of the women who are already involved are typically family members of boat owners or are boat owners themselves. It was emphasized that it is important to create opportunities for women to view fishing as a viable career option and for boat owners to demonstrate a willingness to hire women. In some surveys, it was stated that it was the female boat owners who were not interested or were opposed to the presence of women outside the family group on board. The few fisherwomen that exist, are immediate family members, so we have to bridge that gap, so that you can hire a woman who is going to be equal to a man. We women have to believe that we can do it. . Nevertheless, some women expressed the belief that women's involvement in the fishing industry is limited to onshore roles such as administration and managing boat paperwork. They emphasized that the challenges of balancing family responsibilities hinder women's participation in sea-related activities. --- Limitations for the incorporation of women in the fishing sector Regarding the main limitations for women's enrollment, questions were asked using a Likert scale ranging from 1 to 5 , considering the criteria established in Fig. 5. In this regard, the interviewed women's responses generally agree that family reconciliation and a preference for unrelated salaried jobs are significant barriers to women's participation in the fishing sector. Their views vary on women's inclination toward fishing and the constraints posed by inadequate vessel conditions. They hold a neutral position with a slight tendency towards disagreement regarding women's exposure to knowledge transmission in the sector. However, they strongly disagree with the belief that fishing is too hard for women or that their non-enrollment is due to cost-saving domestic arrangements where they work without formal employment. Many women emphasize the need to formalize their work in the sector and ensure it is recognized and valued rather than considered unpaid assistance to the family economy. Moreover, this data was analyzed considering the diverse roles of the interviewed women and the perspective of men was also taken into consideration. Although only five men were interviewed, their views provided an initial insight into men's perspectives on the challenges women face to get enrolled in the fishing sector. In relation for the perception of the limitation for women's enrollment in the fishing sector, the interviewed women in the extractive sector strongly disagree with the idea that the "domestic organization of women at home" presents an obstacle to women working at sea. They argue that they have successfully managed their domestic responsibilities while acquiring the necessary knowledge and skills for seafaring. On the other hand, most women in other sectors and some men consider domestic organization as a potential limitation. When it comes to the exposure to knowledge transmission, men agree that it is a limitation for women, whereas women themselves disagree. Women assert that they have access to relevant knowledge and skills and believe that their interest and willingness to learn and apply these skills contribute to their engagement in maritime activities. In terms of vessel conditions, most women and men maintain a neutral or disagreeing stance, indicating that the lack of vessel conditioning is not necessarily viewed as a significant limiting factor. However, women in management and advisory roles express concerns that inadequate vessel conditions can hinder women's participation in seafaring, although it is a problem that also affects men, due to the fact that the habitability conditions of small-scale fleets can be improved. Women in the extractive sector strongly disagree with the notion that "fishing work is too hard for women" as a limitation for their enrollment. In contrast, many other women and men hold a neutral or disagreeing position on this matter . Regarding the compatibility of schedules for family reconciliation, women in the extractive sector strongly disagree with it being a limitation, while the majority of other women agree that incompatible schedules can hinder women's enrollment. Men, however, maintain a neutral stance on this statement. Women in advisory and marketing roles disagree with the idea that "lack of interest" is a limitation for women's enrollment. However, the majority of the women in other sectors and some men slightly agree that lack of interest can be a hindrance to women's involvement in seafaring. Lastly, women in the marketing sector strongly agree that women's preference for salaried jobs over seafaring can be a limitation. Most of the women in other sectors and men also agree or slightly agree with this perspective. Based on the interviews, women consider that their incorporation into the fishing sector faces various challenges and limitations. In this context, several specific obstacles have been identified that hinder women's participation in seafaring roles. These limitations include gender inequality, difficulties in conciliating family and work life, the role of family relationships, and the lack of a tradition of women in maritime work. Understanding and addressing these barriers is crucial for promoting gender equality and creating opportunities for women to contribute to the fishing industry. Let us delve into each of these limitations in more detail. • Gender inequality: Limited opportunities for women to be hired due to the perception of seafaring jobs as predominantly male. • Work-life balance: Challenges in balancing family responsibilities with the demanding schedules of fishing, although some women sailors claim it's a matter of organization. Lack of effective commitment to domestic and family responsibilities by men is a key aspect. • Role of family relationships: Women often embark with their families or on boats where they are owners, limiting opportunities for those without existing family connections. • Lack of tradition: Many women do not consider working at sea due to the perception of it being a maledominated job and limited chances of being hired as crew members. Factors such as the difficulty or effort of the tasks are mentioned, but the women who work as sailors and/or shipbuilders insist that the tasks are perfectly manageable. It is more about having the right knowledge to relate to the new technological areas and the people in the social environment, that is, to acquire experience. --- Final remarks The extractive sector faces multiple obstacles to labor regeneration. Economic instability, perceptions of sectoral crisis, inadequate training policies, evolving societal aspirations, and cultural barriers contribute to the stagnation in attracting new workers. Moreover, the limited social scope of the labor market within fishing fleets, primarily relying on close social and family connections, restricts the renewal of labor force composition. The inclusion of women in the fleets represents a potential solution to address these challenges, as it can lead to social normalization and diversification. Women's associations play a crucial role in promoting gender equality and expanding opportunities for women within the sector. Addressing these issues and fostering a more inclusive and dynamic labor market will be essential for the long-term sustainability and development of the extractive sector. Women's experience and contributions in the fishing sector are highly valued as they bring unique perspectives, skills, and knowledge to various aspects of the industry. They possess deep understanding of traditional fishing practices, adeptness in gear handling and repair, and a commitment to implementing sustainable fishing methods. Beyond their physical labor, women play significant roles in the management, administration, and decision-making processes within fishing organizations. Their presence greatly contributes to the overall success and sustainability of the sector by fostering diverse perspectives, promoting effective resource management, and ensuring the inclusion of different voices and experiences. The valuable contributions of women in the fishing sector extend far beyond the immediate tasks at hand, enriching the industry and paving the way for a more inclusive and resilient future. Unlike some prejudices, women who live in a fishing cultural context are sensitive to this tradition of knowledge and attitudes necessary for a work culture of fishing. In conclusion, the perspectives and experiences of women in the fishing sector highlight both opportunities and challenges. Women recognize the value of associations and their potential to address their needs and drive projects. They assert their capabilities to perform tasks traditionally associated with men, while also acknowledging the obstacles they face, such as gender inequality. The resistance from men -and from some women -and self-perception as barriers to enrollment should be addressed to promote gender equality and increase women's participation in the sector. Efforts to provide continuous training, improve economic security, and enhance knowledge dissemination about the fishing industry will further support the inclusion and active engagement of women and men in fishing activities. It is a task that should be undertaken by both government agencies, fishing organizations, and companies. --- Conclusions and proposals The study conducted on the role of women in the fishing sector in Andalusia has highlighted the need to promote greater inclusion and participation of women in this sector. Despite progress in other areas of employment, women's presence in artisanal fishing is limited and faces various obstacles and limitations. However, the study has also revealed opportunities and potential to strengthen the role of women in the fishing sector. The active presence of women in certain ports and fishing modalities has generated a positive demonstration effect, increasing the number of enrollments and gaining greater social acceptance of their participation. This illustrates the potential for change and the significant contribution that women can make to the revitalization and long-term viability of the sector. Given that the fishing labor market is segmented and redundant in the extractive sector , sectoral organizations, including women's associations, could engage in political dialogue to encourage the hiring of women with the support of public funding. Moreover, several studies have demonstrated that co-management structures, facilitates an understanding of stakeholder engagement and the dynamics of power and resource sharing, as well as leadership processes. Therefore, it is not only for the sake of gender equality but also for the potential effective contribution of women to the labor and entrepreneurial renewal of the fishing sector in its various aspects. In terms of training, it is important to promote collaboration agreements with educational institutions to facilitate onboard hours for higher-level training courses, especially for vessel capitains. The women organizations such as AndMuPes could initiate these agreements, promoting the inclusion of women in these courses. There is a lack of content related to environmental values, cultural values of the fishing activity, and the handling and processing of products for commercial value in fishing training. These values are particularly recognized as important by women, so emphasizing them could make training more adapted to the perspectives and expectations of women in the fishing sector. It is crucial to have a policy that transmits the values of maritime and fishing work to society. The work cultures in the sector are based on values such as professionalism, craftsmanship, autonomy resulting from artisanal tasks, lifelong learning, sacrifice, and a sense of belonging to a well-identified community. These values, as experienced and applied by professionals in the sector, should be conveyed through a policy of social communication. AndMuPes should initiate lines of work from a women's perspective and for women, which would have an impact on the entire population, regardless of gender or whether they belong to the fishing sector. AndMuPes can develop a technical support line for women's associations regarding project calls within the framework of European funding. This would help overcome perceived obstacles for women to engage in these frameworks, particularly in the preferred areas identified in this study . Given the issue of renewing the role of net menders and the historical absence of women in these tasks in more intensive fleets , it is urgent to provide economic incentives and institutional support for specific training in these skills for women. This should build upon existing initiatives but needs to be reinforced. --- Appendix El presente cuestionario es una herramienta que permitirá generar información sobre la situación de la mujer en el sector pesquero andaluz, en distintas áreas de participación social y laboral, con el objetivo de explicar los factores que obstaculizan o pueden promover una mayor participación de las mujeres, contribuyendo así a una mejor reproducción social y económica del sector pesquero, incluyendo a las generaciones más jóvenes. La participación de las personas que trabajan y viven a pie de puerto es fundamental para conocer la realidad social de modo que, a partir de esta información, discutir con los agentes sociales y políticos medidas de relanzamiento del relevo social en la actividad pesquera. Se garantiza el anonimato de las respuestas y la única finalidad de la realización del cuestionario es académica. --- Identificación --- Área Valoración Marineras Patronas de embarcaciones Rederas Mantenimiento de equipos y barcos. Motores y mecánica Papel de gestión de empresas Papel de gestión y representación en entidades asociativas Investigación-asistencia técnica en entidades asociativas Patrimonio marítimo: dar a conocer la cultura de las gentes del mar Mejora de la comercialización Otras VI.4. Enumera las principales dificultades que puedan estar obstaculizando la incorporación de los jóvenes al sector pesquero. VI.5. Enumera las principales dificultades que puedan estar obstaculizando la incorporación de mujeres al sector pesquero. VI.6. En el caso de que conozcas a mujeres participando activamente en distintas tareas y áreas dentro del sector de la pesca, valora tu grado de acuerdo con las siguientes afirmaciones, siendo 1, muy poco de acuerdo y 5 muy de acuerdo . --- Afirmación --- Valoración Las mujeres tienen más capacidad para afrontar cambios, se adaptan mejor a las circunstancias Las mujeres tienen una capacidad de lucha y combatividad muy apreciable. Las mujeres no tienen la experiencia suficiente para llevar a cabo las tareas del sector pesquero, tradicionalmente ocupado por hombres. Tienen más capacidad de afrontar y resolver conflictos Su papel ha sido colaborar en las empresas familiares y ése es el rol más idóneo. Las mujeres realizan su papel del mismo modo que los hombres, ni más ni menos. Es necesaria su incorporación a puestos como marineras, rederas, motoristas, etc. pues de ese modo se favorece el mantenimiento de las empresas familiares. Las mujeres realizan los cursos de formación obligatorios, pero luego no son enroladas por decisión de las empresas armadoras. Las mujeres realizan los cursos de formación obligatorios, pero luego no se atreven a dar el paso adelante para enrolarse. Las mujeres que realizan cursos de formación obligatorios, brindan nuevos conocimientos y habilidades que aportarían si se incorporan a la actividad pesquera. Las condiciones laborales de la pesca extractiva, en los barcos, no son atractivas para las mujeres. Las condiciones laborales de la pesca extractiva, en los barcos, no son atractivas ni para mujeres ni para hombres. Existe un problema por la ausencia de rederos especializados, y la incorporación de la mujer en esta área sería una respuesta necesaria. Las mujeres no son diferentes en cuanto a sus conocimientos, aptitudes y actitudes que los hombres, de modo que su incorporación no aportaría nada distinto respecto a los hombres. VI.7. Según usted, ¿qué aspectos se pueden ver mejorados con la incorporación de las mujeres a distintas áreas de la pesca. --- Marque una o varias Aportaría mano de obra en las embarcaciones Aportaría mano de obra en las tareas de armado, cosido y mantenimiento de redes. Aportaría una visión diferente en temas de organización del sector. Mayor interés o apertura para proyectos de innovación: medioambientales, de diversificación, patrimonio, investigación... Aportaría mayor capacidad de diálogo. Aportaría en la gestión empresarial del barco Las mujeres que han recibido cursos de formación obligatorio, brindan nuevos conocimientos. Permitiría mejoras en las estrategias de comercialización Otra. --- VI --- Afirmación --- Valoración Las mujeres asumen responsabilidades y obligaciones en la casa, lo que dificulta su incorporación. Las mujeres no tienen los conocimientos y aptitudes adecuadas, porque no están expuestas a la transmisión de conocimientos necesarios para la actividad pesquera. Hay mujeres que colaboran en algunas tareas, pero es mejor que no sean contratadas porque de ese modo suben los costos de la empresa familiar. En los barcos no se dan las condiciones adecuadas para la convivencia entre hombres y mujeres. El trabajo en los barcos es demasiado duro y siempre ha sido desarrollado por hombres por ese motivo. Los horarios de la pesca profesional son incompatibles con las tareas domésticas, que habitualmente asumen las mujeres. Las mujeres no tienen interés en incorporarse a la pesca profesional, históricamente han estado al margen de la actividad. ---

Andalusia, located in southern Spain, has a long coastline and a rich history of fishing activities that encompass extraction, industrialization, and commercialization. However, women have traditionally been underrepresented in various stages of the fishing chain in this region. Despite this, women have made significant contributions to the sustainability and social well-being of fishing households through their involvement in diverse fishing-related tasks. Women in Andalusia have taken on diverse roles in the fishing sector, including administration and support tasks for fishing units, as well as responsibilities related to care and maintenance of domestic services. In 2018, the Andalusian Association of Women in the Fishing Sector (ANDMUPES) was launched with the aim of promoting the effective incorporation of women in different tasks. The study conducted an in-depth questionnaire across 25 ports in Andalusia to gather comprehensive data on the labor situation of women in the fishing sector, as well as to expose the picture of ideological factors and socio-economic conditions that are hindering the access of women to the fishing sector. The study aims to highlight the role of women in the fishing sector and their potential contributions to the innovation, transformation, and sustainability of small-scale fishing.

INTRODUCTION News spread through online media news reaches the public faster than print media. This is influenced by the habituation of today's society which is all practical. Therefore, it is not surprising that internet users reach 82 million people . If in the past, news connoisseurs had to wait for the arrival of the newspaper courier to find out the latest news, now is the online media comes with all the conveniences offered. Readers just simply open smartphone to access various actual news websites. As described by Suryawati that the characteristics of the online media lies in the necessity of users to be connected to information technology networks by using computer devices, in addition to knowledge of programs to access information or news. As a means to provide actual information, news in the online media required to always provide the latest news so that readers do not miss the information. One of the topics that has been discussed a lot these days is related to the holding of the G20 Summit in Bali. Indonesia as the host not only received attention from the local media, but many foreign media were also interested in discussing the news about this event. G20 or Group of Twenty is a major forum for international economic cooperation consisting of countries with the largest economies in the world consisting of 19 countries and 1 European Union institution. The G20 represents more than 60% of the world's population, 75% of global trade and 80% of world GDP. G20 members consist of South Africa, United States, Saudi Arabia, Argentina, Australia, Brazil, India, Indonesia, United Kingdom, Italy, Japan, Germany, Canada, Mexico, Republic of Korea, Russia, France, China, Turkey, and European Union . As one of the most magnificent national celebrations of the year, the journalists highlighted this activity from various aspects, one of which was the spotlight on several first ladies who were also present in Bali. A proverb says that behind a successful man there is a great woman. This proverb motivates the media online in Indonesia actively discusses the image of first ladies. This current study examines the image of the first ladies of China and of South Korea based on the reports appearing in the online media which portray these two tough women from different perspectives. After conducting a linguistic study, this research is equipped with the explanation of the differences in the imagery that the media builds on the two first ladies. Related to the phenomena of booming news in the online media, many researchers have conducted studies on this topic. The first previous research is conducted by Sukarno and Amurwani who uncovered the frames of online media in Indonesia concerning with the political news of Prabowo Subianto becoming the Minister of Defense in the Kabinet Indonesia Maju . The result shows that there are two types of frames constructed by the online media dealing with the political issue of PS becoming the Minister of Denfence. The supporting frame tends to claim that PS is the right person to occupy the position. In contrast, the criticizing frame tends to position PS as the person who approaches Jokowi because he really wants to occupy the political position. Next, Hartiana, Setyarinata and & Nugraheni examined the images of President Jokowi compared to the former Indonesian President Sosilo Bambang Yudhoyono. This research is qualitative research with Audience Reception Analysis method. The data source is taken from posts on Twitter. The purpose of this study is to explain in detail about the acceptance of meme of political figures. This meme image portrays Jokowi in a positive image but directly leads to a negative image of Susilo Bambang Yudhoyono. The third research was conducted by Prasetiyo which revealed the image of Tri Rismaharini as Minister of Social Affairs in online news using Theo Van Leeuen's model of critical discourse analysis. The result of the shows the figure of Tri Rismaharini provides a special attraction so as to give a good image in the eyes of the public. The fourth research was conducted by Haris, Amalia & Hanafi who studied Anies Baswedan's image in the media. The study applied a qualitative method using Robert N Entman's framing analysis. The results of this study imaged Anies Baswedan as a political figure who is intelligent, polite, and religious. Some of the previous studies above prove that revealing the images of public figures are very interesting. Many studies relating to the image of a figure in several online media have been conducted, but no one has done research on the image of the first ladies who have important roles to support the success of their husbands, such the ladies of China and South Korea. In addition, the previous researches examined the image of domestic political figures only, while this research examines the image portrayed by Indonesian society through online media of two first ladies of China and South Korea. Furthermore, the focus of this study will not only portray their image of the two first ladies during the series of activities at the G20 Summit in Bali, but also their implication of the different images of the two ladies for the Indonesian people. --- RESEARCH METHODS This research is critical descriptive research, which is a type of research that seeks to understand the image of the first ladies of China and South Korea during the G20 Summit activities in Bali. This study describes the contents of the text in several online media, interpret the data obtained, then explain the sociocultural reasons that are the cause of the emergence of different images built towards the two first ladies. The research data are in the form of sentences from text fragments in several online media which reflect the images of the first ladies. The data selection is carried out through critical reading of news texts containing the images of the two first ladies. Having been collected and selected, the data are analyzed using the Fairclough's model of critical discourse analysis. Fairclough views discourse from three dimensions , namely the dimension of text , the dimension of discourse practice , and the dimensions of sociocultural practice . In the text analysis dimension, the collected data will be examined based on elements that are in accordance with the focus of the study on the image built for the two first ladies using linguistic features such as choices of words in the sentence fragments. In the discursive practice dimension, the text production process will be analyzed in terms of the factors that influence the emergence of news about the image of the two first ladies in several online media. Meanwhile, in the dimension of sociocultural practice, we will examine the phenomenon of social reality in today's society which causes different images of the first ladies of China and South Korea. --- RESULTS AND DISCUSSION Result This research was conducted to examine the different images built by several online media towards the China first lady "Peng Liyuan" and the South Korean first lady "Kim Keon Hee". The data analysis was carried out in three stages, namely the stages of text analysis, discursive practice, and sociocultural practice. The following table demonstrate the linguistic data of the two first ladies relating to their images. --- Discussion The results of the study show that there are differences in the image that some online media have built towards the China first lady, Peng Liyuan, and the first lady of South Korea, Kim Keon Hee. The first lady of China is imaged as a tough figure, active in various humanitarian activities, and persistent in her efforts to improve the welfare of the China people. Meanwhile, the first lady of South Korea is portrayed as someone who has a beautiful face. Some of these online media only focus on news about their physical appearance. This also caused a stir among netizens. The difference in these images is illustrated in the news text fragments by the following online media. --- A. Text Analysis The first analysis of the data is the analysis of the text or the analysis of the sentences describing the image of the two ladies. It is the description of the first lady of China and followed by the image of the first lady of Korea . --- The Image of the first lady of China The first lady of China is mostly portrayed by the media online as a woman who cares social welfare, prosperity, health prevention and treatment programs. Each image can be proved by the following data. The two first ladies then talked about a number of things including strengthening public welfare exchanges The news text fragment above shows that the online media dunia.rmol.id gives an image of the China first lady as a figure who is concerned about people's welfare. Next, both ladies discuss how to improve the livelihood and welfare of community, as presented by the following data. Mother Iriana and Madam Peng Liyuan together strive to improve the livelihoods and welfare of the community . The news fragment above shows that Peng Liyuan is a person who cares about improving his people's livelihood, ensuring that his people can live in prosperity. She also cares about social health, especially the efforts to prevent tuberculosis , and HIV/AIDS. Peng highly praised Iriana's enthusiasm and explained to her about China's policies and achievements in the treatment and prevention of tuberculosis and HIV/AIDS Peng herself served as World Health Organization goodwill ambassador for TB and HIV/AIDS since June 2011. Peng recounted her experience visiting a number of medical, school and community facilities since joining the anti-TB effort more than a decade ago. Xinhua online media through dunia.rmol.id gives an image of Peng Liyuan as the first lady who actively promotes TB and HIV/AIDS prevention and treatment programs. The fragment of the news text above shows that Peng Liyuan's involvement in the treatment and prevention of TB and HIV/AIDS is not only within the scope of the country but it has also been made an ambassador of goodwill for preventing TB and HIV/AIDS by the World Health Organization . The text above illustrates to readers that Peng Liyuan's involvement in the prevention and treatment of TB and HIV/AIDS is not only as an induction provider, but also involved by visiting directly various places. In addition, the media online also gives the image that Peng Liyuan is also a friendly and active lady to join the activity G20 Summit for spouses of heads of states in G 20 in Bali. In the meeting the two first ladies exchanged souvenirs in the form of traditional musical instruments from each country. The fragment of the news text above shows that Peng Liyuan is a friendly first lady. Furthermore, the media also portrays Ping Liyuan active participates in the activity conductiy for spouses of heads of states. Peng Liyuan, wife of China President Xi Jinping, took part in an activity held for spouses of heads of state G20 in Bali. . Online media focuses on reporting on Peng Liyuan's active participation in holding the G20 Summit in Bali. Peng appreciated Indonesia's move to integrate handicrafts and women's empowerment, poverty alleviation, and nature conservation. Finally, Peng Liyuan is imaged as the first lady who cares about the condition of other countries. This can be seen through the attitude shown by the China first lady by appreciating the steps of the Indonesian state in empowering women, preserving nature, and others. --- The Image of the First Lady of South Korea As the first lady of China, the first lady of South Korea, Kim Keon Hee, also becomes the central attraction of media online in the G20 in Bali from different perspective. Korea is very popular country for Indonesian, especially the Korean dramas , and choreography. Many Indonesian young people are very interested in watching Korean drama and choreography, not only because of the story but also of the beautiful bodies of the artists. Therefore, the image of the first lady is also related to her physical beauty. Most media more spotlight the beauty of the first lady comparing to her activities during the Summit of G2o in Bali. This image is reflected in the following data. Beautiful! The wife of South Korean President Yoon Suk Yeol, Kim Keon Hee, was in the spotlight when accompanying her husband at the G20 Bali summit . The news text fragment above describes the figure of Kim Keon Hee who received media attention because of her beautiful appearance. Indonesian netizens are busy talking about Kim's appearance and physique, which is still prime at the age of half a century . The first lady of South Korea has become the concern of many Indonesian netizens who are amazed by her beautiful appearance even though she is half a century old. "The first lady of South Korea is 50 years old. Let me wear a mask first," wrote one Twitter user . The news text fragment above highlights the tweet of a Twitter user who feels inferior to see Kim Heon Hee who is youthful at the age of fifty. "Watching the G20 Gala Dinner when I saw the wife of the President of South Korea, I felt like I was watching Drakor," wrote one Twitter account user . The text fragment in bold above shows that the majority of Indonesian people like Korean dramas, which is a country under the leadership of the husbandfirst lady. The wife of the President of South Korea makes netizens fail to focus because her looks are like Korean pop artists. . The news text fragment above shows that netizens were misfocused when participating in the G20 Summit through the mass media, their focus was on the beauty of the first lady of South Korea, not on the context of the event itself. "Is that the first lady or kpop artist anyway? so beautiful," account @tya_sastria said The piece of text above is one of the tweets of Twitter users who are trying to find certainty whether this beautiful woman is really the first lady or a Korean pop artist. "Though she was born in 1972," account @7.110.121.000 said. . The curiosity of netizens led them to find out the year of birth of Kim Heon Hee. Having learned that the first lady of South Korea was born in 1972, netizens were even more amazed because of her very youthful appearance. Kim Keon Hee looked beautiful and charming when she saw the exhibition of original Indonesian MSME products. Online media really focuses on the physical beauty of the first lady of South Korea. Even when Kim Heon Hee attended the MSME product exhibition, what was highlighted was not her concern for Indonesian MSME products but her focus on physical beauty. Referring to the results of the text analysis above, the images built by several Indonesian online media towards the two first ladies are very different. Peng Liyuan is imaged as a formidable first lady who is active in various humanitarian activities. Meanwhile, Kim Heon Hee is imaged as a South Korean citizen who has a beautiful face like most Korean artists who are admired by many Indonesian people. --- B. Discourse Practice The second dimension of Fairclough's critical discourse analysis is discourse practice . In this dimension, the focus of the study is the production aspect and who consumes the text. The producers of the text above are several Indonesian online media, namely dunia.rmol.id,kumand.com, kalbarnews.co.id, cnnindonesia.com, liputan6.com and harijogja.com. Online media are the media most widely used by society today. The ease of accessing various news in online media via mobile phones also influences the shift of print media readers to online media. This also inspires the enthusiasm of news producers in online media to provide actual and interesting news, one example is the news about the holding of the G20 Summit in Bali some time ago. Online media are competing to upload the latest news from the start of the implementation until the event is actually finished. One of the attractions of the online media about the event is the participation of first ladies. Peng Liyuan, who is the first lady of China, is currently a country with rapid progress in various fields, quite a lot of media coverage. Peng Liyuan's presence is a breath of fresh air for news seekers . Many people are curious about the first lady who is currently intensively accelerating the improvement of the quality of life from these various aspects. Through headline Many netizens flocked to the news about the first lady of China, online media to find out the latest news about Peng Liyuan while in Bali. Besides China, South Korea is also a country that many netizens complain about. The recent boom in Korean drama series has had a major impact on the appeal of online media reporting on matters related to the country. Therefore, the momentum of the G20 Summit in Bali was not wasted by the online media. The presence of Kim Heon Hee in Bali has attracted netizens to keep following news update about the G20 Summit. Various of headline which raised the topic of the Korean first lady's facial beauty, increasing internet users to access online news that discusses it. This phenomenon is certainly very profitable for the owners of online media. Apart from that, netizens' curiosity was relieved by the increasingly active online media reporting on the activities of the beautiful first lady while in Bali. --- C. Sociocultural Practice The third dimension of Fairclough's critical discourse analysis is sociocultural practice. Fairclough explained that this dimension focuses on the social context that occurs outside the media and influences the way a discourse text is packaged by the media. The orientation of this dimension can be in the form of political, cultural, social, economic, and ideological aspects. The findings of this study indicate that the aspect that most influences reporting on the image of the first ladies of China and South Korea is the economic aspect. Online media are competing to provide the latest news that is becoming trending topic in society. Thus, the number of internet users accessing news links in the media will increase. Online media owners will benefit through "pay per click". If observed closely, almost every time you open news in online media, advertisements for various products will appear. This is one of the biggest sources of income for the media owner. Every citizen who clicks on the ad will provide an additional AdSense for media owners. Therefore, it's no wonder many controversial headlines appear in various online media. Apart from the economic aspect, the social aspect is no less influential in the incessant reporting on the image of the first ladies of China and South Korea. China is currently one of the most influential countries in the world. It is even predicted to be the United States' toughest competitor. The pearl of wisdom that "behind a successful man, there is a great woman behind him" is one of the reasons many Indonesian people are curious about the figure behind Xi Jinping , who is none other than his own wife. Likewise with the first lady of South Korea, people who love Korean dramas are also very enthusiastic about waiting for the latest news from the beautiful wife of the South Korean president. Humans as social beings have a basic need to continue to try to add relationships. Along with the development of the times, people who access online media to add insight and get the latest news are also part of the social process. That is, through online media, netizens understand what is happening out there. --- CONCLUSION The implementation of the G20 Summit in Bali some time ago made the public eagerly awaiting interesting news from the activity which was Indonesia's biggest event this year. One of the most highlighted by the media, especially online media, is the participation of first ladies to Bali. The image that was built by the media for the first ladies who attended was different. This has also happened to the image that the online media has built for the first lady of China and the first lady of South Korea. Based on the analysis of the text, the first lady of China is imaged as a person who is diligent, persistent, and is heavily involved in humanitarian activities for the welfare of her people. Meanwhile, the first lady of South Korea is imaged as a young first lady with a beautiful face. In the dimension of discursive practice, online media as a news producer has a clear goal of uploading a lot of news that discusses the images of the two first ladies. China is one of the most developed countries, while South Korea is a country that many Indonesian people complain about with its drama series and Korean Pop which are loved by many people. This means that the need for online media as news producers to be widely accessed by the public is directly proportional to the need for news consumers to obtain information regarding the latest news from the two first ladies while in Bali. Meanwhile, based on the sociocultural dimension, economic factors greatly influence the upload of news about the two first ladies which continues to grow every day. The more netizens who access the link, the more benefits they get through the methodpay per click. Besides that, the social aspect also influenced the incessant news about the presence of Peng Liyuan and Kim Heon Hee in Bali. Humans as social beings need a lot of information. Through reporting on the online media, netizens no longer have trouble getting the latest news.

Online media is one of the most influential sources of information for today's society. This aim of this study is to reveal the image of the first lady of China and of South Korea based on the online news collected from the G20 Summit in Bali. The research data were the corpus from the online media texts collected from fragments of several online news which show the different images of the two first ladies (China and South Korea). The data were analyzed using three dimensions of Fairclough's model of critical discourse analysis. Based on the study of text analysis, discursive practice, and socio-culture practice, the study shows that there are different images depicted by the online news for those ladies. Peng Liyuan, the first lady of China, is portrayed as a diligent person, enthusiasm for participating in various social activities. This image is closely related to China's image in Indonesia as one of the most developed countries after the United States. In different perspective, many media highlighted the physical beauty possessed by the first lady of South Korea, Kim Keon Hee. This image is directly proportional to the popularity of Korean dramas, which are very popular in Indonesian society today. One of the attractions of these dramas is the beautiful faces of the actresses.

Introduction Buzzanell and colleagues define leadership as the process by which an individual mobilizes people and resources to achieve a goal. Leaders motivate others to aspire to achieve them to do so. Leadership mediates others to achieve a certain goal for a team, organization or of a nation . It comprises a set of traits that allows a person to persuade others in order to achieve a given aim . The concept of leadership has a religious dimension too. For example, in Islam, leadership is a symbol of trust † that a person acquires to guide the community into the way of Allah SWT . The fundamental sources of Islamic leadership and guidance for the Muslim leaders are Al-Qur'an and Hadith. Nevertheless, different communities may have different perspectives on women's leadership. We people have an image in our minds about what a male leader looks like. But what about a female leader? Do we have an idea or image Rahim in mind? Of course, great leaders possess similar leadership traits. There is currently a lack of powerful women leaders; it is difficult to know how female leadership qualities can be used to their best advantage . In general, women often take up low to middle positions compared to men, and they face stereotypes that limit their power which prohibits them from holding higher positions . Although some women can reach top leadership positions, they still face the leadership labyrinth . In Brunei Darussalam, although women are treated equally as men, the number of women leaders in the country is relatively low as compared to men . This is the reality despite the fact that women in Brunei Darussalam surpass more than men in education . Additionally, in Brunei, women holding seats in parliament are only 9.1% compared to men's 90.9% . As of 2010 until the present, there have only been three women deputy ministers appointed in Brunei: Datin Paduka Hajah Adina Othman as deputy ministers of Ministry of Culture, Youth and Sports, Datin Hjh Elinda Hj CA Mohamed as the Deputy Minister at Prime Minister's Office and Datin Paduka Seri Dr Hjh Romaizah Hj Mohd Salleh as the Deputy Minister of Ministry of Education . Unfortunately, as of today, out of all three female deputy ministers, only one still upholds their position. There is a deficiency of data on women leaders in the private sector. While there is an estimate that women working in the private sector is 58%, the number of women holding leadership positions is inaccurate . This may be because women holding higher positions in Brunei Darussalam are still negligible, and likely that women are sidelined unintentionally, and their full potential is not fully exploited . --- Rahim According to Brunei Darussalam's National Vision plan , the country aims at achieving a sustainable economy in which equal opportunities for women in the workforce and nation-building are encouraged . This research contributes new knowledge about women's leadership to the literature. This, in the long run, inspires more women to take up leadership roles in the country. It was until 26 March 2021 that leadership opportunities for women are brought up during the Legislative council in which YB Khairunnisa have agreed that women have the capability to prove themselves . Countries such as Thailand, Vietnam and China recognize women's ability to engage in their company's best interest to guide others without the help of men . The growing number of women leaders may also motivate other women to advance in their careers. Schipani et al. note that networking is essential for women to build up social capital, which is crucial in obtaining a leadership position. Both mentors and their mentee share their experiences and assist each other in times of need. With an increasing number of women taking up leadership roles, they have more influence, power and recognition, giving inspiration for others to do the same. --- Objectives and methodology Given the relative lack of research on women's leadership in the country, this research aims to discover women leaders' challenges and experiences in Brunei Darussalam. Women are often viewed as unsuitable for leadership in Bruneian society. This research aims to investigate the motivation and problems women experience in becoming leaders and examine the societal pressures encountered by these women leaders in the country. This research is based on a qualitative approach. To collect data, I conducted a semi-structured interview. The respondents were asked about the motivations they had and difficulties they faced as women leaders, and what factors contributed to their success. Tedrow & Rhoads conducted similar research in a community college in the United States. This research inspired me to do this research in Brunei. Women to be interviewed consist of CEOs and Cofounders of their organizations, including government officials. Unfortunately, out of ten intended participants, the researcher could only interview seven women leaders because some of the interviewees rejected the request due to time difficulties. During the interview, notes were taken, and conversations were recorded via mobile phone. The interviewees have given consent to be recorded and used their names for this study. The interview lasted 15 minutes to 1 hour, and most of the interviewees conversed in English, but some also spoke bilingually for better communication. The data collected from the respondents were analyzed using thematic analysis. Women chosen in the study must be women living in Brunei Darussalam and a citizen of the country, and they must also be portrayed as leaders in social media and news report. Finally, they must have at least three to five years of leadership experience which may include managing one department or have experience handling big projects and initiatives that have had an impact in Brunei. --- Factors affecting women leadership Mentorship Mentoring is a professional activity, a trusted relationship, a meaningful commitment . Women leaders need mentors for personal growth through learning experiences . Mentoring may benefit the mentees with experiences, understanding, challenges and support to pursue their own objectives . Lo & Ramayah note that mentoring has a positive relationship with job satisfaction, such as faster promotion, teamwork and helpful supervisors. Hetty van Emmerik shows that with the help of their mentors, employees enhanced their careers. --- Work and family life balance Women being able to lead must also strike a balance between work and family life . Women leaders must be able to see themselves multitasking to maintain familial relationships while advancing the cause of the organizations they work for . Research by Hewlett & Luce revealed that women leaders that achieve success in their careers are seen to have a balanced family life as well. In order for them to bring a balance between work and family, they got to prioritize their family while taking up the work responsibly. Almaki et al. have also supported this that one of the ways to assume a leadership position is by recognizing the role and responsibilities at work and following precise regulations for home. --- Challenges women leaders face Gender Stereotype Historically, the idea that men make better leaders than women has gained prominence . This gender stereotype ‡ has a severe impact in the long run on leadership development. In characterizing leadership, scholars believe that men are more ambitious and risk-takers than women . Women are seen to be more compassionate, caring and emotional . This is, in fact, undermining the capability of women. The roles of gender stereotypes may affect how women lead and result in double binds, in which women are unlikely to receive recognition and undergo specific standards to keep up with men . Omar & Davidson note that in Japan and Singapore, women leaders must display leadership qualities like a man in their organizations. This, in fact forces the potential leaders to adjust to what the organization wants them to be . --- Social culture Society's perceptions on the role of women are often shaped by their cultural practices, social belief and religion , which may affect the potential of women who want to climb up to leadership . The traditional gender norms still remain strong, as seen by McLellan & Uys , where women must maintain their roles as mothers at home. Although the number of women holding higher positions is increasing, they still have the responsibilities as caretakers of their families . An imbalance between work and family may lead to conflict that can affect the mental health of the incumbent . Hence, many women leaders may choose not to have children in order to focus on their careers . Achour et al. found that in Malaysia, female Muslim lecturers are often bombarded with the conflict between work and family life. They are required to work extended hours, and as a result, they cannot take care of their children properly. Family satisfaction and organizational commitment become critical points for their life . ‡ how society sees both men and women should be Current research In Brunei's perspective, K. C. P. Low explained that Brunei Darussalam indeed undergoes a "Father Leadership" style which is referred to as both paternal and maternal leadership. Although now that the social etiquette and gender roles are more open and flexible , this kind of leadership may affect society. This includes gender institutions influenced by patriarchal norms and values in which women were denied to achieve the position of power due to male dominance in the patriarchal society . Although women working in Brunei Darussalam are large in numbers, they still face significant barriers in reaching senior leadership positions . In Brunei, the Sultan's leadership is based on the national ideology of Melayu Islam Beraja § that serves as a beacon for the country's culture . Even though rights were given to women in Brunei in terms of education, healthcare and employment and citizenship , they still struggle for a balanced life . In many Asian countries, it is obligatory for women to be committed to family responsibilities as prescribed by their gender roles . --- Results This study has identified several themes. The first one is the motivation to become leaders in Brunei Darussalam . Of the seven respondents, two have the same motivation, which is to help the youth and community. The first respondent aims to achieve the 'Wawasan 2035' target of creating a local brand called 'BigBWN' to create more entrepreneurship for local youth. While Respondent B is a lecturer in UBD and a youth Geographer herself, she comes to realize that youth in Brunei needs a certain push to help with their vision. She aims to shape the way youth perceives information by encouraging them to become more mobile and critical thinkers. Respondent C is a final year student in UBD. This respondent has been given the responsibility of 'Lamin Warisan' by her elderly. Although she is faced with doubts, she made the decision to overcome her challenges and prove herself suitable to handle the initiative. Respondent D, while taking the course Drama and Theatre, she was motivated, and she becomes passionate about performing arts and decided to create her own theatre production called 'Sutera Memento.' For another respondent , motivation comes when her expectation is low. While working as a doctor in Singapore, her life was very strict and regimental. Once she decided to take the time to shape her personal development, she started to see life differently, and she changed her life in a better direction. She started to share what she learned with others and soon realizes her voice can make an impact. Respondent F is one of the co-founders of her company, 'the scoop'. While working in Brunei, she realizes that news media in the country did not do an excellent job in delivering Bruneians' stories and what interests them. Therefore, she and three of her colleagues took the risk of creating a platform that would reflect thoughtfully on stories in Brunei. Finally, another respondent believes that her motivation came after joining an organization called YEAB 4 in which changed her from being a shy person to the person she is now, more confident and risk-taker. When she was allowed to lead the initiative abroad, she felt it was life-changing to be entrusted with a big decision. From there, it motivates her to keep on going until now. Table 3. Barriers faced by women leaders in Brunei Darussalam --- Barriers experienced Financial support Business sustainability, own financial --- Access to resources Teams, venue As stated in the table above, the barriers experienced by women leaders are divided into two themes, which are: Access to resources and financial support. All three respondents faced the same struggle in finding the right team. One respondent's team was mainly her students who often need parental consent to participate in theatre. Persuading parents is often difficult. This respondent has two permanent members . The other two respondents faced conflict in delegating tasks within their teams. For example, one is having difficulty for her team to understand her goal, and another finds it challenging to find support from friends, especially when handling events. --- Rahim In terms of financial support, respondent F struggles to maintain her media business. She and her co-founders started the company with only BND 2,000, and with all from an editorial background and no business background, she faces difficulty in sustaining her business, knowing that media business is decreasing in revenue due to Facebook and Google taking the line of advertising. Respondent D also has issues funding her initiative. To her, people were still skeptical of 'Sutera Memento'. Therefore, the theatre is built from her own financial source. She often faces hardships in creating astounding events but also wanting to save cost, especially in terms of venue. She said, 'Venue I suppose is another challenge. There is a lot of venue in Brunei, but if you want a theatrical venue, there are only a few.' Two factors that enabled women leaders to succeed. These are Family and spousal support and mentors. All respondents agreed that they found it easy to become leaders due to their family and spousal support, but each has different experiences. Most of the respondents said their family supports them in everything they do. As for another respondent , she was a single mother, and her family supports her in balancing their time taking care of her daughter while she is busy training. As for another respondent , she felt lucky she has her support system to help with her goals. For example, when she travels overseas, her family supports her in taking care of her daughter. She and her husband also agreed to adjust their free time for their daughter. Ford , also agreed her family supports 'The Scoop'; unfortunately, she also admits she sometimes has no time for her families and friends. She said, 'To be honest, I'm not good at balancing things. I work a lot, and I often don't have a lot of time to spend at home, but I have realized that in the past year, that is not a healthy thing to do just because everybody needs time to switch off and everybody needs time to connect with their family and with their friends. So, I think finding that balance is very typical because you always feel torn, especially you know you start your own business because it's like your baby right. You feel like the sense of obligation to it, and I think a lot of women think that way Lastly, agreed that she is very privileged 'Lamin Warisan' is a family initiative, and any events are considered okay. But whenever she needs it, she never hesitates to choose her family first unless given permission otherwise. As for mentors, most respondents agreed that mentoring is essential; even so, how they seek mentors for their organizations is extraordinary. Most of the respondents had a specific mentor that they look up to in their leadership. One respondent sees her mentor as someone she can count on, especially when organizing events and past experiences shared by her mentor in which she can learn and grow. Another respondent is inspired by how her mentor can stay optimistic even faced with challenges in their life. Lastly, respondent F agreed that her mentor has helped shape her writing skills and ability that have led to her success until now. Although three respondents endorsed mentoring, they also claimed that they did not seek only one mentor. Another respondent agreed that mentors helped shape her thoughts, but her mentor functions mainly online through podcasts, videos and YouTube. She explained that "There is no harm in having these online virtual mentors. Most of the stuff made me who I am today, and it is convenient because I can listen to it while driving" fortunately, another respondent claimed that she does not have any mentor and uses her personal experience to brush up her skills and get better. She often learned from other peers' mistakes and made sure she does not do the same and made improvements. She agreed that 'I feel that we are a work in progress and everything should be taken as a learning step.' --- Personal strength of successful women leaders During the interview, time management and determination were recognized in relation to each respondent's personal strengths. Two respondents have claimed that they could manage their time well. One respondent has set her long and short terms priorities. Since she is still a student, she manages her task accordingly, such as assignments first and then manages activities for her initiative. Another manages her time according to her daughter's. She added that her training is always done at night; therefore, it does not disturb her time with her daughter. As for respondent A, her strong desire towards achieving local and community growth have made her very focus on achieving her task and deliverables. Respondent E also has that same focus. Although she is working as an eye surgeon, she still manages to give talks in schools and universities because she has a purpose of sharing her personal development and inspiring everyone around her. Lastly, for B, she has both time management and determination as her strengths. She explained that she always manages her time well for her daughter, especially after school and during weekends. She added that she never loses her focus Rahim because she aims to become a professor by the age of 50. Even though she is faced with challenges, her determination is what keeps her going until now. --- Discussions and Conclusions So far, literature that deals with the motivation and challenges experienced by women leaders in Brunei are scanty. According to Goleman , motivation is one trait that all influential leaders must attain. This study found that women leaders in Brunei have different motivations in becoming a leader. Two respondents said that their motivation is to help the youths and community to create critical thinkers. As for others, initially, they were only handling initiatives based on their own interest, such as self-passion for arts, family requests, and personal development, finding other opportunities and developing the media industries. Fortunately, they become motivated when they see a gap to make a change and decided to go with it. They decided that they have a passion for the work itself, seek out new obstacles and opportunities, learn new things, and eventually take great pride in the job they do. Indeed, the persistent energy they show proves that they have the intention to do things better and grow further beyond what they imagined they would be doing. Despite facing various challenges in reaching leadership positions and carrying out leadership duties, the respondents in the study do not seem to relate to the realities of gender stereotypes. Two reasons may be attributed; one, there is no gender expectation and women leaders are treated the same as men and have the same opportunity. Secondly, and more likely, these women do not realize that it is a fact of their gender that creates these challenges for them in the first place. In Brunei, the challenges identified are only in terms of access to resources and financial support. The majority of the respondent found it challenging to find the right resources they need for their organization, especially the support of teams and venue. Many until now have no permanent members because they are still under the supervision of their parents. Some may have a team but find it challenging to obtain mutual understandings of a specific goal. This can be because there is no proper monitoring of performance in the team. However, the team is not entirely to blame, as the leader can also be questioned. Past research has found that some individuals are more egocentric in preferring to work independently rather than to work in a team . In terms of financial support, as stated by Education Delphi , finance is necessary to acquire physical resources, which are very important to carry business functionalism. In the study, many of the women leaders opened up their organizations using their financial sources. Unfortunately, this limits the expansion of their organization, hence business stability. However, the exciting factor that made them successful is by finding other ways to promote their organization. Due to an increase in technology, many use social media to promote their productions and content, thus attract people to invest in their business. The study found that the presence of mentors may lead to successful women leaders in Brunei. The majority of the respondents agreed that their mentors helped reshape and strengthen their skills and ability. In handling certain events, they added that their mentors often shared personal experiences with them. But they also include that they did not have only one mentor but more than one depending on the initiatives they joined. One respondent claimed that she did not have mentors but uses her personal experience to brush up on her skills and get better. She added that everything she did was taking the risk and making sure not to repeat the same mistake. Studies showed that work-life balance is seen as a way to help female leaders can balance their work and home commitments . Hewlett & Luce conducted a study that demonstrates that women leaders have successfully employed strategies to make more time for their work and their family. This study found that all women leaders can achieve a work-life balance because of the support from their family and partner. The majority agreed that their family has support for them in taking care of their children while they are busy with their professional work. Their parents never question them on what they are doing, and they would never debate on what initiative they are doing. Some may also add that their partners support them in compromising time in terms of familial relationships, such as sending their daughter to school and helping with the logistic side when creating events. This finding also substantiates the study by Cheung & Halpern in which most women leaders agreed that the support they receive from their husbands and family makes their life easier. They added they could not achieve their goals without them. Lastly, the study has identified the fact that that leader obtained certain personality traits or particular skills that make them a leader. In this case, the women leaders have the expertise to manage their time and their determination. Although they are busy at work, with some having children, they would always make it a habit to spend time with their children. Evidently, with compromising time of their family and spouse, these women find it easier to make time for their children. As for their determination, three respondents stayed focused on their aims. Although they are faced with challenges, they always found ways to overcome them. This paper examines the experiences of Bruneian women becoming leaders in their respective organizations. It studies the motivations they had and barriers they experienced, and how they managed their responsibility at home and work. This study showed that women leaders in Brunei Darussalam are often faced with challenges in terms of business stability, especially when funds were --- Rahim mainly from their own and in finding the right team for their initiatives. Fortunately, the motivation these women experienced is what keeps them going. The study found that each women leader had different motivations, but when they see the opportunity to make a change, they are willing to do it. Lastly, these women became leaders because of the personality traits they own.

This article examines the experiences of women leaders and their challenges in Brunei Darussalam. Globally, women are underrepresented in leadership positions. Though the Brunei government does not discriminate against any gender in delivering free education, medical facilities and healthcare, women in leadership positions are as well underrepresented. I try to understand the factors that challenge as well as motivate women to become leaders in Brunei. For this research, I conducted qualitative interviews with some respondents selected purposively. The study results show that women leaders in Brunei face challenges in obtaining financial support and accessing resources. Despite the challenges, there are some women who have been assuming and excelling in their leadership roles. The study identified that personal strengths have also been critical in contributing to their successful leadership.

I. INTRODUCTION A GENT-BASED Modelling [1] has become a mainstream technique for research in Social Sciences. Traditionally, this research requires the gathering of data over potentially long periods of time and about large populations that are not fully controlled, which largely increases its costs. The formal description of social systems with models [2] allows applying analysis techniques based on simulation and verification, reducing the needs of eliciting data for the initial testing of hypotheses. ABM facilitates modelling by providing social and intentional computational abstractions that are closer to the concepts used in this research field than those of other approaches. Despite of its advantages, ABM is limited by its inherent development complexity. The design, implementation and use of an archetypical agent-based model involve different subtasks and roles, which need diverse backgrounds and competences [3]. This situation may lead to misunderstandings between the different stakeholders, which make it hard to guarantee that the model really corresponds to the initial requirements of social science researchers [4]. Available methodologies to develop such ABM models [3], [5], [6] offer little help to address this problem. They focus on the researchers' conceptual models and describe very general tasks for the development. Agent-Oriented Software Engineering also seems not to be a suitable solution, since it mostly deals with the development of standard Multi-Agent Systems [7]. On the contrary, ABM focuses more on the translation of the conceptual models of social science researchers to computational models using agent abstractions [5]. Besides, MAS are usually tied to certain architectural patterns, while ABM deals with an enormous het-erogeneity of structures [1]. There are also differences about implementation. ABM applications usually require centralized monitoring components that can access the internals of agents in order to gather analysis data [8], and sacrifice the individual agent complexity in favour of huge populations [4]. These aspects are often not considered by AOSE methodologies [7]. Model-Driven Engineering [9] with Domain-Specific Languages [10] has been proposed as a way to overcome some of these limitations [11], [12]. A DSL for ABM uses a vocabulary grounded on conceptual frameworks from Social Sciences. The DSL has a formal definition which enables the use of automated MDE techniques to process its models, for instance to generate the code for simulations. Such approach reduces the impact of misunderstandings in ABM in two ways. First, social science researchers are able to perform the modelling themselves using a language they are familiar with. Second, transformations from conceptual models to computational ones can be generalized and reused in different projects. This offers improved opportunities for verification and validation. There are some preliminary examples of this trend, but they are still too biased to their foundations in software development [12] or only present partial solutions for some modelling aspects [11]. Especially issues like the management of large populations or centralized supervision are still open. This paper introduces a DSL for modelling social constructs and organizational interactions. It is part of the ATCAS framework. The Activity Theory [13] is a well-known paradigm for the analysis of human groups. It focuses on the study of activities, which are interactive acts between people and their physical and socio-cultural environment, where both of them act on and mould the other at the same time. ATCAS is intended to provide a general and extensible basis for ABM in social research supporting different applications. The actual representation of AT concepts in ATCAS depends on agent concepts, for instance about inconsistency management [14], code generation [12] and organizational interactions [15]. This last aspect is the focus of the current paper. From the AT perspective [16], the social aspect of groups regards communities of subjects engaged in shared activities. The norms of the division of labour organize these activities and rules emerging from the socio-cultural environment influ-ence and constraint both communities and activities. AT works at the level of abstraction of human societies but the automated analysis of models in MDE requires formalizing them as computational abstractions. ATCAS-IL adopts for this purpose the OperA [15] framework for the specification and analysis of agent organizations. OperA provides predefined modelling primitives for the high-level specification of organizations and the Logic for Contract Representation , which is an extension of deontic temporal logic, for the fine-grained details. The choice of OperA is motivated by the fact that AT explicitly considers social features that are not embedded in individual agents, but apply to the society as a whole. ATCAS-IL extends OperA primitives with AT concepts and introduces a macro mechanism intended to tailor the language for the specific needs of social science researchers. This gradually increases their autonomy in modelling, as they can define their own abstractions. The formal description of ATCAS-IL uses metamodels. This is a common technique for language definition in MDE [9] that facilitates language extension and evolution. MDE manipulates models mainly using standard transformation languages. In this way, engineers do not need to devote effort to the low-level details of model processing, but just to define the transformations for the different purposes. In the case of ATCAS-IL, transformations consider the semantics of the DSL according to AT and agent research. These transformations generate code, for instance, for checking contradictions as [14] and simulation as [12]. This kind of approach has already been successfully tested in other domains [17]. The remainder of the paper further explains the elements in this introduction. Section II presents AT and the case study that guides the discussion in this paper. Section III provides a brief introduction to OperA. ATCAS-IL is introduced in section IV, and section V applies it to the analysis of the case study. Section VI compares the presented approach with existing ABM works. Finally, section VII discusses some conclusions about ATCAS-IL. --- II. ACTIVITY THEORY The Activity Theory [13] is a socio-psychological paradigm for the study of human behaviour. It focuses on the mutual dialectics between people and their physical and social environment: the environment shapes human actions and their execution, and is also changed by these same actions. Hence, human acts cannot be analyzed independently of their context. These contextualized acts constitute the minimal meaningful unit of analysis and are called activities. An activity [18] is a transformation process driven by people's needs. These needs are satisfied with an outcome produced transforming an object. Any element used in this process is a tool. The active component that carries out the activity is the subject. Subjects with a set of common social meanings constitute a community [16], which represents the socio-historical context of the activity. Two bodies of social constructions mediate the relationships of communities in the activity: rules with the subject and the division of labour with the object. Both of them contain similar elements, such as knowledge, implicit assumptions or norms. The key difference is the focus. The division of labour regards task specialization in the community through aspects such as power relationships, goal decomposition or the assignment of responsibilities. On the contrary, rules are guides and constraints not targeted specifically to the activity but affecting it, such as group beliefs, country laws or accepted scientific theories. The different elements can be both physical and mental, so AT considers both types of activities with a unifying analysis. All these elements make up the context of an activity, which is named its activity system. Its traditional depiction [16] appears in Fig. 1. Activity systems always exist in neighbourhoods of interconnected activity systems linked by shared elements. The execution of an activity produces outcomes that become the artefacts needed to execute other activities. Subjects carry out activities in these networks following their own rationality. AT also considers the hierarchical decomposition of activities. Activities pursue high-level objectives that meet people needs. These activities are executed through sequences of actions, which try to achieve low-level goals. These goals do not satisfy by themselves any need, but they contribute or are part of higher-level goals. In their turn, actions are implemented through operations that depend on the specific state of the environment. The evolution of activity systems over time depends on their inner contradictions. These contradictions are conflicts between the elements in the networks of activity systems, and they can appear both inter and intra systems. Subjects try to remove contradictions through the evolution of the involved activity systems, commonly generating new tensions that produce further evolution. As an example of AT analysis, this paper considers the work in [19] about the Challenger crash. In 1986, the Challenger shuttle exploded shortly after launching. This accident has been frequently used as a case study about engineer ethics, communication and group thinking. According to Holt and Fig. 2. Shuttle flight readiness activity system in the 1980s adapted from [19]. Morris [19], NASA began in the early 1960s as a tightly coupled set of subsystems aimed at space research and development. Over the 1980s, it became the home of several competing groups. At the same time, the lowering perceived value of the space research in public opinion put the agency under increasing pressure to cut down costs. NASA could not satisfy all its opposing goals, so a critical situation made it inevitable to violate some of them. The analysis focuses on the readiness activity where the decision is taken to launch the shuttle or not. It discusses how the inner contradictions present in the NASA's organization culture at the end of the 1980s led to the catastrophe, how these contradictions generated the new and culturally more advanced activities in today's organization, and the still existing contradictions in the current forms of activities. Though the case is inevitably simplified, it synthesizes several complex settings and lengthy sources of information. For instance, the report of the Presidential Commission about the accident comprehends 5 volumes of documentation and analysis 1 . Fig. 2 depicts the readiness activity system. The nodes of the triangle show the opposition between the two perspectives about the NASA. The initial NASA from the time of the Cold War had priorities focused on research and safety, and almost unbounded resources. The new cost-aware agency competes with other agencies all over the world for the launching of satellites. The roots of this conflict appear in the object, which is the launching of a shuttle. It opposes the safe shuttle need to the timely and cost-efficient shuttle need. This opposition permeates all the other elements in the activity system. For the subject, it opposes the safety and the cost consciousness of the decision maker. The tools include information, devices and procedures supporting the activity. In this case, they are focused on the safety procedures, confronting a strict adherence to them with a convenient reading, which is only concerned about critical errors and 1 http://history.nasa.gov/Shuttlebib/ch7.html relies on the low probability of certain failures. The social context of the activity shows this same duality. The community includes the NASA defence-dependent with high funding of the Cold War, and the one of decreasing funding in the 1980s aimed at getting a self-sustaining shuttle program. The division of labour regards work organization. In this case, it defines the acceptable tradeoffs between prioritizing security and the flight readiness team. These norms emerge from a wider context of rules where NASA was pushed to have successful and cost-effective timely flights more than to guarantee safety first beyond any reasonable risk. These intra-node tensions scale up to contradictions between nodes. When the decision-maker needed to decide about launching, he/she was confronted with the dual nature of its community, tool and rules. Whatever decision this engineer made, it could lead to failure in some of the objectives. This situation is known as a double-bind contradiction [20]. In it, the system trying to satisfy opposed goals is under growing pressure that it can only overcome through its evolution. This paper uses this case study to illustrate the application of ATCAS-IL to model complex organizations and subsequently analyze particular features of them. Specifically, it shows how its formal specification method supports the automated identification and analysis of the inherent contradictions of the NASA organization. --- III. THE OPERA FRAMEWORK AT considers that the integrated analysis of the individual and social contexts of activities is key for their understanding. Thus, its formalization as a DSL needs to include and define precisely the behavioural aspects common to the artefacts present in both scopes. For this reason, we have chosen OperA as the basis for ATCAS-IL. OperA [15] is a framework for the specification of agent organizations. It considers the requirements regarding the structure, norms and external behaviour of these organizations and their members. Being focused on specification, OperA makes no assumption about the internals of the agents implementing the organization, and represents interactions through landmark patterns. These provide abstract representations of families of protocols. As such, OperA specifications enable different actual instantiations. Three models specify its organizations. The Organizational Model describes the organization requirements using as main concepts, roles and scenes. The definition of a role includes its objectives and their decomposition, the norms applicable to that role, and the rights or capabilities it has. Scenes describe interactions between roles. They specify the participant roles, the ordering of actions through landmarks, the norms governing the scenes, and the results of their execution. This description makes use of information about the domain and the general normative structure applicable in the system. The Social Model specifies the activity of agents in the society. Agents are executable entities able to enact the OM roles. OperA establishes constraints on the behaviour of agents using social contracts, but not their implementation. These contracts indicate under which conditions an agent will play a role, and are used, for instance, to describe the benefits that the agent obtains, additional constraints or potential plans. Finally, the Interaction Model describes how agents enacting roles participate in the scenes. It specifies the actual execution of the scene through a sequence of actions and adds additional norms for it. OperA uses the Logic for Contract Representation for the specification of contracts and norms in agent organizations. As stated in the introduction, it is an extension of deontic temporal logic that includes the stit operator , the temporal operators of the branching temporal logics, and several deontic operators . Its expressions can be linked to deadlines corresponding to events or observed situations. The OperA framework includes a graphical tool, OperettA [21], an IDE built as a plug-in of Eclipse. OperettA has facilities for the visual specification of organizations, syntax checking based on the OperA metamodel, static analysis, and normative verification . The formal logical semantics of OperA form a basis for the formal verification of the new specification language ATCAS-IL, as the paper shows in the next section. --- IV. ATCAS-IL ATCAS-IL extends the underlying OperA framework in several aspects determined by three main requirements. The first one is the foundation of ATCAS-IL in AT. It makes activities the focus of the analysis, which takes in turn the environment becoming a first class citizen of the specifications. The second one is the aim of ATCAS-IL for social analysis through automated tools. This requires specifying not only the constraints for the behaviour of the organization, but also some aspects of the actual behaviour of its members. The last requirement is the need of providing support to deal with the complexity of the description of social systems. Since these specifications account for a wide variety of issues and individuals, the language has to provide mechanisms to work with them at different levels of abstraction. An activity is an act contextualized by its activity system [18]. The activity system includes subjects using tools to transform objects into outcomes, in the context of communities that specify the division of labour and the general rules applicable to that activity. While subjects, communities, and rules and division of labour can be roughly approximated by roles/agents, groups of these, and norms respectively, there are no suitable OperA concepts to describe activities, objects and tools. Objects and tools are elements of the socio-physical environment. In ATCAS-IL, an artefact represents a general element of the environment. Following a widespread perspective in MAS [7], ATCAS-IL characterizes artefacts in terms of an internal state, the operations available to manipulate them, the events they can generate, and the knowledge and norms about their use. An operation is a basic act that can be executed when certain constraints are satisfied, and that generates some results. Both the artefact state, and the constraints and results of operations are defined with OperA formulae. Environments group artefacts adding specific norms. Algorithm 1 shows an example of the specification of an artefact. It introduces as knowledge the different states in which an artefact can be: non-available , available and blocked . Several specific rules limit the behaviour of this basic artefact: rule-0 states that this is an abstract artefact and therefore no instance of it can be created; rule-1 indicates that the artefact must be created before it can be locked for exclusive use ; rule-2 states that an artefact cannot be blocked at the same time by different activities. All the artefacts in the case study satisfy this basic specification. Communities control the access of subjects to environments. A community is a group of subjects that share social meanings and artefacts. ATCAS-IL defines them as an extension of the concept of role group in OperA. It comprehends several sets of elements: roles representing the subjects; the environments it has access to; accessibility rules mediating between subjects and artefacts; and knowledge and norms applicable in it. Roles are standard OperA roles whose capabilities are represented as operations. The accessibility rules indicate that when the states of the role and its environment meet some conditions, the role can use the operations of the artefact and be aware of its events. The previous elements are put together in activities, whose definition following AT [16] includes the following elements: • Subjects indicate who are in charge of executing the 1456 PROCEEDINGS OF THE FEDCSIS. GDA ŃSK, 2016 activity. OperA roles describe them. Objects and tools represent the elements of the environment affected by the execution of the activity. Artefacts represent them. • Communities determine how the subjects can access the artefacts related with the activity system. • Rules and division of labour establish the norms that provide constraints for the execution of the activity. • Outcomes of the activity are modelled as OperA formulae that become true after the activity execution. • Patterns establish intermediate steps in the execution of activities. As seen in Section II, activities are hierarchically decomposed in actions, which are further decomposed into operations. ATCAS-IL only considers operations in its refinement of activities. Patterns allow specifying the preferred order of generation of the results of operations. The previous elements define general patterns of behaviour, and appear at the level of OperA OM. The elements that actually implement them are the agents for the subjects and the instances for the artefacts. SM define these elements. An agent is characterized by its objectives, capabilities , applicable norms and priorities about objectives. Objectives are states of the world that agents pursue defined as OperA formulae. Priorities determine the objectives that agents prefer trying to achieve when their states and that of the accessible environments meet certain conditions. These priorities provide the means to further describe the dynamic behaviour of systems. OM indicate existing objectives and their satisfaction conditions, and the priorities establish a simple way to choose the objective to attempt when several alternatives are available. The instances of artefacts are described in terms of their capabilities, and the knowledge and rules about them. Finally, IM define for each activity the specific agents and instances that act in it, and sequences of operations to execute it. These sequences of operations must be able to satisfy the patterns included in the definition of the activity. The division between the conceptual definition of the model and its actual realization facilitates two goals in ATCAS-IL. First, it isolates the researchers' hypotheses, which appear at the conceptual level of OM, from the variability of individual elements, which SM and IM define. This allows managing the heterogeneity of individuals in populations, which is relevant to check that the obtained results are really a consequence of the hypotheses and not of specific configurations of the population. Second, it enables the independent evolution of the definition and implementation of subjects and artefacts. This evolution is required to adapt the realization of agent-based models to the requirements of different applications, as the already mentioned centralized supervision and different levels of complexity in the implementation of the individual agents depending on the population size. ATCAS-IL also incorporates mechanisms to manage the models complexity. ABM [5] is increasingly being accepted as a tool for social research due to the inadequacy of analytical and traditional computational models to deal with large, heterogeneous and non-linear systems. Nevertheless, ABM models can also become quite complex, in cases that: involve several types of agents; where individuals evolve in different ways; or, large bodies of knowledge and rules affect the society behaviour. The management of such complexity requires that approaches incorporate abstraction mechanisms. Besides the decomposition of specifications in several models present in OperA, ATCAS-IL includes an extension mechanism for modelling primitives and the capability of defining macros. The extension mechanism allows indicating that a given sub-concept extends a super-concept by including all its attributes. This mechanism is available for artefacts, environments, roles, communities, activities, activity systems, agents and instances. It reduces the size of the specifications and allows building conceptual hierarchies of concepts. Such hierarchies highlight the common features of concepts, allow their incremental description, and facilitate the definition of exceptions. For instance, they allow indicating that subjects usually comply with the rules of their communities, but a minority of them are going to break the law, that is, ignoring some of those constraints. The second mechanism is the definition of macros. Deontic logic, used as formal basis of OperA, does not belong to the standard background of social science researchers, so they need the support of experts to use it. In order to improve the researchers' autonomy when using ATCAS-IL, their experiences are gradually crystallized in a set of tailored macros suitable for a domain. That is, researchers initially determine what they want to specify and experts in logics help them to describe these operators with basic logic primitives. After some studies, this early joint effort defines a researcherfriendly specialization of the DSL for that domain. That specialization includes macros for the most commonly used operators and concepts. Then, researchers can specify models on their own, and only need experts in logics to describe new and unusual properties. V. CASE STUDY: NASA SHUTTLE This section uses ATCAS-IL to analyze the conflicts existing in the flight readiness activity system described in Section II. The AT description of this activity system has been presented in Section II and is depicted in Fig. 2. As stated there, the conflicts emerge from the NASA duality between an agency focused on research and safety, and the modern one trying to reduce costs. It affects the shuttle launching opposing a total guarantee of safety and a reduced one but with a timely launching. The first step of this analysis is the formalization of the activity system using the primitives presented in Section IV. Algorithm 2 shows a simple mapping for it. The flight-readinessactivity extends the basic-activity, which includes basic information about activities. Its object has been simplified to focus on the contradictions coming from the community. The two communities reflect the duality of goals in NASA between the in the activity system. Algorithm 3 introduced the decisionmaker as the subject of the flight readiness activity system. As also seen in Fig. 2, it is specialized in the safety-conscious and the cost-conscious decision makers. The definitions of these roles can be seen in Algorithm 5. They have available several operations, such as check-readiness in the decisionmaker. All of them produce pieces of information according to the available state of the shuttle. The operation decide-launch takes this information to approve or deny the launching of the shuttle, which satisfies the objective take-decision. Algorithms 3, 4, and 5 describe the general behaviour of the activity system. Then, the SM specifies the actual individuals implementing it, and the IM how they carry out the activity. In this case, the SM just considers one instance per artefact and one agent for the decision maker playing the two possible roles, the safety-conscious and the cost-conscious decision maker. This models the kind of situation that faced NASA engineers about launching the shuttle in the presence of non-optimal conditions. The final element of the specification is the IM in Algorithm 6. It uses the previous instances to implement the activity system. The protocol establishes the sequences of steps that the roles can perform to execute the activity. The operations have been already discussed in this section. The verification of the previous specification is the second step of the analysis with ATCAS-IL. It finds out the doublebind contradiction pointed out in [19]. When the engineer needs to abort the launching because it does not meet the safety-first condition, there is a violation of the timely-flight condition as no launching time is scheduled before the planned time. If the engineer decides launching anyway, there is a violation of the safety-first condition. This inability to take an action without violating constraints corresponds to the doublebind contradiction [20]. Thus, the verification has been able to automatically find the contradiction and to point out the conflicting properties. Details on the analysis process can be found in [17]. The presentation of the case study has used a textual specification with ATCAS-IL. OperettA supports visual modelling as long as the corresponding Eclipse models with the abstract and concrete syntaxes of the language are available. This step is required to complete a DSL suitable for social science researchers. They can easily grasp the primitives related with AT concepts, as they correspond to a structured textual representation of activity systems. However, the deontic logic used to specify the low-level details do not belong to their standard background. The use of macros and a graphical notation can reduce their difficulties to use it in their models. --- VI. RELATED WORK The field of ABM shares with general agent research many of its limitations. The review of Gilbert and Terna [5] about the implementation of ABM points out the lack of conceptually well-defined blocks for modelling and implementation guidelines. ABM applies models to a large extent as a conceptual tool with only a swallow agreement about what an agent is [3]. Just to discuss some examples, the research in [11], [22], [6], [12] can be considered. The agents in [6] are modelled as sets of simple variables and rules to modify them. There is a set of external parameters not modifiable by agents, and a set of internal parameters that agents can modify with actions. Actions are rules triggered by certain conditions. There are also information links between agents that indicate when changes in their variables are propagated to other agents. The model can consider some noise in the communications to allow non-modelled environmental effects. Works as [22] have a more complex representation of the involved agents. Their condition-action rules include symbolic representations of elements such as goals, tests or capabilities, being therefore closer to traditional MAS. However, they do not represent key elements in MAS like the society, norms or decision making. Some researchers [11], [12] advocate the use of common MAS for ABM. This approach allows for the richer and more abstract modelling of individuals, but overlooks several relevant facts. The first is that MAS abstractions come from software development, and therefore they are not well-suited for social researchers. The second is that ABM usually deals with huge populations that cannot be implemented with computationally expensive agents. ATCAS-IL aligns with research that promotes MAS for ABM, as it allows making modelling as complex as required. However, its language foundation is in AT, and thus in Social Sciences instead of AOSE. Moreover, ATCAS-IL relies on automated transformations of models to manage the implementation and its tradeoffs. This is an approach already pointed out in [12], though it proposes the use of programming languages instead of model transformation languages [9] as ATCAS-IL does. This last approach is intended to be closer to the concepts of the domain. The issue of the lack of implementation guidelines was already mentioned in the introduction. ABM methodologies focus on capturing the information of the social systems at a very abstract level, but disregard how to migrate from these conceptual models to computational ones [3]. From the already mentioned works, [11], [22], [6] only consider the general features of agents in their models and the results of their simulations. The absence of general implementation details suggests that they rely on specific implementations for the problem at hand. This constitutes a major problem to validate the models, as it is not easy to know whether some of the results come from unintended features of the coding [4]. Approaches emerging from MDE and promoting the automated transformation of models [11], [12] facilitate to some extent this validation, as the same transformations are applied to different models. Nevertheless, the full validation of models in these example approaches still requires a complete understanding of the program code involved in the transformation. For this reason, model transformation languages [9], which work at the level of models, seem a more suitable approach for this task. Nevertheless, complete development processes for ABM that guide researcher in the modelling process are still an open issue. --- VII. CONCLUSION This paper has introduced ATCAS-IL as part of the AT-CAS framework for ABM. ATCAS-IL is a DSL to represent social constructions. It is based on two main sources: the socio-psychological AT to define its conceptual framework; the OperA framework for agent organizations for its formal definition. This foundation pursues two objectives. First, it tries to increase the autonomy of social researchers in ABM by directly applying their own concepts. This reduces the misunderstandings that inevitably appear when researchers need to rely on engineers without a background in Social Sciences for the modelling. Second, the formal definition of the language enables the automated processing of its models. ATCAS-IL proposes making these transformations through standard model transformation languages. This implies that the transfer of information between models is partly specified at the level of abstraction of models, and not with code. It is also expected to improve comparability between models: the application of the same transformations to different models of the same hypotheses should generate equivalent results. The paper has illustrated the use of ATCAS-IL with the problem of the identification of the contradictions that led to a well-known failure in the space shuttle program. While the original AT work relies on the human analysis of data, a suitable model allows the automated discovery of the problem and its potential reasons. ATCAS-IL has currently three main limitations. First, the domain-specific primitives are not enough to model a complete system. As the case study illustrates, low level-details have to be expressed with logics, which are not suitable for social researchers. Ongoing work is intended to determine the additional primitives required in the language according to AT and agent research. AT also considers recurrent social patterns [16], [14] that can be described with reusable macros. Second, textual specifications are too verbose given the amount of details required in the models. The extensions of the DSL can help to solve this issue. Besides, the use of the visual modelling capabilities of OperettA can simplify the development of the specifications, hiding the repetitive details of modelling. Third, facilitating the development of the automated model transformations is still an open issue, as social researchers are not expected to be experts in transformation languages. Approaches based on the automated generation of transformations from model prototypes are a potential solution. --- Algorithm 2 ATCAS-IL activity system for Fig. 2. Activity basic-activity , Communities : public-funded-agency, autonomous-agency, Rules : safety-first, timely-flight, DivisionOfLabour : priority-to-safety-procedures, flight-team-readiness, Outcomes : shuttle-flight, Patterns : ) Algorithm 3 Communities of the flight readiness activity system. Community and safety and funding and income and < ), Role-dependencies :, Artefact-accessibility : ) Community basic-community Community basic-community , Rules : , Role-dependencies : , Artefact-accessibility : ) public-funded-agency and the autonomous-agency. The knowledge related with the artefacts in this activity is described as part of their definitions. Algorithm 3 shows a partial definition of the communities involved in the previous activity system. The basic-community contains the role decision-maker, who is the subject of the activity, and a rule that states that the agency must run balancing its expenses and the incoming money . As there are Interaction-contract , Environment : , , , , Clauses :, Protocol : Check-readiness, Check-safety, Check-costs, Decide-launch ) no constraints about artefact accessibility, all the roles have granted full access to all the artefacts of the community. Depending on the type of agency, the specification adds more rules to its description. For instance, the autonomousagency should run without public funding, which is asserted as knowledge of that community. Note that the specification of the autonomous-agency indicates that it constrains the general decision-maker to a cost-conscious-decision-maker, so it does not add a new subject but specializes the existing one. The definition of the rules is illustrated with safety-first and timely-flight in Algorithm 4, which are also examples of the use of macros introduced by keyword DEF. The rule safetyfirst states that the probability of failure cannot be over a given limit, and timely-flight indicates that launchings must adhere to the expected planning. The last type of elements to define is the roles involved

Agent-based modelling is becoming widely used for studies in Social Sciences. However, its application faces limitations coming from its bias to software development, which precludes a more active involvement of social researchers. In order to deal with this problem, this work proposes using domainspecific modelling languages based on the socio-psychological Activity Theory. These languages apply agent research to crystallize that theoretical framework in a formal definition suitable for automated processing but close to Social Sciences. The paper focuses on the language for the specification of group constructions such as organizations, norms and shared knowledge. A case study about contradictory decisions in the space shuttle program illustrates the discussion.

INTRODUCTION Information exchange is pivotal during the disaster management processes and specially response phase. Due to the dynamic and complex nature of the extreme natural or man-made disasters as the rate of communication increases vastly [1,2] and often personal and technical facilities are unable to operate properly as expected and detailed in prepared standard operating procedures and emergency response plans which very often led to the communications failure, which has been claimed as one of the main problems during disaster response management [1]. The enhancement of telecommunication networks and the widespread diffusion of mobile devices and new Web technologies are helping to recover this communication challenge. Recent studies find telecommunication services such as Short Message Services or social media sites enhances the consistent and timely transmission of valuable information throughout the disaster management life cycle which help to establish a disaster resilient community [1][2][3][4]. Disasters are complex in nature, which can have disproportionate effects at varying speed. Social media sites can be used to enhancing a community capacity and preparing for disasters [5]. For example, in a disaster, emergency management authorities may access social networks and blogs to find out the source and severity of the disaster and create situational awareness appropriate to the affected communities [4]. Authorities may monitor online communities to identify emerging trends and potential hotspots which can be flashpoints for disaster. While past studies have provided some mechanisms and recommendations for the use of social media to manage disasters [5][6][7], it is still unclear how social media have been used to support the safety of the community [5]. Identification of various ways of using social media is important, as it helps in providing high level guidance on disaster management to federal, state, territory and local governments and it is first step towards longer term goals of delivering sustained behavior and enduring partnership with the communities [5]. Therefore, this study is an attempt to answer 'in what ways do social media applications contribute to natural disaster management?' To answer this question a systematic review of the social media literature needs to be conducted which would help to synthesize best quality scientific studies on the role of social media in disaster management. Systematic Literature review acts as a methodology of the synthesis. In contrast to expert view, SLR would not just aggregate all the existing evidence on the above research question it would also support the development of evidence-based guidelines for emergency authorities. --- II. METHODOLOGY The objective of this study is to conduct a systematic review of the literature on the characterization of social media response to natural disasters. This study has customized the guidelines for systematic reviews laid by Kitchenham and applied in several reviews [6] and conducted the review in four steps (as shown in Figure 1 --- A. Identification of Resources In order to identify the resources for our data collection, first we need to come up with the relevant phrases or keywords. Therefore, this step begins with the identification of relevant keywords. A trial search on Google scholar was performed by using the terms 'social media' and 'natural disaster' as these terms are the key facets of our research question. A brief overview of the first 200 search results indicated that the terms 'emergency', 'hazard' or 'crisis' can be used in combination with the term 'social media' to identify appropriate resources as shown below: 'Social media' and 'natural disaster' 'Social media' and 'emergency' 'Social media' and 'natural hazard' 'Social media' and 'crisis' Once the keywords were finalised, four databases, IEEE Explore, Science Direct, Springer Link and Google Scholar were searched for studies which resulted in altogether 4840 articles. Distribution of these 4840 articles upon four databases is shown in Table 1. For the initial screening, only titles, abstracts and keywords were considered and the search was contained to the studies published between 2000 and 2014 as from the turn of the century, the development of social media websites and their users increased exponentially and studies into the usefulness of social media particularly in natural disaster started late 2000. --- B. Selection of Studies This step of SLR demands filtering of papers from the initial list of extracted articles from the databases and excludes papers based on the criteria. This process is carried out in three iterations as presented in Figure 1. The first iteration involved searching for the keywords over the four databases and excluding articles based on title, abstracts, keywords and full texts. The articles addressing the following exclusion criteria were not considered: The second iteration has two sub-iterations: search on references of the articles found in iteration 1; and search on articles, which have cited the articles found in iteration 1. These sub-iterations were performed by using Google Scholar and excluded paper based on the criteria provided above. These iterations found 19 new articles candidate of literature review. At the completion of the second iteration, 35 articles were achieved.  The third iteration is a repeat of the second iteration on the articles extracted in the second iteration. This is basically a second level search on citations and references of the articles. This search achieved five more articles and the total was 40 articles. These findings confirm the shortage of studies into the use of social media to manage natural disasters. --- C. Data Extraction and Synthesis In this step, the key details of each of 40 selected papers were retrieved. It includes the context of the study , natural disaster type and information , objectives and results of the study in addition to the demographics of the paper . Studies having similar purposes were grouped to show the streams of research as shown in Table 2. --- III. CRITICAL ASPECTS OF SOCIAL MEDIA USE FOR NATURAL DISASTER MANAGEMENT This section discusses and summarizes the findings from the studies reviewed in Table 2 into six characteristics: --- A. Data Collection Methods The rise of social media and other forms of user-generated contents have created the demand for real-time search on the provided online information [9]. It requires methods and tools which can effectively extract data via APIs and then analysing these data to extract information of interest [3,10]. Social media research has introduced various methods for the effective collection of disaster related posts such as Bloom Filter Chains for real-time tweet search [9], TAKMI technology for content analysis [8] and twitter APIs -'crawl' and 'timeline' [11]. Krishnamurthy et al. used these two methods, both relying on API functions provided by Twitter for the collection of large amount of data through crawl and timeline functions of Twitter. The analysis of the two datasets, i.e. 'crawl' and 'timeline', indicate that there is no considerable difference between the two datasets on the characterization of Twitter users, status updates and validation of crawling methodology. However, the shift in the Internet traffic toward program or machine generated data and consumption by processes on behalf of human users have been concluded as a future trend by the study. --- B. Developing Situational Awareness Social media platforms are overwhelmingly useful to create situational awareness among the affected communities of natural disasters. For many social media researchers a situational awareness view is helpful for anticipating how individuals, groups and communities can use information contributed by others on social media platforms [3,[12][13][14]. Vieweg et al. conducted a study in which they extracted microblogging posts of two natural hazards, Red River Floods and Oklahoma Grassfires both occurred in Spring 2009, and identified and measured features that are useful for situational awareness. The net result of this study is that the activities differ with respect to the location of the event and thus initiates discussion on how these findings can serve as a foundation for designing and implementing systems that can be used in future emergency events for better and targeted situational awareness. --- C. Classifying News and Rumours Mendoza et al argue that if Twitter, as a social media platform, is effective in broadcasting valid information, it is also effective in spreading the baseless rumours which can contribute to the general chaos in the affected areas [15]. Social media researchers such as [16,17] have also questioned the credibility of the social media feeds during natural disasters. They argue that during a natural disaster when information from official sources is scarce, rumours on social media can surface contributing to the chaos on the ground. Studies such as [18] indicate that the users generally have difficulty discerning truthfulness of tweets based on their contents. To develop a deeper understanding of the difference between valid news and baseless rumours, Mendoza et al collected Twitter data, 4,727,524 tweets, on Chilean earthquake from February 27 to March 2 in 2010 and performed two studies to investigate, how the news and rumours propagate through networks and the ability of the network to discriminate between false rumours and confirmed news. The results postulate that the propagation of tweets that correspond to rumours differs from tweets that spread news in a way that false rumours are questioned much more than the confirmed truths by the users [15]. From this perspective, the Twitter community acts as a collaborative filter of information that does not allow rumours to penetrate deep in the social graph [15]. The studies suggest creating a state-of-the-art text classification mechanism at the social media platform level providing instant warnings to the affected community about the information being questioned by the other users. --- D. Distributed Sensor System Crooks et al. argue that the proliferation of social media feeds and Ambient Geographic Information contents fosters the emergence of microblogging as a new type of a distributed system, with bloggers acting as sensors, and their comments in the form of tweets, convey relevant information often with a corresponding geographic footprint [3]. This study analysed the microblogging posts with geolocation footprints about Mineral, VA earthquake of August 23, 2013. One of the critical observations of this study is that the microblogging posts such as tweets can travel faster than the physical event to distant locations and therefore act as an early warning system for large scale incidents [3]. In addition, harvesting geospatial information from microblogging posts can rapidly provide valuable information on the impact of physical event. In this regard, systems that perform real-time trend detection over the twitter stream are suggested to be adopted [19]. Use of adaptive sliding window application of a lexicon-based content analysis solution by Zao, Zhong [20] is one of the examples that detects events particularly in Twitter within 40 seconds of the unfolding event. Integration of such systems with the analysis conducted by this study demonstrates Twitter as an effective distributed sensor system for event detection and impact assessment [3]. --- E. Digital Volunteerism Cobb et al. recognize the importance of processing real-time and first-hand information from disaster-affectedcommunity for a better decision making and designing strategies [4]. Starbird et al. call individuals and groups involved in this task digital volunteers who predominantly monitor incoming feeds to find new, relevant, and actionable information [21,22]. Digital volunteers often work to treat the social media data into useable resources. Cobbs et al. conducted contextual interview with digital volunteers and emergency managers to provide insights into their collaboration and the use of tools for monitoring disasters. The results of the study indicate that the design implications for supporting this work, including coordinating and integrating activities of distributed volunteers exist F. Sentiment analysis Fothergill et al. argue that minorities differ in their risk perception and in their response to emergency warnings, with some having fatalistic sentiments leading to greater fear and less preparedness [23]. These sentiments need to be understood by the emergency authorities to better tailor emergency warnings, preparedness and responses [24]. Public sentiments towards unfolding natural disasters are measured various ways by the social media platforms. For example, Facebook measures the level of public concern through 'likes' and 'dislikes' [6]. Since Twitter has extensively been used for posting opinions and situations, it has become the focus of sentiment analysis [25]. A variety of methods have been introduced to perform sentiment analysis during natural disasters. For example, some scholars use a naïve Bayes classifier trained on unigram features to place tweets into positive and negative classes [24]. Others use Twitter based sentiment analysis algorithms in aggregate form, in which number of occurrences of positive and negative words in each tweet is counted to determine the tweet's sentiment score [26]. Mandel et al. used a similar approach to conduct sentiment analysis on 65,000 tweets referencing Hurricane Irene in 2011. This study used an existing sentiment classification technology that categorises messages based on degree of apprehension, fear or general concern about Hurricane Irene. The results of this study indicate that the sentiments vary based on persons' gender or location however accounts for classifier errors in hypothesized testing and adjusting classification proportions using quantitative methods have been considered future challenges [24,27]. --- IV. DISCUSSION AND CONCLUSION This research began with the arguments that social media is being quickly adopted as it offers ways to retrieve, produce and spread information [12]. The nature of that sharing has been recognized as a life cycle of information production and consumption which is rapid and repetitive [22]. Since social media is a platform for emergency communication it has been considered a place to harvest information about natural disasters to determine what is happening on the ground [28]. Therefore, social media has gained momentous attention both from researchers and practitioners. On the one hand, practitioners are developing skills and knowledge to better utilize social media services, on the other hand, researchers are developing models to examine various aspects of the use of social media in dissimilar settings [13,29]. This study conducted a systematic review of the literature particularly focusing on the use of the social media in natural disasters to determine aspects of interests for the researchers. Initially 4,840 articles were retrieved from four databases. This total was reduced to 40 when article selection and finalization procedures were applied. Evaluation of the 40 finalised articles indicates that the researchers have focused on six aspects of the use of social media based on the analysis of a vast amount of data related to natural disasters around the world. For example, a group of researchers have argued that the selection of an appropriate method is crucial for effective collection and analysis of disaster related data during any stage of the disaster. The majority of the selected articles are in the favour of general perception that social media is an effective platform to create situational awareness among the affected communities before, during and after a disaster. Some of the researchers have looked into the social media data in terms of good and bad information for the communities in the affected areas. They have proposed mechanisms to distinguish rumours from valid news and curb them to avoid further panic among the affected area communities. To some researchers, people from disaster areas, posting location enabled messages on social media platforms, define a distribute sensor system which helps to immediately identify and localize the impact area of the event. Some of the researchers have focused on the importance of collecting and processing real-time first-hand information about an event for the purpose of formulating strategies and decision making. This phenomenon has been recognized as a digital volunteerism in which volunteers use social media to assist in disaster response efforts through filtering and mapping crisis related social media data. Since the social media has been used as a means for emergency communication, to some researchers, it becomes important to understand how people in the affected area feel about the unfolding situation. Sentiment analysis has been focused on by many studies and the results highlight the importance of sentiment of the people of area in relation to devise response strategies by the authorities. All the studies into the six aspects of the use of social media have highlighted challenges and improvements to the existing research. Since social media has become a popular platform to disseminate and harvest information about events, its importance has particularly been recognized for natural disaster management [14]. A systematic review of the literature on the role of social media in natural disasters characterises social media into six aspects: data collection methods, create situational awareness, classify news and rumours, distributed sensor system, digital volunteerism and sentiment analysis. These six aspects have been considered important in the studies related to natural disasters in various parts of the world.

Social media sites are playing a significant role in rapid propagation of information when disasters occur. This effective communication platform is a great useful tool for emergency (disaster) management agencies during all phases of disaster management life cycle: prevention (mitigation), preparedness, response, and recovery. This study has conducted a systematic review of social media use in disaster management literature to identify how social media sites have been used during these four critical phases of disaster management life cycle in order to recommend strategies for government officials. a systematic method has been used to search four major academic databases for this review. The search resulted in 40 articles and categorized the findings in six main themes: situational awareness, data collection methods, distributed sensor systems, news and rumors, sentiment analysis, and digital volunteerism.

Resumo Considerando a gravidez e a maternidade como períodos de maior vulnerabilidade em mulheres migrantes, a fim de caracterizar os cuidados de saúde prestados, procurou-se identificar e compreender padrões de satisfação e procura de cuidados de saúde materna e infantil, avaliando as suas percepções sobre a qualidade deles. O estudo seguiu uma metodologia qualitativa para a coleta e análise de dados e foi realizado no Porto, a segunda maior cidade de Portugal. As participantes foram 25 mães recém-imigradas do Leste Europeu, Brasil, e países africanos de língua portuguesa e seis Portuguesas , contatadas pelas associações e instituições sociais. Os dados sugerem que a saúde depende não só da acessibilidade, mas especialmente das oportunidades sociais. Ações equitativas de saúde pública devem proporcionar aos indivíduos e grupos oportunidades iguais para satisfazer as suas necessidades, que podem não ser alcançadas fornecendo o mesmo tratamento padrão para todos. --- Equidade em Saúde; Desigualdades em Saúde; Bem-Estar Materno; Vulnerabilidade Social; Migração --- Introduction Equity in healthcare has been demonstrated as a relevant factor in reducing disparities in health 1 . Equity in health depends not just on overlapping social factors that affect each other 2,3 . It relates to ethics and moral values of social justice, implying preparedness from public health policies and actions to address and bridge potential inequities created by the conditions that adversely affect the health of individuals and communities 4 . The scientific literature indicates how often migrants tend to be the most vulnerable population groups, especially women of childbearing age, that regularly show greater foetal and neonatal mortality, and more complications during pregnancy and postpartum 5,6,7,8 . Their health indicators are frequently worse, which is explained by interactions between socio-material deprivation factors aggravated by stressors involved in the migration process. Along with issues related to health literacy concerning pregnancy and its assistance, migrants often obtain poorer medical follow-ups 7,8,9,10,11 . Additionally, migrants' expectations about healthcare received during pregnancy commonly differ and conflict with the normative procedures of the host country 12,13,14 . In a moment of current intense reconfiguration of the European societal structure, due to the global economic crisis, Portugal is officially no longer a host country for immigrants 15 and has once again become a country that principally exports human capital. This transition is contributing to an increase in the pre-existing vulnerabilities of populations that were already underprivileged, namely the immigrants who stayed. To contextualize our study, we provide basic information on the structure of the National Health System in Portugal, in order to allow a better comprehension of the potential challenges that migrants may face when encountering healthcare services. The system is based on the concept of accessible care for all that need it. With a large number of local primary healthcare centres run by general practitioners, it establishes a mandatory first contact at this level, except in acute health conditions. Due to organizational limitations, not all individuals have an individually appointed general practitioner, and may therefore be seen by other similarly qualified doctors available at the time. Specialised appointments take place in public hospitals upon referral by general practitioners. Prenatal care in low-risk pregnancies is conducted in primary healthcare centres, while there are national guidelines on high-risk features necessary for referral to specialized obstetric care 1 . It is worth noticing that Portuguese legislation contemplates free healthcare access during pregnancy, irrespective of the mother's documentation status. Our goal was to verify whether there are differences regarding women's perceptions about quality and appropriateness of care received between immigrant and native women, during pregnancy and postpartum. We believe that the users' perspective still offers a sensitive quality indicator of received care 13,16,17,18 , and helps to identify possible gaps emerging from budgetary restructuring in public health policies. Our study was conducted in Porto, the second largest city of Portugal. --- Methods This article relies on qualitative data. We believe that health research is much more robust if exploring the personal and subjective aspects of the user's perspective, essential for a global understanding of the social reality, which have real impacts in health status, care seeking, regular monitoring and treatment adherence. Given the characteristics of the target population, the option for semi-structured interviews as the preferred technique for data collection, proved to be the most suitable methodology, since it only required a single meeting and guaranteed absolute anonymity for each participant. This aspect is relevant since the status of some of the mothers interviewed was as undocumented immigrants. Other strategies would imply not only longer meetings, but also the possibility of more than one interview session. Although other strategies would provide extremely valid and richer information, they are not always applicable and adjustable either to the time constraints of working migrants or to concerns associated with multiple contacts with illegal migrants 19,20 . We designed a specific protocol study in order to understand the role of being from a foreign country: in perceiving quality and adequateness of care. We also evaluated possible differences in medical attention and health outcomes between nationalities. --- Participants: sampling and recruitment A total of 31 participants were recruited, all of low social-economic status, receiving support from civilian associations working with migrants and Portuguese individuals and/or from governmental institutions with the same aim. Approval for conducting the study was obtained from the Governing Boards and from the Ethics Committee of all institutions involved. The participants' recruitment process was initiated through contacts and meetings with the Governing Boards of the institutions and associations, after their approval of the study protocol. The interviews took place in the contacted institutions' locations, after study approval. The participants were recent mothers born in Portuguese-speaking African countries , in Eastern European countries , in Brazil , and six were Portuguese natives , notified about the study and asked to participate in it. Participants were recruited between November 2011 and February 2012 if they met the following inclusion criteria: recent mothers preferably during the first year of a child's life, living in the metropolitan area of Porto. It is usually in the first months of life that children require more attention, from a psychosocial point of view, because of the need to assess the achievement of a series of developmental stages 5,21,22 . This period is also critical for the mother's emotional health and her adaptation to motherhood 23,24 . --- Procedure Semi-structured interviews with similar guidelines were conducted on all women , evaluating the perceived needs and cultural challenges that potentially influence the perceptions of the migrant population 13 , and that determine health demands, treatment adherence, effectiveness of health advice and inequalities in health 20 . Interviews took place in the association or institution where the women were recruited and its duration was about 25 minutes. They were conducted by a well-trained researcher, graduated in psychology. The methodology and general objectives of the study were explained to all participants. Authorization for audio recording of the interviews was requested, and informed voluntary consent was registered. No refusals were disclosed. --- Analysis of data Qualitative content analysis and categorization of emerged information were performed to make a systematic analysis of collected data, which involved transcription of interviews and field notes 25 . Subsequently, a comprehensive in-terpretation of the resulting information was performed. Initial categories were created and later evolved with the analysis of new data . Some initial questions presented at the guidelines included: "Have you detected any issue in accessing family planning services at health centres?"; "How would you evaluate the healthcare you received during last pregnancy surveillance? What about the healthcare provided to your new-born?"; or "Did you notice any problem in attendance and / or in the connection established with health professionals? Please explain in what way, and what were the consequences ?". To maintain confidentiality, socio-demographic data were entered into a coding sheet, and the name of the participant replaced by an alphanumeric code. Quotes were chosen from women's dialogues to best exemplify the main emerging themes. --- Results --- Socio-demographic characteristics of participants Participants were between 20 and 45 years old. African women presented basic to medium levels of education , and almost half were still in the process of legalization. Most women from Eastern European countries completed higher education, and had regular documentation status. Most Brazilian women reported medium level education and legalized documentation. Portuguese women also had basic to medium level of education. Length of stay in the country varied between two and 20 years: five African women were residing in Portugal for more than 12 years, five Eastern European women had lived in Portugal for at least 10 years, and five Brazilian women had been in the country for 7 years or less. It is noteworthy that 12 participants were unemployed at the time of the interview. The collected information emerged in three major themes: maternal and child healthcare ; strategies for managing difficulties; quality and consequences of care by health professionals. --- Maternal and child healthcare Women were requested to express their experiences concerning medical appointments received during pregnancy, as well as monitoring services throughout the postpartum period and subsequent follow-up of the baby. --- • Pregnancy and postpartum We observed that pregnancy surveillance is one of the clinical activities that was met with greater satisfaction, particularly among Portuguese women. Nevertheless, this satisfaction was rarely found for medical attention at postpartum, where the women reported a significant lack of social and affective support, reporting sentiments of abandonment by the NHS. Despite generalized satisfaction, most immigrants report to have had severe reproductive disorders in previous pregnancies, whether they occurred in their country of origin or in Portugal. African women consistently reported more complications during pregnancy, most of them already in Portugal: more infection, hypertensive disorders and gestational diabetes as well as more miscarriage, perinatal and neonatal death in previous pregnancies. "[my baby] had a problem with blood... I don't know why he died because he was born well. He was in the hospital for two months and died. No one ever explained to me" . Brazilian women tend to report high satisfaction with care received during pregnancy. The difficulties they most identified refer to some unfamiliarity from physicians and administrative professionals about pregnant women's free access to healthcare, if undocumented. This has resulted in poor and later prenatal care for some Brazilian women. "In early pregnancy, I went to the doctor and she made my access a bit difficult because she said I had no right to [receive medical care] since I had no social security number" . Women from Eastern European countries claim to be very satisfied with pregnancy consultations especially when they occurred in their hospital of reference, in specialized services. Since most of them are having their second child in the host country, they experienced some progresses over the years concerning medical attention, referring to the humanization of contact with health professionals, and have developed comprehensive skills about the functioning of the NHS. "During the first pregnancy follow-up it was complicated because I had no family doctor. The second was very easy, and I liked [the healthcare services] a lot" . --- • Baby follow-up Regarding medical care for infants, after being discharged from the hospital, several immigrants report difficulties in attaining sufficiently clear medical consultations. Those discourses were not found among African or Portuguese women, unless a family doctor was not allocated. On the other hand, Brazilian and Eastern European women report widespread dissatisfaction with baby follow-up when it was carried out in primary healthcare , since in their countries paediatric services are provided at this level of assistance. This often triggers a higher request for Hospital emergency services. "At the Health Centre there wasn't a paediatrician. So my daughter grew up without consulting a paediatrician! ...only the family doctor could request a consultation, I had no family doctor!" . "The family physician is... well, is good, but not for children!" . --- • Family planning and contraception Family planning services are counselling and information services regarding all aspects of sexual and reproductive life of a couple, since the onset of sexual activity. They focus on health prevention and protection against diseases and women's biophysical preparation for pregnancy . We observed that not all women are aware of the advantages of those services but, irrespective of nationality, the ones who use them are extremely pleased. We only identified one case of an African woman that reported a coercive attitude from the physician that allegedly attempted to impose on her a subcutaneous implant. --- "How can I put one thing if I don't know what it was? She didn't even ask me for my opinion; she didn't ask me if I wanted it! The answer she gave me was 'Oh, it's for you not to become pregnant again'" . Nevertheless, the patterns of receptiveness and use of contraception were highly variable among women of different nationalities. During the interviews, we observed some cultural specificities among Portuguese and migrant women, and we would like to highlight the approach of Eastern European women regarding medicalization overall. These women tend to have a widespread opinion that Western countries use too much drugs, arbitrarily and for all purposes. When concerning contraception, the same posture is detected: they use contraception more rarely, are mostly autonomous in their selection and often prefer condoms or skin adhesives. It is noteworthy that these women find contraception a highly private subject, demonstrating a significant sense of embarrassment and decorousness around it, and not all women felt comfortable discussing it. "For us it is not easy, we initially are very ashamed! In our country we do not talk, it's more intimate. I know people that took them home [birth control pills, condoms] and... just showed them, they didn't use them, you know?... bags of contraceptives as an exhibit in a museum! [laughs]" . --- Strategies for managing difficulties With regard to potential barriers identified by the study participants, we intended to gather information about the level of proactivity and action oriented towards migrant and Portuguese women's strategies to overcome difficulties. Among Brazilian women, only one woman mentioned that she looked for information about her rights when facing a setback in accessing maternal healthcare. We found Portuguese women at the same level of action: despite knowledge about the functioning of the NHS, Portuguese women of low socioeconomic status only carry out verbal complaints. In very serious situations or circumstances with a high probability of medical failure, they tend not to pursue official channels. "When removing the placenta, doctors didn't completely eliminate the remains... And that made me develop an infection that forced me to stay hospitalized for a few days. I wasn't pleased ...but I didn't complain" . Eastern European women, as the most highly educated, were more prompt to be informed and to request medical evidence about their health, frequently claiming to have a shared decision about therapeutic resolutions. Those women report having major initial difficulties, not only being uninformed about the functioning of the NHS, but mostly with regard to key language barriers that they actively try to attenuate with the length of stay. We also observed that some African women are likely to reveal language barriers when considering healthcare use and full comprehension about medical appointments that were not self-identified previously. "...our Portuguese is different from here, and she [the administrative worker] spoke in a way that I didn't understand..." . --- Quality and consequences of care by health professionals As evidenced above, Brazilian and African women assess very positively the healthcare provision that they received. However they tend not to identify potential sub-quality experiences, not relating them to subsequent adverse consequences. Some experiences emerged that compromise the quality of services received, screening obstacles posed by some health professionals . "At the Health Centre I was told that I could not do my pregnancy follow-up there, because I changed my address. I was already three months pregnant! ...in the new Health Centre, I wasn't accepted... We had to go to the hospital" . We observed a more assertive position among Eastern European women when confronted about self-perceptions concerning quality and consequences of care. Several women pointed to the language barrier as a crucial personal gap. Regarding health professionals, these women report that many physicians demonstrate thoughtfulness and tolerance, but some confine themselves to a position of expertise, not looking to actively communicate with the user. This often leads to extremely harmful situations for these women, especially in a period of initial stay in the host country. "My family doctor, she didn't ask me for a few tests, and I had a blood infection during pregnancy, which was not detected, I had a vaginal delivery and infected the baby... She was born and died nine hours later" . Eastern European women also argue that doctors frequently seem unprepared to answer well established questions by informed patients, not appearing to be comfortable in providing technical information or discussing clinical procedures with them. --- Discussion We can anticipate that equity in healthcare depends not only on accessibility but especially on social opportunities. Social risk is widely associated with socio-material deprivation and tends to be reflected in social exclusion to goods and services, including health and education 1,2,26 . Equitable public health action must provide individuals and groups the equal opportunity to meet their needs, which may not be achieved by providing the same standard care to all. Scientific literature and medical history have been showing that the non-mastery of the dominant language determines less adequate treatments and may result in increased risk of health complications 4,26 . Analysing our results, inequities in maternal health concerning immigrants were particularly observed in pregnancy outcomes, dramatically aggravated in previous pregnancies, both for the mother and the baby. We will further explore some examples. Considering African women who consistently reported more complications during pregnancy, two women in our sample have lost a child in the first three months of life and one woman reported having a previous stillbirth at seven months of pregnancy in Cape Verde, evidencing an unquestionable higher risk profile. Nonetheless, they are often satisfied with medical care, demonstrating a largely passive attitude in the approach to healthcare professionals. This inaction tends to result in a perpetuation of unawareness and unaccountability for their health, unless it is sensitively counteracted by the healthcare professional. Women from Eastern European countries demonstrated a more proactive attitude, claiming for themselves a substantial role in therapeutic decisions. However, personal life trajectories and individual experiences often put them in situations of poor pregnancy outcomes and at higher risk of medical lapses. One Russian woman in our sample reported a medical fault severe enough to cause an early neonatal death; one Ukrainian woman expressed that she never felt comfortable with medical follow-ups here because doctors were unaware that she was radioactively contaminated by living near Chernobyl. They never asked her for her medical history, and she has always experienced her pregnancies with anxiety. In the absence of a degree of intimacy between doctor and patient, she opted not to attend prenatal care. She carried out her first ultrasound one month before the baby was born. The failure to identify certain subliminal barriers to equity in clinical practice may result in negative health consequences. They suggest gaps not only in cultural competence by some health professionals but, ultimately, professional limitations in establishing communication and a proper understanding in approaching health behaviours and expectations in immigrants. In fact, health professionals must be alert, not letting themselves be deceived by apparent satisfactory health status . Applying equal healthcare standards may constitute blind clinical tactics in the absence of comprehensive communication between doctor and patient. Patients' perspectives are essential elements to assess quality of maternal and child care and must be taken into account by policy makers and health professionals 17 . We believe that good medical care needs to be an arrangement of clinical quality combined with proper communication, beyond mere access to services. The clinical relation between doctor and patient is the key factor to a successful therapeutic alliance, tackling background inequalities, encouraging compliance and additional differentiated care towards better therapeutic results. Cad. Saúde Pública, Rio de Janeiro, 30:333-340, fev, 2014 --- Resumen Considerando el embarazo y maternidad como los períodos de mayor vulnerabilidad para las mujeres inmigrantes, y con el fin de caracterizar la asistencia sanitaria, se buscó identificar y comprender patrones de satisfacción y demanda de salud materno-infantil, así como la evaluación de las percepciones sobre la calidad de los mismos. El estudio siguió una metodología cualitativa para la recogida y análisis de datos y se llevó a cabo en Porto, la segunda ciudad más grande de Portugal. Las participantes fueron 25 madres, inmigrantes recientes de países de la Europa del Este, Brasil, países africanos de habla portuguesa y seis nativas portuguesas , contactadas a través de asociaciones e instituciones sociales. Los datos sugieren que la asistencia sanitaria no sólo depende de la accesibilidad, sino especialmente de las oportunidades sociales. Las acciones equitativas de salud pública deben proporcionar igualdad de oportunidades a las personas y grupos para satisfacer sus necesidades, que no se pueden lograr al ofrecer una misma atención estándar a todos. Equidad en Salud; Desigualdades en la Salud; Bienestar Materno; Vulnerabilidad Social; Migración Contributors L. M. Almeida participated in the collection, analysis and interpretation of the data, and in the preparation and write-up of the article. J. P. Caldas collaborated in the design, in carrying out the study and in writing the article. D. Ayres-de-Campos and S. Dias contributed towards the preparation, revision and write-up of the final version of the text.

Considering pregnancy and motherhood as periods of increased vulnerability in migrant women, to characterize the healthcare provided to this collective, we sought to identify and understand patterns of satisfaction and demand of maternal and child healthcare, assessing women's perceptions about its quality. The study followed a qualitative methodology (semi-structured interviews) for collecting and analysing data (content analysis) and was conducted in Porto, the second largest city of Portugal. Participants were 25 recent immigrant mothers from Eastern European countries, Brazil, Portuguese-speaking African countries and six native Portuguese recent mothers (for comparison), contacted through social associations and institutions. Data suggests that healthcare depends not only on accessibility but especially on social opportunities. Equitable public health action must provide individuals and groups the equal opportunity to meet their needs, which may not be achieved by providing the same standard if care to all.

Introduction Over one third of youth provide previous or current caregiving support for tasks such as transportation, emotional support, monitoring safety, and dementia care, to an adult relative independently or with another relative and experience lasting impacts on physical and mental health outcomes, school performance, and employment. [1][2][3][4][5][6][7][8][9] However, the vast majority of clinical and/or community interventions focus solely on adult rather than young caregivers. [10][11][12][13][14][15] State and national policies supporting young caregivers are even more sparse and often neglect their distinct circumstances. 16 For instance, youth working while attending school may not benefit from the Family and Medical Leave Act since they are likely working too few hours to meet eligibility criteria. [17][18][19] Young caregivers are increasingly stepping into caregiving roles to accommodate gaps in needed supportive care to adult relatives as more adults are shouldering dual employment and caregiving responsibilities. 20,21 But rather than having specific resources or policies for young caregivers, there is a reliance on resources for adult caregivers presumed relevant for young caregivers. 22 Here, we present findings on the different types of support that youth identify and express as being important for preparing care for an adult relative to facilitate development of youth-informed caregiving policies and resources. 18,23 --- Methods Participants were recruited online to MyVoice, a national open-ended text message poll of youth ages 14 to 24 years, through social media advertising to meet national benchmarks based on weight samples of the American Community Survey. Participants are sent weekly surveys on relevant health topics via text-messaging given high mobile phone access in this age-group . 24,25 Here, we report findings from our use of the MyVoice poll to study caregiving specifically. Demographic information was collected at enrollment. 26 This study was approved by the University of Michigan Institutional Review Board and was fielded in August 2020. Open-ended questions were developed through an iterative process with survey, mixed methods, and caregiving experts, and youth. The study included 5 questions on caregiving experiences and perspectives and was introduced with a prompt: This week is about family caregiving . 27 This report focuses on the question: What would help you prepare to take care of an adult relative in the future? We conducted a content analysis to analyze respondents' descriptions, which were submitted in a narrative format. We first read through all responses and developed a preliminary codebook of identified resources. Then, 2 investigators refined codes, independently coded all responses using Microsoft Excel, and resolved discrepancies through discussions. 28 We assessed the frequency with which respondents identified each type of support among the full sample, then among respondents with caregiving experience and by age-group; summarized demographic characteristics ; and conducted x 2 tests to evaluate the relationship between age-group and identified supports. Caregiving experience was measured by asking whether respondents had ever taken care of an adult relative and was not explicitly defined in the survey, to solicit a range of caregiving experiences. 1 --- Results Of 1076 youth who received the question, we analyzed 905 complete responses . Nearly half self-identified as female and 51.6% were age 14 to 18 years. More than half of respondents had at least a high school degree . Respondents described several different types of resources to prepare them for caregiving responsibilities. These included: education ; financial stability ; workplace support ; social/professional support , time and space , advance care planning , and mental health support . Table 2 provides illustrative quotes of the different types of support. The majority of respondents identifying financial support were female and in the 19 to 24 age-group . Respondents age 19 to 24 years were significantly more likely than respondents age 14 to 18 to identify workplace policies, X 2 , =23.7, P < .001 and social/professional support, X 2 , =10.6, P = .001. Respondents ages 14 to 18 were significantly more likely to identify education than those ages 19 to 24, X 2 , =5.2, P < .05 . Among 349 respondents with complete responses about current or prior caregiving experience who also discussed types of support , education was the most commonly reported type of support followed by financial support . Workplace support was the most common type of support indicated by experienced caregivers in the 19 to 24 age-group followed by social/ professional support and financial support . Among experienced caregivers in the 14 to 18 age-group, the most common response was that nothing can prepare one for caregiving followed by mental health support and education . --- Discussion In this study, youth identified several types of support to prepare them for caregiving: education, financial support, workplace policies, social/professional support, mental health support, advance care planning, and time and space. Approximately 7% of respondents described a lack of clarity around what can help them prepare to be a caregiver; some also acknowledged-or resigned-that nothing can prepare one for caregiving. Respondents evenly endorsed the types of resources identified regardless of age, though respondents identifying workplace policies as a resource were predominantly age 19 to 24, perhaps reflective of their experiences in the transition from education to work environments, while respondents in the 14 to 18 age-group were focused on education and mental health. Moving forward on types of support identified by respondents requires coordinated efforts between policy makers, clinicians, and educators. Lack of clarity around what can help prepare for caregiving was mostly due to a lack of exposure or perceived obstacles to caregiving, and may represent a need for conversations from health, educational, and social support systems that support youth for potential caregiving responsibilities. Future research should continue soliciting insights from youth to ensure that our policy investments are meaningful for youth now and as they transition to adulthood, and are responsive to the needs they identify from their caregiving experiences alongside their developmental and educational circumstances. Coordination between health, education systems at multiple levels, and local government or community organizations could facilitate identifying caregivers and the development and dissemination of support programs, educational guidebooks, videos, and workshops to help youth learn to perform various medical tasks and to learn broadly about caregiving responsibilities and self-care. These materials could be developed by caregivers, care receivers, researchers, caregiving organizations , and health care systems as part of their community outreach programs. These materials could be disseminated in school health classes, advising, or counseling centers, and doctor offices. Policies enabling youth to participate in health care visits could also be critical to their obtaining important information about caregiving, making medical decisions, and managing their mental and emotional health. While involving youth in medical information and decision making could be beyond their developmental readiness, engaging them incrementally could ease the burden associated with the steep learning curve and intense pressure of making sudden medical decisions for a relative without preparation. 29 Meanwhile, some opportunities for minimizing this burden include screening for caregiving and increasing access to remote mental health support services and support groups. 30,31 Screening, for example, could be implemented in the near future through conversations and in the longer-term could be integrated into patient questionnaires or the patient portal. Efforts to foster early engagement in their own health care, particularly during care transitions, could be critical to youth wellbeing. 32 Policy makers should update FMLA and other workplace policies and financial supports to include the circumstances of young caregivers including in state FMLA expansions that offer paid leave. 19 Data from adults 18 and older suggest that FMLA minimum hours and tenure requirements disproportionately exclude workers who are women, Black, Indigenous, and multiracial. 16,33,34 In fact, the policy also excludes consideration of young caregivers who may be financially responsible for their adult relatives through jobs with lower hours alongside educational responsibilities. These caregivers could benefit from the job security offered by FMLA so they can return after their caregiving responsibilities. Investing in the wellbeing of young caregivers early on could be critical to preventing health disparities and inequities stemming from socio-structural barriers, later in life. 35 Although respondents age 19 to 24 were more likely to describe a need for workplace support, extending policies to support those who are younger could be beneficial, particularly given recent state level initiatives to draw on youth to address the labor shortage. 36 Future research should solicit more information about respondents' caregiving experiences, to examine relationships between the types of support identified and the nature of respondents' caregiving experiences. Our study is limited in inferences about caregiving support needs among racial/ethnic minority youth. For instance, youth identifying with communities where mental health disorders are stigmatized may not have identified mental health support. 37 These perspectives should be examined further to ensure the development of policies and other resources that are responsive to the needs of diverse youth. 23 Given the increasing role of family caregivers in supporting the needs of adults who may be aging or affected by chronic or acute health conditions including COVID-19, policies and programs should develop or extend existing policies to specifically meet the needs of youth who may be current or future caregivers. Supporting youth early on is an investment toward promoting caregivers' health later in life. Using youthinformed approaches is critical to developing relevant and effective policies and resources in the evolving area of caregiving. The authors thank the MyVoice team for their assistance with developing questions and data collection. We are also grateful to Eve Rubovits for her assistance with developing figures. To see this article online, please go to: http://jabfm.org/content/ 35/4/814.full.

Background: Youth are increasingly upholding significant caregiving responsibilities. These caregiving responsibilities can have emotional, educational, and professional impacts on youth and young adults. And yet, policies and resources focus on adult caregivers and are limited in supporting young caregivers. The purpose of this study was to describe the different types of support that youth identify as being important to prepare to take care of an adult relative. Methods: We conducted an open-ended, text-message based poll of youth ages 14 to 24 in August 2020. We conducted a content analysis to categorize and describe the different types of support respondents identified in their responses. We compared types of support identified by age-group, gender identity, and prior caregiving experience. Results: Most respondents (42.2%) identified education (eg, skills training) as being an important resource. Other types of support reported included financial support (eg, assistive programs), workplace policies (eg, paid leave), mental health support, and professional support. Discussion: Policy makers should extend existing policies (eg, Family and Medical Leave Act) to include and consider the circumstances of youth and young adults. Policies enabling young caregivers to actively participate in their adult relative's health care visits could be critical to preparing youth for the skills required and the physical and emotional demands associated with caregiving. Coordinated efforts between health and education systems could support youth in learning information about caregiving, medical decision making, and medical tasks. ( J Am Board Fam Med 2022;35:814-820.

nightmares, appetite changes, and separation anxiety . Separation from the parents , age of the child, and parental anxiety have been implicated as affecting the child's reaction after hospital discharge. However, most of this research was conducted prior to 1980 when maternal employment and, thus, experience with separation from the mother and alternate child care at an early age was less common. According to Bowlby ), children look to their primary attachment figures, generally their mothers, in times of stress, such as when they are frightened, tired, or ill. The stress of hospitalization for children is likely due to the fear caused by the child's separation from the mother and care by nonparental adults as well as the child's illness. Experience with separation from the mother and care by nonparental adults when the child is not ill may reduce the novelty and fear caused by hospitalization. Thus, experience with alternate child care could act as a buffer for the negative effects of hospitalization. Conversely, although research suggests that daily mother-child separations due to the mother's employment have no negative effects for children of any age , it also is possible that separation for employment is a low-level stressor that, when added to the major stressor of hospitalization, pushes the child's stress to an unmanageable level. In that case, return to alternate child care after hospitalization might prolong or exacerbate any negative effects of hospitalization. With 50% of single mothers with preschoolers in the workforce and the expected increase in this rate due to welfare reform efforts, more hospitalized children from single-parent families will have experience with nonmaternal care. However, the effects of experience with alternate child care on behavioral outcomes for hospitalized children are unknown. --- Relevant Literature Research on children's reactions after hospitalization clearly indicates that hospitalization is a stressful event for children and that negative effects can continue for months or years after discharge. In the seminal study by Prugh, Staub, Sands, Kirschbaum, and Lenihan , a sizable proportion of preschool and school-age children who experienced "routine" ward conditions had "significant" negative reactions immediately after discharge and at 3 months . Most subsequent studies have found posthospitalization upset at 1 to 4 weeks after discharge . However, two other studies have found normal behavior in children after hospitalization . Age of the child frequently has been related to posthospitalization upset, with younger children having more severe negative reactions than older children . Wolfer andVisintainer found no age effects in their first experiment, but saw age-by-treatment effects in a later experiment. Tiedeman and Clatworthy found that older children experienced a decrease in anxiety through 7 to 14 days after hospital discharge, but 5-to 8-year-old children did not. Several investigators have found that posthospitalization behavior of preschool or school-age children was not related to the number of previous hospital admissions . However, Elander, Nilsson, and Lindberg found that 4-year-old children with a history of two or more hospitalizations displayed more somatic symptoms; oral tendencies; and dependent, aggressive, and hyperactive behaviors than children with only one hospitalization. Findings regarding the effect of length of hospital stay also vary. Although Menke and Simons et al. found no relationship between length of stay and child behavior, Tiedeman and Clatworthy found greater anxiety in 5-to 11year-old children with shorter lengths of stay. Two studies found significant differences in behavior and coping strategies based on experience with separation. Elander et al. found an interaction effect of daycare attendance and hospitalization on behavior of 4-year-old Swedish children. Daycare children who had been hospitalized sometime since birth displayed more aggressive, hyperactive/ impulsive, dependent, and regressive behaviors than daycare children who had not been hospitalized. Children had these reactions even though 116 of the 231 admissions had lengths of stay of 3 or fewer days; 35 of these were overnight stays. Children experience separation when they start school, around the age of 5 or 6 years. In a sample of 3-to 7-year-olds , Corbo-Richert found that older children and those with school experience used more information-seeking coping strategies to deal with chest tube removal than younger children and those without school experience; previous experience with hospitalization did not influence choice of coping strategies. However, it is not clear whether these results are due to age of the child or to the effects of separation from having entered elementary school. In summary, the negative behavioral effects of hospitalization for children are well-known. However, whether experience with alternate child care and, therefore, routine separation of the child from the mother, has a moderating effect on the relationship between hospitalization and child behavior is not known. A sample of preterm and full-term preschoolers living in femaleheaded single-parent families from a larger, longitudinal study on the effects of maternal employment for preschoolers in single-and two-parent families was used to examine this question for several reasons. Single mothers are twice as likely to have a preterm birth than married mothers . Research suggests that preterm children are more negatively affected by environmental stressors . Preterm infants and children are more likely to experience hospitalization than full-terms. The research question for this secondary analysis was: Does experience with separation from the mother due to alternate child care moderate the negative effects of hospitalization on preschool children's behavior? Thus, the purpose of this study was to examine the interaction effects of alternate child care experience on the relationship between hospitalization and behavior of preschool children living in female-headed single-parent families. --- Method Sample The sample for this study consisted of 121 female-headed, single-parent families, with 60 preterm and 61 full-term preschoolers . Families with preterm preschoolers were identified from the admission records of three Level III neonatal intensive care units ; families with full-term preschoolers were identified from birth records of two normal newborn nurseries. A systematic random sample of families with full-terms and all families with preterms from four consecutive birth years were sent a letter that briefly described the study. Interviews were conducted within a month of the child's third, fourth, or fifth birthday. Single-parent families were eligible to participate if the mother had not lived with a man in the father role for at least 6 months prior to recruitment. Five index children were being raised by a single woman other than the birth mother . In each case, the child had lived with the family for most of the child's life and the woman was performing the role of mother to the child. Thus, these women are referred to here as "mothers." Inclusion criteria for the preterm child in the family were: born prior to 36 weeks gestation weighing less than 2,500 grams; appropriate weight for gestational age; and hospitalized for at least 1 week in a Level III NICU. Inclusion criteria for the full-term child in the family were: birth between 38 and 42 weeks gestation; discharged home with the mother after birth; and no preterm siblings that were born within 10 years of the study child's birth. Children whose mothers were told the child had more than a 2-year developmental delay were excluded. None of the families contacted were excluded on this basis. Of the families contacted and eligible to participate in the larger study, 71% agreed to participate. Sample demographics are listed in Table 1. The sample had approximately equal numbers of male and female children. Birthweight and gestational age of the preterm and full-term groups reflected study inclusion criteria. In the preterm group, there were 25 low birth weight , 23 very LBW , and 12 extremely LBW children. Average length of stay in the NICU at birth was 46.1 days . Few had experienced complications of prematurity. Thirteen had had an intraventricular hemorrhage . Two children had been diagnosed with cerebral palsy and two with bronchopulmonary dysplasia . At the time of the study, one child with BPD was using an aerosol machine for asthma, and the other child with BPD did not require any special equipment. Most of the mothers were African American , never married , high school graduates , and living in poverty . Number of children in the family ranged from 1 to 8, including the study child . --- Instruments Child behavior-The Child Behavior Checklist was used to measure child behavior. It contains two behavior problems scales . The internalizing scale contains three subscales: anxious/depressed; somatic complaints; and withdrawn. The externalizing scale contains two subscales: aggressive and delinquent behavior. Mothers rated their child on each of 118 behaviors on a 3-point scale ranging from not true to often true. Scale and subscale scores were obtained by summing the mothers' ratings. Normed scores, based on child age and gender, then were obtained with the Revised Child Behavior Profile sheet. The CBCL has been widely used with a variety of populations, including LBW children and African American children, with documented evidence for its construct validity . Achenbach reports moderate to strong correlations between the CBCL's internalizing and externalizing scales and similar scales of the Quay-Peterson . Achenbach reports test-retest reliabilities of .82 to .95 and internal consistency reliability estimates of .90 and .93 for the internalizing and externalizing scales, respectively. For this sample, the alpha coefficients were .83 and .91 for the internalizing and externalizing scales, respectively, and ranged from .58 to .89 for the subscales. Experience with hospitalization and alternate child care-Mothers indicated when the index child was in the hospital and the lengths of stay for each hospitalization on a Life History Calendar . From that data, number of hospitalizations, total number of days hospitalized, and age at last hospitalization were derived. Additionally, mothers used the LHC to indicate the child's alternate child care experiences, including timing, duration, and types of alternate child care providers. Alternate child care experiences were defined as situations where routine child care was provided by adults other than the child's parent, primarily for the mother's employment. Episodic baby-sitting to allow the parent to go out for a few hours were not considered alternate child care experiences. From these data, number of different child care arrangements and proportion of the child's life in child care were derived. In addition to these interval-level variables, two dichotomous variables also were created: experience with hospitalization and experience with alternate child care . Validity and reliability of the LHC is enhanced through its use of memory cues, relating one event to other events that occurred at about the same time. When Freedman et al. compared data obtained in 1980 about the respondent's current situation with data obtained retrospectively with the LHC in 1985 , agreement ranged from 72% to 92%. The LHC constructed for the current study contained five segments , each with 12 blocks . Major life events, such as residential moves, births, and deaths also were recorded to aid the mother's memory. Recording began with the month and year of the study child's birth. --- Procedure The study was approved by the appropriate human subjects review committees at the university and each of the three hospitals. Interviewers were trained to administer a set of parent interview and self-complete instruments and developmental tests for the larger study. Within 2 weeks after an introductory letter was mailed, interviewers contacted each family to screen for inclusion criteria, answer questions, and schedule a home visit for data collection. Interviewers offered to read self-complete instruments to parents; parents and children were encouraged to take short breaks as necessary. --- Results More children experienced alternate child care than hospitalization . Children with alternate child care experience had spent an average of 28.5 months in up to four different arrangements . Types of child care arrangements included care in the home by relatives and nonrelatives , family daycare , and an organized daycare facility . Number of hospital admissions ranged from 1 to 18 , days in hospital ranged from 1 to 355 , and time since last admission ranged from 3.4 to 59.7 months for those who had been hospitalized. Children who had experienced only hospitalization did not differ from children who had both hospital and child care experiences on number of admissions or number of days hospitalized, t = .79 and t = 1.19, respectively, p = NS. However, children with only hospital experience had been hospitalized more recently than children with both hospital and alternate child care experience , t = 2.25, p =.03. Gestational status was related to child care-by-hospital experience group , chi square = 15.27, p < .01, primarily because more preterm than full-term preschoolers had been hospitalized . The numbers of preterm and full-term preschoolers who had experienced alternate child care were similar . Number of different child care arrangements and months spent in alternate child care were similar for preterms and full-terms, t = .88 and .92, respectively, p = NS. However, preterm preschoolers experienced a greater number of hospitalizations and spent more days in the hospital than full-terms, t = 2.63 and t = 2.27, respectively, p < .05. Separate variance estimates of the t values were used for these two comparisons due to unequal group variances. Time since last hospitalization was similar for preterm and full-term preschoolers, t = 1.50, p = NS. Two-way ANOVA 1 was used to examine the interaction effects of experience with hospitalization and child care on the subscales of the CBCL internalizing and externalizing behavior problems scales, controlling for gestational status and race. The interaction effect of experience with alternate child care and hospitalization was significant for somatic complaints, F = 8.40, p = .005, and aggressive behaviors, F = 5.70, p = .02. Children who had experienced hospitalization but not alternate child care had higher somatic complaints than other children . Children who had experienced both hospitalization and alternate child care had lower aggressive behavior scores than other children . To further explore the effects of hospitalization, additional analyses were conducted with data collected from the children who had experienced a hospital stay. Correlations between child behavior subscales and aspects of the child's experience with hospitalization and child care were examined . More recent hospitalization and older age at last hospitalization were related to more aggressive behaviors. However, greater proportion of the child's life in alternate child care was associated with fewer aggressive behaviors and somatic complaints. --- Discussion In this study, experience with alternate child care and hospitalization had an interaction effect on behavior problems of preschool children regardless of the child's gestational status at birth. Elander et al. also found an interaction effect of these two factors. Children in our study who had a history of hospitalization had more somatic complaints. Since these children also had been hospitalized more recently than those with both hospital and child care experiences, this finding could reflect the effects of the child's more recent illness or an underlying chronic illness. However, because children with both experiences had scores similar to those who had not been hospitalized at all, this finding may reflect a protective effect of experience with alternate child care. That is, experience with child care and, therefore, experience with routine separations from the mother may decrease the negative effects of hospitalization on the child. Preschool children who had experienced both hospitalization and alternate child care displayed fewer aggressive behaviors than other children in our study. In contrast, Elander et al. found that 4-year-old daycare children with a history of hospitalization displayed more aggressive behaviors than daycare children without a history of hospitalization. The effects for aggressive behaviors and somatic complaints are in the same direction, although the mean for children with hospital experience only is similar to the mean for children without a history of hospitalization regardless of child care status. Thus, something about having both experiences makes the child less aggressive. Although effects of hospitalization on preschool children's behaviors were not related to number of hospitalizations or length of hospital stay, timing of the last hospitalization was important. Children whose last hospitalization was at a younger age and less recent displayed less aggressive behavior. Menke , Simons et al. , and Tiedeman and Clatworthy also found no effect of number of admissions or length of stay, although Elander et al. found more behavioral dysfunction in Swedish children with more hospitalizations. In our study, experience with alternate child care again appears to be protective. For children who had been hospitalized, amount of aggressive behavior and somatic complaints decreased as the proportion of the child's life in alternate child care increased. Thus, greater experience with 1 An alternate method for testing the interaction effects is to use multiple regression. However, the creation of interaction terms with variables that are not continuous for use in regression analysis is somewhat controversial. Violation of regression assumptions would invalidate the results; therefore, we have chosen to use two-way ANOVA because it assumes that the independent variables are categorical. alternate child care may ameliorate some of the negative psychological effects of hospitalization. A limitation of this study is that effects of different types of alternate child care situations as well as the quality of these child care arrangements were not considered. The emphasis of the programming in organized daycare has been found to improve cognitive performance, especially for children from low-income families . Quality of the child care arrangement has been found to affect cognitive development . However, research on the effects of quality of child care on the behavior of children from lowincome families is lacking. In summary, preterm preschoolers were more likely to have been hospitalized than full-term preschoolers; however, they were equally likely to have experienced alternate child care. Experience with alternate child care had a protective effect for preschoolers who also had been hospitalized, regardless of gestational status. For those with hospital experience, number of hospitalizations and duration of stay were not related to child behaviors; however, children showed less aggressive behavior as time since last hospitalization increased. Although more research in this area is needed, we speculate that experience with routine separation from the parent for child care diminishes the negative effect of separation from parents during hospitalization and perhaps other types of separation. Aggressive behaviors for children with and without child care experience.

Background-With more single mothers entering the workforce due to welfare reform efforts, more hospitalized children from single-parent families will have experienced alternate child care arrangements where routine care is provided by adults other than the child's mother. Objectives-To investigate with secondary analysis of data whether experience with alternate child care has a moderating effect on the relationship between hospitalization and behavior of preschool children living in female-headed single-parent families. Method-A sample of 60 preterm and 61 full-term children who were 3,4, or 5 years old was recruited for the larger longitudinal study. Behavior problems were measured with the Child Behavior Checklist. History of hospitalization and alternate child care arrangements were measured with the Life History Calendar. Results-Preschool children who experienced hospitalization without alternate child care experience had more somatic complaints, but those with both hospital and alternate child care experience had fewer aggressive behaviors than other children. For children with a history of hospitalization, aggressive behaviors decreased as the proportion of the child's life in alternate child care increased.with alternate child care may ameliorate some of the negative effects of hospitalization, and potentially other novel and negative experiences, for preschool children. This could be due to child care providing positive experiences with separation from the mother, a peer group with which to talk about the novel experience, or actual instruction about the novel experience.child behavior; child care; hospitalization; single-parent families Hospitalization of a child is a stressful event for both parent and child. In his review of the research, Thompson (1985) concluded that children have negative reactions to hospitalization, including night terrors, feeding disturbances, loss of newly acquired developmental skills, and intolerance of separation from the mother, that persist for months after the child's discharge. Despite efforts to make the hospital more child-and family-focused, preschoolers in a recent study also demonstrated distress 1 week after hospital discharge, including irritability,

INTRODUCTION Breast cancer is the most frequently diagnosed cancer and the second leading cause of cancer mortality among US women, with an estimated 252,710 new cases and 40,610 deaths in 2017 . Notable disparities in breast cancer mortality have been documented among different racial groups, with mortality rates being 42% higher among Black women compared with Whites, regardless of similar incidence. 2 This has been attributed to a combination of factors that include differences in stage at diagnosis, existence of comorbidities and tumor characteristics, as well as access and adherence to high-quality cancer treatment and prevention. [3][4][5] Racial disparities in mortality have been documented for young breast cancer survivors , a clinical group that includes women with a cancer diagnosis at a young age . 6 YBCS constitute a special population within the 3.6 million US breast cancer survivors. 1 These women are at a higher risk of being carriers of germline pathogenic variants associated with hereditary cancer syndromes. 7 Black YBCS experience a higher incidence of triple-negative breast cancer, a subtype with poorer survival, and are twice as likely to die from the disease compared with White YBCS, contributing to racial disparities in breast cancer mortality. 3 Genetic testing for breast cancer susceptibility targets primarily germline pathogenic variants in the breast cancer 1 and breast cancer 2 genes, which confer a three-to sevenfold increase in the lifetime risk of breast and/ or ovarian cancer. 8 Knowledge of genetic status is important, as it can inform clinical management options that include more intensive surveillance, chemoprevention and prophylactic risk-reducing surgery . 9 However, cancer genetic services that provide credible information about these options are not equally utilized by all US women. Several studies have shown that Black women are accessing cancer genetic services disproportionately less than White women, mainly due to lack of awareness about the availability of genetic testing, 10 lack of access to cancer genetics experts, 11 geographic barriers, 12 and other socioeconomic factors, including lack of insurance, lower education, and lower income. [13][14][15] Racial segregation and poverty have been traditionally identified as concordant barriers for accessing health care, leading to poor management of health risks among underserved communities. [16][17][18] Yet, despite existing information, many more questions remain unanswered regarding the availability, accessibility, and acceptability of genetic services in these populations. The United States Preventive Services Task Force recently identified this knowledge gap, and called for actions to improve the understanding of disparities in genetic service utilization among racial and ethnic minorities. 19 The purpose of this study is to address this knowledge gap by examining reasons that explain use of cancer genetic services in YBCS, a population of cancer survivors where there is sparse information. The specific aims of the study were to compare levels of genetic services utilization between Black and White/Other YBCS, identify attitudes and knowledge of breast cancer risk factors, and identify possible barriers that explain lack of genetic counseling and testing uptake, in a random sample of YBCS recruited from a statewide cancer registry. In a previous publication, we examined use of genetic services in Black YBCS. 20 In this paper, we examine novel barriers for observed disparities in genetic services utilization between Black and White/Other YBCS, which have not been previously examined. --- MATERIALS AND METHODS --- Design, sample, and procedures We used baseline data from a randomized trial conducted in the state of Michigan that tested the efficacy of two interventions designed to increase breast cancer surveillance and use of cancer genetic services among YBCS and their high-risk female relatives . Methodological details of the study and the recruitment process have been described in detail elsewhere. 21,22 In brief, a racially stratified random sample of YBCS was selected from the Michigan Cancer Surveillance Program . The MCSP is a population-based cancer registry that has been in operation since 1985, and has been certified by the National Association of Central Cancer Registries as to data quality, completeness, and timeliness since 1999. The initial sample of 3000 potentially eligible YBCS included 1500 Black and 1500 White/Other YBCS, with oversampling of Black YBCS to ensure their adequate representation in the study. 22 Approximately 7% of YBCS of other racial backgrounds were grouped with White YBCS, because they could not form a separate stratum. YBCS were eligible to participate if they were diagnosed with invasive breast cancer or ductal carcinoma in situ between 20 and 45 years of age, Michigan residents at the time of diagnosis, alive at the time of sampling, 25 to 64 years old at the time of the study, able to read English and provide informed consent, and not pregnant, incarcerated, or institutionalized. Prior to contacting the YBCS, the MCSP mailed the reporting facility and the physician of record a letter asking if there was any reason that the YBCS should not be contacted for the study. An invitation letter explaining the study, a consent form, a self-administered baseline survey, and a stamped return envelope were mailed to YBCS. Eligible YBCS received up to three mailed invitations to participate in the study. Collection of baseline data from YBCS took place between September 2012 and January 2013. Institutional review boards of all organizations involved in identifying and recruiting YBCS approved the study protocol. 21 Informed consent was obtained from all participants. --- Measures A survey collected self-reported demographic information and clinical characteristics . An adapted version of the Theory of Planned Behavior guided data collection about knowledge and attitudes . 21 Items assessing knowledge of breast cancer risk factors and knowledge of breast cancer genetics were answered as "True," "False," or "Do not know" to generate a knowledge score. 23 Perceived breast cancer risk was assessed with a single item asking participants to rate their chances of getting cancer on a 10-point Likert scale with verbal anchors . 24 Fear of cancer recurrence was assessed with 4 items, on a 7-point Likert scale ranging from 1 to 7 . 25 All measures were reliable with Cronbach α greater than 0.71. Distance from the nearest board-certified counseling center was considered a potential barrier for accessing genetic services and was calculated using ZIP code information. Minimum distance from YBCS place of residence to the nearest board-certified genetic counseling service was calculated using the great circle distance formula, 26 D = 3963.0*sin + cos *cos*cos], where T i is the latitude and G i is the longitude of the specified locations in radians. Residential and genetic counseling center ZIP codes were geocoded as latitude/longitude coordinates using the R package "zipcode." Board-certified cancer genetic counseling centers, operating at the time of the survey, were identified from the National Society of Genetic Counselors , the American College of Medical Genetics and Genomics , and the National Cancer Institute Cancer Genetics Services Directory. --- Statistical analyses We used descriptive statistics to inspect the data, identify missing values, and assess distributions. Frequencies and percentages described categorical variables. Means, ranges, and standard deviations were used for continuous variables. Bivariate analyses and inferential statistics, such as chi-square tests for differences in proportions, and t tests for differences in means examined relationships in demographic and clinical characteristics, use of genetic services, and knowledge and attitudes among YBCS. Linear regressions with simultaneous variable selection identified predictors of knowledge of breast cancer risk factors and genetics. Similarly, logistic regressions identified factors associated with the use of genetic counseling and testing. Two-tailed p values were calculated with the level of significance set at p < 0.05. A post hoc analysis in G*Power v.3.1 revealed that >90% power could be achieved for all regressions. We used R software for data analyses. --- RESULTS Of the 3000 randomly selected YBCS, 2960 were eligible to participate, and could be contacted. In total, 95 Blacks and 179 White/Others formally declined participation. Moreover, 832 Blacks and 663 White/Others who were eligible to participate and were invited to the study did not respond after three mailing attempts. In total, 883 of contacted YBCS accepted participation; 353 were identified from the cancer registry as Black, and 530 as White/Other. The response rate was 38.6% for White/Other participants and 27.5% for Blacks. Most consented YBCS resided in Michigan at the time of the survey and the remaining resided in 23 other states. The most common reason for no response was lack of a current address , especially for Black YBCS . Other known reasons for nonparticipation were YBCS was deceased , reporting physician advised against contact , YBCS was incarcerated . From the 883 YBCS who consented, 24 were ineligible because they were not diagnosed younger than 45 years old , were pregnant , or returned their survey too late . The final sample consisted of 859 eligible YBCS, of whom 519 were White/ Other and 340 Black. The mean age of YBCS at the time of the baseline survey was 51.0 ± 5.9 years. There was no difference in age between Black and White/Other participants. White/Other YBCS had higher education, were more likely to be employed, had higher income, and were less likely to be in "single" relationship status than Black YBCS . Moreover, Black YBCS were less likely to have a routine source of care , lacked access to care due to high out-of-pocket costs , and lacked health insurance . Black YBCS reported having Medicaid , or Medicare health-care programs more frequently than White/Other YBCS. On the other hand, the latter were more likely to have private/other type of insurance . The distance from the nearest board-certified counseling center was greater for White/Other than Black YBCS , as indicated by the Wilcoxon rank sum test . Cancer genetic counseling centers were operating predominantly in urban areas, where the population density of Black YBCS in the state of Michigan is high. At the time of the study there were 21 cancer genetic centers with board-certified/eligible genetic professionals in the state of Michigan. Of n = 353 participating Black YBCS, approximately 1 of 3 lived in Detroit at the time of the study. Of n = 473 White/Other YBCS who resided in Michigan at the time of the study, the largest residential cluster was in Grand Rapids , and only a few White/Other YBCS lived in Detroit. We also identified more than 200 genetic services for participants living in other states at the time of the survey based on their ZIP codes. Because clinical presentation of breast cancer, personal health history, and family health history influence referral to genetic services, we examined self-reported clinical characteristics between Black and White/Other YBCS. There was no difference in the years since diagnosis, or age at first diagnosis between Black and White/Other YBCS . A larger proportion of White/Other were diagnosed with a DCIS compared with Black YBCS . Invasive breast cancer and ovarian cancer were reported at similar rates. Reported family history of breast cancer was greater in White/Other compared with Black YBCS . There was no difference in reported family history of other types of cancer including ovarian, or male breast cancers. As we presented in our previous publication, a greater proportion of White/Other reported having genetic counseling than Black YBCS, and more White/ Other reported having genetic testing than Blacks. 20 Among 58 YBCS who reported having been identified as germline carriers of pathogenic variants, more White/Other reported a BRCA1 or BRCA2 pathogenic variant than Black YBCS . There was no difference on reported rates of lumpectomy or mastectomy between Black 3). The results of the regression indicated that 11 predictors explained 17.4% of the variance = 16.45, p < 0.001), regarding knowledge of breast cancer risk factors, and 14% = 13.01, p < 0.001) regarding knowledge of breast cancer genetics. Black race was associated with less knowledge of both breast cancer risk factors and breast cancer genetics . YBCS with more family members affected by breast/ovarian cancer were more likely to have higher knowledge of risk factors . Higher education and having genetic counseling were positive predictors of higher knowledge of breast cancer risk factors. They were also positive predictors of higher knowledge of breast cancer genetics . Having genetic counseling and having genetic testing were highly correlated among YBCS . That was true for both Black YBCS and White/ Other . However, a greater proportion of White/Other YBCS who reported having genetic counseling also reported having genetic testing , compared with Black YBCS . Logistic regressions assessed potential reasons that explain use of genetic services among YBCS . Black race , high out-of-pocket costs , older age , and years since diagnosis were negative predictors of genetic counseling uptake. On the other hand, higher education , family history of breast/ovarian cancer , and fear of cancer recurrence were positively associated with genetic counseling. The same predictors were identified for genetic testing uptake. The baseline survey also collected data for YBCS who did not use genetic services. In our previous publication, we reported reasons that Black YBCS did not use genetic services. 20 Here we provide reasons for not using genetic services for the entire sample . Potential answers included an extended list of options, such as provider-related factors , availability of services , accessibly of services , and acceptability of services . The most frequent reason for not using genetic services among all YBCS was lack of physician recommendation . Moreover, a higher proportion of Black YBCS reported no physician recommendation compared with White/Other . --- DISCUSSION Racial disparities in genetic services utilization are an important problem of health equity in the United States. In this study, only 26.2% of Black YBCS reported receiving genetic counseling, and 17.6% reported genetic testing, as opposed to 37% and 32.8% of White/Other YBCS, respectively. These figures are in agreement with what has been reported in the literature. 13,15,28,29 Findings from this study suggest that out-of-pocket costs are negatively associated with use of genetic services. Others reported that low-income women are less apt to receive genetic testing, even after controlling for race, ethnicity, and other sociodemographic factors. 30 Taken together, these findings indicate that even modest cost-sharing requirements can prove burdensome for some YBCS. Additional expenses that usually follow a genetic diagnosis can be prohibitive, and out-of-pocket spending for surveillance needs, chemotherapy, and risk-reducing surgery can act as an additional barrier for accessing genetic services. 28,31 Racial segregation and poverty have been proposed as the main reasons for health disparities. 15,18 One of the main factors that deprives minority communities of quality care is distance to health-care services. 16 However, in our study the median distance from the nearest board-certified genetic Our findings suggest that YBCS who were older, less educated, did not have a family history of breast and/or ovarian cancer, or were not aware of their family history were less likely to receive genetic counseling and testing. Several studies have shown that use of genetic services is associated with awareness of familial risk. 20,28,32 Black YBCS in our sample reported lower rates of family history of breast/ovarian cancer compared with White/Other, suggesting that Black women may tend to underreport, or fail to recognize family history of cancer. 15,33,34 Black YBCS in our sample also had lower knowledge of breast cancer risk factors and breast cancer genetics compared to White/Other. The most important predictor for this finding was educational level. As expected, genetic counseling increased knowledge of breast cancer risk factors and genetics, suggesting that genetic counseling is an important first step to using cancer genetic services. Participants were also asked to report reasons for not using genetic services. Lack of physician recommendation was the single most common response, and this was reported more frequently by Black than White/Other YBCS. This finding has also been reported in other studies. 14,15,29,35 Some studies have shown that minority-serving physicians were more reluctant to order genetic testing for Black women, because they were more concerned about cost and insurance coverage for these patients. 36,37 Our findings are in agreement with others who have previously reported that minority-serving physicians were significantly less likely to have ever ordered a genetic test for breast cancer, colorectal cancer, or Huntington disease, or to have ever referred a patient for genetic testing compared with those serving fewer minorities. 37 Our findings also suggest that low income and low educational attainment are complex barriers to receiving genetic services both at the individual and at the health-care system levels. 15,18 Our study demonstrated this with a population-based sample that included a large number of Black YBCS, which addresses the knowledge gap identified by USPSTF about understanding disparities in genetic service utilization among racial and ethnic minorities. 19 Other studies have reached similar conclusions. 14,15,29,35 Longitudinal follow-up is critical to determine whether these disparities persist, or widen with time. The study has several limitations, such as self-reported data, a cross-sectional design, and residential patterns, which may not be representative of other states besides the state of Michigan. Results of positive genetic tests were based on self-reports and were not verified with data from medical files. A smaller proportion on Black YBCS accepted participation compared with White/Others. Given the lack of detailed demographic and clinical data for nonparticipating YBCS, it was not possible to assess how this may have affected the outcome of the study. There is a possibility for selection bias among responders. Moreover, participants who were not identified as Black were collapsed into the same category as Whites. This could have potentially decreased the observed difference in racial disparities, especially if these participants shared the same barriers as Black YBCS. However, a sensitivity analysis was conducted, and showed that this did not affect the reported outcomes. Despite these limitations, this study has many strengths. Our study was one of the few that examined the impact of distance from genetic counseling services as a barrier, using geocoding of genetic counseling centers across the country, and residential ZIP codes of all participants . Our findings indicate that distance to genetic services does not explain observed racial disparities when using these services, and that increasing availability of genetic services may not be adequate to close this gap. Few registry-based studies have achieved significant representation of Black YBCS. 15,29 Having a larger sample of Black YBCS enabled racial comparisons to be conducted with enough statistical power. Furthermore, this study can serve as a reference for future comparisons. This may be particularly important for assessing the impact of the Affordable Care Act on genetic services utilization. 38 The Healthy Michigan Plan was put in action in April 2014, and has granted access to genetic testing and cancer preventive measures to more than 650,000 previously uninsured individuals, most of whom belong to minority groups. 29,38 Thus, our study can serve as a standard for comparing genetic service utilization in a model state, and provide evidence on its broader social and economic impact. 39 This is particularly important following discussions concerning ACA continuation, or proposed dismissal after the fiscal year 2020. Additional initiatives that include interventions tailored to the unique needs and special characteristics of racial groups need to be considered. The primary focus of system-based interventions should be on mitigating social and health-care inequalities created by disparities in income and education. Once the financial barriers have been bridged, more attention should be given to empowering minority communities by strengthening their knowledge of genetic cancer risk, and increasing appropriate use of genetic services. 40 --- DISCLOSURE The authors declare no conflicts of interest.

Purpose: Increasing use of genetic services (counseling/testing) among young breast cancer survivors (YBCS) can help decrease breast cancer incidence and mortality. The study examined use of genetic services between Black and White/Other YBCS, attitudes and knowledge of breast cancer risk factors, and reasons for disparities in using genetic services. Methods: We used baseline data from a randomized control trial including a population-based, stratified random sample of 3000 potentially eligible YBCS, with oversampling of Black YBCS. Results: Among 883 YBCS (353 Black, 530 White/Other) were significant disparities between the two racial groups. More White/ Other YBCS had received genetic counseling and had genetic testing than Blacks. Although White/Other YBCS resided farther away from board-certified genetic counseling centers, they had fewer barriers to access these services. Black race, high out-ofpocket costs, older age, and more years since diagnosis were negatively associated with use of genetic services. Black YBCS had lower knowledge of breast cancer risk factors. Higher education and genetic counseling were associated with higher genetic knowledge.Racial inequalities of cost-related access to care and education create disparities in genetic services utilization. Systembased interventions that reduce socioeconomic disparities and empower YBCS with genetic knowledge, as well as physician referrals, can increase access to genetic services.

Introduction In order to understand contemporary society, it is necessary to look at the influence of mass media on people, particularly in terms of aspects linked with the creation of values and behaviour patterns in younger populations [1]. Accordingly, authors such as Hong and Kim [2] and Wearn and Shepherd [3] found that audiovisual media have a persuasive effect on emotions, psychosocial aspects such as subjective wellbeing and social-behavioural patterns. Likewise, the educational context should take on great importance in the early stages of life in order to establish healthy physical habits and the responsible use of new technologies, as this has been shown to benefit individuals' physical and mental health and quality of life [4][5][6][7][8]. In fact, when a habit or lifestyle is acquired in the early stages of education, it will rarely be forgotten in adulthood [9][10][11]. Specifically, mass media present a positive impression of iconic personalities through publicity, encouraging people to change their lifestyle, such as physical activity motivation or eating habits, thus promoting healthy habits [12,13]. However, not all mass media have a positive effect, as highlighted by Francisco et al. [14] and Stirling et al. [15], as excessive mass media exposure generates dissatisfaction, setting beauty standards and causing society's demands to fit within established parameters. Therefore, these factors may cause eating disorders, usually among younger women, who aim to respond to the standards fixed in their culture and society, including by excessive physical exercise. Thus, this leads individuals into an unhealthy lifestyle and reduces their emotional wellbeing [16,17]. Age and socio-economic factors are therefore a key issue in the adult population, as it has been shown that increasing age advocates a healthy state and improved quality of life, responsible use of mass media, consolidation of mental and emotional processes and improved psychological well-being [18][19][20][21][22]. Based on the above, it should be noted that low levels of physical activity and unhealthy behaviours among young adults and adolescents are a cause for concern and lead to significant socio-health problems [23]. In this sense, Meganto-Mateo et al. [24] showed that physical activity and an optimal diet are associated with greater psychological well-being, because these people perceive themselves as healthier, have lower levels of stress and have a better mood compared to people who do not engage in physical exercise. For the use and implementation of healthy physical activity, the guidelines proposed by the World Health Organization [25], which state that around 27.5% of adults and 81% of adolescents do not comply with them, which causes serious problems in terms of non-communicable diseases and mental health, should be considered. Therefore, a balanced diet that provides all the nutrients necessary for the body should be adopted [26], with the Mediterranean diet being one of the best-known diets that meets all the requirements for healthy eating [27,28]. However, physical activity and adherence to the Mediterranean diet are associated with individual factors such as gender, age, socioeconomic status, educational level, living location and personality features [19,29]. In relation to the problems described above, in this study, a structural equation model is developed including physical activity and Mediterranean diet adherence as endogenous variables, taking into account the effect of mass media pressure, psychological well-being and age. Positive and negative relationships between the variables studied is found, highlighting the importance of lifelong health education. Furthermore, this association between variables can differ according to gender; thus, a multigroup analysis was carried out. Based on the above, the following aims were set out: to develop an explanatory model regarding associations between mass media, psychological well-being, physical activity, Mediterranean diet and age; to contrast the structural model by multigroup analysis according to gender. --- Materials and Methods --- Subjects and Design A descriptive, non-experimental and cross-sectional design was used in this quantitative study. One single measurement was carried out for a single group using convenience sampling for participant selection. The study sample was composed of 634 participants, providing a homogeneous distribution by gender. Women accounted for 55.5% of the sample and men for 44.5% . The age range of the participants was from 18 to 66 years . Spanish subjects were invited to participate if they fulfilled the criteria of being over legal age and not exceeding ordinary retirement age. A total of 53 questionnaires were invalidated because they did not fulfil the inclusion criteria or because they were incorrectly completed. --- Instruments and Variables A self-reporting and self-completion questionnaire was used to compile socio-demographic information. Participants' gender and age were recorded. In addition, physical-sports aspects were recorded, showing how much time subjects spent on physical activity or sport. A questionnaire was used to assess social pressure regarding physical appearance. The original questionnaire "Sociocultural Attitudes Towards Appearance Questionnaire-4 " was developed by Schaefer et al. [30] and translated into Spanish by Llorente et al. [31]. It is composed of 22 items using a Likert-type scale with five response alternatives . The questionnaire assesses the dimensions of internalising thin/low body fat ; internalising an athletic/muscular build , family pressure , peer pressure and media pressure . In particular, the dimension of media pressure was used in this study, with an example of its items being "I feel pressure from the media to improve my appearance". The internal reliability consistency in the present study was α = 0.916 across all items; more specifically, for media pressure, α = 0.967 was obtained. In order to evaluate psychological well-being, a shortened version of the Psychological Well-Being Scale [32] was used, adapted to Spanish by Díaz et al. [33]. This scale is composed of 29 items answered on a Likert-type scale with six response options , including 10 items that are inversely formulated. This instrument makes it possible obtain a summary of psychological well-being emerging from the following six dimensions of psychological well-being: self-acceptance , positive relationships , autonomy , environmental mastery , personal growth and purpose in life . The internal reliability of this scale in the present study for all items was α = 0.918. The Mediterranean diet quality index questionnaire developed by Serrá-Majem et al. [34] was used to measure adherence to the Mediterranean diet. This instrument is made up of 16 dichotomous items that represent Mediterranean diet standards. Four of the items are negatively rated if answered in the affirmative . All other items are scored positively if the answer is yes . An overall score ranging from -4 to 12 is obtained by summing the items, allowing the total of the scale to be obtained and to categorize adherence to the Mediterranean diet . --- Procedure As shown in Figure 1 of the study procedure, scientific research was carried out in order to gather information and clarify some of the current problematic situations in society. Afterwards, the Department of Didactics of Musical, Plastic and Corporal Expression of the University of Granada created a Google form with the specified instruments, describing the study aim, as well as accepting participation by sending the form. Several routes were used to administer the questionnaire, including social network dissemination. Two questions were duplicated in order to check that they had not been filled in randomly and to ensure the reliability of the answers. Therefore, the correct bias of the answers was ensured. Even so, 53 questionnaires were excluded because they did not comply with the inclusion criteria or were incorrectly completed. Research ethics principles established by the Declaration of Helsinki [35] were followed in this study, ensuring anonymity and respecting participants' rights. In addition, research approval was granted by the Ethics Committee of the University of Granada . --- Data Analysis The statistical software SPSS 25.0 was used to process the data in order to establish the frequencies and means of the basic descriptive analysis. The Cronbach's coefficient was used to determine the internal consistency of the instruments, establishing the reliability index at 95%. Structural equation multigroup analysis was performed with AMOS 23.0 software . SEM analysis was used to establish the relationships between the variables composing the theoretical model for both groups . One general model was constructed for the study sample and two differentiated models in order to verify the relationships between variables according to the participants' genders. The SEM developed for this analysis was based on five observable variables that provided an explanation for the relationships. For endogenous variables, causal explanations were made by considering the observed associations between indicators and measurement reliability. Therefore, the measurement error of the observable variables was included in the model and could be directly controlled and interpreted as multivariate regression coefficients. Unidirectional arrows represented influence lines between latent variables and were interpreted from regression weights. A significance level of 0.05 was established using Pearson's Chisquare test. The Mediterranean diet is an endogenous variable that is affected by mass media pressure , psychological well-being , physical activity and age. Thus, PA is affected by MMP, PWP and age, in the same manner as PWP affects MMP and age. Furthermore, in this case, they act as endogenous variables. In order to verify the compatibility between the developed model and the empirical data obtained, the model fit was examined. Following the criteria proposed by Marsh [36], the reliability model was obtained according to the goodness-of-fit. For the Chi-square analysis, values associated with a non-significant p-value indicate a good model fit. Since this statistic is very sensitive to sample size effects, other fit indices should be used [37]. Additional parameters such as the comparative fit index , normalized fit index , incremental fit index and the Tucker-Lewis index were used. The values obtained needed to be be above 0.90 to represent an acceptable fit, and values above 0.95 represented an excellent fit. Moreover, the root mean squared error of approximation was used, where an acceptable fit was determined by values of 0.08 and excellent fit with values below 0.05. --- Data Analysis The statistical software SPSS 25.0 was used to process the data in order to establish the frequencies and means of the basic descriptive analysis. The Cronbach's coefficient was used to determine the internal consistency of the instruments, establishing the reliability index at 95%. Structural equation multigroup analysis was performed with AMOS 23.0 software . SEM analysis was used to establish the relationships between the variables composing the theoretical model for both groups . One general model was constructed for the study sample and two differentiated models in order to verify the relationships between variables according to the participants' genders. The SEM developed for this analysis was based on five observable variables that provided an explanation for the relationships. For endogenous variables, causal explanations were made by considering the observed associations between indicators and measurement reliability. Therefore, the measurement error of the observable variables was included in the model and could be directly controlled and interpreted as multivariate regression coefficients. Unidirectional arrows represented influence lines between latent variables and were interpreted from regression weights. A significance level of 0.05 was established using Pearson's Chi-square test. The Mediterranean diet is an endogenous variable that is affected by mass media pressure , psychological well-being , physical activity and age. Thus, PA is affected by MMP, PWP and age, in the same manner as PWP affects MMP and age. Furthermore, in this case, they act as endogenous variables. In order to verify the compatibility between the developed model and the empirical data obtained, the model fit was examined. Following the criteria proposed by Marsh [36], the reliability model was obtained according to the goodness-of-fit. For the Chi-square analysis, values associated with a non-significant p-value indicate a good model fit. Since this statistic is very sensitive to sample size effects, other fit indices should be used [37]. Additional parameters such as the comparative fit index , normalized fit index , incremental fit index and the Tucker-Lewis index were used. The values obtained needed to be be above 0.90 to represent an acceptable fit, and values above 0.95 represented an excellent fit. Moreover, the root mean squared error of approximation was used, where an acceptable fit was determined by values of 0.08 and excellent fit with values below 0.05. --- Results The model developed through the evaluation of the variables assessed in an adult sample aged 18-66 years as a function of gender showed a good fit for all indices. The Chisquare analysis showed a significant p-value . However, these indicators could not be interpreted independently due to the influence of the sample size and susceptibility [36]. Therefore, other standardized fit indices were employed that were also less sensitive to the sample size. The comparative fit index analysis obtained a value of 0.995, which represented an excellent fit. The normalized fit index analysis obtained a value of 0.991, the incremental fit index was 0.995 and the Tucker-Lewis index obtained a value of 0.953, all of which were excellent. The root mean square error of approximation analysis also obtained an excellent value of 0.040. Figure 3 and Table 1 shows the regression weights of the theoretical model, with statistically significant relationships at p < 0.01 and p < 0.001. Mass media pressure was negatively associated with psychological well-being , physical activity and Mediterranean diet adherence . The strength of the regression weights was medium-low for this case. However, PWB was positively related to PA and MD , showing regression weights with a medium effect strength. In addition, MD was associated directly and positively with BP and age , with a low-medium strength of the regression weights. No statistically significant differences were found between the existing relationship of age with BP and PWB. --- Results The model developed through the evaluation of the variables assessed in an adult sample aged 18-66 years as a function of gender showed a good fit for all indices. The Chi-square analysis showed a significant p-value . However, these indicators could not be interpreted independently due to the influence of the sample size and susceptibility [36]. Therefore, other standardized fit indices were employed that were also less sensitive to the sample size. The comparative fit index analysis obtained a value of 0.995, which represented an excellent fit. The normalized fit index analysis obtained a value of 0.991, the incremental fit index was 0.995 and the Tucker-Lewis index obtained a value of 0.953, all of which were excellent. The root mean square error of approximation analysis also obtained an excellent value of 0.040. Figure 3 and Table 1 shows the regression weights of the theoretical model, with statistically significant relationships at p < 0.01 and p < 0.001. Mass media pressure was negatively associated with psychological well-being , physical activity and Mediterranean diet adherence . The strength of the regression weights was medium-low for this case. However, PWB was positively related to PA and MD , showing regression weights with a medium effect strength. In addition, MD was associated directly and positively with BP and age , with a low-medium strength of the regression weights. No statistically significant differences were found between the existing relationship of age with BP and PWB. The developed model according to male gender showed a good fit for all indexes. Chisquare analysis showed a significant p-value . Comparative fit index analysis obtained a value of 0.993, the normalized fit index obtained a value of 0.989 and the incremental fit index was 0.993, representing an excellent fit. In the Tucker-Lewis index analysis, a value of 0.927 was obtained, showing in this case an acceptable fit. The root mean square error of approximation analysis obtained a value of 0.079, which is acceptable. The developed model according to male gender showed a good fit for all indexes. Chi-square analysis showed a significant p-value . Comparative fit index analysis obtained a value of 0.993, the normalized fit index obtained a value of 0.989 and the incremental fit index was 0.993, representing an excellent fit. In the Tucker-Lewis index analysis, a value of 0.927 was obtained, showing in this case an acceptable fit. The root mean square error of approximation analysis obtained a value of 0.079, which is acceptable. Figure 4 and Table 2 show the regression weights of the male model, showing statistically significant relationships at levels p < 0.05, p < 0.01 and p < 0.001. The male model fitted better than the female model. MMP was negatively associated with PWP in the case of a high strength in the regression weights; with a medium relationship strength, it was associated with MD , and with a low strength, it was associated with PA . Conversely, PWP was shown to Figure 4 and Table 2 show the regression weights of the male model, showing statistically significant relationships at levels p < 0.05, p < 0.01 and p < 0.001. The male model fitted better than the female model. MMP was negatively associated with PWP in the case of a high strength in the regression weights; with a medium relationship strength, it was associated with MD , and with a low strength, it was associated with PA . Conversely, PWP was shown to be positively related to BP , MD and age , showing a medium-strength regression effect. Finally, a direct relationship of MD with BP was shown, determined by a low-strength regression weight. be positively related to BP , MD and age , showing a medium-strength regression effect. Finally, a direct relationship of MD with BP was shown, determined by a low-strength regression weight. For females, the model fit was adequate, although this condition was not fulfilled for all indices. Chi-square analysis showed a significant p-value . However, the comparative fit index obtained a value of 0.879 and the Tucker-Lewis index obtained a value of 0.827, indicating that these parameters did not fit. However, the normalized fit index analysis obtained a value of 0.903 and the incremental fit index was 0.909, representing an acceptable fit. The root mean squared error of approximation analysis obtained a value of 0.126. These results show that the female model did not fit as acceptably as the male model. For females, the model fit was adequate, although this condition was not fulfilled for all indices. Chi-square analysis showed a significant p-value . However, the comparative fit index obtained a value of 0.879 and the Tucker-Lewis index obtained a value of 0.827, indicating that these parameters did not fit. However, the normalized fit index analysis obtained a value of 0.903 and the incremental fit index was 0.909, representing an acceptable fit. The root mean squared error of approximation analysis obtained a value of 0.126. These results show that the female model did not fit as acceptably as the male model. Nevertheless, as the fit values were close to acceptable values, some statistically significant relationships were obtained. Figure 5 and Table 3 show the regression weights for the female model. In this case, the most influential variable was MD, as it was positively related to BP , determined by a medium regression weight strength, and it was also associated with PWB and age , showing a low-strength regression effect. For the other relationships proposed in the model, no statistically significant differences were observed. Nevertheless, as the fit values were close to acceptable values, some statistically significant relationships were obtained. Figure 5 and Table 3 show the regression weights for the female model. In this case, the most influential variable was MD, as it was positively related to BP , determined by a medium regression weight strength, and it was also associated with PWB and age , showing a low-strength regression effect. For the other relationships proposed in the model, no statistically significant differences were observed. --- Discussion The current study carried out among older adults addressed aspects of special relevance today, such as psychological well-being and mass media pressure, as well as their relationship with factors associated with a healthy physical education, such as the level of --- Discussion The current study carried out among older adults addressed aspects of special relevance today, such as psychological well-being and mass media pressure, as well as their relationship with factors associated with a healthy physical education, such as the level of physical activity and the Mediterranean diet. The principal aim was to establish a structural model to explain the associations between the above variables, as well as to carry out a multi-group analysis to identify relationships according to participants' genders. Some research with similar aspects or features has been carried out by Afshin et al. [38], Peter and Brosius, [39], Cranmer et al. [40] and Troncone et al. [41]. At present, there are few studies providing answers to the associations proposed in this research. Therefore, this study's novelty is to provide information about the effect of mass media pressure and psychological well-being on the physical-healthy habits of people. Mass media are a valuable tool for organizing and informing citizens; however, they frequently place a direct pressure on viewers [42,43]. The main results of this model show that the pressure applied by the mass media was negatively associated with psychological well-being, the practice of physical activity and adherence to the Mediterranean diet. Similar data are found in the study of Mieziene et al. [44], who found that mass media pressure generates a beauty ideal that seeks extreme thinness, which leads to greater personal dissatisfaction and less physical activity. Likewise, Vilhjalmsson et al. [45] argue that dissatisfaction provoked by mass media does not act as a motivating factor towards acquiring healthy habits. Based on the above, Griffiths et al. [46] show that mass media pressure is greater on younger adults who show less psychological wellbeing. As a consequence, there is a worsening of their eating behaviour, especially based on restriction, which can lead to eating disorders. However, findings showed that higher psychological well-being leads to an increase in healthy parameters such as physical activity and adherence to a Mediterranean diet. Sofija et al. [47] showed that wellbeing, which is manifested as a positive self-reported experience, promotes personal self-care and keeps individuals away from risky behaviours, favouring the development of behaviours focused on the protection and promotion of health [48]. Finally, an important factor, age, was included in the general model presented. In this case, it was shown that this variable had a direct impact on adherence to the Mediterranean diet. This fact can be justified and related to the fact that the adult population establishes food preferences and their consumption improves as a result of greater exposure to different food varieties [29,49]. In terms of gender differences, it should be noted that the results for men showed that mass media pressure has a negative impact on psychological well-being, leading to a decrease in physical activity and a worsening diet. According to the results obtained in the study by Izydorczyk et al. [50], the mass media pressure for men to conform to an ideal standard is evident, leading them to suffer great emotional dissatisfaction if their body size does not coincide with society's ideal. Similarly, Tylka [51] notes that mass media pressure acts as a promoter of physical activity among men, who use it to improve their muscle tone, causing an imbalance in their eating behaviour. Likewise, Van Hecke et al. [52] and Pfister and Sisjord [53] found that mass media disseminate physical-sports activities of interest to men, which may promote physical exercise in this population sector. However, males showed a positive personal satisfaction associated with healthy physical activity habits. These data are similar to those found by Zhang et al. [54], who found that men's physical activity is associated with higher levels of happiness, mood and vitality, highlighting that physical exercise increases their levels of self-efficacy, which is a key factor in people's life satisfaction [55]. Similarly, in the study by Grao-Cruces et al. [56], it is shown that physically active men show high levels of emotional well-being, as well as a positive self-concept, which favors a healthy diet that provides them with improved well-being. In addition, psychological well-being was positively influenced by age. In particular, men have been shown to perform better than women in terms of self-acceptance, purpose in life, autonomy and mastery of the environment [57]. In fact, authors such as Steptoe et al. [58] show a close relationship between wellbeing and health, as well as the increasing importance of maintaining well-being in older age. Regarding women, it should be noted that no significant relationships were found between mass media pressure and psychological well-being and healthy habits. However, the study by Jiménez-Boraita et al. [59] shows that the mass media have a greater impact on women, deteriorating their emotional well-being. Nevertheless, it was found that higher levels of well-being in women are associated with better healthy habits. Women showing higher levels of well-being were reported to have higher levels of healthy physical activity, as well as better eating behaviours. González-Valero et al. [60] noted that women have an extrinsic motivation to practice physical activity, as they have a specific purpose in mind; i.e., improving their physical condition, slimming their body shape, reducing their fat levels, among others. Physical exercise reduces stress levels, increases energy and improves emotional coping and control, with positive repercussions on the acquisition of healthy eating habits, as well as improved psychological well-being [61,62]. Furthermore, age was positively associated with the Mediterranean diet in this group. In other words, those individuals who were older had a better adherence to the Mediterranean diet. In fact, the study by Álvarez-Fernández et al. [63] showed that the consumption of healthy foods in a group of young women was significantly lower than in older women. Taking the above into account, it should be noted that this is a new study, showing relationships between mass media pressure, psychological wellbeing and the acquisition of healthy habits. This enables a series of associations to be established that allow society to observe an underlying reality, as well as to verify the importance of a physically healthy education over lifetime. However, this study was not exempt from limitations, including the descriptive and cross-sectional nature of the study, which only allowed us to analyze variables at the time when they were extracted. Furthermore, due to the number of participants, the results cannot be generalized to the whole population. Based on the results and practical applications, as future perspectives, it is intended to develop interventions that have an impact on physically healthy education from an early age, with the aim of developing emotionally stable people who do not perceive the mass media as a pressure they have to respond to, but rather as a way of promoting healthy behaviours. --- Conclusions Overall, an acceptable value was obtained for all parameters of the general equation model. This work shows that mass media have a direct influence on the well-being of the society, having a negative impact on the practice of physical activity and adherence to the Mediterranean diet. Psychological well-being is reduced as a result of these factors, leading individuals to develop eating disorders, obesity, sedentary habits or personal dissatisfaction. Regarding the gender model, males obtained a better fit in the indices than females, although this was also acceptable. For males, it was shown that mass media pressure leads to lower emotional well-being, as well as physical activity and adherence to the Mediterranean diet. However, a better adjustment of psychological well-being was related to higher levels of regular physical exercise and better adherence to the Mediterranean diet. As for women, no relationship was found between mass media pressure and psychological well-being and physically healthy habits. However, women showed greater personal satisfaction and reported more physical activity, as well as better eating patterns. Therefore, in general terms, it can be concluded that the mass media have a direct influence on people's psychological well-being. Thus, an inadequate use of mass media has a negative influence on the practice of physical activity and adherence to the Mediterranean diet. Furthermore, age will be a key factor in the development of healthy habits and the well-being of society. Therefore, the correct use of new technologies should be encouraged from an early age and special attention should be paid to physical activity and dietary recommendations in order to avoid physical and mental health problems. ---

Background: The influence of mass media on emotions, subjective well-being and behaviours in society should be clearly understood. Physical-health education has an important role to play as a preventive tool. The aim of this study was to develop an explanatory model regarding the relationships between mass media, psychological well-being, physical activity, Mediterranean diet and age and to compare the model with multi-group analysis according to gender. Methods: A descriptive, non-experimental, cross-sectional design was used, with 634 participants between 18 and 66 years old (M = 35.18 ± 9.68). Results: Structural equation modeling was found to be satisfactory for all parameters. Results show that mass media have a significant direct influence on well-being, with negative effects on physical activity and adherence to a Mediterranean diet. The model fitted better for males in terms of gender differences, showing a better fit of psychological well-being being associated with higher levels of physical activity and better adherence to the Mediterranean diet. Among women, no relationships were found between mass media pressure and psychological well-being and healthy physical habits, but higher personal satisfaction was associated with better physical activity and better dietary patterns. Conclusions: Thus, the study approaches society to a perspective influenced by mass media and physical-health education, reporting and emphasizing the importance of healthy lifestyles.

INTRODUCTION As a country with a population of around 1.4 billion, China currently has around 400 million patients with chronic diseases. Management of chronic diseases outside the hospital exists as a tricky issue. Meanwhile, treatment of chronic diseases is a long-term process, which requires lifelong treatment, long-term medication and health management . Under the traditional medical care model, patients need to communicate with doctors about their disease in hospital. Because of difficulty in making an appointment with some doctors and the expensiveness of medical treatment, patients' medical experience has been low and inefficient . The problem should be mainly attributed to the inadequacy and uneven distribution of overall medical resources in China . Specifically, this problem is reflected as inadequacy of overall medical resources, including doctors, nurses, medical resources, and per capita disposable medical fees . With the development of information and technology, people can seek medical advice from the Internet . From the Internet, patients can search for information related to their need of medical health and consulting doctors or other patients online on health issues . Internet medical care can not only reasonably serve part of patients who do not have to seek consultation in hospitals offline, but also reduce the waste of medical resources and medical fees . Moreover, the time cost for patients to receive medical treatment face to face is efficiently brought down, and the limited number of quality medical resources are left for those who are genuinely in need . Online health communities or online health communities exist as a community for health information exchange, where users can have health consultation, experience sharing, knowledge sharing and emotional sharing . So far, a majority of OHCs have developed functions like health consulting, online socializing, themed recommendations, etc. These functions can help users realize long-term health management and establish a long-term communication system. Information sharing behaviors of users on OHCs community platforms can not only help improve the efficiency of users' health information acquisition efficiency and alleviate the asymmetry of health information, but also help community operators increase the activity of community users, thus providing more comprehensive information services to satisfy users' information sharing needs . Kingod et al. and Petrič et al. thought that OHCs can create a supportive space for daily self-care of patients suffering from chronic diseases under effective support of doctors, and at the same time, OHCs can construct the offline relationship between patients and doctors and contribute to the harmony of the relationship between doctors and patients. Additionally, the rapid communication of information, suggestions or advice among OHC community members is conducive, since the OHCs usually deliver referral information related to diseases and health, suggestions, individual experience, etc. . Emotional sharing can facilitate the expression of sentiments, such as understanding, encouragement, sympathy and care among OHC members, through which the pressure and anxiety imposed on OHC members can be mitigated. As a new platform for knowledge sharing and exchange in the modern medical field, OHCs ensures community members to contribute their role to efficient diffusion of medical knowledge via the information exchange . Not only can this help people deepen their knowledge of health, but also this can promote the transformation of the focus of the health concept from treatment to prevention, thus allowing people to have access to more medical and health care information and more reasonably pursue self-health management . Characterized by user-generated content , OHCs requires deep participation of community members to accumulate mass data of health and medical care . However, a majority of users remain an inactive status, who mainly acquire health knowledge and information from the community, lacking the motive and intention to share . Nevertheless, current literatures generally focus on the information search and adoption stage of users' participation. This has resulted in OHCs' lack of theoretical guidance on how to protect users' knowledge acquisition and knowledge contribution. On the other hand, OHCs is an information exchange and relation network made up of community members, including doctors, patients, family members and passers-by. Users' behavioral decision-making is subject to the influence of not only community quality factors, such as system quality, information quality, service quality, but also structural, relational cognitive dimensions, which actually refer to social capital. For example, other users' likes, comments, replies, number of followers, activeness and joint vision can all affect users' participatory behaviors . Therefore, an in-depth research into factors influencing OHCs' user participation is necessary to address the issue of how social capital can prompt users to adopt knowledge and information from OHCs and how users and other members can be encouraged to share knowledge and medical data. Therefore, this research adopts social capital and its components to explore their influence on information adoption and sharing. At the same time, informational support and emotional support are introduced to the research model to improve the explaining capacity and applicability of this model. This research aims at examining how people's assessment and experience of social capital affect their knowledge acquisition and knowledge contribution. To sum up, this research can not only help people acquire more medical care information and manage their own health more reasonably, but also provide a scientific basis for the sustainable development of OHCs. The paper is structured as follows. In the next section we describe the concept of OHCs, literature review and theoretical background. Section "Research Model and Research Hypotheses" reports we propose the research model and hypotheses. Sections "Materials and Methods" and "Results" reports the instrument development, data collection process, data analysis, and results. In section "Discussion" we discusses these results. Then, we present the theoretical and practical implications in section "Theoretical and Practical Implications." We conclude the paper by summarizing the limitations of the study and suggesting avenues for future research in section "Limitations." --- BACKGROUND AND LITERATURE REVIEW Online Health Communities In the narrow sense, OHCs can be divided by user types into online doctors' communication communities and online doctorpatient communication communities . In the broad sense, OHCs is mainly divided into professional health and medical care websites represented by "Patients Like Me" and "Online Good octor, " health and medical care sections of largescale social networks represented by Baidu Post Bar and Douban , and temporary chatting groups spontaneously organized by patients or enthusiasts of health knowledge . The OHCs discussed in this paper tend to adopt the classification in the broad sense, which is a set of online communities that organize activities related to health information and aiming at providing a platform for users to express their opinions, share their experience, exchange their health information, offer or seek medical and health care information. Management of chronic diseases after patients are discharged from hospital remains a tremendous challenge. Since the treatment of chronic diseases is a lengthy process, patients need long-term medication and health management control. OHCs are a relationship network established based on information sharing of community members, including doctors, patients, family members, passers-by, etc. This has led to a large amount of health and medical care data created, record or inferred by community members, which can promote the efficient diffusion and sharing of online medical health knowledge. Thereby, patients can have an easier access to health information. This can not only smoothen the information interaction between patients and doctors, but also improve the patient-doctor relationship, and continuously attract patients to receive treatment in accordance with doctor's advice . A main challenge facing every online community is the user activity, that is, how to increase the information search frequency and promote the information sharing between users and other members . Therefore, to encourage users to use community information actively and provide knowledge sharing is essential to the survival and development of OHCs . For example, tieba.baidu has set up multiple discussion columns for chronic diseases, such as diabetes and hypertension. All these columns fall under the category of UGC, but a majority of users are at an inactive status, suggesting a lack of motive and intention to participate among them. This will finally impair the vitality for the community development. Hence, it is necessary to study users' participation in OHCs from the perspective of knowledge acquisition and knowledge contribution. --- Users' Participation Users' participation is described by many literatures as a psychological status. When the system can attract users' attention, the user will be attracted by the system to obtain internal rewards . Literatures of the early stage tend to associate participation with the network traffic theory. An eye-catching system can effectively attract users while interacting with it . A research by Di Gangi and Wasko defined users' participation in social media as "one kind of user psychology, which can not only ensure the improvement of users' participation, but also bring benefits meaningful to individuals." Users' participation is a critical index of OHCs . OHCs can provide medical and health care service consumers with instant access to information, massive medical care information, and other medical and health care services . In addition to allowing users to search for health information, OHCs also facilitates information sharing . Individuals can easily share their medical and health care knowledge, experience and treatment plans, and post comments on healthcare products, drugs and doctors' comments . In spite of the aforesaid advantages of OHCs as part of virtual communities, challenges facing OHCs in terms of users' participation cannot be ignored . To learn predictive factors of users' participation can shed light on how to promote the sharing of online community information . --- Social Capital Theory Social capital can reflect the total resources included in the relation network of an individual or a social network, and users can have social interaction based on resources possessed thereby to promote information exchange . Social capital is first use to study community relations, and people's intention to share information and knowledge is subject to the impact of social capital . When social interaction is friendly, people are usually willing to share their knowledge and information . Social capital theory holds that social resources can integrate some elements in the social network, and that social capital can not only achieve a consensus among individuals, but also force them to realize the shared objective . Research suggests that social capital has three dimensions, including structural dimension, relational dimension and cognitive dimension, and that all these dimensions can control individual behaviors and interaction . First of all, structural capital stands for the contact model among individuals, which can reflect the users' position in the social system and identify users' ability to access resources . In other words, structural capital can decide whom an individual can reach, and how the individual can reach the person, and can indicate the connection and availability of network friends . Whether network relation or any proper organization exists or is established among individuals constitutes the most important aspect of this dimension. Organizations of the kind reflect the relation models in terms of the scale, density, connection, hierarchical structure, etc. . Proper organizations refer to the network that is set up for one shared purpose . Second, relational dimension can describe the trust among members in the communication process . This dimension can reflect the attribute of the cooperation with other members and the quality of the interpersonal relationship, which is a kind of interpersonal relationship that is formed because of the interaction history among individuals . This concept focuses on the special relation among individuals, such as respect for friendship, which can affect their behaviors . Besides, cognitive dimension refers to a shared understanding and opinion of something among individuals . The formation of the virtual community has a close bearing on the values and shared interests of individuals, the latter of which can be named as "shared vision" . In a virtual community, the higher the similarity is between individuals, the more likely they will see the information interesting them, and the stronger their motive to seek the information will be . At the same time, the virtual community is a platform for individuals to share similar experience and interest. The more the values and objectives are shared by individuals, the more confident the individuals will be to share opinions and comments . --- Social Support Social support can reflect the experience of users to receive care, help and attention from the online community . Social support is provided from different dimensions. Liang et al. noticed that social support exchanged in social commerce includes informational support and emotional support. This measurement outcome has been applied to the follow-up research . Social support can include instrumental support and expressive support , which is similar to the informational support and emotional support, respectively. Prior research also points out that social support covers friendship, assessment and respect . Additionally, Yan and Tan pointed out that social support for online medical care community communication includes the informational support, emotional support and friendship. Swift development of OHCs has benefited patients in many ways, enabling patients to have access to social support, which informational, emotional, etc. In this research, social support refers to an individual's experience of getting support, care or help from a group. Informational support refers to providing suggestions, explanations or information that is conducive to problem-solving to help users resolve problems or organize their thinking to make a wise decision. Users' information behaviors include sharing and exchange of health information, which can create favorable information sources for continuous value adding and communication of health information . Emotional support can fuel up sharing behaviors of community information . In this research, emotional support is defined as psychological support, such as sympathy, care or understanding, which can comfort users psychologically and help resolve problems indirectly. Hence, social support, in addition to enriching information sources, also warms individual's heart . The favorable experience can enhance users' willingness to seek social interaction with other users, while support providers can also obtain the sense of identity from interaction of the kind . In the OHC, users hope that they can have access to not only valuable suggestions, but also emotional attention and encouragement . Research has identified that social support is an important factor that affects OHC user behaviors . Therefore, this research includes social support into the model to verify its impact on social capital. --- RESEARCH MODEL AND RESEARCH HYPOTHESES Research Model Set against the background of OHCs and adopting social support as an antecedent variable, this research proposes a comprehensive model to verify the impact of three dimensions of social capital, including structural capital, relational capital and cognitive capital, on users' participation . Figure 1 displays the research model of this paper. --- Based on the Research Hypotheses of Social Support Social support consists of informational support and emotional support , of which informational support can provide suggestions which can facilitate problem solution . When users acquire valuable health information from other members, they might perceive the ability and credibility of these members. This is conducive to developing users' trust for OHCs, because the information can help them improve health. Besides, a frequent information interaction can enhance users' familiarity with other members and strengthen their bilateral relation, such as social interaction and relation. Users might form the habit of social interaction in OHCs. Moreover, informational support can also affect relational capital, for it can promote users' understanding of other members' opinions. Last but not least, a frequent information interaction can contribute to the formation of a shared language among users. Users can create their commonly used terms and jargons through a frequent interaction, thus increasing their bilateral exchange. Therefore, we make the following hypotheses: H1a. Informational support is positively related to structural capital. H1b. Informational support is positively related to relational capital. H1c. Informational support is positively related to cognitive capital. Emotional support is the care, encouragement and sympathy expressed by other members . As an intermediate support, emotional support, though incapable of directly addressing problems, can otherwise warm users spiritually, thus FIGURE 1 | Research model. stimulating users' interaction with other members. So emotional support is critically important to users' behavioral decisionmaking. Particularly to users of OHCs, they might be stuck in negative emotions, such as depression and anxiety . When users acquire emotional support from other members, they might perceive kindness from other members, and generate trust and recognition of OHCs . In the opinion, they have established the emotional link with their community. In addition, emotional support can promote the formation of a shared language among users, through which users can understand each other better. Thus, we suggest: H2a. Emotional support is positively related to structural capital. H2b. Emotional support is positively related to relational capital. H2c. Emotional support is positively related to cognitive capital. --- Based on the Research Hypotheses of the Social Capital Structural capital can show the social interaction among users . As one of major driving factors of user behaviors, social interaction can promote information and resource exchange among users . OHCs make ubiquitous interactions possible for users, which might increase users' interaction frequency and duration, and promote social interaction among them . So users with frequent social interaction with more frequently communicate with others about health information and share about medical care experience. In other words, a close connection can enhance users' participation, for they hope that they can maintain a close connection with other users. Therefore, we suggest: H3a. Structural capital is positively related to knowledge acquisition. H3b. Structural capital is positively related to knowledge contribution. Relational capital can reflect the quality of the relation of exchange, which includes trust and reciprocity . As an important factor that decides the quality of relations, trust is fostered through the frequent interaction among individuals . In the network environment, trust for others is a basic factor that drives users' behavioral intention . Since the online community involves uncertainties and risks, such as the risk of information privacy leakage, it is necessary for users to establish trust with other community members so as to reduce the perceived risk . Besides, if there is a strong awareness of reciprocity, that is, if members believe resource exchange is reciprocal, they will show more intention to participate . In other words, users disclosing information are also expecting information. Otherwise, they might deem the exchange as unfair and stop participating in the exchange. Thus, we suggest: H4a. Relational capital is positively related to knowledge acquisition. H4b. Relational capital is positively related to knowledge contribution. Cognitive capital is a mirror of shared language among users . A frequent interaction can increase users' mutual understanding and help them form a shared language, such as common terms and jargons . In addition to enhancing the shared perception and understanding among individuals, shared language can also reduce every individual's cognitive barriers . This can improve the communication efficiency and promote users' participation. Because of the difficulty facing users' entering in the information on the mobile terminals, a shared language is critical to their communication and interaction. Thus, --- MATERIALS AND METHODS The research model covers seven factors, each of which is measured by three or four items. In order to improve the content validity, all items are adapted from the existing literatures. The questionnaire on structural capital, relational capital, and cognitive capital is adapted from that of Zhao et al. and Xu and Zhou . The questionnaire on knowledge acquisition and knowledge contribution is adapted from that of Zhou and Chen et al. . The questionnaire on informational support and emotional support is adapted from that of Liang et al. andZhou . The unit of analysis focused on the individual and the responses were measured using a 5-point Likert scale on an interval level ranging from "strongly agree" to "strongly disagree" . Based on a comprehensive review of research findings by scholars at home and abroad, and in order to ensure all items to adapt to the OHC environment, we invite six experts and professors from the chronic disease and information system management field to assess the selection of questions and whether bias exists in these questions. Finally, items consistent with the OHC environment are chosen. After that, in order to test the instrument, a preliminary test is conducted on 20 users with OHC use experience. According to pre-survey results and suggestions of professors, the wording of questions is modified moderately under the prerequisite of not changing the connotation of questions so as to improve the clarity and comprehensibility of questions. At last, the questionnaire survey is designed. The final version of these items is translated by the professional translator independently into Chinese, which is then translated by another translator into English again to ensure the equivalence of translation. Data of this research are collected through online surveys. Considering the dramatic difference potentially existing in the demand of chronic disease patients and non-chronic disease patients for medical information, and to more accurately predict OHC members' participating behaviors, this research adopts chronic disease patients as research samples to measure participating behaviors of OHC members. We post survey links on platforms, including WeChat, Weibo, Baidu, etc. We have also invited users with the OHC use experience to fill in the questionnaire. The survey lasted for 4 months. Finally, 1309 copies of questionnaire were collected from 20 March 2021 to 20 July 2021. We also scrutinized all responses and dropped 32 that had too many missing values. As a result, we obtained 1277 valid responses. Table 1 shows the demographic information of the respondents. --- RESULTS Structural equation modeling together with SPSS 23.0 and AMOS 23.0 is adopted to analyze the research data. --- Measurement Model First, we conducted a reliability and convergent validity. As listed in Table 2, each loadings are larger than 0.7. Dijkstra-Henseler's rho and Jöreskog's rho exceeds 0.7, and each average variance extracted exceeds 0.5. In addition, all Cronbach alpha values are larger than 0.7. This indicated the excellent reliability and convergent validity . Table 3 lists the square root of AVE and factor correlation coefficients, for each factor, the square root of AVE is significantly larger than its correlation coefficients with other factors, suggesting excellent discriminant validity . --- Structural Model We adopted structural equation modeling software AOMS 23.0 to estimate the structural model. Table 4 lists the recommended value and actual values of structural model fit, all fit indices have better actual values than the recommended values. Table 5 presents the results. Except H3a and H3b, other hypotheses were supported. Factors performance expectancy have high loadings on the continuous usage intention. In Figure 2, the explained variance of, structural capital, relational capital, cognitive capital, knowledge acquisition, and knowledge contribution is 4.8, 40.1, 31, 52.3, and 63.2%, respectively. The square root of AVE and factor correlation coefficients. --- DISCUSSION As the awareness of health is strengthening among Chinese nationals, online medical care information has gained increasing popularity, which has driven the development of OHCs. A critical factor that affects the development of OHCs is users' participation. Pitifully, the motive for users to participate in OHCs is often lacking. Therefore, this research examines decisive factors that influence OHC users' participation from the perspective of social capital theory. As presented in Table 5 and Figure 2, H1 is fully substantiated. Results show that informational support has a significant impact on structural capital , relational capital , and cognitive capital . This finding coincides with that of Zhou and suggests that users are more willing to adopt opinions of other community members, if they can acquire valuable information from the interaction. This also indicates that users who are willing to use OHCs are all pragmatic. For example, users suffering from certain disease aim at seeking valuable information. If professional knowledge acquired from other members can give some help to alleviate their symptoms, then users will be more willing to share their knowledge with patients suffering from the same disease. Hence, informational support can show the value of OHCs, and enhance users' sense of belonging and cohesion, which can then promote users' participation. Similarly, H2 is fully substantiated. Emotional support has a significant impact on structural capital , relational capital , and cognitive capital . This finding shows good agreement with that of Zhou . This suggests that emotional support given by other OHC members can comfort users, because a majority of them share the experience of their disease. The comfort from users with the same or similar experience can easily strike a chord in each other's heart, and the comforted user will generate a sense of belonging toward the community deep down inside. Therefore, emotional support can help users perceive more warmth, who will them internalize the community values into their code of conduct. The comfort, encouragement and care provided by emotional support can effectively ease users' negative emotions , which can boost users' establishment of structural capital, relational capital, and cognitive capital of OHCs. Relational capital can directly exert a positive impact on knowledge acquisition and knowledge contribution . This finding shed light on the impact of other community members' trust and reciprocity on users' participation in OHCs. Users are willing to accept reliable and reciprocal suggestions and recommendations from other OHC members, and share their own knowledge of health. Participating in OHCs might be deemed as a risky behavior. Thereby, trust and reciprocity have been decisive factors that can directly predict users' participation in OHCs. This research result is consistent with prior research finding. For example, Luo et al. discovered a positive correlation between trust and trading intention in e-commerce. Like relational capital, cognitive capital can directly exert a positive impact on knowledge acquisition and knowledge contribution . This finding is consistent with that of Chen et al. , and also indicates that users' participatory behaviors in OHCs are decided not only by trust and reciprocity, but also by the shared language among community members. This suffices to show the importance of a shared language in OHC. For example, the public terms and jargons can promote communication and interaction among users, and improve their communication efficiency. At last, the impact of structural capital on knowledge acquisition and knowledge contribution is not significant. This suggests that where there is a strong interaction between one user and other members, they might establish a close connection and deepen their participation, which is contradictory to our prediction. The contradiction is probably caused by OHCs as a platform for commenting, adding likes to and searching for professional knowledge online. Compared with the real-life communities, virtual communities have one striking defect, that is, the fragile connection among their members, which can negatively influence user behaviors. This research result is consistent with that of Xu and Zhou and Yang . --- THEORETICAL AND PRACTICAL IMPLICATIONS This study proposed an extended information system success model to gain a deeper understanding of the factors affecting continued intentions to use a mobile workplace. The results suggested that the system quality, information quality, and collaboration quality have all made significant contributions to promoting performance expectancy and satisfaction. The performance expectancy and satisfaction could in turn boost the continuance intention. The empirical results also provided solid evidence for the good explaining power of the informational support success model for this model. This substantiated its validity in predicting users' continued intention to use mobile workplaces. Therefore, the IS success theory was a favorable model to explain the mobile workplace. Besides this, we combined the IS success model, collaboration quality, and performance expectancy to verify its feasibility. By emphasizing the feasibility of extending the publicly acknowledged model theory, this paper could provide an enriched insight and contribution to the literature on information systems. To researchers, this research could provide a solid foundation for the further improvement of the IS success theory and be regarded as a springboard for future research. We are convinced that this model could be further tested in different countries and among different age groups and information systems. --- Theoretical Implications First of all, prior research has extensively tested consumers' behavior intentions using social capital theory. For example, Yang adopted social capital theory to verify consumers' purchase intentions in social commerce. Another example is the verification of the impact of social capital on consumer value and consumers' loyalty in online outshopping. Nevertheless, little research attention has been paid to the correlation between social capital and OHC users' participation. Considering the above research gap, this research adopts social support theory as an antecedent variable based on social capital theory to verify OHC users' participation. Second, in order to clarify the complex correlations of different dimensions of social capital theory, this research explores the impact of different dimensions of social capital theory on users' participation. Specifically, relational capital and cognitive capital have a significantly positive impact on knowledge acquisition and knowledge contribution, while knowledge acquisition and knowledge contribution are free from the impact of structural capital. This finding verifies that three dimensions of social capital theory have different functioning mechanisms under different research backgrounds. Under the C2C background Huang et al. found that social capital can cause different changes along with different roles of individuals. Chiu et al. observed that the impact of social capital on the quantity and quality of knowledge shared is varied. Hence, our research findings have enriched the understanding of how the three dimensions of social capital, namely structural dimension, relational dimension, and cognitive dimension, influence users' participation. --- Managerial Implications This research can also provide some managerial implications. Research results show that OHC managers should encourage community members to develop a strong social capital. It is observed that social support can significantly influence social capital. Therefore, OHC managers need to create a supportive atmosphere to develop social capital and promote users' participation in OHCs. On the other hand, relational capital and cognitive capital have a significant impact on knowledge acquisition and knowledge contribution. In other words, trust for colleagues, reciprocity and shared language are major decisive factors of users' participation. This means that only when users can acquire valid information from OHCs will they share their own knowledge. This necessitates the establishment of regulations to encourage users to communicate with each other honestly and sincerely, which can benefit the creation of a reciprocal atmosphere. OHC managers can encourage users to exchange social support with other members via rewards, such as integrated points and rewards. They can also choose some critical opinion leaders as examples. Besides, OHCs should take measures to promote the formation of a shared language. They can analyze user texts to acquire commonly used terms and jargons, and list these terms for users' reference. Service providers should also build a credible environment for users and develop the recognition between them and the community. --- LIMITATIONS Our research limitations suggest the necessity of future research. First, the correlation between the OHC users' participation and social capital might be a dynamic process, while this text adopts the cross-sectional data to test the model. Therefore, the future research can examine dynamic changes of user behaviors. Second, OHCs can be divided into different types. User behaviors from different types of communities can be varied. Hence, the future research can pursue a comparative analysis of different types of OHC user behaviors, such as examining the difference of users' participation in diabetes communities and hypertension communities. --- DATA AVAILABILITY STATEMENT The original contributions presented in this study are included in the article/supplementary material, further inquiries can be directed to the corresponding author. --- ETHICS STATEMENT The studies involving human participants were reviewed and approved by the Jiaxing University. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements. --- --- 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. Copyright © 2022 Tian and Wu. This is an open-access article distributed under the terms of the Creative Commons Attribution License . The use, distribution or reproduction in other forums is permitted, provided the original author and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. --- APPENDIX Appendix A | The study's measurement items. --- Dimensions Texts of items Structural capital I maintain close social relationships with other members on OHCs. I spend a lot of time interacting with other members on OHCs. I know some members on OHCs at a personal level. I have frequent communication with other members on OHCs. Relational capital The relationship is characterized by mutual respect between me and other embers on OHCs. The relationship is characterized by personal friendship between me and other members on OHCs. The relationship is characterized by mutual trust between me and other members on OHCs. The relationship is characterized by high reciprocity between me and other members on OHCs. Cognitive capital When interacting with other members on OHCs, we use common terms or jargon. I feel members in the OHCs have values similar to mine. When communicating with other members on OHCs, we use understandable narrative forms. I feel members in the OHCs have experiences similar to mine. Knowledge acquisition I often visit this community to acquire knowledge on a particular health topic. I come to the OHCs to get information on a particular health subject. When faced with health difficulties, I will seek help in the community. I will consider coming to the OHCs to seek related information when I need to know a particular health topic. Knowledge contribution I'd like to share my commentary with other members of the OHCs. I visit this community to share my health skills and experience with other members. I will disclose more health information in the community in future. I visit this community to give other members information I know about a particular health subject. Informational support In the community, some people would offer suggestions when I needed help. When I encountered health problems, some people in the community would give me information to help me overcome the problem. When faced with health difficulties, some people in the community would help me discover the cause and provide me with suggestions. Emotional support When faced with health difficulties, some people in the community comforted and encouraged me. When faced with health difficulties, some people in the community listened to me talk about my private feelings. When faced with health difficulties, some people in the community expressed interest and concern in my well-being. Frontiers in Psychology | www.frontiersin.org

As the national awareness of health keeps deepening, online health communities (OHCs) have achieved rapid development. Users' participation is critically important to the sustainable development of OHCs. Nevertheless, users usually lack the motive for participation. Based on the social capital theory, this research examines factors influencing users' participation in OHCs. The purpose of this research is to find out decisive factors that influence users' participation in OHCs, enrich the understanding of users' participation in OHCs, and help OHCs address the issue of sustainable development. The research model was empirically tested using 1277 responses from an online survey conducted in China. Data was analyzed using the structural equation modeling (SEM). We found informational support and emotional support to have significant direct effects over the structural capital, relational capital and cognitive capital of OHCs. Meanwhile, it is observed that relational capital and cognitive capital degree have a significant influence on knowledge acquisition and knowledge contribution of OHCs. For researchers this study provides a basis for further refinement of individual models of users' participation. For practitioners, understanding the social capital is crucial to users' knowledge acquisition and knowledge contribution that achieve high participation in OHCs.

Introduction Diaspora deals with the problems of a person who feels isolated not only in the place he is working but also within his own family. Isolation does not necessarily mean living separately. It has another face which affects mental health and can't be cured easily. In most families, at least one person, especially, children feel isolated or lack enough love from their family members. Even the family is not dysfunctional, the problem is in this mechanical world which makes humans nothing but a machine. Family means unity. It gives strength to the family members. If one person in the family gets isolated or not given enough love and care then the family becomes meaningless. Even if half of the family gets separated for a few medications for a child it will affect other children in the family as they don't get affection from the parents. The immigrated persons who work for their family in an alien land or people living far from their homeland undergo a lot of struggles and obstacles in their life. It will affect their mental health and daily life. This not only suits migrated people but some unlucky people who live their lives in loneliness even within their own family like immigrated people. For instance, in the novel Drums, Girls, and Dangerous Pie, our hero, Steven Alper, feels alone and isolated from his own family. His younger brother, Jeffrey Alper, is diseased with leukaemia. So, his parents often go out with him for his treatments and medications. The concept of this paper is about a thirteen-year-old teenage boy, Steven Alper, from the novel Drums, Girls, and Dangerous Pie, who is deprived of his parents' love. His mother is always with his younger brother and his father pays less attention to Steven. This novel elucidates the loneliness and the feeling of disappointment in Steven's life which in turn distresses his mental health. --- Review of Literature Deborah Stevenson in her review on Drums, Girls and Dangerous Pie, explores the throbbing and isolated life of the teenager, Steven Alper. She also adds how much the isolation within his home has affected his school life and his passion for playing the drums. Jordan Sonnenblick in his interview on his novel, Drums, Girls and Dangerous Pie, tells that he has experienced isolation and loneliness because of his parent's divorce. And it is reflected in the character of Steven Alper who feels two types of isolation physical and psychological isolation. In Teen Book Reviews, Mary Kates gives her review on Jordan Sonnenblick's Drums, Girls, and Dangerous Pie. She analyzes and describes the life of Steven Alper, a thirteen-year-old boy whose life turns upside down when his brother is diagnosed with Leukaemia. Also, she recommends this book to everyone, especially, young readers which can make them feel all sorts of emotions at once. She also adds that Vol. 13, No. no one knows what is happening in one's house. We don't know their hard feelings or whether they are being isolated even living with a family. Sue Jackson a freelance writer and reviewer has reviewed Jordan Sonnenblick's Drums, Girls, and Dangerous Pie in one of her blogs. She explains the mentality of Steven Alper whose brother has cancer. This book gives a rollercoaster experience of emotions. She describes Steven's pain of isolation and also tells that he yearns for his parent's love and care as they spend their time mostly with his younger brother. Jennifer Rees an editor of "Scholastic Books for Children and Adults" depicts how Jordan Sonnenblick dwells in the mind of Steven who feels isolated in his own house. His family concentrates on his sick brother while Steven yearns for his parents' love. According to her, Drums, Girls, and Dangerous Pie is a story that promises and never disappoints, delivers and never falters. In the article, "Parental coping with children's negative emotions: Relations with children's emotional and social responding", Richard A Fabes et.al., examine the relationship between parents and children. This study also investigates the reaction of parents to children's negative emotions and social competence. The findings of the study concluded that children having distressed parents will have low social competence and become socially detached. Pamela Li in her article, "Coping Skills for Kids -Strategies That Work", tells about coping skills that help children to deal with stress. There are various kinds of stress which children undergo such as conflicts with their parents, academic pressure, bullies in school, problems with friends, etc., The author suggests that children can learn positive coping skills from their parents to avoid such pressures or problems. Tami M. Videon in his article, "Parent-Child Relations and Children's Psychological Well-Being: Do Dads Matter?" talks about the importance of the father's role in adolescents' life and how they influence adolescents' psychological well-being. Grade seven students who are living in intact homes are taken for the study. The results show that adolescents frequently change their behaviour in relation with fathers more than to mothers. This study also emphasizes the unique contribution of fathers to their children's psychological well-being. Yoon Phaik Ooi, et.al., in their article, "The Impact of Parent-Child Attachment on Aggression, Social Stress and Self-Esteem", examines how parent-child attachment influences child behaviour. For this study, they have taken ninety-one boys ranging between 8 to 12 years of age. Through their analysis, they find that the boys who are all having less attachment to their parents are often involved in fighting and teasing others while boys having good parental attachment show high self-esteem and low aggression and they socialize with everyone. --- Methodology Researchers use the Qualitative method to analyze the current study. The selected text for this study is Jordan Sonnenblick's Drums, Girls, and Dangerous Pie. The select text is explored, interpreted and analyzed on the concept of Diaspora concerning the psyche of children. The characters selected for the current study are Steven Alper, the protagonist and his brother Jeffery. It further explores the relationship between parents and children, children and their psyche and children and their association with society. --- Research Gap So far, no research work has been carried out on the novels of Jordan Sonnenblick. There are only a few articles and interviews on the selected text and the author. Various studies have been carried out on the topic of Diaspora. For this study, the researcher is analyzing and interpreting the selected work by incorporating the characteristics of diaspora. --- Theoretical Framework Researchers employ Lazarus and Folkman defined coping as, "constantly changing cognitive and behavioural efforts to manage specific external and internal demands that are appraised as taxing or exceeding the resources of the person" . Coping strategies help to solve problems and reduce stress. According to H.W. Krohne , "Lazarus and co-workers distinguish eight groups of coping strategies: confrontative coping, planful problem-solving, distancing, self-controlling, seeking social support, accepting responsibility, escape-avoidance and positive reappraisal". Social coping is an effective coping strategy identified by Lazarus and Folkman who are pioneers in coping theory. People under mental stress and frustration can use coping methods to overcome their depression. A strategy to relieve themselves from stress by seeking social support is known as social coping. Social coping methods or social support include friends, neighbours, teachers, or any other distant relations, etc. There are various kinds of social coping methods. The two most significant methods are emotional support and informational support. Emotional support is sharing positive concerns for an individual. To a stressor, it offers care, affection and nurturance. Informational support is the timely guidance and advice given to the stressor to solve a problem or to overcome his stress. In this novel, Renee Albert and Samantha help Steven to overcome his loneliness and make him a better person. Renee Albert has helped him to raise funds for his brother's treatment and is a good friend of Steven throughout his struggles. Samantha has made him realize the significance of his brother's feelings and that has made him develop a lovable bond with them. Both the characters Renee Albert and Samantha have helped him emotionally and informatively. Because of these characters, he overcomes the feeling of diaspora which he has experienced within the family. Vol. 13, No. --- Discussion The word 'family' defines loveliness. It is where the beautiful bond of happiness starts. It is where we can find a person who loves us deeply and unconditionally. Parents work hard to bring joy and delight into the life of their children. Their love towards them never fades. Parents should not compare their children with others. If there are two children in a family, they should not show a difference in sharing their love. If one of the children feels deprived of his parent's love it will affect him/her mentally. Diaspora is living far away from the native land. It includes struggles, hardships, and depression people undergo in an alien land. Even if not about immigration but some people feel lonely and isolated within the family which gives them a diasporic feeling. One such character is Steven Alper, a thirteen-year-old teenage boy from the novel, Drums, Girls and Dangerous Pie. It is a heartrending novel written by Jordan Sonnenblick. The themes of this novel include finding ourselves and coping with love and care. It also deals with the hardships faced by young adults in their life. Although it is a heartbreaking novel it acts as a guide for overcoming stress and difficulties in life. Steven Alper, the protagonist of the novel is an eighth-standard student. Having a brother is horrible. Having any brother would be horrible, I suppose, but having my particular brother, Jeffrey, is an unrelenting nightmare. It's not because he's eight years younger than I am, although that's part of it. How would you like to be a King of the Planet for eight glorious years, and then suddenly get demoted to Vice-King?" . He feels irritated because his brother follows him everywhere. His parents shower love on his brother and spend more time with him. This makes him angry and jealous and as a result of his pain, he becomes isolated. Later, when he came to know about the seriousness of the disease his brother is undergoing, he starts to support his brother. Though he understands his brother's sickness still he feels neglected and lonely. It is evident from the following lines, I started crying, but when my mom started coming over to hug me, I ran upstairs for bed. If I had known that this would basically be the last time I'd have both parents paying attention to me at once, I probably would have taken the hug. I could hear my parents talking for a long time before I fell asleep, but nobody came up to check on me. Jeffy groaned in his sleep once or twice in the next room but never woke up. I was alone…. And I muttered to myself in the darkness, Guess what, Mom? I'm going to be the star of night spring concert. As Steven is feeling low and alone, he started to talk to himself in the dark. The isolated feeling makes him more stressed and it affects his psyche. This is because his parents spend less time with him. His father has stopped talking to him. This makes him slowly lose interest in his studies and gets detached from his friends. Because of his lack of interest in studies he gets fail marks in mathematics. Jeffery's condition becomes worse and he undergoes chemotherapy. Though he is too young he is confident and has the mental strength to bear the painful treatments. But Steven who is elder than his brother feels lonely and distressed as he doesn't get enough attention from his parents. The situation of the family becomes worse. They spend money on treatment and medications. In this situation, the family faces a money crisis. Because of this Steven decides to quit drums class. But luckily his drums teacher allows him to continue the classes because playing the drums is the only stress buster for Steven. Renee Albert notices the changes in Steven. She helps him to raise funds for Jeffery's treatment through a musical concert. At a certain stage, Steven feels guilty. It is clear from the following lines, There is another thing for which I will feel guilty forever: Jeffery was driving me so nuts with his need attention that I was often relieved when he went back into the hospital. I mean, I needed the time off from him so I could catch up on schoolwork, but how hideous was that? I wished for my own baby brother to be hospitalized just so I wouldn't have to play some one-sided Chutes and Ladders games. The above lines picture his thinking and it shows how much it has affected his psyche because he has wished a lot for his parent's love and attention. But all his pain has ended in vain. Often, he feels annoyed by his brother. But at some point, he realizes the situation and starts to accompany his brother to the hospital. There he meets a girl named Samantha who is also diseased with leukaemia. She tells him that her sister abandoned her after knowing about her disease. She also tells him how much she is depressed because of her sister's actions towards her. So, she asks him to be with his brother. "Stay with your brother, Steven. Stay with him. No matter what. Do you promise?" . Then only he realizes how important is to look after his brother and not to hurt him anymore. It is through her words he decides not to let his brother feel lonely and be a strong support to him. Timely guidance from the right people helps the stressor to move on the right path and make his life more meaningful. Vol. 13, No. 5;2023 Published by Sciedu Press 442 ISSN 1925-0703 E-ISSN 1925-0711 --- Conclusion The word diaspora may seem simple but it encompasses all sorts of emotions which can change a happy or an active person into a sad or a passive one. A person feels isolated when he or she lacks love from others and longs for a happy life in the past. Normally, diaspora means people living in foreign countries by leaving their native land. Meanwhile, a person can also encounter diasporic feelings even within the family. Mostly, children are the victims in the family. This happens because of parents' carelessness or irresponsibility. Family is a treasure. It is the foundation of positivity, inner strength and delight. It differs from the characteristics of a dysfunctional family. Family should teach morals and good behaviour to children. It should not affect their mental or physical health. Whatever the situation or problem is parents must always share their love equally with their children. Parents should spend time and listen to their children. Parents must be aware of the change in children's behaviour, daily life and also mental health.

Diaspora deals with leaving the native place due to life changes. It not only deals with the external self but also affects the psyche of any person who has moved from their homeland. Home and family are the roots of self-development. It teaches us the qualities of life, real-life struggles, love, and much more of this sort. Home plays a major role in nurturing one's life. It is the starting point of learning all the basic qualities that enhance one's behaviour. The qualitative method is used in this research paper. It attempts to study the psyche of the characters who get isolated and abandoned within the family with reference to the novel, Drums, Girls, and Dangerous Pie by Jordan Sonnenblick. People are in need to earn for their survival. Thousands of people immigrate to other places leaving their beloved families. Though they work hard for the welfare of the family they too own struggles and sufferings. In a few cases, the characters move within the native but they get affected. Even if half of the family separates it affects the other members of the family. Though they live in their native place it feels as if they are in an alien land. The findings of this research paper will establish the positive impact and the power of coping strategies through which every stressor can subjugate any negative state of mind.

Introduction The terrorism threat level in Europe is critical [1,2]. Terrorism is defined as the use of intentionally indiscriminate violence as a means to create terror, or fear, to achieve a political, religious or ideological aim [3]. The concept of radicalisation differs, as there is a temporal dimension: it is a process that may lead to terrorist actions. For Khosrokhavar, radicalisation is a 'process by which an individual or group adopts a violent form of action, directly linked to an extremist ideology with a social or religious political content that undermines the established political, social or cultural order' [4]. Beyond the repressive and security apparatus, many professionals and researchers from the fields of anthropology, political science, sociology, psychology and psychiatry have been involved in understanding terrorism and radicalisation. Attempts to explain terrorism have explored the links among Islamism, the Muslim religion, delinquency and immigration. Sageman, a psychiatrist working for the Central Intelligence Agency, showed that terrorists are educated and mostly from the upper or middle classes [5]. Several literature reviews have confirmed that there is no predefined pathway leading to radicalisation: radicalised individuals come from various backgrounds, have different origins, different family beliefs, social status or gender [6][7][8][9]. The pyramidal model of radicalisation has been widely developed. This model emphasises the idea that only a few individuals would be likely to commit a violent act after undergoing a whole step-by-step process [10][11][12][13][14]. However, various authors have highlighted the difference between radicalised individuals who commit violent actions within a radical group and individualsthe 'lone wolves'-who act in a more isolated manner, who are radicalised more quickly and for whom the pyramidal model does not apply [7,15,16]. The latter represent only a minority of the individuals involved in terrorist activities and are more likely to suffer from psychiatric pathologies [7]. Sageman reported no indicators of mental illness among the terrorists he studied [5]. One may wonder why certain individuals are more likely than others to go through the different steps of the pyramidal model process of radicalisation. Some authors have put forward various predisposing factors, such as depressive tendencies or suicidal thoughts [17][18][19]. The feeling of injustice or humiliation has been also highlighted [20]. Others insist on notions of identity and belonging, emphasising that being part of a radical group and embracing a cause gives a comforting sense of a 'significance quest' around a dedication that has an 'empowerment effect' for the radicalised individual [8,21,22]. Moments of 'existential fragilities' have also been mentioned as elements of vulnerability that can foster radical commitment [23]. Moghaddam noted the importance of the radical group effect, together with the major influence of leaders on the individual throughout the radicalisation process [13]. Literature reviews suggest that the environmental, political, religious, social and cultural context plays an important role, making it difficult to compare the phenomenon of radicalisation from one context to another [8,14]. The studies mentioned above focus on terrorist movements of the 1990s in which terrorists target a foreign country or are fighting for national liberation. However, the understanding of terrorist acts has changed in recent years in Europe with the emergence of 'homegrown' terrorists, born and raised in Europe, who adopt the ideology of violent radical Islamism. The recruitment methods and the methods of action have changed and given rise to new models of radicalisation. Various political scientists, sociologists and governments have made this observation. According to Sageman, the increasing use of the Internet by jihadist movements since the 2000s has led to an organisational change: today's radical groups are less organised and less centralised compared to previous hierarchical organisations such as Al-Qaida. Khosrokhavar also asserts that a new model of radicalisation has appeared in Western countries since the 2010s that is different from the pyramidal model. This new model has the following characteristics: the groups are smaller , more discrete , younger, and composed of more-fragile individuals who have been influenced by recruiters. Although the number of radicalised individuals remains marginal at the level of the general population, it has greatly increased since 2014 under the influence of Islamic State propaganda. In July 2014, the French Ministry of the Interior services listed 899 French people who either joined IS in Syria, returned from Syria, were on their way to Syria or said they wanted to join IS. An increase of 58% in six months was observed between January 2014 and July 2014 [24]. In August 2016, 364 minors were registered with the French judiciary authorities by the police because of objective and worrying signs evoking a radicalisation process [25]. Existing literature reviews on radicalisation have mainly focused on adults and have scarcely explored the question of radicalised adolescents. However, since 2010, it appears that radicalised individuals in Europe are younger than they used to be and that the number of young women involved is increasing [4]. How do these young people shift from a symbolic affiliation with a European country to an organisation that was originally foreign to themselves and that advocates hatred and destruction of the environment in which they grew up? We formulate the hypothesis that there are similarities between the mechanisms at stake during the radicalisation process and the psychopathological manifestations of adolescence: the attraction towards an ideal place and the rejection of their symbolic affiliation could be reflected in the issues of separation and individuation that occur during adolescence and young adulthood. In this systematic literature review, we focused on understanding the profiles of the European adolescents and young adults who have embraced the cause of radical Islamism since the beginning of the 2010s. Various organisations have observed that the profiles of the radicalised have changed since 2010 in Europe, and this new context deserves to be examined. We searched relevant data from medical and psychological search engines and obtained articles from various fields: psychological, sociological, educational, medical and anthropological. We chose to perform a multidisciplinary review, having noted in the scientific debates that each researcher tends to reduce the comprehension of this phenomenon towards his field of knowledge thus providing an incomplete panorama of the phenomenon. This broad approach aims to make a comprehensive inventory in order to return afterwards to a more specific psychiatric or psychological approach. --- Methods We performed a search in the PubMed, PsycINFO and Psychology and Behavioural Sciences Collection databases from January 2010 to July 2017. All papers containing the terms 'Radicali*' or 'Terror*' or 'Violent Protest' in the abstract AND 'Adolescen*' or 'Juvenil*' or 'Teen*' or 'Youth' or 'Young People' or 'Young Person' in the text were identified. Moreover, the French database from the MIVILUDES was screened in order to identify relevant French articles on radicalisation. Because of its experience with sectarian hold, the MIVILUDES was one of the first governmental organisations commissioned to study radicalisation in France, and it played a pioneering role in this field of research. For the selection of the relevant studies, we used the following criteria: the study population included adolescents and/or young adults: subjects aged between 12 and 25 at the time of the radical engagement; the subjects lived in a Western European country: the International Centre for the Study of Radicalisation and Political Violence showed that the majority of the youth who left for Syria and Irak since 2013 came from Belgium, Denmark, France, Germany, Netherlands or Great-Britain [26] the study was about radicalisation at large; the study included empirical data and not only theoretical information or views; the study was recent . The PRISMA diagram flow maps out the number of records identified, included and excluded, and the reasons for exclusion through the different phases of the selection process. Of the selected studies, two co-authors selected the relevant information independently: authors , study design, the number of individuals, age ranges, gender distribution, country, radicalism criteria, features of the sample, assessment tools used in the studies, and the main findings. --- Results In total, we found 22 publications. Their main characteristics are summarised in Table 1. As expected, they came from different fields and used different methodologies . Among the 22 articles, 20 were about radicalisation related with Islam. The two remaining articles concerned affiliation to an extremist ideology or extremist group without any religious specificity or any link with Islam. Regarding the sample of interest, Table 1 distinguishes studies focusing on terrorists or individuals involved in violent actions , studies focusing on individuals who intend to join the Islamic State or present extremist ideology , studies focusing on adherence to radicalisation in general population samples , and studies focusing on theoretical aspects based on single case reports or small series . To obtain a broad view of the complex phenomenon of radicalisation, we decided to present the main results of this review according to 4 axes: the different classifications that have been proposed to delineate radicalised individuals; the individual risk factors of radicalisation; the micro-environmental risk factors of radicalisation; and the societal and cultural risk factors of radicalisation. --- Existing categorisations and typologies Various authors have suggested categorisations of different forms of radicalisation trying to decipher the phenomenon and to identify subgroups or profiles of interests within radicalised youths. These different categorisations are presented in Table 2. These categorisations are not comparable, because of the great variability of methodologies, approaches and populations from which they are developed. However, they portray the complexity of the profiles of these radicalised individuals. To detail one example: the usual categorisation used by law specialists to described terrorists, which distinguishes mainly 2 profiles , does not seem to fit radicalised young Europeans seen in prison contexts, where 3-4 profiles are proposed . Generally, these categorisations do not allow any correspondence with current classifications of mental conditions . --- Individual risk factors Psychiatric disorders are rare among radicalised youths. Individuals with a psychiatric condition represent a minority of the individuals involved in terrorist activities. Bazex and Bénézech found that only 10% of the individuals under the hand of justice for radicalisation presented a psychiatric disorder [27]. Besides, the average age of these individuals with psychiatric disorders is the highest among the 4 categories proposed . For the rest of the individuals, they note various personality traits without a formal psychiatric diagnosis: anti-social, obsessional, and histrionic traits . Rolling corroborates this finding: she reports only 3 individuals with mental disorders among the 25 included in her study [28], knowing that this study is based on a young population received in child psychiatry and not necessarily known to justice for radicalisation. In contrast, several authors describe trait vulnerabilities or psychological vulnerabilities. Several types of psychological vulnerabilities have been highlighted as risk factors for youth radicalisation. A depressive dimension is often reported among radicalised youth, with a frequent feeling of despair that does not qualify as a major depressive episode . Radical commitment can be thought of as a way to fight against this depressivity [28,29] the power of jihadist attraction comes from the 'exalting promise' that offers the adolescent a 'soothing of the pain of living' [29]. Several authors hypothesise that radical ideology could often hide a suicidal intentionality. The notion of martyrdom is attractive because it gives meaning to one's existing fragilities and suicidality. The issue of suicidal intentionality remains complex. While some authors insist that the promise of a 'true' life in the hereafter gives to death a saving status [4,29], others evoke preexisting suicidal intentionality for certain subjects [30]. The relationship to death probably varies from one subject to another and is somehow reflected in the radical ideology that gives death a central place. This echoes Durkheim's notion of 'altruistic suicide' by excess of integration [31]. In total, we lack empirical evidence to address the issue of suicidal intentionality among radicalised youths. A history of addictive behaviour is often reported. Dependence on the radical group may act as a substitute product and replace the previous addiction: some subjects reported that the positive and rewarding effect of religious commitment allowed them to get rid of an addiction to a substance [7,32]. Radical commitment can also be conceptualised as a risky behaviour, with all the appeal it has for teenagers. For radicalised youths, going to Syria represents an initiatory quest [4,33]. Addictive behaviour, risky behaviour and sensation seeking attitudes are common conduct disorders found among delinquent adolescents [34]. Radical commitment channels these behaviours by giving to the subject a firm framework within the group. Finally, early experiences of abandonment are found in most of the radicalised youth's life trajectories [33]. A perceived fragile family structure and painful parental representations are vulnerability factors for youth radicalisation [28,33]. Adolescence per se is a risk factor for radicalisation. Adolescence is a phase of turbulence and reorganisation. For some adolescents, the inherent detachment from primary care givers and finding one's own identity bring a loss of security and sometimes a fear of loneliness and of being abandoned [28,32]. Belonging to a radical community conveys a sense of belonging, a sense of meaning and comfort [35]. Based on the psychoanalysis of a radicalised subject and the content of jihadist propaganda, Leuzinger-Bohleber claims that the IS permits the satisfaction of pre-genital drives, which are rekindled in the early phases of adolescence [36]. Violent actions advocated by radical groups unconsciously offer an enormous satisfaction of archaic drive-impulses and can be experienced as an omnipotent victory over the fear of death [36]. Also, finding love objects outside of the family is another major issue of adolescence and is simplified by the organisation, which guarantees a reassuring marriage [36]. Motivation towards marriage is found more frequently in young girls than in young men who are generally more attracted by the fascination exerted on them by armed combat [30,37]. The changes in identifications during adolescence and the quest for an ideal open the way to radical ideologies [32]. Thus, the message sent by the IS may become attractive for some adolescents. The adolescent characteristics of the radicalisation process had not been put forward in the terrorist movements of the 1990s and 2000s. Bazex and Bénézech showed an age effect on the reason of justice control: subjects under judicial control for acts of apology or for acts of terrorism are much younger than those condemned for ordinary law crimes. On one hand, there are young people for whom radical engagement leads them to be under judicial control, and on the other side there are condemned adults who meet, by availability, radicalisation in prison. Individuals in prison can encounter radical commitment in a very variable way [27]. Personal uncertainty is another individual risk factor for radicalisation. Identity and identification issues and idealisation processes are central during adolescence. For several authors, these issues are at the core of the radicalisation process. The neoidentity associated with the radical group and ideology may give a new and reassuring meaning to the young person's experience [32]. Some add a dimension of imaginary affiliation to fallen Muslim origins that the subject glorifies again through his/her commitment [29]. Radicalisation is viewed as an 'act of recovery of identity' or 'recovery of lost dignity' [37]. Members identify with the leader's power and prestige which compensate for the failures of their individual narcissism [7]. Theories of narcissism and grandiosity in groups highlighted the fact that the figure of the leader and the ideology becomes for the members of the group their ego ideal [38]. This mechanism acts as a megalomaniac protection against anguish that reinforces the subjects' fantasies of immortality [38]. Middle-class youths can also experience this feeling of 'victimisation' by suffering from anonymity and nonbelongingness. The latter could therefore also engage in jihadism because of a lack of identity firmly anchored in reality. The fact that adolescents from middle class and without family affiliation with Islam undergo radicalisation process underlines the importance of 2 case reports of radicalised adolescents who do not come from a Muslim family. 2 Profiles: The Male adolescent is 14 years old and the Female adolescent is 15. 1. Young Muslim who hates society 2. Young middle-class convert who breaks with his original filiation the adolescence process in this choice of radical engagement [32]. For Doosje, personal uncertainty is one of the three main determinants of a radical belief system, along with perceived injustice and perceived intergroup threat [39]. This finding is based on Hogg's 'uncertainty-identity' theory: the more individuals are uncertain in their environment, the more likely they are to identify themselves massively with groups [40], and the more the properties of this group form a unit where individuals seem interchangeable, the more effectively this group reduces uncertainty [41]. Perceived injustice, or the feeling of injustice, is another determinant of radicalisation. It is cited in several studies, whatever the vocabulary or the methodology used: 'Perceived injustice' [39], 'Perceived Oppression' [14], 'Frustration' [27], 'Despair', 'Sense of Unworthiness' or 'Perceived contempt' [37]. All these terms illustrate a deep malaise of the subject who tries to give sense to this 'existential failure'. This 'injustice' is often put forward by the radicals themselves to justify their commitment and designate the culprits. Benslama also explains that this feeling of injustice, often provoked by personal experience, is then covered by radical ideology, giving a stronger value to its commitment [29]. The link between perceived injustice and the reality of a socioeconomic discrimination is developed below . Several authors state the importance of a triggering event as a determining factor in acting out or at least reinforcing the radical commitment. The list of events includes the occurrence of a brutal trauma concerning a loved one [30,42]; the occurrence of a love disappointment [33]; the viewing of a video of a battered woman who reminded him of his mother [42]; or the viewing of a video that reactivated a profound suffering linked to family history [33]; and a recent experience of discrimination or the diffusion of violent and unsustainable videos [30]. Finally, several authors have described psychopathological mechanisms that are at stake during radicalisation and that reinforce radical engagement. The idea here is to explain the psychological mechanisms involved once the individual has started the process of radicalisation. First, projection is active: radical ideology offers the subject the possibility to project onto his enemies all the evil he feels inside and to soothe his guilt by what is designated as an act of purification [43]. This operation may be accompanied by a paranoia mechanism and a use of splitting of the ego. This paranoia mechanism acts as a defense mechanism and does not account for a psychiatric pathology or personality organisation [7,27,29]. The splitting process is highlighted by several authors and explains how these individuals can set aside the moral values they had in the past [15,30,43]. Various authors have reported that obsessive compulsive habits are also frequent among radicalised youth [29,30]. These symptoms often have a function of 'purification', such as found in radical ideology or the practice of a rigorist religion. Radical engagement may soothe pre-existing anxiety symptoms by offering a reassuring framework. For Khosrokhavar and Ludot, today's European jihadists of the middle classes suffer from a lack of authority and are in need of an authoritarian normativity that shows them rules to follow [32,37]. These elements show the importance of the radical group that welcomes the candidates to radicalisation, and they suggest how the influence of the micro-environment is decisive in the process of radicalisation. --- Micro-environmental risk factors Fragility and failure of the family group is a risk factor for radicalisation. Studies focusing on the families of radicalised youths often portray deficiencies, traumas and/or distress during the childhood and adolescence of these subjects [4,28,33]. In Bazex and Bénézech's study, which included 112 individuals under judicial control for radicalisation, they report a large proportion of individuals who experienced 'a childhood marked by significant parental difficulties, a father often absent and a mother whose integrity is often attacked ' [27]. In their view, this contributes to a major inconsistency in identificatory processes . Among the members of the Hofstadgroup, a Dutch terrorist organisation, Schuurman found that one of the risk factors that makes the radical group particularly effective is the isolation of the subject and the absence of countervailing opinions in the family [42]. The lack of a corrective answer to the subject's radical positions was also reported by Van San et al among 16 Islamists and extreme right adolescents who exposed on the internet their radical ideas [44]. He described permissive parents with little response to their child's radical opinions. Friendship or admiration towards a member of the radical group is often reported. This seems obvious in studies focusing on the mostextremist cases seen in judicial contexts. Most of the radicalised individuals had a role model, an inspirational figure in the radical group who initiated the process of radicalisation. Schuurman highlighted this finding for individuals among the 'Hofstadgroup' in Holland, and Bazex confirmed it among French radicalised youths returning from Syria [33,42]. Several authors have found similarities between the techniques of radical jihadist groups and sectarian community methods to recruit new members. Based on his knowledge of sectarian methods, Dayan described common mechanisms such as narcissistic gratification, moral debt, real or imagined threats and the progressive setting aside of the family/friendship network [43]. Rolling described two phases in the radicalisation process of the youth encountered in child psychiatry units: a first phase during which the radical commitment has a soothing function on psychic suffering and a second phase during which the ideological indoctrination leads to the self-effacement of the subject in favour of the group [28]. Bouzar studied the way the youth was enlisted on the internet and explained that the 'recruiters' of radical groups adapt the propaganda media to the 'sensitivity' of each subject [30,45]. Youths are subject to psychological pressure from the recruiters to adopt the radical beliefs and to break with friends and family [46]. Bazex & Bénézech highlighted one category of radicalised individuals who have a charismatic character associated with proselytic and manipulative behaviours: 'the proselytic networker' [27]. Radicalised individuals with other profiles may be vulnerable to indoctrination by 'proselytic networkers'. Khosrokhavar also reported this form of relational dissymmetry between a charismatic individual who exerts an influence on more-fragile individuals [4]. The radical group instigates the subject's dehumanisation. This is a key process to understanding how mainly normal individuals can engage in terror activities. For Bazex & Bénézech, the radical group and ideology help to legitimise a pre-existing violence in the subject [27]. The aggressive and destructive instincts that dominate the fantasy life of these individuals find a way of justification and ideal expression in projection towards the outside. Dayan states that there is a 'narcissistic contract' between the individual and the group and that this contract is crucial: the group offers a place and a role to the subject, and in exchange, he must repeat the same statements and ensure the permanence of this transmission [43]. Khosrokhavar studied the case of young Europeans who wanted to go to Syria [4]. He explained that this desire was more often driven by humanitarian concerns than by violent radicalism. He described these young people as 'pre-radicalised' and explained that once within the group, they learn to become insensitive to the suffering of others and becoming 'real jihadists'. Bouzar identifies four steps used by recruiters in order to effect the 'dehumanisation' of the subjects: isolation of the individual from his environment; destruction of the individual as a unique personality for the benefit of the group; adherence to the IS ideology; dehumanisation of the subject and of his future victims [46]. These steps are consistent with Bandura's work on 'moral agency' and dehumanisation: 'The moral disengagement may center on the cognitive restructuring of inhumane conduct into a benign or worthy one by moral justification, sanitising language, and advantageous comparison; disavowal of a sense of personal agency by diffusion or displacement of responsibility; disregarding or minimising the injurious effects of one's actions; and attribution of blame to, and dehumanisation of, those who are victimised' [47]. These mechanisms are also described by Zagury [15] and Bouzar [46] when they explore the link between the 'recruiters' and the youth exposed to radical theories. --- Macro-environmental risk factors The evidence of macro-environmental factors is more complex. Social polarisation is one such risk factor. Unequal or discriminatory socio-economic conditions are highlighted by many authors as contributing to the phenomenon of radicalisation. Khosrokhavar explains that the 'ghettoized existences' associated with the sense of dehumanisation subjects feel through social disregard lead them to the desperate conviction of being in a deadlock [4]. When this conviction is not associated with an ideology, these young people may take the path of delinquency. If this conviction finds religious ideological support, the hatred of society is sacralised and the goal becomes to 'save Islam'. Khosrokhavar noted that most French people who have committed terrorist acts were born in disorganised families, had grievances against society, had a feeling of social injustice and a denial of identity [4]. Moyano studied Muslim and Christian high school students in Spain and showed a significant polarisation between the two groups when they came from the same neighbourhood [14]. When asked what defines them best, Muslims mostly chose identification with religion , while Christians chose their country. Similarly, French authors reported that young people who have returned from Syria or young people under the grip of justice for radicalisation often had difficulties in social integration, particularly at school, with a history of exclusion [27,33]. Doosje et al identified perceived group threat as a major contributor to maintaining a radical belief system [39]. The perceived group threat may take three different forms: symbolic threat, realistic threat, and intergroup anxiety. Symbolic threat refers to threat to the Islamic culture. Realistic threat refers to threat to the economic status of one's group. Intergroup anxiety is defined as the fear one can experience when one has to interact with a person from another group. Once again, these results underline the importance of group polarisation and its consequences for feelings of oppression or perceived injustice. Bhui supports the use of a public health approach to understand and prevent violent radicalisation, arguing that anti-terrorism approaches based on the judicial system or on criminology do not prevent radicalisation, since they intervene after the act [48]. These latter actions would even tend to stigmatise Muslim communities and undermine social cohesion, while better social cohesion is associated with a reduction of violence, better public health, and a more equal and just society. To support this view, Bhui et al. studied vulnerability factors and resistance factors to violent radicalisation in a cross-sectional survey of a representative population sample of men and women aged 18-45 of Muslim heritage in two English cities [49,50]. They showed that sympathy for violent protest and terrorist acts was more likely to be articulated by those under 20, by those in full time education rather than employment, by those born in the UK, by those speaking English at home, and by high earners . Anxiety and depressive symptoms, adverse life events and sociopolitical attitudes showed no associations. On the other hand, resistance to radicalisation measured by condemnation of violent protest and terrorism was associated with a larger number of social contacts, less social capital, unavailability for work due to housekeeping or disability, and not being born in the UK [49,50]. Interestingly, here again, sympathy was higher among the youngest individuals and those born in the UK rather than immigrants [49,50]. In another cross-sectional study in the UK, Coid et al showed that men at risk of depression may experience protection from strong cultural or religious identity [51]. These results seem paradoxical and highlight the importance of a consideration at all three levels : some factors can act either on the side of protection or of vulnerability to a radical system. The link between religious fundamentalism and radicalisation is complex. Many authors mention the importance of religion in the process of radicalisation. Moyano reported that Muslim religious identity was stronger than Christian identity in Spain [14]. Dayan noted recent claims of religious identity among French youths [43]. Coid showed that religious affiliation and practice were protective against antisocial behaviours but could determine targets of violence following radicalisation [51]. Benslama emphasised the importance of the history of the Arab world and Islam in explaining the discourses of Islamic fundamentalists and the creation of a new identity figure that he called the 'super-Muslim' [29]. According to Benslama, there is a 'constraint under which a Muslim is led to outbid the Muslim he is by the representation of a Muslim who must be even more Muslim'. The subject is encouraged to identify himself with 'the perfect Muslim, the Prophet and the ancestor', according to the belief that 'the good has already happened, the promise has been fulfilled, there is nothing left to do but to return to the past, while waiting for the end of the world, or better: to hasten it' [29]. Benslama illustrates how religious fundamentalism provides an identity figure in which youths can find a totalising meaning. It is crucial to differentiate Muslim religion from radical Islamist ideology. Some political scientists have argued that for ISIS, religion is more a way to justify their actions, rather than a cause to defend, unlike other forms of religious fundamentalism [52]. The geopolitical context also has a major influence. Several authors have noted that the proclamation of a caliphate by the IS added legitimacy to this radical ideology [4,29,30]. Contrary to the Al-Qaeda network, the IS is anchored in a territory, and its propaganda describes the territory as an idyllic place, a community utopia where the Muslim 'brothers' are most welcome. Different terms have emerged, conveying the message of ideological propaganda associated with this territory: the 'Hijra' , the 'Land of Al-Sham' or 'Ummah' . Geopolitical events undoubtedly influence the phenomenon of radicalisation whether in its rise in 2014 during the proclamation of the caliphate or during its decrease in intensity since the IS was defeated in Syria and Iraq in 2017. Societal changes may affect the radicalisation process. According to various authors, the mutations of modern societies may favour the appearance of the phenomenon of radicalisation. Khosrokhavar uses Durkheim's term 'Anomie' [53] to explain that thedissolution of the moral, religious or civic values of modern societies leads to a feeling of irresolution that makes youths more inclined to turn towards religious fundamentalisms and possibly towards radical ideologies [4]. These considerations are in line with Zygmunt Bauman's more modern concept of 'liquid modernity', according to which the individual is integrated only by his act of consumption [54]. Social status, identity or success are defined in terms of individual choices and can fluctuate rapidly depending on socio-economic requirements. This fluid modernity brings at the same time freedom but also uncertainty and insecurity. Khosrokhavar [37] adds that among these young people, who suffer from the 'deliquescence of politics' and the 'dispersion of authority between several parental bodies', radical Islamism offers tangible and reassuring norms carried by an unequivocal authority. Benslama also discusses this change in the social model [29]. He explains that the traditional model, in which filiation defines the identity of the subject, is endangered in favour of a new social model where each subject needs to forge his own place and identity, with all the anxiety it can generate. Some individuals have the resources to face this challenge, but those who fail to do so may be tempted by the solutions offered by the radical ideologies by attacking the society that has placed them in this insecure situation. Radicalised individuals believe that radical ideology, by destroying the existing society, offers them the possibility of a promising new societal model. Also, the society is believed to convey all kinds of conspiracies by forces that stole power from the people , and IS recruiters use conspiracy theories to destroy the world of candidates to radicalisation to make them commit to a new one [30,45]. --- Discussion --- Three-level model for the radicalisation process among European youths In an attempt to summarise the main results of this review, we have developed a three-level model to explain the phenomenon of radicalisation among young Europeans since 2010. This model is shown in Fig. 2, and it follows the proposal of Doosje et al. [55]. However, it also includes some differences and the idea that some factors should be regarded as interactive factors between an individual who commits to radicalisation and a recruiter who tries to favour this process. As in Doosje et al.'s model, we distinguish individual, micro-environmental and macro-environmental factors . We preferred to use the usual distinction employed in child and adolescent epidemiology of at-risk behaviour [56]. The red circle encompasses the different factors that show the interaction between the subject and the radical system whatever the level: mechanisms at stake during the process of radicalisation at the individual level, similarities with the sectarian communities and the use of dehumanisation to justify the use of violence at the micro-environment level, and the proposal of a new societal model at the macroenvironmental level. --- Is it possible to formulate recommendations? The multifactorial aspect of the radicalisation process implies that the proposals imagined to prevent this phenomenon are varied. Using the three levels previously described, we synthesised the different recommendations that have been formulated inside and outside the literature review in order to prevent radicalisation. We also emphasised the proposals that have been discussed through empirical data. At the individual level, we only find two empirical studies describing a de-radicalisation program. The first study used an anthropological perspective with a large number of French individuals [30]. They proposed a combination of socio-educative support, therapeutic groups of de-radicalisation to help individuals experiencing novel emotional associations with previous life trajectory, and family therapy [46]. No outcome data are available to date. The second study was based on a Dutch radicalisation prevention program, the DIAMANT program [57]. The training consists of three modules conducted --- INDIVIDUAL FACTORS MACRO -ENVIRONMENTAL FACTORS --- MICRO -ENVIRONMENTAL FACTORS --- Psychiatric condiƟon --- Psychologic vunerability --- Experiences of abandonment --- Personal Uncertainty --- Adolescence --- Perceived injusƟce Triggering event over a period of 3 months. The first goal of DIAMANT is to help participants find a job, internship, or education in order to reduce feelings of relative deprivation and social disconnectedness. Regarding prevention, Feddes et al. reported a longitudinal study showing that improving empathy and agency skills, as well as balancing self-esteem , could prevent the risk of violent radicalisation [57]. Based on case reports and series, Ludot et al reported the importance of mental health professionals meeting these young people to analyse their psychological vulnerability, taking into account their political, social and cultural context, and to try to grasp the meaning radical commitment holds for them [32]. Zagury insisted that it is important for mental health professionals to offer young people psychological support without judgment [15]. The lack of judgment from professionals is a necessary condition for deradicalisation. Bazex & Bénézech point out that for mental health professionals in prison, it is crucial to search for dissimulation strategies, because they hinder psychotherapeutic work [27]. --- Mechanisms at stake during radicalisaƟon --- Friendship with radicalised individuals --- SimilariƟes with sectarian hold --- Family dysfuncƟon --- DehumanisaƟon to justify the use of violence --- Societal polarisaƟon --- Religiosity --- GeopoliƟcs --- Proposal for a new societal model At the micro-environment level, Van San et al. underlined the benefits of a pedagogical approach, in which a benevolent educator accompanies a young person by setting limits for him while offering him a discourse that counterbalances extremist ideologies, an approach that could prove to be relevant for prevention [44]. Khosrokhavar proposed the formation of groups of imams, city officials, police officers, neighbourhood authorities and psychologists in order to show to youth accustomed to violence that a constructive dialogue exists among these people [4]. According to him, repression will never be enough to combat these violent ideologies, so a preventive approach is essential. Leuzinger-Bohleber reported the case of a radicalised young man who identified a determining event in his distress: a teacher he appreciated and identified as a father figure humiliated him in class [36]. Although the case per se illustrates a triggering event, she proposed setting up supervision groups for teachers, educators and social workers to raise their awareness of transference phenomena. This supervision would strengthen the ability of these professionals to offer the youth a sense of belonging and anchoring in another group than the radical group. We believe that supervision and formation should not be limited to transference phenomena but should also include social exclusion phenomena of all kinds, including first signs of radicalisation [58] and in-group out-group perspectives [59]. At the macro-environmental level, Bhui et al. showed that there are modifiable risk and protective factors for the earliest stages of the radicalisation process [49,50]. They proposed a global approach centred on social inclusion/exclusion, cultural identity, acculturation, stigmatisation and political commitment [48]. The fight against the emergence of radicalisation also involves the prevention of major depressive symptoms and the promotion of well-being and social capital. The fight against these societal risk factors would, in any case, be beneficial on a large scale, since these risk factors are also those of violence and poor health in general. Moyano explained that the polarisation between Christian and Muslim groups is certainly the 'antechamber' of radical commitment [14]. Thus, he insisted on the need to pursue a policy that fights against these in-group polarisations reinforced by sociocultural inequalities. Leuzinger-Bohleber emphasised the importance of defending freedom of expression and open dialogue around societal issues such as migration, trauma or sexuality in order to offset simplistic radical ideology [36]. Khosrokhavar wondered whether non-violent Islamist fundamentalism could be a protective factor against radical violence [4]. For him, it is important to fight against the amalgam and to mark the difference between religious fundamentalism and radical violent jihadist movements. --- Implications for adolescent psychiatry The literature review points out that radicalisation cannot be directly linked to mental illness. Individuals with psychosis are the exception rather than the rule. The resulting psychiatric treatment are suitable only for a minority of situations and there is no specific profile leading to radicalisation. However, the multiple psychological vulnerabilities as well as their similarities with the psychopathological mechanisms of adolescents with mental and behavioural difficulties suggest a key role for adolescent psychiatry in terms of secondary and tertiary prevention. The recent report published by the French Psychiatric Federation on the link between radicalisation and psychiatry thoroughly supports the role of psychiatry [60]. The authors explain: there is 'an encounter between a banal process of painful adolescence and the power of the religious who, by abolishing doubt, meets the conditions of an efficient narcissistic repair'; and 'there is no specific psychopathology, but there is a specific form of expression of a common psychopathology' of the adolescent period. As a consequence, mental health professionals can intervene in two ways to face radicalisation [28,32,60]. First, they have a role to play in helping adolescents or young adults to find the meaning the radical commitment has for themselves. Psychotherapeutic intervention can enable the young person to understand his personal functioning, to adopt a different point of view, beyond the political, cultural or judiciary approaches, and to find other paths than radicalisation. The links between radicalisation and adolescence psychopathology invite to consider this phenomenon from the eyes of psychiatrists or psychologists familiar with adolescence issues. Second, mental health professionals could be involved in deradicalisation programs in order to drag subjects out of their radical engagement. Knowing the specificities of adolescents and young adults in psychological treatment, they can propose adapted therapeutic devices which are 'likely to rehumanise the functioning of the subject by reducing the factors which induced its narcissistic regression and its reaffiliation in the process of radicalisation' [60]. We believe that this dynamic apprehension of the adolescent mind offers a way out of the radical commitment for a number of young subjects, whose dehumanisation and dilution within the radical group is not yet too advanced. --- Study limitations and future research Radicalisation and terrorism are major issues in our society. This article is the first multidisciplinary systematic literature review focusing on the radicalisation of young people in Europe since 2010. It tried to encompass both qualitative and quantitative studies. There is no decisive explanatory factor for radicalisation but rather a multitude of vulnerability factors that are difficult to handle in a prevention perspective. We developed a comprehensive three-level model of the radicalisation process. We believe that this broad three-level vision can be useful in understanding and apprehending the complexity of the phenomenon. Although this openness is necessarily weakening in terms of theoretical consistence of our subject, it seemed necessary to offer a holistic model in order to bring together the different fields working together on such an important subject. Besides, the fact that there is no explanatory factor is certainly explained by the epiphenomenal aspect of radicalisation whose sudden appearance is part of a complex entanglement of contextual factors. One of the specificities of adolescent psychopathology is to manifest itself in the social field [61]. However, this literature review has several biases due to the complexity of the phenomenon and the variety of youth characteristics. First, there are few empirical studies on the subject, and the inclusion criteria of the populations are highly variable . Most studies are qualitative studies, with a small number of subjects. This raises questions about the representativeness of some studies or the generalisation of some results outside the specific context in which the studies were performed. Second, many studies focused on subjects who had committed violent actions and who had a judicial history, which is not representative of today's radicalised population. Third, the majority of the subjects included in the studies are men , while it is known that the number of young radicalised women is significant. Further research is needed to broaden the scope of subjects to radicalised young women and to radicalised subjects who never committed violent actions. Finally, although our aim was to focus on young individuals, many studies selected also included adults older than 25 years questioning the generalisation of the results. It is also fundamental to develop more empirical research using multidisciplinary approaches to the phenomenon. Different scales have been developed in order to assess sympathy for radical movements in the general population , but none of them has been validated. It would be useful if field professionals who have met radicalised youths were to develop an evaluation grid specifically designed for individuals perceived as at risk by the healthcare, judicial and penitentiary institutions. It is crucial to develop specific tools to study the phenomenon and to evaluate programs aimed at preventing radicalisation in order to combat this phenomenon that threatens our society.

Background: A new model of radicalisation has appeared in Western countries since the 2010s. Radical groups are smaller, less hierarchical and are mainly composed of young, homegrown individuals. The aim of this review is to decipher the profiles of the European adolescents and young adults who have embraced the cause of radical Islamism and to define the role of psychiatry in dealing with this issue. Methods: We performed a systematic search in several databases from January 2010 to July 2017 and reviewed the relevant studies that included European adolescents and/or young adults and presented empirical data. Results: In total, 22 qualitative and quantitative studies were reviewed from various fields and using different methodologies. Psychotic disorders are rare among radicalised youths. However, they show numerous risk factors common with adolescent psychopathologies. We develop a comprehensive threelevel model to explain the phenomenon of radicalisation among young Europeans: (1) individual risk factors include psychological vulnerabilities such as early experiences of abandonment, perceived injustice and personal uncertainty; (2) micro-environmental risk factors include family dysfunction and friendships with radicalised individuals; (3) societal risk factors include geopolitical events and societal changes such as Durkheim's concept of anomie. Some systemic factors are also implicated as there is a specific encounter between recruiters and the individual. The former use sectarian techniques to isolate and dehumanise the latter and to offer him a new societal model. Conclusion: There are many similarities between psychopathological manifestations of adolescence and mechanisms at stake during the radicalisation process. As a consequence, and despite the rarity of psychotic disorders, mental health professionals have a role to play in the treatment and understanding of radical engagement among European youth. Studies with empirical data are limited, and more research should be promoted (in particular in females and in non-Muslim communities) to better understand the phenomenon and to propose recommendations for prevention and treatment.

Introduction It is now widely recognised that family caregivers looking after patients with advanced cancer play an essential role by helping them meet both their medical and nonmedical needs [1][2][3][4][5][6][7][8][9][10]. Family caregivers' experience has been studied mainly from the perspective of the psychosocial cost of caregiving to a family member living with a patient with advanced cancer [10][11][12][13][14][15]. Family caregivers' quality of life usually deteriorates [16][17][18], as does their physical and mental health [19][20][21][22][23][24][25]; they also sufer fnancially, professionally, and socially and experience occupational problems [26,27]. To alleviate the burden of caregiving, as well as caregivers' loneliness [28], several psychosocial interventions designed to improve the quality of life of adults with cancer and their family caregivers have been developed during the last few decades [29][30][31][32][33][34][35][36][37]. Some of them have targeted various forms of psychosocial support for family carers in charge of palliative care patients, including psychoeducation, psychosocial support, symptom management, family meetings, and training in carer coping and means of improving sleep [38][39][40][41]. A deeper understanding of caregivers' burden is essential in order to enhance the benefts of these interventions. It was therefore proposed to complement and extend previous fndings on caregivers' burden by analyzing more closely the content of what has been called "the soul of care" [42] and the moral dimension of this interpersonal experience. Tis qualitative study based on an anthropological approach to lived experience [43] was conducted in France with patients with advanced cancer and their family caregivers. We adopted the defnition for experience proposed by Kleinman and Seeman [44]: "the intersubjective, felt fow of events, bodily processes, and life trajectory which always take place within a social setting." Te aim of this study was to explore the socioeconomic and cultural factors responsible for shaping the complex personal experience [9] of family caregiving, the meaning given to these caregiving practices [45] by the caregivers and the patients, as well as the hitherto undervalued visible and invisible aspects of family caregiving, especially the various skills and moral qualities required. For this purpose, we used Tronto's "ethic of care framework" [46], in line with previous authors in the feld of health research [47,48]. According to Tronto, caring is a practice and a process consisting of four interrelated phases: caring about ; caring for ; care giving ; and care receiving . Each of these phases requires moral qualities: caring about requires attentiveness ; caring for involves responsibility ; care giving requires competencies that are not only technical issues but also moral ones; and care receiving requires responsiveness on the part of the recipient of the care. It was therefore proposed to analyse both patients' and carers' lived experience of informal home care using Tronto's "ethic of care framework" as a guide to identifying the various levels of responsibility, the relationships and efects at work, and the moral dilemmas with which primary caregivers have to cope in order to provide appropriate quality informal home care. --- Methods --- Study Design. In this qualitative study, in-depth faceto-face interviews were conducted with patients and family caregivers from October 2017 to March 2020 with the approval of the French Institute of Medical and Health Research Ethics Committee and the Institutional Review Board . --- Sample. Based on a purposive sampling strategy, 20 patients and 19 primary caregivers were interviewed . Te enrolment process stopped when data saturation was reached [50]. Te patients were recruited with the help of oncologists and palliative care specialists at two comprehensive cancer centers in the South of France. Tey were all treated with frst-line chemotherapy for locally advanced or incurable metastatic cancer, age ≥18 years, life expectancy ≥1 month, and performance status ≤ 2. Fourteen of them had been referred to outpatient palliative care services. Family caregivers were defned here as any family member or friend who usually took care of these patients and provided them with help and moral support. Tey were designated by the patients interviewed. Te main exclusion criterion adopted for both patients and caregivers was the presence of severe cognitive impairments. In-depth face-to-face interviews were carried out with 15 caregiver/recipient dyads either separately or both together . In addition to these dyads, 5 patients and 4 caregivers were interviewed who were involved in caregiving relationships but not with each other . None of the patients subsequently refused to participate in the study, withdrew their consent, or dropped out. However, a few caregivers refused to participate. --- Data Collection. Te interviews were conducted in French by an experimented medical anthropologist . Te respondents chose the place where the interviews were to take place: the patient's room, a neutral hospital setting, or their own homes. Te carers' interviews lasted 43 minutes on average . Te patients' interviews lasted 33 minutes on average . All possible precautions were taken when discussing painful situations with patients and carers to avoid upsetting them. We applied the ethical principles of research in the feld of social science, namely, guaranteeing the anonymity of the data collected, the informed consent of the respondents, the possibility of ending the interview whenever they wished, and presenting them with the results of the study. Te semistructured interview guide used here covered the following topics: respondents' knowledge and understanding of the disease and its prognosis; patients' care pathway since the onset of the disease; the various tasks involved in informal caregiving; and the reasons why these caregivers undertook these tasks. In addition to these themes, the interviews with caregivers covered topics relating to the caregivers' psychosocial burden: the signifcance of looking after an ailing relative; what aspects of the caregivers' everyday lives were afected by this work; and the efects on caregivers' own well-being, mood, quality of life, fnancial resources, and family relationships. Te interviews were audio-recorded and transcribed. All personally identifable details were removed from the transcripts. In the excerpts from interviews quoted here, patients were coded P plus a number, whereas carers were coded C plus a number. Tese excerpts have been translated by a professional translator . --- Data Analysis. An inductive thematic approach and an iterative process [50] were used to analyse the interviews. Te frst step was to code the interviews. Te transcripts were coded manually by two coders working independently; the coding of each transcript was compared, and any discrepancies between the two were discussed until the fnal coding scheme was consolidated. Te data were managed with Excel software. We applied the following strategies to assure the rigour of the analysis [51]: saturation, refexivity, and a peer review. Data collection and analysis were conducted simultaneously and until no further news items emerged [50]. Core categories and themes were discussed at interdisciplinary team meetings and examined in relation to the existing literature in order to spot any inaccuracies and misinterpretations. Preliminary results were presented at two research conferences to obtain feedback from peers. Te last step of the analysis was to refne the thematic fndings using Tronto's "ethic of care" framework [46]. Tronto's four moral qualities required for caring were associated with each category of task identifed during the frst stage of the coding process. Figure 1 represents these associations. --- Findings --- Attentiveness to and Empathy with Patients' Experience. Family caregivers provided logistic assistance with the aspects of daily life with which patients could no longer cope themselves or only with difculty, such as cooking, housework, shopping, and administrative tasks. Attentiveness, i.e., attending to their loved ones' need for care and anticipating their medical and nonmedical needs, was frequently mentioned as a means of improving the patient's living conditions, as well as providing emotional support: "I help him psychologically as well as doing the shopping and the housework, disinfecting everything and keeping him clean I actually help him all day long When he perspires, I bring him a damp cloth, and when he vomits, the bucket is full so I go to empty it in order to get rid of the smell, then I disinfect it It's just improving his everyday living conditions. " Psychological support could take many forms: accompanying patients in activities such as walking, distracting them from their disease by engaging in relaxing activities, and staying by their side in their everyday lives and their medical care to show that they are not alone and not abandoned. Tis presence was perceived by caregivers as a means of alleviating the patients' anxiety and emotions to make them feel better. Almost all the interviewees mentioned the emotional support involved, which was perceived by both patients and caregivers as being the most important and most valuable aspect. [46], attentiveness requires suspending one's own goals, ambitions, plans for life, and concerns in order to meet the needs of others. Tis is what the family caregivers interviewed did. Te time devoted to informal care was taken at the expense of the carers' own professional, family, and leisure activities. European Journal of Cancer Care 3 Some caregivers were obliged to reorganise their working, social, and family lives. As one caregiver said, "I have put my life on hold" . Although most of them did not complain, some of them expressed the feeling that their lives had been invaded. Among the 19 family caregivers interviewed, 13 were unemployed or had retired by the time of the interview and 6 were still working. Only three of them were able to reduce or postpone the periods spent at work. Te other three stopped their occupational commitments and devoted their time entirely to their ailing relatives. --- Downside of Attentiveness. According to Tronto However, although informal caregiving was a timeconsuming activity, some respondents perceived informal care as an opportunity to spend more time with a parent whose life expectancy was shrinking: "[Caregiving is] helping her with her formalities, accompanying her to see the doctor, taking her to the hospital, enjoying the time left to her, that's it . . . Yes! It's important to be with her, yes! I feel I am lucky despite my misfortune because I can enjoy being with her now, before she passes away! " Some caregivers declared paradoxically that this time of care was a "lucky opportunity" to stay with their loved ones during their last few weeks or months of life. --- Taking Responsibility for Patients' Health. Family caregivers took on the responsibility for meeting the patient's identifed needs, especially the health-related ones, by coordinating various tasks as well as ensuring continuity of care and ftting patients' and medical professionals' timeschemes. In addition to making medical appointments, almost all the carers questioned said that they often accompanied their loved ones on medical visits, at admission to hospital or for chemotherapy sessions. Although the cost of patients' transport from their homes to hospital for the latter reasons is entirely covered by the French public health insurance system, some caregivers preferred to drive the patients to hospital themselves. Even when the patients travelled by ambulance or cab, the caregivers stayed with them. Teir presence on these occasions mainly meant giving the patients physical and emotional support. Support of this kind gave these carers an additional facilitating role in the management of their relative's disease, by informing the hospital doctors about the patient's state of health and about any after-efects of the drugs prescribed, as well as making up for any forgetfulness on the patient's part. Staying with the patient during chemotherapy sessions also gave caregivers an opportunity to spot and deal with any defcits in the coordination of their medical care: "Te last time , we didn't have the right prescriptions for the blood tests, the right nurse, or the right drugs First I went to see the dispatcher, who referred me to a nurse, who sent me all over the place. . . In the end, there was no solution and I had to go and see the head nurse! How do people manage when they are all alone? " It is worth noting that this study was conducted before the implementation of biosecurity measures due to the COVID-19 pandemic. Patients must have missed this important form of support during the COVID-19 pandemic because family caregivers were not allowed into French hospitals or medical consultations at that time. Family caregivers are still not admitted to chemotherapy sessions. By supervising and coordinating the whole treatment plan, primary caregivers contribute actively to the management of their relatives' care. But ensuring continuity of care and supervising the timing of medical interventions to meet the patient's needs are burdensome tasks: "I am thinking about Mummy's appointments, I am thinking about keeping house for her. Tat's the problem, thinking all the time about doing the shopping, always wondering whether her blood tests have been done in time, remembering to print out the results, thinking about. . . hundreds of things which need to be thought about, all of which relieve her, but put me myself in a state of stress. " Tese cumulative responsibilities and tasks imposed a mental load on some of them. --- A Responsibility Shaped by Cultural and Societal Norms. Te notion of responsibility also carries the idea of obligation or duty that is embedded in a set of cultural practices [46]. In this study, informal caregiving was found to be shaped by social norms based on the moral contract between spouses and solidarity within families and between generations. When the patient was the carer's father or mother, the commitment to help also responded to a wish for reciprocity, i.e., to repay to a parent the help and love received during childhood or at least to show their gratitude: "I only do what's normal. What children ought to do for their parents. Parents have brought up their children during their whole lives, at least up to the time when they left home, and even then, some parents are still there for them. So it's sort of up to the children to repay them. " Several caregivers expressed a sense of purpose in accomplishing their duty, as explained by one caregiver: "I do it because she's my mother, and then I do it in order to have a clear conscience. I want to be able to say, "I have done what I had to do." Personally, I have done what I had to do. Nobody will be able to say one day, "Ah, but you didn't do this or that for your mother," or "you didn't stand by your mother". . .. " Although our sample of caregivers was not intended to be representative, we noted that the proportion of women was much higher than that of men . Some respondents mentioned the cultural tradition of allocating the work of care to women, as illustrated by this person who was caring for her mother and could not delegate any of her tasks to her brother: "I have a brother who lives near Mummy's home But he's a male and he's not very useful. Neither am I, but I do my best! I too often feel scared, but what can you do? I can't rely on him, no, no, no, not at all. I actually asked him to help, but he said, "I can't handle the enemas the way you do, I couldn't stand seeing Mummy like that. . --- ." " As in many other cultural settings, caring is shaped in French society by the gender-based distribution of domestic work. --- Developing Competences. Most of the informal care work provided required experiential knowledge and specifc skills, such as the ability to perform general nursing tasks . Te carers interviewed often carried out these basic nursing tasks in order to reconcile the timing of the health professionals' interventions with the patients' requirements as suggested by this carer: "the nurses come in three times a day, but they don't always come when we need them most" . In addition, their daily presence enabled them to monitor their ailing relatives' health and to warn the nurses or doctors when their condition deteriorated. Some of these primary informal caregivers have had to make decisions about an emergency hospitalisation, making a change of medical team, or even administering treatment, as explained by one woman who decided to administer oral chemotherapy to her mother without the oncologist's approval because of the poor communications with the hospital team: "For three weeks they kept me waiting, three whole weeks! And when we went along for an appointment , it so happened that we did not see my mother's own oncologist, but a diferent one. Ten they scolded me for having given her the next chemo without consulting them frst. So I said "Listen here, I called you on the phone! I asked to be called back and told what to do . . . you should just have answered! I looked at the results of her latest blood tests and I saw they were the same as the time before, so I took the responsibility of giving my mother her chemo. Nobody answered my calls and I didn't know whether I should do it or not!" I felt really all on my own ." As this excerpt suggests, efcient informal home caregiving requires building a dialogue and trust between professionals and carers so as to alleviate carers' responsibility and strengthen their self-esteem. --- Responsiveness: Primary Caregiving as a Lucky Opportunity. Te patients interviewed recognized the help received and expressed their gratitude to their caregivers. Te carer's presence was regarded as a token of afection: "Te main thing is not feeling alone. Because the most important part of the care received is the other person's presence and afection. . . It's knowing that one is loved which is the most important thing. " Many of the patients interviewed said that they were lucky to have somebody to help them meet their daily needs and break the social isolation into which their illness had plunged them: "I am sorry for people with serious illnesses who have no caregivers! Well frankly! Nobody to deal with the housework, the shopping, the doctor's prescriptions, providing psychological support and all that, it must be very, very hard! I mean, caregivers play an essential role And having somebody coming in every single day is wonderful too! " However, most of the patients interviewed were aware of the impact of their illness and the burden placed on the caregivers' health and well-being. Tey often expressed moral awareness of this burden and wanted to protect their caregivers from having too much work, as explained by this patient whose caregiver was a friend: "She has just turned 89 but she is a wonderful woman. now I try to protect her a little bit even so, I do protect her ." --- Discussion Te present qualitative study helps in understanding how informal care is carried out in practice. Te concept of care developed by Tronto [46] provides a useful analytic framework for documenting carers' experiences of caregiving. It takes us beyond the caregivers' daily life and lifestyle changes, their perception of their role, and their difculties and needs, and it highlights the mental load resulting from the moral dimension of care, resulting in particular from the cumulative efects of the attentiveness devoted and the responsibilities assumed. Providing informal home care means showing concern for patients and constantly making sure that all their basic and care needs are met. Tis attentiveness is required to improve the patient's living conditions, as well as providing psychological support. In line with previous reports [2,6], the most important and most valuable part of family caregivers' work described by the respondents was the emotional support and presence they provided. Kleinman has described presence as being central to caregiving [52] and constituting "the soul of care" [42]. However, this presence disrupts caregivers' everyday lives [10], imposes several restrictions, and leads to a form of "emotional devastation" [53], which often afects their own physical and mental health. Taking responsibility for meeting these needs leads family caregivers to perform a multitude of tasks that pile up, often on the same day. Te present fndings also show that caregiving is a socially and culturally situated activity. It has been shaped by societal norms in the moral contract between spouses and the solidarity within families and between generations, especially placing women in the role of wives, mothers, and daughters. Te moral responsibility for home care certainly rests on women [54,55], especially when the work involved is belittled because it is "dirty work" [56]. Like those in previous qualitative studies [5,10], the present respondents took on the responsibility for monitoring their ailing relatives' health at home and alerting the medical team when their health was deteriorating. Tey also dealt with planning and coordinating the patients' care. Tese tasks facilitated the management of patients' cancer care and palliative home care and made family caregivers full partners in treatment of patients experiencing advanced cancer. Caregiving gave them the feeling of contributing actively to the patient's management [10,57] and helped them cope with the sufering caused by a close relative's illness and their own helplessness to deal with its evolution [58]. However, this involved mobilizing their experiential knowledge of coordination, monitoring, and care planning as well as skills about how to dispense basic nursing care. As described by McDonald et al. [4], these primary caregivers had undergone an active process of learning via the health professionals encountered during the patients' disease and had thus acquired the essential skills required for this job. However, previous studies have suggested that caregivers felt they had inadequate information and preparation [14] and had important caregiver learning needs [59]. As previous authors have suggested [60], those of our respondents who had some professional experience of caregiving had greater ability to provide home care and navigate within the healthcare system, which "highlights a potential inequity of access to service and resources between caregivers" [60] . Moreover, in line with previous reports [59], our respondents needed to be reassured about their ability to care at home for their loved one by being supported by the 6 European Journal of Cancer Care healthcare system [59], especially by enhancing direct communication with clinicians [61]. Te last topic analyzed here was responsiveness, which refers to how recipients respond to the care provided. Te patients interviewed expressed the feeling that the presence of a caregiver and the assistance provided were of great importance, and they qualifed this accompaniment as "a lucky opportunity." As previously reported, social support from family and friends has positive efects on patients' quality of life [62]. Tis raises the issue of patients' potentially unequal access to informal caregiving, especially in the case of socially isolated patients or those whose family ties have weakened. Heathcare professionals could help correct these inequalities by paying special attention to socially isolated patients. However, responsiveness requires balancing the needs of care receivers with those of their caregivers [47]. As noted in previous surveys, informal caregiving reinforces the bonds of afection and trust between patients and their caregivers [57,63,64]. "Caregiving is an existential action afrming a moral commitment" [42] that satisfes caregivers' need to fulfl their marital or flial moral duty, thus also enhancing their self-esteem [14,27,58,65,66]. As care is a gift-sharing process [42,67], caregivers expressed gratitude for being able to spend time with their loved ones [68] and "to take advantage of the time that still remains." Tis feeling may have been perceived more deeply when the patient was sufering from a cancer with a rapid progression that altered their quality of life, such as pancreatic cancer which afected ffteen of the patients in our sample. --- Study Limitations. Te authors acknowledge the limitation that the study was conducted in the French context and that there may exist signifcant diferences from one country to another. Despite these limitations, the fact that the present fndings are consistent with previous reports certainly suggests that our conclusions are also applicable to other national contexts. --- Conclusion Te concept of care developed by Tronto [46] makes it possible to think of the care provided by family caregivers in charge of persons with advanced cancer as valuable work and not simply as an activity restricted to private spheres, involving qualities such as altruism and compassion. In line with the political stand taken by Tronto, whose aims included denouncing the relegation and discrediting of care, the present study sheds light on the moral dimensions of caregiving on the basis of caregivers' own personal experience and shows the important social contribution made by them. Te present fndings could serve to guide healthcare professionals as a means of developing best practice resources and guidelines to support family caregivers and promoting a coordinated public health approach to address the needs of caregivers [23]. Tese eforts are particularly important as present health policies tend to promote the shift from inpatient to outpatient treatment, which increases the importance of family caregiving and the burden it involves. Future research could pay particular attention to this last point in order to assess the possible increase in the burden of home care and, in particular, the increase in the responsibility and mental workload of family caregivers. --- Implications for Practice. In their treatment plans, healthcare professionals should place greater stress on the value of informal care work [42]. Tey should also be made aware of the underestimated efects of caring in order to alleviate these efects. Supporting family caregivers is certainly a key aspect of palliative care and family medicine. Healthcare professionals should be aware of the risk of caregiver burnout and be able to provide multidisciplinary support. However, the support provided should not be only psychological [3]; caregiver burnout also results from the piling up of the tasks involved in informal care. Although this study was conducted in a European welfare state context, health professionals everywhere should be made aware that informal primary caregivers must be able to delegate some of their tasks and have their own respite times [61]. --- Data Availability All the data sets on which the results of the study were based are available on request from the corresponding author. --- Conflicts of Interest Te authors declare that there are no conficts of interest regarding the publication of this paper. --- Supplementary Materials

Background. Family caregivers in charge of patients with advanced cancer play an essential role. Te psychosocial cost of these caregiving activities has been studied, and psychosocial interventions have been developed to improve the quality of life of family caregivers. A deeper understanding of caregivers' burden is essential in order to enhance the benefts of these interventions. Te aim of this study was to explore the socioeconomic and cultural factors responsible for shaping the complex personal experience of family caregiving and to analyse the moral dimensions of the caring experience so as to understand its efects on family caregivers more clearly. Materials and Methods. A qualitative study based on in-depth interviews was conducted with patients with advanced cancer (n � 20) and their family caregivers (n � 19) from 2017 to 2020. Tese interviews were analysed using an inductive approach and an iterative procedure. A thematic analysis was then performed using Tronto's "ethic of care" framework in order to identify the various levels of responsibility and the relationships and efects involved. Results. Providing patients with advanced cancer with informal care is highly valuable work requiring various moral qualities, including attentiveness, responsibility, competence, and responsiveness. Te mental load resulting from the moral aspects of care results from the cumulative efects of carers' attentiveness and the responsibilities they have taken on. Conclusion. Te present fndings could guide healthcare professionals to develop best practice resources and guidelines in order to alleviate the hitherto underestimated efects of caring and promote a coordinated public health approach addressing the needs of caregivers. Tese eforts are particularly important as contemporary health policies tend to promote the shift from inpatient to outpatient treatment, which increases the importance of informal caregiving and the burden involved.

In recent decades, declines in fertility have been documented in younger women in the majority of Western countries . It has been suggested that these trends indicate a postponement of first childbirth to older ages, rather than an intention to remain childless . From a public health perspective, older maternal age has a raft of well-known health consequences for women and children, including increased risks of gestational diabetes, miscarriage, and birth defects . Furthermore, the use of assisted reproductive technologies associated with older childbearing carries risks in pregnancy and adverse outcomes in offspring, in addition to considerable financial and emotional costs . Thus, it is imperative to understand barriers to childbearing at ''optimal'' ages. Because evidence also suggests that women now have fewer children than they would like to have , it is particularly important to investigate underlying structural determinants of older motherhood and foreshortened reproductive careers. Recent fertility theory suggests that individual-level economic uncertainty, brought about via the impact of globalization on national institutional structures, is associated with a delay in having children . In this vein, factors likely to contribute to economic uncertainty for young people in Western countries include nonpermanent employment arrangements for individuals and, where relevant, their partners; higher education debt; and the high cost of home ownership . Few such factors have been considered empirically with regard to their influence on childbirth timing. Extant studies do not adequately consider economic exposures across adulthood, either disregarding or not sufficiently capturing the role of life-course timing and duration of time spent in economically uncertain states -both of which are likely to be influential in this context. We designed the Life Journeys of Young Women Project to address the major methodological limitations of studies conducted to date. The project will provide the first empirical evidence of the contribution of several dimensions of economic uncertainty to the timing of first childbirth. This paper discusses study objectives and design, fieldwork procedures, planned statistical analyses, and recruitment outcomes. --- MATERIALS AND METHODS --- Study objectives and design The Life Journeys of Young Women Project aimed to examine the influence of economic uncertainty in adulthood on age at first childbirth. The study design was guided by 2 complementary life-course approaches. The cumulative approach considers life-course exposures as incrementally accumulating through episodes of illness and injury, adverse environmental conditions, and healthdamaging behaviors . The sensitive period approach examines exposures within windows of susceptibility during which excess risk may be present . The concomitant research questions were as follows: Does a longer exposure to economic uncertainty increase the likelihood of older age at first childbirth, and is there a particular period of time in which economic uncertainty is especially potent in influencing the timing of first childbirth? Figure 1 presents the conceptual models that guided this study. The models are not relevant to women with very low socioeconomic status; these women are unlikely to have high material aspirations and are therefore unlikely to delay childbearing in the presence of economic uncertainty . In fact, for this group of women, early childbearing may be central to identity formation . The LJYWP was a cross-sectional study in which much of the exposure data was collected retrospectively in the event history format ; that is, exposure status was ascertained in continuous time, rather than at a single point in time or as successive point-in-time measures. The study population was drawn from a cohort study that investigates the prevalence and social and biologic determinants of a range of women's reproductive health issues . The LJYWP was approved by the Ethics Committee of The University of Adelaide, and all participants gave written consent at wave 1 to be contacted for further study interviews. Participants were not paid for their participation in this project. --- Study population The study cohort was based on all consecutive female babies born from January 1973 to December 1975 at the Queen Elizabeth Hospital in Adelaide, South Australia, who survived to discharge. Information about the stages of recruitment and cohort representativeness can be found elsewhere . Cohort members participated in wave 1 of the study between 2003 and 2007. All cohort members were subsequently invited to participate in the LJYWP, which corresponded to wave 2 of the longitudinal study. Cohort members were excluded if they were not able to participate in an interview because of illness or disability. Interviews were conducted between December 2007 and July 2009, when participants were aged 32-35 years. Figure 2 shows a flow diagram of participation in the LJYWP. --- Constructs and variables The 4 key constructs identified in the development of this project were economic uncertainty , first childbirth , ''stable'' relationships, and socioeconomic status. The relative contribution of different aspects of economic uncertainty with regard to the timing of first childbirth has not been reported to date. A model for this construct was designed, and data were collected on all components. The model included employment precariousness , education, income, and occupation; presence or absence of a stable relationship and, when relevant, partner's level of education, income, occupation, and employment precariousness; home ownership status; level of assets and debts, including higher education debt; and presence and level of financial and social support from family. All components of the model are capable of change over time and were conceptualized and measured as such in this project. Schemas were designed for the more complex dimensions of economic uncertainty in order to guide construction of exposure variables. For example, Figure 3 shows the schema for point-in-time employment precariousness. Each month of observation will be categorized according to 1 of the 7 activity states shown in the figure . Because literature suggests that women tend not to have children while engaged in courses of study , periods of study were designated as such, irrespective of concurrent employment status. Of the 7 shaded states shown in Figure 3, casual employment is of greatest interest in this study. Age at first childbirth was the primary outcome variable in this study. However, the focus on livebirths ignores a multitude of social and biologic dimensions of the pathway to childbearing, such as pregnancies that do not result in a livebirth and ''unplanned'' or ''unwanted'' pregnancies. Thus, the ages at first pregnancy and at first ''wanted'' pregnancy were also collected. Although fertility theory implies that a ''stable'' or ''committed'' relationship is an important prerequisite for childbearing for most women , little is known about how individuals construe relationship stability or how this impacts on their experiences of starting a family. The most obvious proxy for a stable relationship that could be collected retrospectively was a relationship in which partners live together ; thus, information on periods of time spent in live-in relationships was collected in the LJYWP. Socioeconomic status will be gauged according to the highest level of educational attainment at the time of study interviews. The role of this variable is 2-fold. First, it will provide a means of identifying the very low socioeconomic group to whom the conceptual models shown in Figure 1 do not apply. Second, analyses will be stratified by this variable, because educational attainment is a well-known determinant of fertility behavior . --- Survey instruments An event history calendar was developed to collect retrospective data in this study. Web Figure 1, which is posted on the Journal's Web site , shows an abbreviated version of the event history calendar used for the LJYWP. Event history calendars are acknowledged to have positive effects on the recall accuracy of life-course information , and they consist of a large calendar on which the timing of key events or states of interest is recorded by an interviewer, using a predetermined system of symbols. Advantages of event history calendars stem from interviewers being able to easily prompt participants according to information already collected and to cross-check across domains for consistency of information . In the LJYWP, information was collected from participants from the age of 15 years to the time of the study interview, a period of 18-20 years. Fifteen years of age was chosen as the start point so as to capture the timing of schoolleaving for all participants, and this was the minimum age of school-leaving in Australia in the 1980s and 1990s, when cohort members were teenagers . Information on key constructs was collected at the month level . A conventional questionnaire was used in the LJYWP to record information not suited to the event history calendar format. To enhance data quality, we sent a simplified version of the event history calendar to participants to fill out prior to study interviews . Because participant-completed calendars may deter participation in research interviews , substantial efforts were directed to ensuring that the participant-completed calendar was visually appealing, that associated instructions were straightforward, and that it took under 20 minutes to fill out. To our knowledge, this is the first time that an event history calendar and In A, ''þ'' indicates a positive association, and ''À'' indicates a negative association; in B, ''þ'' indicates a positive association, ''À'' indicates a negative association, and ''?'' indicates an unknown association. a participant-completed calendar have been used in the same study. All survey instruments were piloted extensively prior to the commencement of study interviews, with attention paid to participant burden for women with varying life-course experiences. --- Interviews The order of topics in the structured interview was standardized for all participants. In line with recommendations in the literature , the simpler recall tasks were completed before those likely to be more complex; for example, educational and partnering histories were taken before work histories. In order to accommodate geographically dispersed participants and to reduce travel costs, interviews were conducted primarily via the telephone. When participants attended an interview without having filled in a participant-completed calendar, interviewers assisted this process before commencing a formal interview. Interviewers were trained to manage participants' fatigue levels throughout interviews, which was particularly important given the greater complexity of the LJYWP interviews compared with standard survey interviews. --- Facilitation of interview participation Substantial efforts were directed to facilitating cohort participation in this study. Cohort members were approached in a systematic order, such that participants who were likely to find the interview most challenging were interviewed late in the fieldwork period, when interviewers were most experienced in the conduct of the interview. These women were also offered a faceto-face interview if they lived locally, so that interviewers could better facilitate the use of the participant-completed calendar instrument with women who may potentially find it most challenging. --- Interviewer selection and training High-quality interviewers are integral to high-quality data collection in retrospective life-course studies. The recruitment strategy for LJYWP interviewers included carefully worded selection criteria, a higher salary than usually offered for casual fieldworker positions, a targeted advertising campaign, and thorough discussions conducted with shortlisted applicants. A small team of 5 interviewers was employed, allowing for more intensive training and a higher level of interviewer support throughout the fieldwork phase than if a larger team had been employed . Interviews involving event history calendars require extensive interviewer training . Interviewers were paid to participate in a part-time training program over a 3-week period for this study. The structure of the program was based on that provided in Freedman et al. and modified to meet the needs of the LJYWP. Training began with an exploration of the project rationale, theoretical underpinnings, and important concepts being quantified in the interview. This was supported by a series of readings that interviewers undertook as their ''homework.'' The next and main component of training was a step-by-step process that engaged interviewers at levels of increasing complexity, moving through observing mock interviews and recording answers to conducting interviews with volunteers with progressively more complex life histories. Each interviewer completed approximately 4 face-to-face interviews and 3 telephone interviews with volunteers who were not in the cohort, throughout the training program. --- Quality control procedures It has been suggested that life-course retrospective interviews warrant a detailed quality control system . The first author observed training and fieldwork interviews until a stage of adequate comfort, ease, and accuracy was reached for each interviewer . A procedure was developed for checking and editing all completed questionnaires and event history calendars. Interviewers spent 20-30 minutes checking interview documents immediately following an interview. A second check of documents was then undertaken by another member of the interview team. In addition to the identification and correction of any mistakes, the quality-monitoring processes provided an opportunity to identify issues that required further interviewer training. Interviewers were encouraged and supported to monitor and manage their own fatigue throughout each interview and the immediate postinterview period. --- Statistical analyses The event of interest for initial analyses will be age at first childbirth. Time-constant and time-varying survival analysis techniques will be applied to examine the effects of different dimensions of economic uncertainty on the likelihood of having had a child by the time of the study interview, within educational strata and taking into account other influential factors such as stable relationship histories. The period of observation will start in the month that participants turned 15 years of age and end in the month of first childbirth minus 1 year for women who are mothers or month of study interview for those who had not had a child. The period of observation is lagged for mothers in order to capture the timing of conception as opposed to the timing of the birth itself; lagging by 1 year is consistent with data from a number of studies that investigated the structural determinants of childbearing . Taking the example of precarious employment as the exposure construct, investigators are to label all months in the period of observation according to the schema in Figure 3. For the cumulative analyses, within each educational stratum, the effect of the total number of years under observation spent unemployed or employed with a casual, fixed-term, or permanent contact on the outcome variable will be examined through fitting separate Cox proportional hazards regression models. For sensitive period analyses, within educational strata, the effect of the number of years spent in each exposure state per 3-year period on the outcome variable will be examined through fitting separate Cox proportional hazards models. Age bands in 3-year periods are to be used . These age bands were chosen to complement life-course phases that could be easily communicated when discussing study results-for example, early twenties, mid twenties. Future analyses are planned to explore other options for construction of sensitive periods. STATA, version 10.0, software is planned for use in the conduct of all analyses. As noted, the relations between employment arrangements and age at first childbirth are to be explored within groups with different levels of educational attainment. The cutpoint of educational attainment below which the conceptual models shown in Figure 1 no longer apply will be identified. A variable representing partnering status will be included in models. The total time in the labor force will be included to recognize that participants are at risk of forms of economic uncertainty only during the periods in which they are in the labor force. These variables will be time constant in cumulative analyses and time varying in sensitive period analyses. Time-constant variables representing parental ethnicity and childhood socioeconomic status will be included in both analyses, because they are thought to be determinants of fertility behavior . Because the study cohort is a birth cohort, period and age will not need to be accounted for in the models. --- RESULTS --- LJYWP cohort participation rates An overview of LJYWP participation is presented in Figure 2. Of 974 cohort members, 970 were eligible to participate, and 663 completed an interview, resulting in overall participation of 68%. Among the 290 eligible nonparticipants, 26 withdrew from the cohort study, 223 declined participation in the LJYWP but continued their cohort membership, 11 could not be contacted, and 30 did not participate for other reasons. Among the LJYWP participants, 451 had at least 1 child at the time of interview. --- Description of LJYWP participants Table 1 describes some of the demographic characteristics of LJYWP participants and nonparticipants, as well as the overall study cohort, according to data collected at wave 1 interviews. Results of statistical tests comparing participants and nonparticipants are presented. A variable corresponding to the Socio-Economic Index for Areas indicator of relative socioeconomic disadvantage was constructed from the 1986 SEIFA indicator of relative socioeconomic disadvantage that corresponded to each participant's postal code at primary school . A lower SEIFA score reflects higher relative disadvantage. At wave 1, the majority of LJYWP participants were employed and in a relationship . Approximately 10% of participants did not finish high school and had not received further qualifications. Over one-third of participants had attained a technical and further education qualification, and a further third had received a university qualification. Participants and nonparticipants were similar in terms of marital status and parents' birthplace; however, there were significant differences in labor market status, highest level of education, and SEIFA index of socioeconomic disadvantage, with participants more likely to be employed, to have a university qualification, and to have lived in a less disadvantaged area in primary school. Table 2 describes, for LJYWP participants, women who had and had not had a child, as well as the median age at first childbirth, according to highest level of education at the time of LJYWP interview. As expected, participants in the ''some high school'' and ''completed high school'' groups were more likely to have had a child than those with technical and further education or university qualifications . The median age at first childbirth was lowest for the some high school group and highest for the university group . --- DISCUSSION The LJYWP will contribute to a growing international literature investigating the impact of economic uncertainty on the timing of childbirth in women's lives. The study responds to limitations of existing literature by collecting data concerning a broad range of economic factors across the time dimension and by applying an analytical framework enabling investigation of the duration and timing of economic uncertainty across the life course. Moreover, this will be the first Australian study to investigate the influence of any form of economic uncertainty on the timing of first childbirth, to our knowledge. The LJYWP is methodologically innovative within the paradigm of life-course epidemiology in several ways. First, the collection of fine-grained life-course data is more sophisticated than cross-sectional or successive point-in-time measures. Second, the project builds on existing techniques aimed at collecting high-quality retrospective data, in terms of type of survey instrument and fieldwork procedures. Third, the development of a framework for the use of survival analysis techniques to complement life-course theories is an important contribution of this study. Finally, the project is one of only a handful of crosssectional studies collecting an event history data set to have been conducted, and the first Australian study of this design, to our knowledge. Several limitations of this study are of note. As is the case for all cross-sectional studies with a major retrospective component, a key methodological limitation is related to recall of life-course information. This is especially relevant to the LJYWP given the detailed data collected in combination with the relatively long period of recall. However, a major aim of project investigators was the development of strategies to ensure the collection of data of the highest possible quality through the use of event history calendar and participant-completed calendar survey instruments, careful choice and comprehensive training of interviewers, and rigorous quality monitoring of interviews. In combination, the cluster of strategies is the most comprehensive attempt to collect high-quality retrospective data among published studies of this design. Future analyses may include checking the agreement of retrospective information collected in the LJYWP with data collected at wave 1 . A limitation of the LJYWP is the age of cohort members at the time of project interview , because they had not yet completed their reproductive life period. Thus, a number of study participants who have not yet had a child will do so in the years to come. It is possible that patterns of employment experiences and other forms of economic uncertainty up to the midthirties may differ between women who will have a child in the future and women who will never have a child e.g., because the desire to have a child in the future may shape employment trajectories). Thus, considering these 2 groups in combination may blur associations between the exposure and outcome. Investigators plan to follow the cohort until the end of their reproductive life course, so it will be possible to conduct analyses with complete reproductive data in the future. Sending participants a simple calendar to fill out prior to study interviews may have decreased response rates for the LJYWP. However, the 68% response rate attained in this project is considerably higher than that attained in a published study in which a participant-completed calendar was also used . Of greater methodological concern are the socioeconomic differences between participants and nonparticipants in the LJYWP, as noted in Table 1. However, because analyses will be stratified by educational attainment, resultant bias in study results will be minimized. An extensive research program based on the LJYWP data set is planned. As has been discussed, this will include investigating the influence of a range of economic factors on age at first childbirth, from a cumulative and sensitive period approach. Analyses may also consider age at first pregnancy and age at first ''wanted'' pregnancy, as alternative outcome variables. Ultimately study findings have the potential to help shape policy that enables women and their partners to have children at a time in the life course that is most conducive to their health and to the health of their children. Irrespective of study results, it is hoped that the methodological and analytical frameworks developed will provide guidance to researchers wishing to collect event history data sets and to conduct analyses based on such data. a Known in some countries as polytechnic, vocational technical college, or community college. --- Conflict of interest: none declared.

The Life Journeys of Young Women Project is the first population-based study to examine the role of economic uncertainty throughout early adulthood on age at first childbirth. A retrospective cross-sectional component was added to an existing cohort study that is based on a birth cohort of women born during 1973-1975 in Adelaide, South Australia (n ~1,000). An event history calendar instrument was used to obtain data regarding a range of life domains including partnering, educational attainment, home ownership, higher education debt, employment, and pregnancies over a 20-year period (sometimes as detailed as at monthly intervals). Interviews were conducted between 2007 and 2009. An analysis framework applying time-varying and time-constant survival analysis techniques within a life-course framework was developed that will guide analyses to examine the role of duration and life-course timing of economic uncertainty on age at first childbirth. This paper discusses study objectives and design, fieldwork procedures, planned statistical analyses, and recruitment outcomes, focusing on novel features that would facilitate analogous epidemiologic research.

Background --- Syrian crisis Since the start of the conflict in 2011, the situation in Syria has remained one of the worst humanitarian crises in the world [1] with 5,5 million Syrians having fled the country as of June 2020, and another 6 million people internally displaced [2,3]. Lebanon hosts the largest number of Syrian refugees per capita with approximately one million displaced individuals, and refugees now account for greater than a fifth of the country's population [2,4]. Although not a signatory to the 1951 United Nations Refugee Convention, Lebanon has a long history of hosting refugees including 200,000 Palestinian refugees in addition to 18,500 refugees from Ethiopia, Iraq, Sudan and other countries [5]. As a consequence of the Lebanese government's decision to not establish formal camps for displaced Syrians, refugees are dispersed across the country, living in informal tented settlements or integrated within local communities, either living with family or in rental accommodations, as opposed to being hosted in formal, supported refugee camps [6,7]. The highest numbers of Syrian refugees are located in the governorate of Bekaa followed by North Lebanon and Beirut [5]. Nine years into the crisis, Syrian refugees in Lebanon are now more vulnerable than ever, with more than half of them currently living in extreme poverty and over three quarters living below the poverty line [8,9]. Daily economic challenges significantly affect parents' ability to adequately care for their children. The increased stress of not earning adequate income to cover the family's living expenses, can lead to negative mental/physical health consequences, while also contributing to an ongoing cycle of poverty [10]. Having itself experienced war and years of political instability, Lebanon's precarious infrastructure and resources are significantly strained by hosting more than one million Syrian refugees. This in turn has resulted in tensions within the host communities, which at times has been manifested as violence and discrimination towards Syrian refugees [11]. Displaced Syrian families in Lebanon are also confronted with burdensome governmental policies and regulations. For example, those registered with UNHCR must pledge that they will not work in Lebanon, while other Syrians are only permitted to work in agriculture, construction and environment. Although the Lebanese Government agreed in 2016 to ease work restrictions for Syrian refugees, discussions are still ongoing around implementation of these procedural changes [12][13][14]. --- Sexual and gender-based Violence in humanitarian settings Lebanon currently hosts about 500,000 Syrian children between the ages of 3 and 18 [9] and it is estimated that 54% of the Syrian refugee population in Lebanon is under the age of 18. While many of these children face risks related to poverty, food insecurity, lack of access to healthcare, and forced labor, girls are also vulnerable to additional gendered risks including child marriage, domestic violence and intimate partner violence , sexual exploitation and assault, as well as intimidation and fear of violence within their communities [15][16][17]. In humanitarian crises women and girls are at risk of various forms of sexual and gender-based violence [18]. SGBV is defined as "any act that results in harm, suffering or humiliation to women and girls and includes rape and sexual abuse, forced pregnancy, dowryrelated abuses, female infanticide and domestic violence" [19]. Although a violation of human rights, SGBV has remained widespread with at least one in every three women being beaten, coerced into sex, or otherwise abused in her lifetime [20,21]. In conflict settings, women are at an even higher risk of SGBV [22,23] with different manifestations of it sometimes being used as a "weapon of war" to conduct ethnic cleansing [24,25] and to terrorize people as a method of systematically destroying communities [26]. Mass rape in war has been documented in crises regions such as Bosnia, Somalia, Uganda, the Democratic Republic of Congo and Myanmar [24,27,28]. As of today, Rohingyan women in Myanmar and Yezidi women in Iraq face systematic campaigns of sexual violence based on their ethnicity and religious affiliations [24,29]. At the time of displacement, refugees are also recognized to be at particular risk of sexual victimization with many women being forced into "commercial sex work" or subjected to sexual exploitation and/or torture to pay for their migration [30]. In post-migration settings, women frequently experience social and cultural isolation as well as dire economic need, making them again vulnerable to sexual abuse and exploitation as well as commercial sex work [31]. While SGBV against women and girls during the Syrian conflict, mainly perpetrated by armed actors, is well documented [32], data about SGBV in Syria before the crisis is lacking as it was rarely discussed [33]. --- SGBV and gender norms in pre-conflict Syria versus postdisplacement Prior to the war, Syria was very patriarchal and men dominated almost all aspects of society [34]. According to the feminist perspective, men dominated the public sphere whereas women were essentially relegated to the private sphere. This public/private dichotomy served to entrench the patriarchal system and ensured the oppression of women. For instance, women found it difficult to raise issues that impacted them and the political process was biased towards the public sphere whilst largely ignoring the private realm. By marginalising the private sphere, men maintained their dominance of the political process from one generation to the next [35,36]. Earlier evidence has suggested that gender norms are changing due to the Syrian crisis and its resultant displacement. For instance, Syrian women report feelings more empowered and describe having increased responsibilities with regards to work outside the home and providing for their families [37]. These altered gender norms have been accompanied, however, by increasing reports of SGBV perpetrated by men who use violence as a coping mechanism and to reassert hegemonic patriarchal roles [16]. Under these circumstances, men's low selfesteem and feelings of disempowerment have contributed to violence against the family as some men have felt stripped of their function in society [34]. Increased rates of child marriage within Syrian refugee populations have been identified [38][39][40] and studies in Lebanon have highlighted that financial hardships, lack of educational opportunities and safety concerns are the main drivers of early marriage among Syrian refugee populations [38,39]. Sexual exploitation, forced prostitution and, "survival sex" have also been described as a negative coping strategies for some Syrian women and girls who have no other means to raise the funds needed to cover living expenses in Lebanon [41,42]. Economic vulnerability also puts many girls and women at risk of sex trafficking [42]. In addition, Syrian women and girls in Lebanon are exposed to physical and verbal abuse, sexual harassment, robbery and fear of being kidnapped [16], often resulting in parents' reluctance to permit their daughters to leave the house unaccompanied [43]. While many SGBV risks have been well documented and are common across a variety of settings, less is known about the SGBV threats faced by displaced women and girls who are largely integrated into their host communities rather than hosted in formal refugee camps. It is also less common that research on SGBV includes the perspectives of men and boys in addition to that of women and girls. Therefore, what the current study contributes is two-fold: i) an analysis of threats and experiences of SGBV among Syrian refugee women and girls who are integrated into their host communities; and ii) a perspective of SGBV that is broader and more nuanced because it also includes insights from male community members. The ultimate goal of the research is to reduce SGBV risks and to improve assistance to those who have been affected. --- Methods --- Study design and participants The study was conducted by Queen's University in collaboration with the ABAAD Resource Center for Gender Equality. Data was collected across three locations in Lebanon hosting a majority of Syrian refugees: Beirut, Beqaa and Tripoli. Twelve interviewers were selected by the ABAAD Resource Center for Gender Equality and completed a four-day training seminar immediately ahead of data collection. The team included six Syrian female interviewers to collect the perspectives of women and girls, as well as three Syrian and three Lebanese male interviewers to collect the perspectives of Syrian and Lebanese men respectively. To represent a broad spectrum of perspectives on the lives of Syrian girls in Lebanon, a convenience sample within each of the following participant subgroups was recruited: married and unmarried Syrian girls, Syrian, Lebanese or Palestinian men, Syrian parents, and community leaders. Within each location, research assistants approached potential participants from their naturalistic community settings including: markets, post offices, commercial settings, bus stops, etc. Only participants aged 13 years and older were eligible to participate in the study. Narratives were audio recorded in Arabic and then transcribed and translated from Arabic into English by native Arabic speakers. Because the story prompts were open-ended, participants could share narratives about any experience of Syrian girls in Lebanon and thus, multiple themes arose from the data. The research presented here is a qualitative analysis of the subset of narratives about experiences of SGBV among Syrian women and girls. All first-person stories by female participants from married and unmarried Syrian girls, Syrian mothers and female community leaders were screened and those about experiences of SGBV were included for analysis . To ensure that male perspectives were also represented, every fifth story out of 701 stories from male participants was also screened and those about experiences of SGBV were included for analysis. --- Instrument The dataset is derived from a larger mixed-methods study, investigating the experiences of Syrian refugee women and girls in Lebanon [38]. For our purposes, Syrian girls referred to Syrian females under the age of 18, and this was clearly defined for all participants at the beginning of the survey. In the original 2016 study, data were collected using SenseMaker®, a mixed-method data collection software tool that extracts meaning from a repertoire of stories shared about people's experiences. SenseMaker was developed by Cognitive Edge and has been used historically for understanding corporate culture, corporate restructuring, etc. Using SenseMaker, participants selected one of three open-ended prompting questions to share an anonymous story about the experiences of Syrian girls in Lebanon: 1. Tell a story about a situation that you heard about or experienced that illustrates the best or worse thing about the life of a Syrian girl in Lebanon, 2. Provide a story that illustrates the biggest difference between life for Syrian girls living in Lebanon in comparison to life for Syrian girls in Syria, 3. Suppose a family is coming to Lebanon from Syria, and the family has girls under the age of 18. Tell a story about a Syrian girl in Lebanon that the family can learn from. After telling a story, participants interpret their shared narrative by responding to predefined questions on handheld tablets, smartphones or computers. SenseMaker then quantifies each of the responses, providing statistical data backed up by the explanatory narrative [44]. The Sense-Maker survey intentionally did not ask direct questions about experiences of SGBV but instead enabled stories about SGBV to become apparent from a wide range of Syrian girls' lived experiences. In total, 1422 selfinterpreted stories from 1346 unique respondents were collected. More detailed information on the survey design and data collection procedure has been previously published [45]. --- Analysis Two researchers independently screened all 327 first person female narratives as well as the 159 systematically selected male narratives to identify main themes and subcategories around SGBV. Please see below for the definition of SGBV used for screening and coding purposes. Screening discrepancies between researchers were reviewed by a third researcher and inclusion of the narrative was decided by consensus agreement between all three researchers. Thematic analysis was considered appropriate as it allows rich and detailed meaning to be drawn from data through the emergence of patterns [46,47]. Analysis consisted of familiarization with the data by three researchers through reading all narratives multiple times, followed by independent coding to create the set of initial codes. Initial codes were then grouped into potential themes representative of the data by researchers SR and SG. Themes were reviewed by researchers SR and SG and final themes decided by consensus agreement after discussion between researchers SR, SG, HB and SB. --- Definition of SGBV and its use within the study context For the purpose of this study, stories that discussed any form of physical, sexual, psychological or verbal violence perpetrated against women and girls and seen as a direct result of their gender were screened in and later coded as being about SGBV. Additionally, harassment based on gender, domestic and intimate partner violence of a physical, sexual or emotional nature as well as sexual assault, commercial sex work and exploitation were all included as forms of SGBV. While men and boys are also affected by SGBV, because this study focused on the experiences of Syrian girls, SGBV perpetrated against men and boys was not captured in the current dataset. While early, forced and child marriage is widely recognized as a form of SGBV, these narratives were only included in the current analysis if they also described one of the above forms of SGBV since a separate analysis of early, forced and child marriage from this data has been previously published [38]. --- Results In total, 70 of the 327 first person narratives shared by females discussed some form of SGBV while 42 of the 159 male stories included dialogue on some form of SGBV. Demographics of study participants are presented in Table 1. --- Qualitative data analysis Two main themes around SGBV among Syrian refugee women and girls in Lebanon were highlighted in this analysis: 1. Continuum of SGBV risks in the public and private spheres; 2. Gendered differences and perspectives around SGBV. --- Theme 1: continuum of SGBV risks for Syrian women and girls in Lebanon Different threats and experiences of SGBV in both the private and public spheres were evident for Syrian women and girls in a continuum across the adolescent and early adult years. By private sphere, we refer to acts of violence that occur in the home, typically behind closed doors and often perpetrated by either a family member, friend, or someone else known to the woman or girl. By public sphere, we refer to those acts of SGBV that occur outside the home, in public areas such as schools, streets, shopping areas, parks, etc., and more likely involving an unknown perpetrator. The continuum described in our analysis refers to life stages from unmarried adolescent girls living in the family home, to child brides living with her husband and/or parents-inlaw, and in some cases to being divorced or separated after a marriage ends. From this chronological perspective, early SGBV risks refer to those primarily affecting unmarried adolescent girls, while later SGBV risks refer to those experienced more so after being married and leaving the family home. As adolescents, girls are perceived to be at high risk of sexual harassment and sexual violence in their communities. In an effort to protect them from these public sphere threats and to protect their honor, some parents choose to marry their daughters earlier than they perhaps would have under different circumstances. Child marriage is itself a form of SGBV but the risks are further compounded by the IPV that many girls experience after they are married. In addition to IPV within their marriages, some girls also describe DV by their husbands' relatives and in a few cases girls were subjected to abuse after they fled violent relationships and returned to their parents' home. --- Public SGBV risks for women and girls: harassment and assault Several forms of SGBV were reported by women and girls as part of their experiences in the public sphere. A majority of the first-person female narratives about SGBV referred to various forms of harassment directed towards women and girls in public spaces, and street harassment was often mentioned as the reason why girls' mobility was forcibly limited. Harassment was most commonly reported by girls while travelling to and from school as well as in school, creating a sense of unease and fear among some participants. Fear of harassment and heightened concerns around safety for girls led some parents to engage in negative coping mechanisms in an attempt to protect them. For instance, an unmarried Syrian girl based in Tripoli reported that she was not allowed to attend school. " … When our family moved to Lebanon, I lived in a very unsafe neighborhood, so I was not allowed to leave the house to attend school, fearing that the Lebanese men would take sexual advantage of me or subject me to verbal violence and threats. Moreover, my sister and I are not allowed to stay at home by ourselves because our parents fear that someone might come in and harass us." Some participants described being bullied and harassed not only by boys, but also by teachers and school officials. An unmarried Syrian girl living in Beirut (age [13][14][15][16][17] reported being harassed by her teacher. "...One time in class the teacher would ask me to sit right in between two guys, and I would ask him whether he would be ok with his wife sitting in between my father and uncle and he would say "no". I told him that he has to move me from between the two boys or I would go home, and he would ask me to go home. The teacher kept going on with his hitting and making me sit between guys and I told him that if my father was here he would have never treated me like that … " Living in ITSs or integrated into the host community, many respondents felt unsafe due to experiences and / or perceived threats of sexual harassment and sexual assault on the streets. One Syrian mother based in Beqaa (age [25][26][27][28][29][30][31][32][33][34] shared her personal story about being sexually harassed by a male driver while she was walking with her daughter, and discussed concerns for her daughter's safety in Lebanon. "My daughter and I were going to the pharmacy once, when a man in a car started following us and honking. My daughter and I ignored him completely, so he stopped the car and opened the window, only to find him naked in the car. I was very scared for my daughter. He was a disgusting man and what he did was even more disgusting. If the man had done this while she was alone, the man could have kidnapped her. I advise every girl to leave the house with an adult, someone who can protect her." Some female respondents reported being followed and verbally harassed on the streets by passing drivers or while commuting on public transportation. A 22-year old married Syrian girl from Beirut talked about her experiences of harassment and commented that all women were at risk of harassment. "When I go to Beirut, the taxi drivers would harass and catcall any women passing by. Regardless if the woman was with her child, married, or pregnant. Some of them would say some very dirty words. When she is walking with a man, they wouldn't dare to say anything. But they will harass a woman walking alone a lot." One married 18-year-old Syrian girl, who had been in Beqaa for 5 years ago reported being sexually assaulted in the public sphere and discussed misconception and the disrespect she faces as a refugee girl in Lebanon. " … I worked in a chicken restaurant, they disrespected me and assaulted me. The owner was paralyzed and harassed me sexually, he always tried to get close to me. He was old. They think that the Syrian girl has no dignity. I felt really offended and we argued a lot. He gave me a wrong idea of the Lebanese men, they just want us sexually, they think that we sell our dignity for cheap." An unmarried 13-year old Syrian girl living in Beqaa blamed herself for being sexually assaulted on the street and felt that she could not tell her parents about the incident. "After some time alone, the boy cornered me in a high place, and then he started holding my hand, touching me in places, and getting his body close to mine. He then wanted to take me somewhere, but I refused, so I wanted to break free but he wouldn't let me. He kept pulling me towards him, making our bodies collide and touch. After freeing myself from his grip, I ran home and I have not been outside the house since. I know that it is my fault and I am responsible for such action. I cannot tell my parents about this because they will make a big deal out of this. They will blame us and get away with it because we are Syrians. I advise every girl to never play with boys and to never leave the house by herself or without an adult." One married Syrian father living in Beqaa valley shared a story about a Syrian refugee girl who had lost all her family back in Syria and needed to work to finance herself in Lebanon where she has been sexually assaulted at her workplace. " … She had to work here to support herself so she started by selling flowers. As the days passed she started staying out late for her job. One day, a group of men promised her money if she'd walk with them. They assaulted and raped her. From that day on, the barrier that was once there in her mind was gone so she started working as a prostitute. The girl is a minor, she is 16 years old." Moreover, given the dire financial situation, some women and girls worked in order to support their family's finances. The employment market in Lebanon, however, often restricts women's labor opportunities to domestic house work, posing an increased risk of SGBV due to the lack of accountability and reporting mechanisms. Some female respondents reported unwanted sexual advances by their supervisors and in some cases, employers were benefiting from the financial dependency of female employees. For instance, one married Syrian man living in Beqaa described how an 18-year old girl was harassed by her employers. "I know a girl who came to Lebanon with her husband. While they were entering Lebanon, they took her husband on the border. So, she had to work in houses here. Some people used to pay her and others didn't. She has a little boy who she needs to secure food for. Some people used to harass her while working. So, she never returned to these people's houses. She is only working for the sake of her son. And she doesn't know what to do." Private SGBV risks: child, early and forced marriage SGBV risks for Syrian refugee girls were described as one of the main drivers for parents to marry their daughters early, since marriage and having a husband were perceived as protective against SGBV. One 16year-old married Syrian girl in Beqaa was married at the age of 14 as a protection measure due to safety concerns. " … I got married because my parents feared for my safety. There were a lot of kidnapping cases happening in our village. … I have marital problems as well. I advise parents to reject the idea of early marriage. It is not right. The girl should be allowed to live through her whole childhood. I will not approve for any of my daughters to get married at an early age; especially after this experience that I am going through." This Syrian girl in Beqaa chose to get married at the age of 14 and gave up education due to a fear of being harassed and assaulted. "When I was 14 years old, I chose to get married. This was due to my former school principal; he used to harass the students and teachers. He constantly harassed them. After a while, he started to harass me. Therefore, I hated school, and I accepted to get married to the first man who proposed to marry me. I was relieved after I got married. I was relieved from men's harassment. I preferred to get married. Now, I am very comfortable. I hated being harassed. I saw how the teachers harassed the girls at school. … More than one girl from my school decided to get married. Even though I would have preferred to continue my education, I was compelled to marry in order to get rid of the harassment." One Syrian girl who was displaced with her parents to Beirut at the age of 12 was forced into marriage to protect her from further harassment after being targeted by a man in the community. " … In Lebanon, I was sexually harassed from an old aged man. When my parents knew, they forced me to get married. Now, I have a child, and I am pregnant. I am unhappy." Early, child and forced marriage is a form of SGBV because it forces girls to have sex before they are old enough to consent and before their bodies are physically mature. In some cases, the outcomes are extreme as in this example shared by a married Lebanese man in Beirut . "...My daughter is only 9 or 10 years old. The employer sees my daughter, and he asks to either marry her or to buy her. I accept since I don't have any other solution, but my wife starts shouting and crying. We got a clerk and married the employer to our child, and he took her as his wife … Our daughter died the next day from an extensive hemorrhage that was the result of brutal intercourse. He is a monster. Our child died. This is a true and very sad story. I hope, by documenting and sharing this story, to spread awareness to the public." Additional private SGBV risks: intimate partner violence and domestic violence within marriages In addition to experiencing early marriage as a form of SGBV, married Syrian girls were sometimes exposed to violence in their homes after being married at a young age. In the private sphere, intimate partner violence against women and girls in their own homes was commonly reported. Other girls experienced domestic violence perpetrated by other family members, while a few girls were sexually exploited by their parents, often under the guise of marriage. IPV was a prominent form of SGBV discussed by respondents in this analysis. Much of the narratives around IPV were linked to a power differential since the bride was a child/minor and/or was forced into a marriage. For instance, one 15 "They got her married to a Lebanese guy. After a while of their marriage, which lasted less than a month , the guy and his family started hitting her, she ran away to her parents' house, but they did not welcome her because she is a married woman now. However, the neighbors pressured her parents to let her into the house, but after 3 days the girl disappeared. A neighbor they know saw her in Beqaa, but she acted as if she didn't know or see him. And he finds out that she is working in prostitution; her parents are trying to get her back now." Theme 2: gendered differences discussing SGBV risks SGBV risks for female Syrian refugees were discussed differently by female and male respondents. For instance, experiences and fears of sexual harassment were mainly reported by women and girls, with 48 of 70 female narratives presenting sexual harassment as the primary form of SGBV in Lebanon. There were no accounts of sexual exploitation shared by women or girls in this study. This unmarried Syrian girl in Beqaa and her school friends were harassed while waiting for the bus (age [13][14][15][16][17]. " … My friends and I, who are waiting for the bus, are all girls and there were children amongst us as well. During the walk, a car with black tinted windows started following us; every now and then it would shine the front lights and honk. When we reached our camp, the car drove off. My friends and I were terrified." Another unmarried Syrian girl in Beqaa (age [13][14][15][16][17] reported how unsafe she felt as a girl in Lebanon due to harassment, comparing her situation in Lebanon to that in Syria. " … We are living in tents and I cannot go out and walk around. I can't even smell the fresh air . I am not even sad that I cannot go out in the city because there are young men that sit outside and I am scared to go outside because they bother us and I am a girl, I do not want to go out." In contrast, only 8 of 42 male respondents discussed sexual harassment against Syrian girls. Instead men were much more likely to talk explicitly about sexual assault and sexual exploitation . For instance, this unmarried Lebanese man from Beirut reported a story on sexual abuse where one man wanted to marry a Syrian girl, but her father refused. "The guy started threatening to kill her father if she does not marry him. Every day, a different man would sexually abuse her until the girl, 17 years old by that time, broke down mentally and she started talking to herself, prompting everyone to call her crazy. Furthermore, this man and his group of friends surrounded the house, banning anyone from coming in or out. Her father tried leaving the house once, but they would not let him. The girl contemplated suicide because of what she has suffered." One Syrian man (age [25][26][27][28][29][30][31][32][33][34] in Beqaa shared a story about a 16-year old girl who was forced into commercial sex work to financially support her family. "I know a Syrian family that displaced to Lebanon. They have three daughters. When one of their daughters was 16 years old, she was compelled to work in sinful routes due to the family's circumstances. She got raped, and she worked in prostitution. Now, she is a drug addict, and she is 18 years old." In some cases, the dire economic need led to coerced commercial sex work. For instance, a married Syrian man located in Beqaa (age [25][26][27][28][29][30][31][32][33][34] shared a story about a 17-year old Syrian girl who was forced into commercial sex work in order to pay for rent: "They came here after the war. She lived with someone here in this area. He abused her age and bad situation and made her work in prostitution. It started as a secret, but then we found out what was happening with this girl. He is abusing her in return of a little amount of money, let's say for rent maybe." Similarly, a Syrian father in Beqaa shared a story about Syrian girls who were promised job opportunities but ended up as commercial sex workers. "I know three girls who came from Damascus. They are students at school; the youngest one is 18 years old to 16-17 years. They are under the age of 20. They came to work at restaurants and cafes, but they used them and took them to prostitution places, for the sake of money. Several girls came, not only one." Sexual exploitation in the form of short term, contractual marriages was also discussed as illustrated by this Syrian father from Beqaa (age [35][36][37][38][39][40][41][42][43][44]. " … She is getting married for two weeks or a month, and then she'll get divorced. I asked one of the girls how is she? She told me that she resents herself. She is being exploited. They are continuously getting married, and they are still minors. Even in her few weeks with one of her husbands, she'll be abused. … I know a girl who has been married 6 times. This is wrong." --- Discussion In this analysis, SGBV was identified as an important issue by both women and men, even though the study design and methodology did not include any prompting about the topic, neither in the study introduction nor in the questions. Given the methodology, experiences and threats of SGBV arise naturally from a broad spectrum of lived experiences. Sexual harassment in the public sphere was the most common form of SGBV identified in this study, followed by sexual abuse and exploitation, which is consistent with previous research [16,17,48]. Despite facing a range of SGBV threats, Syrian women and girls also demonstrated their agency by making strategic decisions to protect themselves from SGBV threats . It is important to recognize their strength and resourcefulness despite living with very constrained choices. In Lebanon, Syrian refugees are largely integrated within their host communities, as opposed to living in formal refugee camps, where SGBV risks have been better documented [6,7]. Hosting Syrian refugees has strained Lebanon's resources and infrastructure, which has created tension, and at times hostility, within the host community towards displaced Syrian families [12,49]. What this analysis adds to the existing literature is a more nuanced understanding of SGBV risks faced by refugees who are integrated into such a host environment. This is especially relevant since most of the perceived SGBV threats for girls were derived from the host community itself . For instance, since work permits are restricted in Lebanon [50,51], some Syrian women and girls are forced to work to meet their families' basic needs. Since this work is usually in the informal sector, it can increase their risk of sexual abuse and exploitation [52]. Furthermore, since some Syrian women and girls are not registered in Lebanon, they are understandably hesitant to report sexual assault since reporting could lead to detention and deportation [15]. Women and girls also faced SGBV risks in their private lives with significant threats reported from husbands, from the husband's family, and from the girl's own family. IPV perpetrated by husbands was the most common form of violence in the private sphere, consistent with other existing evidence [18,53]. IPV is thought to be exacerbated among displaced Syrian families by the frustration that some men experience when their lives are disrupted, when their pre-existing roles as breadwinners within the family are altered, and when traditional gender norms are challenged [16,54]. Girls who marry young and have a large age gap with their spouses seem to be more at risk of IPV related to systems of power [39,55]. Restricted educational opportunities and insecurity increases girls' dependency on men, thereby limiting their agency and perpetuating patriarchal gender norms [39,56]. While Syrian men may have lost some of their patriarchal power due to displacement, women and girls are often challenged to take on more responsibilities, both inside and outside the home [37]. These new gender roles, as well as an unsafe environment and severe economic strain may increase stress levels and contribute to men perpetrating IPV. Increased levels of IPV have been reported during other periods of armed conflict in the region included during the second Intifada in the West Bank and Gaza [29] as well as among Palestinian refugee communities in Lebanon [57]. --- Continuum of SGBV risks Our analysis highlights that female Syrian refugees are at heightened risks of SGBV in their host communities. However, the types of SGBV experienced appear to differ across the adolescent and early adult years. As earlier risks of SGBV, sexual harassment and sexual assault were experienced in the public sphere limiting girls' mobility as well as access to education and work. It has been documented that if a girl experiences sexual abuse, her risk of subsequent sexual and physical victimization is significantly increased [58] and this phenomenon was evident in the current research. As a result of shame and dishonor, these SGBV threats and experiences in the public sphere lead to other SGBV risks in the private sphere. For instance, unmarried Syrian girls who are perceived to be at risk of sexual harassment and sexual assault in the community are sometimes pressured or forced into child marriage as a means of "protecting" them. Child marriage is itself a form of SGBV since girls are forced to have sex before they are able to give consent, and sometimes before their bodies are physically mature. In essence, some families exchange the threat of sexual harassment or sexual assault in the community for another form of SGBV, namely child marriage, as they believe that it will protect the honor of the girl as well as the honor of the family [59]. The cultural and gender norms that render girls unmarriable and their families dishonourable in the event of sexual harassment [60], sexual assault or pre-marital dating, directly contribute to girls experiencing the next type of SGBV, i.e. child marriage. After their marriages, sometimes to older men with the age inequality further compounding the gender inequality, many child brides face the additional threat of IPV in the form of physical and emotional abuse. Again, this phenomenon has been well documented in previous research [34,61]. Leaving violent relationships can be challenging particularly for young brides as there is sometimes family and societal pressure to maintain the marriage despite violence. If women attempt to leave their abusive relationships they may then be at risk of abuse by their own families including physical and emotional abuse as well as sexual exploitation, as they are often blamed for their failed marriages [62]. Children born into this cycle of violence are more likely to witness IPV and to experience DV [63], with this exposure increasing the risk that the children will personally experience and perpetrate violence later in life [64][65][66], thus continuing an intergenerational cycle of violence. --- Gendered differences discussing SGBV risks Although SGBV was noted to be an important issue for both men and women there was a gendered difference in how SGBV risks were discussed. Female respondents mainly described experiences and/or fears of sexual harassment in the public sphere and talked about the resultant limitations in their daily movement and freedom. Among women and girls, the notable lack of dialogue around IPV and DV aligns with previous portrayals of IPV and DV as private issues. Challenging the privacy of SGBV and attempts to address/report it has sometimes been perceived as a betrayal of family and a violation of social cohesion [67,68]. In contrast, men more explicitly and openly discussed sexual assault and sexual exploitation. In some cases, male respondents identified commercial sex work as a form of sexual exploitation, often related to the family's dire economic needs similar to other research [69]. It is noteworthy that there were no cases of sexual exploitation mentioned by women or girls despite interviewing men and women from the same communities. This may be secondary to the stereotypic differences in gender socialization whereby women are raised to shy away from talking about sex, but boys are socialized with less taboo around sex [70]. If the analysis had only included narratives from female participants, the more overt forms of SGBV such as sexual exploitation and abuse would not have been appreciated. It is also likely that SenseMaker contributed to this finding since it asks broad, open-ended questions, thus allowing participants to decide what they would like to share. This observation may be relevant and have implications for the design of other SGBV research. Since many SGBV studies focus exclusively on women and girls, they may be missing important findings and our experience in the current work would suggest that researchers should consider additionally including men and boys, particularly if stereotypical taboos around sex are anticipated to prohibit women from sharing their experiences / concerns openly. --- Strategies to address SGBV The study's findings demonstrate a varying array of SGBV threats faced by Syrian women and girls in Lebanon, and these threats require a multi-pronged prevention and protection strategy. --- Consider protection strategies in the context of refugee integration into the host community First, protection strategies must consider particular SGBV risks arising from the integration of Syrian refugees into host communities, and how inherent tensions likely contribute to feelings of insecurity and perceived SGBV threats. In Lebanon's ITS, Syrian women and girls have limited access to essential SGBV programs due to lack of transportation, as well as cost and gendered expectations around mobility and domestic responsibilities. Recognizing that Syrian women and girls are a hard-toreach population, a few organizations have piloted mobile service provision and an initial evaluation of one model showed promising potential to increase refugee's social networks, knowledge and self-confidence [71]. A replication of such mobile services nationwide may therefore have the potential to reduce barriers to SGBV services. Additionally, these programs should serve as one-stop shops for women and girls where they can access different services all in one place, while also following the GBV Call to Action 2016-2020 in emergencies by providing a minimum package of legal services, case management, and referral. --- Addressing patriarchal social norms Second, addressing SGBV must challenge patriarchal social norms and the socialization of men towards the use of violence, in addition to raising awareness about the negative effects of SGBV. Engaging men in SGBV prevention programming is critical to challenging social norms, attitudes and practices that increase the risk of SGBV while also harnessing positive male power to prevent violence and promote safety. Veale et al. [72] found that, through guided discussion on masculinity and gender roles, Syrian men talked about their changed circumstances as refugee men and fathers. In this work, a safe emotional space was created for men to collectively talk about their problems, to become more attuned and reflective about their relationships with their wives and children, and to engage in better emotional regulation. Through dialogue, men started to distance themselves from earlier responses of anger and frustration and became more empathic and reflective on their own needs as well as those of their families [72]. Several other initiatives have been piloted in Lebanon targeting men and boys as a means to alleviate the burden of some forms of SGBV, and in 2014 the Men Engage Lebanon Country Network was launched [73] to harmonize these efforts, while ensuring minimal duplication of interventions. Little evidence-based curricula are currently available to guide front-line staff on how to engage with men and boys and future efforts should prioritise this gap to produce context specific and culturally sensitive tools. --- Addressing structural drivers of SGBV including legal reform Lastly, in line with Heise's ecological model, which conceptualizes SGBV as being perpetuated through multilayered factors including intra-individual level, microsystem, exosystem and the macrosystem, programming efforts should also address the overarching structural problems that contribute to SGBV [53]. In their adapted social ecological model for GBV against Syrian women in Lebanon, Yasmine and Moughalian [74] argue that interventions designed to focus on the intrapersonal level fall short from effectively influencing a change. They recommended that programming should aim at challenging existing power dynamics and urged organizations to look at structural determinants that contribute to SGBV during forced displacement, such as lack of employment, dire economic situation, as well as security and protection concerns [74]. For instance, Lebanon adopted legislation in 2014 aimed at better protecting women from SGBV, but feminist and women's right organizations argue that the law falls short of addressing all forms of SGBV and more robust legislation is called for [48,75] especially when it comes to the protection of Syrian and other refugee populations. Further legal reforms are also necessary including laws against street harassment and work protection laws, as well as modifying the age requirement for marriage to 18 years. Additionally, policy reform to facilitate legal registration and obtaining official documents is urgently needed in Lebanon since women and girls are vulnerable to sexual exploitation and assault, yet currently have little mechanism for recourse without risking detention and deportation. --- Strengths and limitations This qualitative study has several limitations. First, the sample is not representative of the Syrian refugee population in Lebanon and therefore the results are not generalizable. Girls less than age 13 were excluded and since this study is qualitative in nature there is no consideration of the subgroups' balance. Second, due to time and human resource constraints we were not able to analyze all male narratives, and this may have also introduced a bias. However, a systematic sampling method was utilized in an effort to be objective. Third, the SenseMaker narratives are often quite brief, and without the follow up questions typically asked in qualitative interviewing, the shared stories sometimes lack the detail and richness expected of in-depth qualitative interviews. Finally, it was sometimes not clear whether the SGBV experiences described were perceived to result from being a Syrian refugee, from being a female or from a combination of the two. The study also has a number of strengths worthy of mention. Since the SenseMaker survey did not ask directly about SBGV, participants shared their stories more naturally as part of their broader experiences. Moreover, the current analysis offers a variety of perspectives from women and girls of different ages as well as from men and it illustrates the various SGBV risks faced by female refugees who are integrated into their host communities rather than housed in formal refugee camps. --- Conclusions Our findings shed light on the importance of recognizing the impact of SGBV on the family as a whole in addition to each of the individual members, and to also consider the cycle of SGBV not only across the woman's lifespan but also across generations. Furthermore, this analysis contributes to our understanding about the SGBV risks faced by female refugees integrated into the host community rather than hosted in formal refugee camps, highlighting how SGBV is intimately related to other aspects of social welfare as well as being part of the everyday lives of Syrian girls. Syrian girls are exposed to early SGBV risks in the public such as harassment and assault, which sometimes leads to child marriage as the next form of SGBV experienced. Within early / forced marriages, some girls face additional SGBV risks in the private sphere in the form of IPV and/or DV. In this analysis, female respondents were more likely to share stories about sexual harassment, while male respondents were more likely to discuss explicit forms of SGBV such as sexual exploitation, and this may have implications for the design of future research focused on SGBV prevention and response. --- --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: A myriad of factors including socio-economic hardships impact refugees, with females being additionally exposed to various forms of sexual and gender-based violence (SGBV). The aim of this qualitative analysis was to understand and to provide new insight into the experiences of SGBV among Syrian refugee women and girls in Lebanon.The data are gained from a larger mixed-methods study, investigating the experiences of Syrian refugee girls in Lebanon, using an iPad and the data collection tool, SenseMaker®. The SenseMaker survey intentionally did not ask direct questions about experiences of SGBV but instead enabled stories about SGBV to become apparent from a wide range of experiences in the daily lives of Syrian girls. For this analysis, all first-person stories by female respondents about experiences of SGBV were included in a thematic analysis as well as a random selection of male respondents who provided stories about the experiences of Syrian girls in Lebanon. Results: In total, 70 of the 327 first person stories from female respondents and 42 of the 159 stories shared by male respondents included dialogue on SGBV. While experiences of sexual harassment were mainly reported by women and girls, male respondents were much more likely to talk explicitly about sexual exploitation. Due to different forms of SGBV risks in public, unmarried girls were at high risk of child marriage, whereas married girls more often experienced some form of IPV and/or DV. In abusive relationships, some girls and women continued to face violence as they sought divorces and attempted to flee unhealthy situations. Conclusions: This study contributes to existing literature by examining SGBV risks and experiences for refugees integrated into their host community, and also by incorporating the perceptions of men. Our findings shed light on the importance of recognizing the impact of SGBV on the family as a whole, in addition to each of the individual members and supports considering the cycle of SGBV not only across the woman's lifespan but also across generations. Gendered differences in how SGBV was discussed may have implications for the design of future research focused on SGBV.

social care and support, which may involve multiple formal and informal caregivers [8][9][10]. In the community setting, care and support for frail older adults is often fragmented and uncoordinated, leaving them at risk for negative health outcomes [11,12]. Care should ideally be based on a formal assessment and tailored to older adults' needs and preferences, as well as integrated [13], whereby interprofessional collaboration and coordination between all relevant caregivers is leveraged to support frail older adults to age in place [14][15][16]. To help the aging population remain living at home despite their high care needs, and to avoid costly institutional care, there will be an increasing need for both health and social care from formal and informal caregivers [8,9]. Health care services include "acute, chronic, preventive, restorative and rehabilitative care", delivered by various providers [17], while social care includes a wide variety of services which provide "physical, emotional and social support to help people live their lives" [18]. Formal care at home includes health or social services provided by paid and trained professionals, such as home care nurses or household services [19]. Informal care occurs when care is provided without payment or formal training, typically provided by a spouse, children, family and friends or neighbours [10,19]. Informal care includes assisting with activities of daily living , or instrumental , assisting with medical or nursing tasks, or providing emotional support [10]. Due to the challenges which result from fragmentation between health and social care, integration has been widely promoted [20][21][22]. Both formal and informal care are well-researched; however, only more recently have researchers and policy-makers considered the intersection of these two approaches to caregiving for home-dwelling frail older adults [19,[23][24][25][26], an area of growing interest [24]. One study suggests that while nonfrail older adults used informal care often as a substitute for formal care, frail older adults appeared to use both in compliment [8]. Although formal and informal caregivers should ideally work together, this is another gap recognized in community-based care for older persons [19,24]. Bridging these "problematic divides" between health and social as well as formal and informal care is important when moving towards care integration [16,27], i.e., optimally collaborating and communicating on aspects such as shared decision-making and care planning [28,29], and all caregivers fulfilling their key roles in supporting the older person according to their needs and preferences [5,16]. Therefore, when planning future care services for the aging population, it is helpful to first understand the specific sources and contributions from formal health and social care as well as informal care and social support used and preferred in future by the frail population. Such insight can help allocate resources and organize services which are coordinated and delivered around the needs and preferences of older adults [4], as well as identify and collaborate with local stakeholders who will become increasingly involved in caring for older adults in the community. The various individuals involved in providing care and support to help meet the needs of home-dwelling frail older adults have often been studied as "care networks" [5,8,26,[30][31][32], or more recently as "care convoys" [23,33]. Researchers identified the diversity within the structure of care networks or convoys [26], reporting multiple different combinations of informal and formal care use [33,34], and occasionally explored whether frail older adults perceive their care and support to meet their needs [5,33]. As shown by Verver et al. 's study, 33.7% of frail individuals living independently did not have the care and support that they needed e.g., social contacts or domestic help, even though they had more formal care providers and were more likely to have informal care providers involved than their non-frail counterparts [5]. Lambotte et al. also noted that a frail person's satisfaction with his/her care convoy did not necessarily mean they had sufficient help, and vice versa [33]. Although these needs are bound to increase over time and would likely need to be iteratively re-assessed, it is important to understand in what ways the care and support of frail older adults are meeting their needs and to detect any gaps. Identifying these gaps and determining how to engage and support those living with unmet needs should be a priority given the risk and vulnerability associated with frailty [35,36]. The present study is part of the larger INSPIRE parent study taking place in one canton, Basel-Landschaft , an area in the Northwestern part of Switzerland. A cantonal care law enforced in 2018 ensued that older adults living at home will have access to a new information and advice center for advice related to care and nursing in old age, as well as an assessment of needs, and either mediation of care or potential nursing home referral [37,38]. The INSPIRE project aims to develop, implement, and evaluate an integrated care model for these IACs [38]. During the development phase, a contextual analysis was conducted which included a population survey [39] followed by interviews with older adults [40] to create an IAC care model which was suited to local health and social needs and preferences. More information on the parent study can be found elsewhere [38][39][40]. Using a quantitative approach, we investigated the type and frequency of formal health and social care as well as informal care and social support that frail older adults are currently using and their future preferences, and to what extent the older adults perceive their support in place meets their current needs. Subsequently, in the qualitative arm, we aimed to gather a more in-depth understanding of their experience and preferences with their care and support, and explore appearance of integrated care concepts . Using a mixed methods approach, we unified this data to explore the types of care and support used by home-dwelling frail older adults as well as their experience and future preferences. --- Methods --- Study design We used an explanatory sequential mixed methods design [41]. First, we conducted secondary analysis on a subset of data from the cross-sectional INSPIRE population survey [39] . To help explain and expand on these results [41], we used data from interviews with older adults in the INSPIRE parent study [40] . --- Phase 1: Quantitative Sample The current study included a sub-sample of frail older adults from the INSPIRE population survey in the parent study. In the population survey, the Groningen Frailty Indicator [43,44] was used to assess the geriatric risk profile of participants , as the GFI considers frailty to be a multidimensional construct which includes physical, psychological, social, and cognitive domains [44,45]. Using this broader definition to measure frailty and determine those at risk of institutionalization based on their geriatric risk profile implies that this sub-sample represents a group experiencing frailty in more diverse areas than solely biomedical indicators [46]. Regardless of whether participants answered all 15 GFI questions, the quantitative arm of the present study included only those with a GFI score of 4 or more , resulting in a sample of n = 2314 frail older adults. --- Variables and measurements We selected variables from the INSPIRE population survey [39] . Except for sample characteristics, variables are presented according to adapted domains and concepts of the SELFIE framework, to stay consistent with the organization of results [28]. The SELFIE framework has been commonly used in European studies to support development, description, implementation and evaluation of integrated care initiatives, and emphasizes important concepts within integrated care, such as presence of a care coordinator [28]. Socio-demographic information and frailty status Participants' socio-demographic characteristics were summarized, including age , gender, education level, monthly household income, and household members. Additionally, the geriatric risk profile was determined by the individual's GFI score. The GFI is a reliable and valid 15-item instrument for frailty screening [43][44][45]. A GFI score of greater than four indicates frailty [44]. Individual with multi-morbidity and their environment Needs : one investigator-designed question with a "yes/no" option measured whether their support received in everyday life meets their needs. Social network: availability of social support was assessed using the validated German version of the Brief Social Support Scale [47]. There are six items to be rated on a 4-point Likert scale, divided by tangible support and emotional-informational support . Responses were dichotomized for the analysis. Use of community services was measured through two questions to capture the types of community services needed or used in 2018 and services which would be considered if they become more in need of help in the future. Workforce domain Informal caregiver support in everyday life and preferences for future was captured through two questions designed by the research team which contained different options for sources of support . Whether participants were caregivers themselves was determined through one binary question designed by the research team which asked if they looked after, cared for or supported another person . Use of health care professionals in 2018 was measured using three questions which examined frequency of General Practitioner visits, specialist visits, and an open-ended question for other medical services used. Use of care organizations was measured through two questions to capture the current care organizations providing regular support in everyday life and care organizations preferred in future if they become more in need. --- Statistical methods Consistent with the INSPIRE population survey in the parent study [39], selected socio-demographic variables were descriptively analysed and reported as frequencies, percentages, medians, and interquartile range to describe the sample of survey participants. We also included a description of household members to support interpretation of care use. Frequencies were reported for dichotomized or categorical survey variables. However, many of the survey questions were set-up to allow each participant to give multiple responses per question. For these questions with multiple response options, the proportion of respondents accounted for by each item were reported. When respondents provided inconsistent answers , we excluded them from analysis within the respective survey domain . To analyze three of the survey variables, use of community services, informal caregiver support, and care organizations, we first dichotomized the responses to indicate whether each type of support was currently used or not. We then selected only those who provided an answer for both current use and future preferences in order to compare them, excluding those who did not provide an answer for future preferences. Following principles of a sensitivity analysis, we analyzed our data with and without the excluded individuals to confirm that the results were not impacted. Finally, we separately analyzed those who did not currently use the support, but provided data on future preferences, to further inform future predictions. Missing data was assumed to be missing at random for all questions and therefore excluded, but amount of missing data was reported throughout. A brief sensitivity analysis was performed to address how we calculated the GFI score. We calculated the total GFI score by summing each score of "0" or "1" to the 15 items that comprise the GFI. We categorized anyone with a GFI score greater than or equal to four as frail, regardless of how many total GFI-items they answered, to avoid losing power. We analyzed to see if the results would have been different had we required an answer to all 15 questions, but values remained consistent. Analyses were performed using IBM SPSS Statistics . --- Phase 2: Qualitative Sample Using purposive sampling, semi-structured interviews were conducted in the INSPIRE parent study between September 2020 and August 2021 with eight homedwelling older adults with multimorbidity, defined as the occurrence of two or more chronic diseases [48]. Furthermore, individuals had to be using health services provided by at least two care organizations, or three or more different health services provided by one organization [40]. In the current study, we included interview data from individuals who indicated frailty based on the PRISMA-7 frailty screening assessment [49] and considered the sample size to be adequate based on information power , i.e., whether the sample size was sufficient to contribute knowledge in response to the research question, determined by aspects such as the study aim, sample specificity and whether the study is theoreticallyinformed [50]. The PRISMA-7 includes seven dichotomous questions, and was completed by the researchers according to the participants' responses and interactions with them during the interviews, as well as the demographic data collected. The lead author screened the original interview participants' anonymous PRISMA-7 scores to identify eligible participants for the present study. The PRISMA-7 assessment instead of the GFI was used in the interviews to reduce burden on the participants, given the majority of the information collected through the PRISMA-7 was easily observable by the researchers or already embedded within the interview questions. --- Instruments The interview guide was developed in the parent study to build on the INSPIRE population survey findings and further explore older adults' health and social needs and experience of their care and support. We incorporated additional concepts which are key to integrated care from literature such as the SELFIE framework into the interview guide, to get a sense of the presence of these concepts in their current care [28]. --- Analysis Before the anonymous interview transcripts were translated into English , the original German transcripts were cleaned from all filler words and Swiss-German nuances were translated to High-German. Validation of content was performed only for discrepancies between the two German dialects, or between the German and English languages. Applied Thematic Analysis was then used to analyse the transcripts [51]. First, the lead author created a research map to establish structural codes based on the domains in the interview guide. Next, the lead author performed structural coding on the data using NVivo [52], to organize the data by the structural codes , which led to a coding report for each structural code [53]. Thereafter, content coding was performed, to analyse the data within each structural coding report [53]. A separate codebook was created for each structural code to contain all content codes. For each structural coding report, a memo was written to describe the content codes and help to derive themes. --- Techniques to enhance trustworthiness Given the nature of this study, we considered the following techniques to enhance trustworthiness [54]. Context The original interviewer made the context of participants' responses fully available through providing access to the transcripts and a thorough written description of the interview setting. Consistent with participants' preferences, most interviews took place in the individual homes of the participants in Canton BL, with one interview taking place at a nursing home during a short stay. Some participants had a family member present during the interview, while others were alone. In most cases, there were two INSPIRE interviewers present. Rigor In aiming for consensus, initial results were presented to the larger research team and original interviewers throughout the analysis, to gather feedback and input based on their methodological, clinical and local expertise. A precise codebook was also developed. To maintain an audit trail, all notes, memos, changes to the codebooks, and analytical documents which were prepared during the study have been archived. --- Data integration For analysis, we organized the quantitative and qualitative findings according to the adapted domains and concepts of the same theoretical framework, SELFIE [28]. At the interpretation and reporting phase, the quantitative and qualitative findings were first integrated through a joint display table and later synthesized via weaving [41,55,56]. With the goal and principles of conducting a value-adding qualitative analysis in mind [57], the original interview findings from the INSPIRE parent study were occasionally included in the results of this paper, to help explain the survey findings and allow for discourse and reflections targeting integrated care. --- Ethical consideration For ethical review, the population survey was submitted to the Ethikkommission Nordwest-und Zentralschweiz in Switzerland, BASEC Nr Req-2019-00131. It did not meet the definition of a research project requiring further review as per the Human Research Act ART.2, and was able to move forward as it met the general ethical principles for research involving humans . Data collection for the interviews was approved by the EKNZ under Project ID: 2020-01755. To conduct the mixed methods study, a clarification of responsibility was submitted to the EKNZ, however it did not require further review . --- Results --- Survey participants In total, 2314 INSPIRE population survey participants were eligible for this study, but not all participants e a monthly income of 2459CHF was used as the threshold to consider a person at-risk-of-poverty in Switzerland for 2018 for a single person [58] f Multiple responses possible, therefore percent of respondents shown g combined response for other adults, professional help or other h possible GFI score range: 4-15 --- Characteristics % Median [IQR] Age 83 [79- Live with siblings 0.2% Live with adult children 3.2% Live with other g 3.4% GFI Score h 5.0 [4][5][6] responded to each survey question . There were 594 participants who provided a response to all survey questions discussed in this paper. Participant ages ranged from 75 to 107 years with 60.3% being female . The median GFI score was 5. --- Interview participants The characteristics of the seven interviewees in this study are reported as part of Esser et al. 's study [40]. In summary, four of the interview participants were men; most were single and living alone; had an education level of vocational training or higher; and had a range of four to eight chronic diseases. The mean PRISMA-7 score was 5.6 . The integration of quantitative and qualitative findings through a joint display table informed the presentation of results below. --- Individual with multi-morbidity and their environment Needs and preferences Overall, most survey respondents perceived that the support they receive in everyday life meets their needs . Interviewees described care and support from multiple services or professionals as going well, such as physiotherapy, social, or home care services, the latter described by one as "impeccable" . As also acknowledged by Esser et al. , all discussed their strong desire to continue living at home [40], "for me it has always been my goal in life to avoid the nursing home" . Interviewees were pleased when they received personal care which helps them to meet their goals "it's nice when you find the person who can treat you individually. That is a gift" , but also recognized that personal care "is not easy at all, because every person is special in their own way" . Interestingly, while a lack of continuity was occasionally discussed as challenging, others experienced this not to be problematic. One participant pointed out the gaps she sees for others "But where I actually have a big problem … is that I know people who are in a similar situation to mine, who are alone and have no one to help them" . Interviewees mostly felt that an overall assessment of their needs was often not performed but two could see this would be beneficial, while a few believed this had been performed by their home care service. Social network Availability of social support was lower for the 'tangible' support items ranging from 44 to 70%, and higher for 'emotional-informational' items . Interviewees gave examples of receiving both types of tangible and emotional-informational support . Moreover, it was discussed that sharing experiences and maintaining a social network was important, but it was easier for some than others, "I have a lot of visitors. And I didn't know that when you look after your friends that it would come back one day. And it does come back" . Use of community services Use of at least one community service was 58% ; meanwhile a large proportion reported that they did not need help in 2018 . Of those using services, most participants appeared to be using multiple different services. Among current and future service users, the highest frequency of responses for current use was for help with the housework . Care and assistance at home was the most preferred choice in future , even for those who did not use current services in 2018 . Interviewees often described their need for support with housework and home care, which also benefits them in additional ways, such as motivating them "if they didn't come tomorrow … I don't get dressed or I don't wake up and I don't open shutters" . Most types of service use doubled between 2018 and future preferences. Interviewees described various purposes for transportation services. There was currently a low use of meal services despite that many reported they never had someone to help prepare food for them when they are not able to in the Brief Social Support Scale questions, yet strong demand for meal services in future . --- Workforce domain of the SELFIE framework Informal caregiver support Three-quarters of the survey respondents indicated having at least one source of informal care . Of those receiving informal care, 71% of respondents relied on one source while 29% relied on two or more sources listed, and interviewees likewise described diverse informal care sources. Support from family members of the same age will continue to be sought in future by those who will still want informal care, but the largest increase was seen for younger family members. Of those who did not use informal care in 2018, 55% preferred younger family members provide them support in the future if needed, with many also desiring support from others e.g., neighbours. As similarly identified by Esser et al. the roles of informal caregivers were diverse, with neighbors helping with mail, transport or checking-in on them, while relatives provided the widest support [40]. Some interviewees indicated complete reliance on informal caregivers, such as "so if I didn't have her [the daughter], and the son-in-law, then we would be, I would be lost" . Two described how their daughters help encourage their mobility or independence, and one interviewee described his appreciation for his daughter emotionally, "sometimes when she comes and when she leaves I have tears. Because she always helps me so much and does everything for me" . Yet on the other hand, some interviewees stressed the importance of avoiding burden on their family "I'm very happy that I have family but I didn't want to involve them in my care and in my dependence... I want my children to be free of this burden" . Meanwhile, M4 also implied that informal care would come first before being dependent on help. With respect to their own role in informal care, one-fifth of the survey participants reported also caring for someone else. Use of health care professionals Participants were asked how often they visited their family doctor and a specialist in 2018. The majority had 1-6 visits to the family doctor and a large proportion had 7-10+ visits. During the interviews, GPs appeared quite central in the older adults' discussions about their care, and were often reported as the main contact person for their health questions, as also seen in Esser et al. [40]. Likewise, for specialist visits, the majority also had 1-6 visits, with a decent share having more than 7 visits. Thirty percent indicated they used physiotherapy services in the past year, and many provided at least one response when asked in an openended question about any other medical services they used in 2018, such as a dentist or eye doctor. All interviewees used physiotherapy, who on top of mobility support and exercises, could provide unique value in helping them to practically cope with their situation and for example, taking them for walks in the forest where they would otherwise be restricted to get to. This same interviewee mentioned her appreciation for the approach taken by her physiotherapist "But that she looks at my physiotherapy holistically … she always asks how I'm doing, she really asks, and this morning she said I should show her how I get out of bed and back in again" . Interviewees often discussed the importance of having interesting or meaningful conversations with professionals , and also highlighted when this relationship or obtaining support for tasks has been problematic for them due to language barriers or lack of continuity. --- Use of care organizations The survey question about current use of care organizations had the largest proportion of missing data, as 56% of participants did not answer; however, we assume this is because there was not an answer option available if no organizations were needed. There were 44% of survey participants using at least one organization in 2018, and of those, most reported only using one. An interviewee recognized the unique situation they are in to be aging in Switzerland, "we in Switzerland are actually very well provided for with these organizations and associations" . Yet, another recognized that even a good system can be improved. Private help was most commonly used in 2018 for those indicating both current and future organizational use, although slightly less preferred in future . There was an increase in responses for almost all care organizations, with nonprofit aid more than doubling in value, and still anticipated by almost all who did not respond about current organization use. Interviewees described many formal caregivers from organizations who were involved in their care, providing support with basic and instrumental activities of daily living, as well as household services or taking them out for a walk. Interviewees discussed how they can also count on these care organizations, and their dependence on these organizations, as one described they need accompaniment now to go out for a walk, "so I walk with this bodyguard" . Two of the interviewees brought up challenges during the initial phase of receiving support from care organizations, however expressed that these problems improved over time. Multi-disciplinary team While there was no survey data on the remaining topics, additional integrated care concepts were touched on during the interviews. There was no emergence of strong themes in this section; however, the occasional references made to these concepts shed light on integrated care. When asked about cooperation between providers, one interviewee could imagine challenges with providers trying to cooperate together "there only has to be a family doctor who has no time and says: 'Who pays me? What's going on? Now I have to go out to the house' . … 'and then we have scheduling difficulties, and who pays for that?'" . Still in relation to cooperation, one imagined that their professionals communicated and two felt there was good cooperation between their home care and their GP. With respect to coordination, interviewees could envision the benefits of bringing all providers together to discuss planning their care, "I could even invite the priest, the oncologist and the psychiatrist … and of course the doctor and home care. That they come and that this [future care] is managed" . Named coordinator Esser et al. also determined that many interviewees felt no one, except for themselves perhaps, had an overview of their situation [40]. While some individuals could imagine this to be beneficial, many were still capable of booking their appointments by themselves or in collaboration with professionals or family members. One gave an example of a care coordination office offered by their home care support provider. Interviewees each mentioned someone different who initially organized their home care support, varying from a family member, their GP and a hospital social worker. In comparison to fragmented care one interviewee experienced in the hospital, she provided an analogy of how an architect builds something to describe how she could imagine coordinated care planning: "...when he had to build something, that he always had the carpenter, the electrician, the plumber, the bricklayer, all at one table. And he said: 'So, how is it going? Scheduling: you are dependent on that, he is dependent on that' . This is the only way to get to their goal" . --- Leadership and governance domain of the SELFIE framework Individualized care planning One interviewee discussed an interest in having a transparent discussion about care with all stakeholders: "questions come, 'nursing home? In need of care? Can the relatives still handle it?' . I think that should be discussed transparently. And it should be done at a round table" . --- Discussion Given the pressing need to organize community-based care and support which helps frail older adults continue to live at home [59,60], this study aimed to first understand their current and anticipated health and social care and support in context [61], using a population survey and interviews from the INSPIRE parent study. The synthesized data indicates home-dwelling frail older adults reported being supported from various formal and informal caregivers. Nevertheless, it is concerning that there remains a small subset of this population who have unmet needs. This is especially true given that the population of older adults in this Canton and Switzerland overall is anticipated to increase [62,63], along with the demand on formal and informal care as dependency increases [8,9]. When combining our current findings with existing literature, it points to focal elements which merit further attention to optimize integrated care [28], benefiting frail older adults, the system and reducing potential burden on informal carers. Most home-dwelling frail older adults in our study appeared to receive support which met their perceived needs. It should be noted that this was a subjective selfreported assessment of needs which we did not further evaluate objectively in this study, and that the question also did not differentiate between health versus social support, nor whether they deemed this support as satisfactory versus sufficient [33]. This low rate of unmet needs may be attributed to the reported formal and informal care which also helps them pursue their goal to live independently at home. There is an abundance of care organizations and services available for older adults in this Canton which is consistent with Switzerland generally [64] -a country renowned for its good health care system [65]. We postulate that this has contributed to why less than one-tenth of our survey participants perceived a need for more care and support, even when facing frailty. However, the 6.5% of frail older adults with unmet needs are of major concern. When comparing this to a study in the Netherlands, a much higher percent of home-based frail older adults in their survey reported more need for support, albeit their study included a smaller sample and a different measure of frailty [5]. Our results also showed that there are several frail older adults who are not using care organizations or services, suggesting that they are less dependent. A fraction of frail older adults themselves were even providing help to others, a finding that has been observed before in Switzerland and believed to "be a good indicator of people's health" [66] and helps maintain their "sense of independence" [33]. This finding also supports the notion of maintaining a strengths-based approach to frailty by also considering the strengths/resources of older adults living with frailty [33]. If dependence increases in our population, our findings indicate a potential increased demand on care organizations and services in future to meet the needs of frail older adults. Therefore, to avoid duplication of services and fragmentation in future, this implies that coordination and communication should become progressively important for them to function together effectively as a care network and also meet the needs of the frail older adult [26,28]. Use of information and communications technology could help support care coordination [28,67]. The synthesis of our findings support insight into the dynamics with involving informal caregivers in frail older adults' care. Some frail older adults reported reliance on their sources of informal care. However, despite increased need, others did not intend to use informal care in the future, due to a fear of burdening informal caregivers, as expressed by interviewees. Our finding corroborated with existing literature which has frequently reported older adults' fear of burdening informal caregivers or asking for help [5,[68][69][70]. Given the diverse roles of informal caregivers noted in our study and the expected demand in future, coupled with the well-known concern that informal caregivers are at risk of burden [10,28], informal carers -especially relatives -are an important target. It is imperative to intervene as caregiving can impact them in many ways, such as financially, emotionally and psychologically [71,72]. We therefore support Ambugo et al. 's suggestion that informal caregivers' needs should also be assessed when assessing the needs of frail older adults, and connected with support [73]. Informal caregivers have an important role from an integrated care perspective in contributing to the care planning and shared-decision making processes for older adults with multimorbidity and/or frailty [28,74,75]. Furthermore, integrated care may positively impact informal caregivers of frail older adults [76]. Some of the burdensome responsibilities of informal caregivers could be alleviated through professionals in the system, for example by the Information and Advice Center in the INSPIRE parent study, through helping them to find appropriate services within the system [71]. There needs to be a balance between following best practices with involving informal caregivers, yet ensuring frail older adults' preferences are taken into consideration for the level and timing of involvement, and the type of information shared with the informal caregiver. This points to a consideration that professionals or coordinators providing care to frail older adults should be aware of and consider how to navigate and manage. Our qualitative results regarding integrated care concepts were quite heterogenous and did not generate strong themes. For example, care coordination was a topic which seemed to be of greater familiarity to the researchers [40]. Nevertheless, the findings raised a few concerning points from an integrated care perspective, such as suggesting that the needs of home-dwelling frail older adults are not consistently evaluated, yet two interviewees could imagine that someone assessing their needs would be helpful. Integrated care guidance for older adults confirms that a comprehensive assessment of health and social needs is a key first step in the intervention of older adults with frailty [75]. Further, this assessment can help to identify priority conditions associated with declining intrinsic capacity, the type of care needed, and lead to creation of a care plan which is coordinated and tailored to their needs [28,75,77]. Aside from assessing their original needs, the literature recommends a key person has an overview of their situation, or that there is one consistent point of contact who manages referrals and coordinates care, among other duties [28,67,78]. Consistent with our quantitative findings, our qualitative data indicated these are relevant gaps in frail older adults' care, from our view as health researchers. Informal caregivers have also suggested that they would rather turn to one person to arrange care [71,79]. Our study largely confirms what is existing in the literature on these topics, but brought more light to the power of understanding each individual older adults' care situation in context, which is fostered by person-centred integrated care. These results have meaning on a local level for the IAC, and could also be applied to similar community-based services aiming to assess older adults' needs and support care coordination and integration, if appropriate in context. In the context of our research where the care law has requirements for the IAC , the IAC staff will also need to collaborate well with the many formal and informal carers involved, could maintain an overview of the frail older adults' situation, and help to relieve the potential burden on informal caregivers. Furthermore, by identifying each individual's formal and informal caregivers involved and understanding their roles, a named coordinator can for example "map a care network, " as a starting point for care planning discussions between the older person as well as all relevant caregivers, as suggested by Grol et al. [14]. As a next step from our research, we support building from Janse et al. 's work in aiming to capture the evolving dynamics between informal and formal care when studying integrated care of home-dwelling frail older adults [24]. Moving forward, the findings from our study support that the network of all formal and informal caregivers providing health and social care could indeed be viewed as a "care convoy", as put forward previously by Kemp et al. [23] and Lambotte et al. [33]. The "care convoy" concept acknowledges the overlap between formal and informal care, and embraces the complex and dynamic nature of care networks for this population [23,33]. When considering the properties of these care convoys, we also observed diversity in the structure and function , and that an individual's care convoy will likely evolve in future depending on their health and social situation and needs [23,33]. As highlighted by previous authors, given the outcomes tied to these care convoys , it is important to consider the contextual influences on these care convoys, for example, societal influences on role expectations which may differ [23]. --- Strengths and limitations Given the complexity and diversity of frail older adults' care, our study is one of the first mixed methods studies to gather a more comprehensive understanding of the types of care and support used, as well as experience and future preferences of home-dwelling frail older adults, as a precursor to implementing an integrated care model. Provided a rigorous approach is taken towards data collection and analysis during both arms of the study , this leveraged the strengths of both quantitative and qualitative approaches to address our research question [41,55]. Gathering a more in-depth perspective from the interviews helped to explain and expand on the survey findings and give us a better overall picture of the contributions and care dynamics involved with frail older adults' health and social care from formal and informal caregivers. However, our study comes with limitations which need consideration. First, as mentioned in the INSPIRE parent study, the non-random sampling strategy used for the population survey could result in biases, weakening the generalizability and transferability of our results [39]. Due to recruitment issues, perhaps related to the Covid-19 pandemic, the sample size for the interviews was limited [40] though as previously described in the methods, was deemed to be sufficient for our purposes. In the quantitative part of the study, we considered the percentage of missing data to align with the rates expected from research with our target population [80], except for the question on use of care organizations, which we assumed was a result of the answer options available. We also chose to analyze the data in a more segmented way to assess those who currently used care/support separately from those who were not currently relying on care/support, although this resulted in lower denominators for some questions. Furthermore, all interview respondents were using multiple services , which appeared to always include home care services, therefore not representing the segment of the frail older population with no home care support. While appropriate for our research question, this leaves uncaptured voices of those who need help and support but are not receiving it, which may also be a relevant point about our survey respondents. In addition, given that care is dynamic whereby different caregivers are relied on throughout time [81], this study is only capturing a snapshot in time. Nevertheless, it gives some insight into perspectives of our target population and points to areas which could benefit from future research. In hindsight, it would have been ideal to be able to capture more questions specific to integrated care during the INSPIRE population survey, but we were limited with the survey length in accordance with stakeholder input. The interviews provided the opportunity to expand and collect some qualitative information on topics where we were not able to collect quantitative data first. Finally, use of some formal care providers were not assessed in the survey. Future research that explores care networks of frail older adults should include all possible constituents. --- Conclusions Most frail older adults in Canton BL appear well-supported, receiving formal health and social care as well as informal care and social support from various sources. Given the anticipated demand for future care and support of home-dwelling frail older adults, we recommend that efforts are in place to prevent fragmentation between health and social care as well as formal and informal care. Further research could also explore those living with unmet needs and how integrated care models impact the dynamics within the care networks of homedwelling frail older adults. --- --- Abbreviations --- --- --- Funding --- --- --- Competing interests ---

Background: Home-dwelling frail older adults are often faced with multimorbidity and complex care needs, requiring health and social care systems that support frail older adults to age in place. The objective of this paper was to investigate the types of formal health and social care as well as informal care and social support used by home-dwelling frail older adults; whether they perceive their support as sufficient; and their experience with and preferences for care and support. Methods: Using an explanatory sequential mixed methods design, we first conducted a secondary analysis of a subset of cross-sectional data from the ImplemeNtation of a community-baSed care Program for home dwelling senIoR citizEns (INSPIRE) population survey using descriptive analysis. Subsequently, we analyzed existing data from interviews in the parent study to help explain the survey results using applied thematic analysis. Results were organized according to adapted domains and concepts of the SELFIE framework and integrated via a joint display table.Of the parent population survey respondents, 2314 older adults indicating frailty were included in the quantitative arm of this study. Interview data was included from 7 older adults who indicated frailty. Support from health and social, formal and informal caregivers is diverse and anticipated to increase (e.g., for 'care and assistance at home' and 'meal services'). Informal caregivers fulfilled various roles and while some older adults strongly relied on them for support, others feared burdening them. Most participants (93.5%) perceived their overall support to meet their needs; however, findings suggest areas (e.g., assessment of overall needs) which merit attention to optimize future care.Given the anticipated demand for future care and support, we recommend efforts to prevent fragmentation between health and social as well as formal and informal care.

Background Large variations in cancer survival exist between countries across a range of cancer types. Survival rates are generally lower in Denmark than in comparable countries [1,2]. Even within countries, survival rates vary much between patient groups with the same type of cancer, and for most cancers people with lower socio-economic position have poorer outcomes than their socioeconomically more affluent counterparts [3,4]. These variations between and within countries are undoubtedly multifactorial and complex, but a growing body of research suggests that differences in stage progression at the time of treatment initiation may explain some of this variation; thus, differences in the time that passes from the first symptom is experienced until diagnosis and treatment seem to play a crucial role [5,6]. Recent years have seen a stronger focus on cancer awareness and its possible effect on the 'patient interval' [8]. Across a range of different cancer types, both quantitative and qualitative studies have found an association between low awareness of cancer symptoms and risk factors, and a long patient interval [9,10]. Studies have also indicated that the patient interval accounts for a substantial part of the time to diagnosis and treatment [11,12]. Cancer awareness accordingly seems to be a potentially modifiable contributor to the variations seen in healthcare seeking and, ultimately, survival [8]. It is therefore important to assess cancer awareness among the general population and to investigate possible associations with different subgroups. Few studies have explored cancer awareness in the general population, and they find that being a man, living alone, belonging to an ethnic minority group and having a low level of education are independently associated with a lower level of cancer awareness [13][14][15]. In these studies, all SEP indicators are based on self-reporting. Owing to the existence of a Civil Registration System in Denmark with complete, updated information on all Danish citizens [16], a range of highly valid and complete SEP indicators can be linked to survey data at the individual level in this study. The aim of the present study is to assess awareness of cancer symptoms, risk factors and perceived 5-year survival from bowel, breast, ovarian, and lung cancer in a Danish population sample and to analyse the association between these factors and several register-based SEP indicators. --- Methods --- Study population and data collection Data on cancer awareness among the general population in Denmark were collected as part of the International Cancer Benchmarking Partnership , Module 2 [17]. The survey consisted of a 20-minute computer-assisted telephone interview undertaken by trained native-language interviewers from the market research company Ipsos MORI using the Awareness and Beliefs about Cancer measure [18]. In Denmark, a target study population of 1,000 respondents aged 30-49 years and 2,000 respondents aged 50 years and older was initially defined. Using the Danish CRS [16], a random study base was selected consisting of 20,000 persons aged 30-49 years and 40,000 persons aged 50 years and older. Among these 60,000 persons, a total of 6,570 persons were excluded because of research protection . For the remaining 53,430 persons, full name and complete address were obtained from the CRS and used by the Danish market research and consulting firm NN Markedsdata to obtain the phone number belonging to the identified person. Phone numbers could not be obtained for 6,309 persons, who were therefore excluded. Lastly, another 55 persons were excluded just before the data collection began, either because of a newly established research protection status, emigration from Denmark, or because the person had passed away. In total, 47,066 persons were thus eligible for being contacted to answer the ABC measure . The Danish survey was conducted from 31 May to 4 July 2011, and each person was contacted on up to seven occasions at different weekdays and times. Interviews were not performed if the person was unable to speak or understand Danish. --- Survey measure The ABC measure was applied; the development and the validation of the ABC measure is described elsewhere [18]. It explores awareness of cancer symptoms by using recognition as well as recall [19]; awareness of risk factors for cancer; awareness of growing risk of cancer with age; awareness of perceived 5-year survival from cancer; access to a doctor; anticipated healthcare seeking for cancer symptoms; anticipated barriers to healthcare seeking; beliefs about cancer in general; beliefs about cancer screening and actual screening behaviour. In addition, the measure explores smoking status, self-rated health and personal experience of cancer . The English version of the ABC measure was translated into Danish in accordance with the WHO guidelines for translation procedures, which included forward and backward translations [20]. --- Dependent variables Data reported here include awareness of cancer symptoms using the recognition method, awareness of risk factors for cancer, awareness of growing risk of cancer with age and awareness of 5-year survival from four different types of cancer. --- Awareness of cancer symptoms This awareness was measured by asking respondents whether they thought that a specific symptom could be a warning sign of cancer. In total, 11 different possible warning signs were stated in a rotated order with yes/no response options. These warning signs were unexplained lump or swelling; persistent, unexplained pain; unexplained bleeding; persistent cough or hoarseness; change in bowel or bladder habits; persistent difficulty in swallowing; change in the appearance of a mole; sore that does not heal; unexplained night sweats; unexplained weight loss; and unexplained tiredness. Don't know was not indicated as a response category, but such a response was noted by the interviewer. A score of 1 point was given for a correct answer , while an incorrect answer was given 0 points. Don't know was classified as an incorrect answer. The total score of cancer symptom awareness was computed and dichotomised into low and high awareness using the median split. --- Awareness of risk factors for cancer In a rotated order, respondents were asked whether they thought that a specific factor could increase their risk of getting cancer. Respondents could answer strongly disagree, tend to disagree, tend to agree, or strongly agree for 13 different risk factors: Smoking; exposure to passive smoking; drinking more than 1 unit of alcohol a day; eating less than 5 portions of fruit and vegetables a day; eating red or processed meat once a day or more; being obese; getting sunburnt more than once as a child; being over 70 years old; having a close relative with cancer; infection with human papillomavirus ; not doing much physical activity; using a solarium; and exposure to ionising radiation from, for example, radioactive materials, x-rays, or radon. The answers tend to agree and strongly agree were given 1 point; and strongly disagree, tend to disagree and don't know were given 0 points . On the basis of the median split, awareness of risk factors for cancer was categorised into low and high awareness. --- Awareness of growing risk of cancer with age This element was assessed by asking the respondents the following question: "Over the next year, which of these groups of people, if any, do you think is most likely to be diagnosed with cancer?". Four possible response categories were given: 30-year-olds, 50-year-olds, 70-year-olds, or people of any age are equally likely to be diagnosed with cancer. Again, a response of don't know was accepted, but was not mentioned by the interviewer. The answer 70year-olds was coded as correct, while all other answers were coded as incorrect. --- Awareness of 5-year survival from four different types of cancer Respondents were asked to estimate the 5-year survival rate from four different types of cancer: "Out of 10 people diagnosed with bowel/breast/ovarian/lung cancer, how many do you think would be alive five years later?" The interviewers recorded the stated number of people , and the following answers were coded as correct: Bowel , breast , ovarian and lung [21,22]. To analyse the possible association between SEP and awareness of survival from the four different types of cancer, the data were dichotomised into correct estimation and underestimation/overestimation. --- Independent variables Information on SEP indicators was obtained by data linkage to Statistics Denmark [23]. For each person in the study population, we obtained information on seven SEP indicators: gender ; age ; marital status ; ethnicity ; level of education according to UNESCO's International Standard Classification of Education groups [24]; occupation ; and, lastly, OECD-modified disposable household income adjusted for number of adults and children in the household [25]. To level out yearly variation, this SEP indicator was calculated as an average for the preceding three years and categorised as low, middle and high income based on the 20%, 60% and 20% income distribution among the 60,000 persons in the study base. To examine the association between previous personal experiences with cancer and cancer awareness, we retrieved data from the Danish Cancer Registry on all registered cancer diagnoses for each person within the past 10 years [26]. For these register-based SEP indicators, missing data ranged from 0% for age, gender and information on cancer diagnosis to 3.9% for information about educational level for the study base of 60,000 persons. Data on close relatives with cancer and self-rated health were obtained from the ABC survey. --- Statistical analysis Analyses were carried out using Stata version 13.1. Statistical analyses were performed with unweighted data, because weighting may introduce additional bias. Prevalence ratios with 95% confidence intervals were used to determine the association between SEP indicators and awareness of cancer symptoms, risk factors for cancer, growing risk of cancer with age and 5-year survival for four different types of cancer. Unadjusted analyses were carried out with each of the independent variables, and an adjusted model was used to control for possible confounding. PRs were chosen over odds ratios as ORs may overestimate the associations when there is a high prevalence of the dependent variables [27]. --- Ethics and approval The study was approved by the Danish Data Protection Agency and the Danish Health and Medicines Authority. In accordance with the Central Denmark Region Committees on Biomedical Research Ethics, the study needed no further approval . --- Results --- Response To obtain inclusion of 1,000 respondents aged 30-49 years and 2,000 respondents aged 50 years or older, we approached a random sample of 11,297 persons. A response rate of 36.7% was achieved ; this was estimated as the number of completed interviews divided by the number of eligible persons made contact to. The SEP of the respondents and of the study base are shown in Table 2. A higher proportion of the respondents than of the entire study base of 60,000 persons were females, younger, married/cohabiting, ethnic Danes, had a highlevel education, a high household income and were in the labour force. The differences were statistically significant at the 0.01 level. --- Awareness of cancer symptoms The two most well-known symptoms of cancer were a change in the appearance of a mole and an unexplained lump or swelling. These two symptoms were recognised by 97.2% and 94.3% of the respondents, respectively. The lowest awareness of cancer symptoms was found for unexplained night sweats and a sore that does not heal . The median number of cancer symptoms recognised by the respondents was nine out of 11. The associations between SEP and recognition of less than nine symptoms of cancer are presented in Table 3. In both the unadjusted and adjusted analyses, several of the SEP indicators were statistically significantly associated with awareness of cancer symptoms. Especially men, immigrant/descendants, people with low-level education, people outside the labour force, people with a low household income and people with no close relatives with cancer were more likely to recognise less than nine symptoms of cancer than women, ethnic Danes, people with a high-level education, people in the labour force, people with a high household income and people with close relatives with cancer, respectively. Sensitivity analyses revealed similar social gradients in awareness of cancer symptoms based on recognition of both less than five and less than seven symptoms, but the PRs were generally higher in these analyses. For example, the PR of recognising less than five symptoms was 3.81 when people with a low-level education were compared with people with a high-level education. The corresponding PR was 1.57 when the cut-off used was fewer than nine symptoms. --- Awareness of risk factors for cancer The highest awareness of risk factors for cancer was found for smoking and using a sunbed , while the lowest awareness was found for infection with HPV and intake of less than five daily portions of fruit and vegetables . The median number of risk factors recognised by respondents was nine out of 13. Certain characteristics were strongly associated with recognising less than nine risk factors for cancer; these include being a man, older, having a low-level education, or having a low household income . Again, sensitivity analyses showed a stronger social gradient . --- Awareness of growing risk of cancer with age In total, 24% of the respondents were aware that 70-yearolds are more likely to be diagnosed with cancer than 30-year-olds, 50-year-olds and people of any age. However, the majority responded that people of any age are equally likely to be diagnosed with cancer or that 50-year-olds are more likely . Being a woman, having a low level of education, and a low income were associated with nonrecognition of growing risk of cancer with age . --- Awareness of 5-year survival from four different types of cancer The 5-year survival from bowel, breast, ovarian, and lung cancer was correctly identified by 42%, 49%, 9% and 19% of the respondents, respectively. For ovarian and lung cancer, a large majority of the respondents overestimated the 5-year survival, whereas almost half of the respondents underestimated survival from breast cancer. The distributions of the respondents' estimations are shown in Figure 2. Table 6 shows the associations between the SEP indicators and underestimation/overestimation of the 5-year cancer survival. People outside the labour force were more likely than people within the labour force to wrongly estimate the 5-year survival from breast cancer, and men were more likely to wrongly estimate the 5-year survival from lung and ovarian cancer than women. Furthermore, people with a low and middle income and people with no close relatives with cancer were less aware of the 5-year survival from bowel cancer than people with a high income and with close relatives with cancer. --- Discussion A strong socio-economic gradient was found in cancer awareness; thus, people with a low educational level and a low household income were more likely to have a lower awareness of possible cancer symptoms, factors that can influence the risk of getting cancer, and the growing risk of cancer with age than people with a high-level education and people with a high household income. The sensitivity analyses showed that the associations between SEP and the respondents' awareness of symptoms and risk factors were independent of the median cut-off; thus, the findings appear to be robust. We also saw a trend that men and people outside the labour force were less aware of these factors than were women and people in the labour force, respectively. However, women were more likely than men to lack awareness of the relation between age and cancer. No clear associations were found between SEP and lack of awareness of the 5-year survival from bowel, breast, ovarian, and lung cancer. Our study supports findings from previous studies that people with a low SEP are generally more likely to be less aware of cancer than people with a high SEP [13,15,28,29]. The findings also mirror the findings that cancer survival has a social gradient [3]. However, the mechanisms underlying the association between SEP and cancer awareness are not well understood. It has been suggested that, to some degree, the association may be related to health illiteracy and thus a lower capacity among people with lower SEP to obtain, process and understand health information [30]. It has also been rightly questioned whether a heightened awareness in itself may lead to the desired change in behaviour [31,32]; knowledgeable people do not always make wise decisions [14,33]. Recent research has also emphasised the role of other factors in the link between cancer awareness and cancer-related behaviour. Among others, it has been suggested that anticipated barriers to healthcare seeking and beliefs about cancer may mediate this link [33][34][35]. Although the role of cancer awareness as a determinant of behaviour should not be overemphasised, cancer awareness will often be an important step towards healthcare seeking and screening attendance [19,36,37]. The present study found that the two most commonly recognised symptoms of cancer were a change in the appearance of a mole and an unexplained lump or swelling and that smoking and sunbed use were the most wellknown risk factors. On the other hand, unexplained night sweats and infection with HPV were the least recognised symptom and risk factor, respectively. These findings may reflect that Danish national campaigns have focused strongly on breast and skin cancers [38][39][40]. Thus, campaigns addressing cancer symptoms and risk factors may help the population evaluate these more accurately. Accurate evaluation of cancer symptoms and risk factors may reduce the patient interval [41,42], increase screening uptake [43,44] and encourage cancer risk-reducing actions [45,46]. Our findings may also reflect the fact that a lump is a specific symptom, while unexplained night sweats, for example, are a less specific symptom that may be more readily associated with conditions such as menopause and infections than with cancer [47], and may therefore not immediately be considered a symptom of cancer. Likewise, in a comprehensive review by Macleod et al. [10], vague, ambiguous and more common symptoms were associated with a longer patient interval. Cancer is primarily a disease of the elderly, and for most cancers the incidence rate increases with age [48]. However, the majority of the respondents tended to think that people of any age were equally likely to be diagnosed with cancer. This was a surprising finding; but as implied by others [37,43], individuals may not conceptualise nonmodifiable factors as risk factors, whereas modifiable factors may be more easily seen as part of the conceptual framework for cancer risk among laypeople. Nevertheless, awareness about both modifiable and non-modifiable risk factors is important because awareness may facilitate healthcare seeking [28,49]. Awareness of the 5-year survival from bowel and breast cancer was fairly high; however, only a small percentage of the respondents correctly identified the 5-year survival from ovarian and lung cancer. This may be due to inadequate communication about the chances of survival from these cancer types. However, the results for lung cancer may also be partly explained by end-aversion bias, i.e. the tendency to avoid the extremes of a scale. --- Strengths and limitations A key strength of the present study was the use of the Danish CRS. All Danish residents are registered in the CRS which contains complete information on any Danish resident's date of birth, gender, migration, etc. Owing to our use of the CRS, we were able to define a study base of 60,000 persons, a representative sample of the entire Danish population aged 30 years and older. Furthermore, the use of the CRS and the data linkage to a range of register-based SEP indicators provided us with precise and valid insight into variables that may be related to cancer awareness. Naturally, the SEP indicators capture correlated aspects. Still, since the correlation is not a hundred percent, each indicator contributes with unique information about the association with cancer awareness. To analyse associations between SEP and cancer awareness of symptoms and risk factors, cancer awareness was categorised into low/high using the median split procedure. One of the shortcomings of this procedure is that the median is contingent upon the particular sample on which it is based [50,51]. Thus, respondents categorised as having a low cancer awareness in this sample may be categorised as having a high cancer awareness in another sample. However, sensitivity analyses using both awareness of less than five and less than seven cancer symptoms and risk factors showed a similar, but intensified social gradient in cancer awareness. A limitation of the study was the modest response rate. Only 36.7% of the persons whom we made contact to agreed to participate in the study. Unfortunately, response rates have been declining over the past decades and telephone surveys have been particularly affected by this decline [52]. However, by collecting data using a telephone interview, the respondent did not have the possibility to look for information elsewhere. This advantage could not have been achieved with paper-based or web-based surveys. The respondents completing the ABC measure were more often females, younger, married/cohabiting, had a high-level education and a high household income than people in the study base. As a consequence, selection bias may in some way affect the generalisability of the findings since women and persons with a high-level education and a high household income were generally more aware of cancer symptoms and risk factors than men and persons with a low educational level and a low household income. Consequently, the actual awareness level in the population is most probably lower than estimated here. --- Conclusion and implications The results of this study indicate that people with a low educational level and a low household income are less aware of cancer than people with a high-level education and a high household income, respectively. Awareness about possible cancer symptoms, risk factors for developing cancer and survivability has shown to be positively associated with cancer-related behaviour, such as healthcare seeking and screening uptake. Thus, consideration must be given to tackle the current inequality in cancer awareness and to address this issue in future public health strategies. It is important that these strategies are targeted and tailored to the intended recipient groups. Otherwise, strategies may unintentionally increase social inequality in cancer awareness, as individuals with higher SEP often acquire and adapt to new health information at a faster rate than individuals with lower SEP [53]. In conclusion, decisions on healthcare seeking for potential cancer symptoms is a complicated process that is shaped by much more than simply awareness. Thus, the present study should be seen as part of a larger framework of research examining possible associations between SEP indicators and other factors that may influence cancerrelated behaviour, such as beliefs about cancer and psychological and practical barriers to healthcare seeking. --- Competing interests The authors declare that they have no competing interests. Authors' contributions AFP, CNW and PV conceived the study. All authors contributed to the design of the study and in its data collection. LH performed the statistical analyses in consultation with the other authors. LH drafted the manuscript, and all authors contributed to critically revising the paper. Finally, all authors read and approved the final version of the manuscript.

Background: Differences in cancer awareness between individuals may explain variations in healthcare seeking behaviour and ultimately also variations in cancer survival. It is therefore important to examine cancer awareness and to investigate possible differences in cancer awareness among specific population subgroups. The aim of this study is to assess awareness of cancer symptoms, risk factors and perceived 5-year survival from bowel, breast, ovarian, and lung cancer in a Danish population sample and to analyse the association between these factors and socio-economic position indicators. Methods: A population-based telephone survey was carried out among 1,000 respondents aged 30-49 years and 2,000 respondents aged 50 years and older using the Awareness and Beliefs about Cancer measure. Information on socio-economic position was obtained by data linkage through Statistics Denmark. Prevalence ratios were used to determine the association between socio-economic position and cancer awareness. Results: A strong socio-economic gradient in cancer awareness was found. People with a low educational level and a low household income were more likely to have a lower awareness of cancer symptoms, cancer risk factors and the growing risk of cancer with age. Furthermore, men and people outside the labour force tended to be less aware of these factors than women and people within the labour force. However, women were more likely than men to lack awareness of the relationship between age and cancer risk. No clear associations were found between socio-economic position and lack of awareness of 5-year survival from bowel, breast, ovarian, and lung cancers. Conclusions: As cancer awareness has shown to be positively associated with cancer-related behaviour, e.g. healthcare seeking, consideration must be given to tackle inequalities in cancer awareness and to address this issue in future public health strategies, which should be targeted at and tailored to the intended recipient groups.

Introduction Respondent driven sampling is a popular means of sampling difficult to survey populations. The ISI Web of Science database currently tags 642 academic articles with RDS listed as the topic [1]. These papers have been cited 10,217 times by 4,897 unique articles. A search of the NIH RePORTER database shows that the National Institutes of Health has awarded more than $180 million to 448 projects and subprojects with "respondent driven sampling" as a topic [2]. Much of this popularity owes to the fact that RDS is a cost effective and rapid means of sampling hard to reach populations, which have received increased attention across the social and health sciences. There are two key components to the RDS approach. The first concerns sampling and recruitment, where respondents themselves are asked to find new survey participants through their social network connections with members of the target population, which are tracked with anonymous codes or coupons [3]. This is encouraged through a dual incentive structure where recruiters are paid for participating in the study and for recruiting others. The second component of RDS is inferential. Recruitment through social networks is complemented by a set of estimation techniques. Many of the estimation techniques used in RDS derive from the mathematics of random walks on graphs [4][5][6], because when RDS sampling and recruitment conforms to theoretical assumptions it mimics a simple random walk on an undirected, connected graph [7][8][9]. Under ideal conditions [10][11][12], RDS estimators of the population mean are asymptotically unbiased and generalizable to the population of interest, even absent a conventional sampling frame [7,13]. In this paper, we focus on an aspect of RDS inference that has received only limited attention in the literature to date: variance estimation. Most prior work on RDS inference focuses on estimating population means. Some have noted that RDS assumes sampling properties that are not followed in practice , which can lead to substantial biases [10,[13][14][15][16]. Others have evaluated the precision of RDS mean estimates, or, more precisely, the variance in the sampling distribution of mean estimates . An important recent finding is that RDS mean estimates may exhibit very high sampling variance compared to simple random sampling , even when assumptions are met [18]. This is an alarming finding for practitioners who typically collect only one sample, because their mean estimates may be far from the population mean, even if the average value from repeated sampling would converge to the population parameter. Prior work has not thoroughly addressed the accuracy of RDS estimates of sampling variance, however. There are two commonly used estimators of sampling variance in RDS, known as the Salganik bootstrap estimator [19], which uses a bootstrapping procedure to obtain variance estimates, and the Volz-Heckathorn estimator , which obtains variance estimates algebraically [7]. These approaches are quite similar, as both attempt to account for sample-induced correlations between cases that are close together in the referral network [14]. Such correlations lead the sampling variance of RDS to be larger than what would be obtained via SRS, yielding design effects greater than one, much in the same way that the design effects of cluster-based sampling increase as a function of intra-cluster correlations between units [20]. It is possible to obtain an exact variance estimator for random walks by incorporating data on the entire population's social network structure to account for these correlations [6]; we refer to this exact estimator as the Bassetti and Diaconis estimator. However, the RDS variance estimators lack data on the population network-they have only a sample-and as such need to approximate it. With a poor approximation, however, these variance estimators will be biased. To date, despite attention to the general issue of sampling variance in RDS, the actual estimators of sampling variance used by researchers have escaped evaluation. The most thorough prior treatment was by Neely [14], who diagnosed fundamental similarities between the SBE and VHE and limitations of both. Only two prior works have explicitly considered biased variance estimates in RDS [14,18]. Initial inquiries suggest that researchers should be wary when reporting confidence intervals and hypothesis tests based on commonly used RDS variance estimators. One study used simulated sampling and found that the SBE underestimates the empirical sampling variance using a set of school-based social networks [18], and two additional studies report that known population means are frequently much larger than the upper bounds of confidence intervals produced using the SBE [21,22]. However, the generality of these findings is uncertain. Whether they are limited to the cases studied or are endemic remains an open question. In general, the sampling variance of random walks on graphs, and by extension other chain referral methods like RDS, depends on the specific network structure of the population under study which determines the closeness of nodes in the network, and, hence their covariance [6], and researchers do not know how closely the VHE or SBE can approximate it [14]. Because of this, it is important to assess why the RDS variance estimators might not reflect the population sampling variance, and whether this is a general problem or one that is primarily confined to specific types of "problem" networks. In this paper, we diagnose when and why the current RDS variance estimators are biased and assess the generality of that bias across different types of networks. We build on Neely's [14] observation that both the SBE and the VHE rely, at a fundamental level, on a First Order Markov assumption. This assumption holds that RDS recruitment can be modeled as a FOM process on the nodal attribute of interest, where transitions between states depend solely on the prior state and not a higher order sequence of prior states [23]. It is a convenient assumption for estimating RDS sampling variance, because a single RDS sample consists of a sequence of observed cases rather than the whole network. Though the FOM assumption allows the VHE and SBE to estimate sampling variance from RDS sample data, researchers have not evaluated whether it is justifiable or assessed the potential consequences of its violation. We organize the remainder of the paper as follows. We first provide a technical discussion of variance estimation in link tracing samples. We next extend this to RDS, wherein we explain the FOM assumption. We then offer two intuitive and general examples which demonstrate the VHE is biased when the FOM assumption is unwarranted. The first of these is a proof of concept and demonstrates the limitations the FOM assumption places on network structure. The second offers a simple model of correlated homophily which shows the VHE underestimates sampling variance when the FOM assumption is incorrectly applied, a problem which is likely to be quite general, at least whenever homophily based on an unaccounted for factor is present. After this, we test five variables in 215 empirical social networks for violations of the FOM assumption. We find that it is violated in every case for the full network data. However, when simulated samples are drawn from these data, these samples pass the test for FOM in the majority of cases. These results mean that, were a researcher in the field to have collected one of these simulated samples, she would not know, on the basis of the data collected, whether the RDS sampling variance estimators are likely to be biased. A fact which we confirm with our next set of analyses wherein we test whether sample-level FOM violations indicate especially poor estimates of the sampling variance . This is a grave situation for RDS, whose mean estimators are known to exhibit high-sampling variance, because it indicates that researchers are unable to detect situations where mean estimates would have high variance and produce correspondingly large confidence intervals and other indicators of statistical uncertainty. Finally, we propose and test two modifications to the VHE that account for the branching structure of RDS data and higher-order Markov transition matrices and afford closer approximation of the actual RDS sampling process. These modifications improve sampling variance estimation in RDS, but they do not offer a complete solution. This paper contributes to the literature on drawing inferences from network sampling designs by demonstrating that RDS sampling variance estimators are biased because their assumptions are invalid for many social networks and not just "unusual" or "difficult" ones. Our results also defy heuristic notions that situations where RDS estimates will exhibit high sampling variance can be easily detected on the basis of observed homophily, perceptions of network choke points, or sample level indicators of whether the network is FOM. They show that homophily is not strongly correlated with biases in variance estimates, that even though the networks we examine do not have choke points they still have high sampling variance, and that sample level indications that the network is not FOM give little indication of the extent of bias in estimates of sampling variance. Taken together, we show that RDS researchers using currently available estimators are unlikely to know-a priori or even after RDS data has been collected-whether a given network would exhibit high sampling variance. Such uncertainty calls for further development of procedures for accurately estimating RDS sampling variance. --- Sampling Variance Estimation in RWS vs. SRS We can calculate nodal sampling probabilities at any sample size under a simple Markovian random walk model, which has the features of random referral, non-branching recruitment and with-replacement sampling assumed by RDS under ideal conditions [7,24]. These can be used to express the exact sampling variance of random walk sampling . In this section, we contrast how this calculation differs from analogous estimation in simple random sampling ; we rely on a similar discussion in reference [14]. Let the matrix G represent a population's underlying social network, which we assume comprises a single undirected connected component with at least one triangle. Ties between nodes i and j are indicated in cells G ij and G ji with a value of 1, while unlinked nodes have values of 0 in G ij and G ji . A person's degree measures how many ties they have; the degree of person i is given by d i ¼ P N j¼1 G ij , where N is the population size. If D is a square diagonal matrix with 1/d i along the diagonal, the Markov transition matrix is defined as M = DG, where M ij = G ij /d i . If P 0 is a vector containing probabilities of starting the random walk in each node of the network, then P 1 = P 0 M gives the probability of being in each node after one random walk step through the network, and P s = P 0 M s gives probabilities of the random walk being at each node in the network after s steps conditional on the starting probabilities [5]. It is convenient to express the transition matrix M in terms of its eigenvalue decomposition. Eq presents M in terms of a set of eigenvalues and corresponding eigenvectors: M ¼ P N k¼1 l k D 1=2 v k v 0 k D À1=2 ;ð1Þ where λ k is the kth orthonormal eigenvalue of M ordered in terms of magnitude 1 . . . N, v k is the kth eigenvector of a symmetrized version of M, and D is the diagonal matrix of inverse degrees described above . The eigenvectors of M will not necessarily be orthonormal because M is not symmetric. However, as suggested by others [4], consider the matrix M Ã = D -1/2 MD 1/2 where D is the diagonal matrix of inverse degrees and M is the Markov transition matrix defined above. This matrix is symmetric, as each cell M Ã ij ¼ G ij d 1=2 i d 1=2 j ¼ M Ã ji . M Ã can be decomposed as M Ã = VLV 0 where L is a diagonal matrix consisting of the eigenvalues of M Ã along the diagonal and zeros off diagonal, and V is a matrix of M Ã 's eigenvectors. M and M Ã share the same eigenvalues. A key feature of the eigenvalue decomposition of M is that the largest eigenvalue λ 1 = 1, and its corresponding eigenvector ν 1 has elements n 1j ¼ ffiffiffiffi p j p , where π j = d j /2m is the steady state probability of being in node i, and m = 1/2∑ i d i is the total number of edges in the network [4]. After expressing M in terms of the eigendecomposition in Eq , we can find the transition matrix after s steps: M s ¼ P N k¼1 l k D 1=2 v k v 0 k D À1=2 h i s ¼ P N k¼1 l s k D 1=2 v k v 0 k D À1=2 ;ð2Þ where ν k is the kth eigenvector of M Ã = D -1/2 MD 1/2 . Using Eq , we see that lim s!1 M s ¼ l s 1 D 1=2 v 1 v 0 1 D À1=2 ¼ p j because the first eigenvector has elements v 1i ¼ ffiffiffiffi p i p ¼ ffiffiffiffiffiffiffiffiffiffiffiffi d i =2m p [5] . Importantly, π j is a probability vector where each element represents the probability of being in node j in the steady state equilibrium. This is an important result because it means that the probability vector P s does not depend on P 0 when s gets large. This is the source of the argument that the particular seeds from which an RDS sample begins will not bias estimates from large RDS samples, provided the sample size remains a small fraction of the population [10]. Sampling variance in RWS differs from the sampling variance of SRS in two important ways. The first owes to non-uniform sampling probabilities. The effects of this are most clearly understood by considering the probability model underlying SRS in terms of the matrix algebra introduced above. To get the sampling variance of SRS, we first arrange members of the population into a vector Y with elements Y = {y 1 , . . ., y N }, and define Y Ã as its mean centered version Y Ã ¼ y 1 À P N i¼1 y i N ; ð. . .Þ; y N À P N i¼1 y i N . We denote the sampling variance of mean estimates of Y from simple random samples as s 2 b m , which can be estimated with the population variance and sample size as c s 2 b m ¼ 1 S P N i¼1 Y Ã 2 i N , where S indicates the sample size. If units are selected with given probabilities instead of with equal probabilities , then the sampling variance is defined as: c s 2 b m ¼ 1 S P N i¼1 p i Y Ã2 i ¼ 1 S p 1=2 Y Ã ðY Ã p 1=2 Þ 0 :ð3Þ in this way-breaking out the squared deviations into two pieces and taking the square roots of π -to simplify algebra presented later). The most important thing to note about Eq is that it uses population values of Y Ã and π, despite being an estimate based on a sample. However, with a sample, researchers almost never know these values, which must be approximated. Denote the elements selected in a sample of size S as Y SRS ¼ fy SRS 1 ; . . . ; y SRS S g and their sampling probabilities as p SRS i ; assume . The estimated sampling variance on the basis of the sample is then: P S i¼1 p SRS i ¼1 c s 2 b m ¼ 1 S À 1 P S i¼1 p SRS i ðY ÃSRS i Þ 2 ¼ 1 S À 1 p SRS 1=2 Y ÃSRS ðY ÃSRS p SRS 1=2 Þ 0ð4Þ Eq resembles the presentation in Eq , except that it now has sampled squared deviations Y ÃSRS and sampling probabilities π SRS . Due to the Law of Large Numbers, Eq The estimation of sampling variance in link-tracing designs also differs from the analogous estimator in SRS because if there is homophily-the tendency for individuals with similar attributes to be friends with one another [25]-on the basis of some variable, then individuals in the network who are connected will tend to have similar values of that variable. As a result, values of the variable of interest from two cases in a random walk will be correlated if the number of steps between them is small. This results in non-zero covariances between cases in link-tracing style samples, which must be accounted for when estimating sampling variance. It is important to remember that homophily can exist on any sort of characteristic, observed or unobserved . The variance of a RWS on a network can be analytically derived from the eigenvalue decomposition in Eqs and [6]. First, working from Eq , we can express the population variance 1/2 Y Ã ) 1/2 Y Ã ) 0 in terms of an orthonormal eigenvector basis {v k : K = 1 . . . N}, P N k¼1 a 2 k v k n 0 k ¼ P N k¼1 a 2 k , where α k is a scalar constant that maps the kth eigenvector of the transition matrix M Ã onto π 1/2 Y Ã . Note that n k n 0 k ¼ 1 and n j n 0 k6 ¼j ¼ 0 because they are orthonormal. With this, we can denote the covariance between the ith and jth step of a random walk on G as: covði; jÞ RWS ¼ P K k¼2 a 2 k l jjÀij k ;ð5Þ where α k is the mapping of the variable Y Ã onto the kth eigenvector of the transition matrix, λ k is the kth eigenvalue of the transition matrix, and |j -i| is the number of steps between the ith and jth cases sampled by the random walk [6]. In general the covariance between two steps of a random walk is affected by all three components: α k , λ k , and |j -i|. Using Eq , we can write an estimate of the sampling variance of a size S random walk sample as the average of all the possible covariances among the population that the walk could take on G: d s 2 b m RWS ¼ 1 S 2 ðS À 1Þ P S i¼1 P S j¼1 covði; jÞ RWS ¼ 1 S 2 ðS À 1Þ P S i¼1 P S j¼1 P K k¼2 a 2 k l jjÀij k ¼ c s 2 b m þ 1 S 2 ðS À 1Þ P S i¼1 P S j¼1;i6 ¼j P K k¼2 a 2 k l jjÀij k :ð6Þ When |j -i| = 0, this reduces to the estimated simple random sampling variance of Y, c s 2 b m . Eq highlights that the sampling variance of a RWS will be greater than the sampling variance of a same sized SRS if the variable Y Ã maps onto the eigenvectors such that the contribution in the sum from the positive eigenvalues outweighs that from the negative eigenvalues. The other critical difference between Eqs and is that the latter for SRS operates on a sample and estimates the sampling variance, while the former for RWS requires the underlying social network and dependent variable data from the entire target population. A researcher in the field, for example, would be unable to use Eq without knowing the entire social network connecting individuals in the target population, which will obviously not be the case with a hidden population. In the next section, we discuss the VHE, which is an approximation of the Bassetti and Diaconis estimator used by the RDS literature to estimate the sampling variance of a respondent driven sample absent complete knowledge about the underlying social network. --- Variance Estimation in RDS In RDS the researcher does not have data on the network itself, only data on the sequence of recruitments that occurred and the characteristics of those recruited, i.e., a sample. We present the VHE in the same framework as Eqs - above. The key difference between the VHE and the exact estimator of Bassetti and Diaconis discussed above is that the estimator developed by Volz and Heckathorn uses the patterning of recruitments and characteristics in the sample in place of the entire population network. Given a RDS sample on the population connected by the social network G, let Y RDS 2 Y indicate the cases sampled from the Y values of the population. If Y is dichotomous, let the matrix C be a 2 × 2 matrix showing observed transition probabilities among values of Y RDS between respondents and those they referred to the study. If Q t is a 2 × 1 row vector indicating the probability that the Y value of the jth respondent is 0 or 1, then we can estimate the Markov transition probability between any two steps j and i of the survey using observed categories of the dependent variable rather than nodes in the network as we did in Eq above: Q s ¼ Q r C jjÀij :ð7Þ We can then estimate the covariance between the jth and ith steps by modifying Eq as: d covðj; iÞ RDS ¼ P 2 k¼1 ða VHE k Þ 2 ðl VHE k Þ jjÀij ;ð8Þ where the superscript VHE on alpha and lambda indicate that we are using α and λ from the eigendecomposition of sample observed category transition probabilities from the 2 × 2 matrix C instead of the population node transition probabilities from the N × N matrix M Ã . This yields the VHE as: d s 2 b m RDS ¼ c s 2 b m þ P S i¼1 P S j¼1;i6 ¼j d covði; jÞ RDS ¼ 1 SðS À 1Þ P S i¼1 p RDS i ðY ÃRDS i Þ 2 þ 1 S 2 ðS À 1Þ P S i¼1 P S j¼1;i6 ¼j P 2 k¼1 ða VHE k Þ 2 ðl VHE k Þ jiÀjj ;ð9Þ where p RDS i is the RDS sampling probabilities. Connecting Eq to the notation used in Volz and Heckathorn [7], we can write 1 SðSÀ1Þ P S i¼1 p RDS i ðY ÃRDS i Þ 2 þ 1 S P s i Y RDS i =d RDS i P s i 1=d RDS i 2 1 À S þ 1 P s i Y i P S i¼1 P S j¼1;i6 ¼j C jjÀij 11 , where Y ÃRDS i is the Y value of the ith case selected mean-centered via the Volz-Heckathorn "RDS-2" mean estimator, p RDS i is the corresponding case's sampling probability, d RDS i is its degree, and C jjÀij 11 is the estimated transition probability between Y j,j 6 ¼ i = 1|Y i = 1; this presentation can be found in reference [14]. Thus, 1 --- SðSÀ1Þ P S i¼1 p RDS i ðY ÃRDS i Þ 2 is analogous to the estimated sampling variance of SRS , P s i Y RDS i =d RDS i P s i 1=d RDS i is the Volz-Heckathorn estimate of the sampling mean, and 1 À S þ 1 P s i Y i P S i¼1 P S j¼1;i6 ¼j C jjÀij 11 is the expected correlation between sampled units if the process is FOM. Using C instead of M equates to making the FOM assumption because it assumes the likelihood of transitioning between categories of the Y variable in question only depends on the category of the currently sampled node, one of several assumptions in RDS variance estimators described in detail in prior work [14]. Importantly, this is the same assumption made by the other commonly used RDS variance estimator, the SBE [14], which we evaluate via simulation later in the paper. The SBE is defined in the literature [14,19] as using a bootstrap sampling procedure to simulate synthetic RDS chains from the FOM approximation embedded in the C matrix. Using C to approximate the node-specific Markov transition probabilities is a simplification, as pointed out by its developers [7]. It may be a reasonable one to make because the M matrix is unknown; however, the validity of this assumption has rarely been examined or tested in the RDS literature [14]. In the remainder of the paper, we explore the implications of the FOM assumption in greater detail. --- What Happens to the VHE when FOM Is Violated? In this section, we provide several examples of what happens to RDS variance estimation via the VHE if the FOM assumption is violated. --- Illustration 1: Intuition We begin with an illustration that highlights how the relationship between sampling variance and sample size differs between FOM and non-FOM networks when using random walk sampling . This example is intentionally stylized so readers can see what is occurring and intuit the effects of network structure on the VHE of RDS sampling variance. This is important as it means that both networks have identical levels of dyadic homophily and that they share the C transition matrix used by the VHE; that is, C A ¼ 18=27 9=27 9=27 18=27 " # ¼ C B . Because C is identi- cal in these two networks, they will produce identical estimates of RDS sampling variance based on the VHE as shown in Eq . However, there is one key difference between these networks: Network A is FOM with respect to Y while Network B is not. This difference can be described in terms of the conditional probability of the current node's Y value given the Y values of prior nodes visited [23]: PrðY s ¼ 1 j Y sÀ1 ; Y sÀ2 ; . . . ; Y sÀ1 Þ ¼ PrðY s ¼ 1 j Y sÀ1 Þ:ð10Þ Eq holds for network A, and therefore network A is FOM, while for network B it does not hold . Does this difference in network structure matter for RWS sampling variance? Given that the VHE estimates of sampling variance for these two networks will be identical , if there is a large discrepancy between these networks in terms of the empirical RWS sampling variance , then it suggests potential problems with the VHE. Fig 2 shows how large an impact network structure can have on the sampling variance of random-walk based designs and that the VHE cannot detect these differences. Because we know the complete network structure, we can calculate the exact sampling variance of RWS estimates for these two networks using Eq . Alternatively, we can approximate it using the variance in the distribution of proportion estimates produced by repeatedly sampling the network with random walks starting from random points proportional to their stationary distribution probabilities a large number of times; these approaches yield indistinguishable results. The striking result in Fig 2 is that while the sampling variance in Network A approximates to the sampling variance of SRS as the sample size increases, the sampling variance in Network B is much higher at every sample size. More importantly, while estimates from the VHE accurately describe the sampling variance of network A, they fail to do so for network B . In fact, the VHE estimates for network B are identical to those obtained for network A, which makes sense as both networks have the same first-order transition probabilities. In terms of design effects , the design effect from sampling S = 5,000 cases from Network A is 1.9997 while the same sample size in Network B yields a design effect of 29.2601 . In substantive terms, we can say that mean estimates based on RWS samples with 5,000 cases from Network B will be more almost 15 times more variable than the same estimates from Network A, yet, even with perfect information, the VHE will estimate that they have identical sampling variance. This finding is problematic for RDS because it indicates that if the FOM assumption is violated, researchers may have no idea whether their estimates of sampling variance-and, hence, their confidence intervals and hypothesis tests-are accurate or not. --- Illustration 2: Generality Though Figs 1 and 2 indicate that the VHE may produce estimates that are quite far from the population sampling variance of a network if the network is not FOM, one may wonder whether this result is produced by some feature of the two networks we considered. For instance, these are low degree networks and Network B has "choke points" , which makes it the type of network that RDS heuristics suggest should be avoided . Because of these issues, we now turn to a more general illustration showing that the VHE will be downwardly biased-i.e., underestimate the population sampling variance-in all cases where there is homophily on an unobserved variable that is correlated with the Y variable of interest. Imagine a population composed of two groups categorized by a dichotomous Y variable which are connected via a social network but which exhibit homophily on Y. We wish to take a RWS or RDS sample to estimate the population proportion of Y = 1. In addition, imagine there is a variable, Z,-unobserved by the researcher-which organizes a portion of the social ties in this network. As has been well documented, homophily exists on a wide range of dimensions, some of which may not be observed or anticipated by researchers or even research participants [25,26]. In this case, Z indicates a propensity for forming cross-group ties: individuals for whom Z = 0 only have ties with those within their Y group, while those for whom Z = 1 have ties across Y groups. For instance, if Y indicated a dichotomous measure of race, then Z might indicate an individual preference for interracial friends. Other examples that may generate a lack of conditional first-order independence are given on page 75 of reference [14]. For the sake of simplicity, assume an equal number of people of each Y value in each of the Z groups, and that the total degree of each Y|Z combination is the same. The number of friends among and between the different Y|Z groups is depicted in Table 1 the whether the ego and alter share Z values and which Z value they have. Because the total degree is the same for each Y|Z combination we have = , ; D = . Y = 0 Y = 0 Y = 1 Y = 1 Ego Values Z = 0 Z = 1 Z = 1 Z = 0 Y = 0 Z = 0 D E 0 1 0 2 Y = 0 Z = 1 E F H 0 3 Y = 1 Z = 1 0 1 H F E Y = 1 Z = 0 0 2 0 3 E The transition matrix M can be written as M ¼ 1 À a a 0 0 a 1 À a À b b 0 0 b 1 À a À b a 0 0 a 1 À b 2 6 6 6 6 6 4 3 7 7 7 7 7 5 , where a ¼ F EþFþH and b ¼ H EþFþH . M represents a situation where there is heterogeneity in the level of dyadic homophily on Y. Individuals with Z = 1 form cross Y friendships, while those with Z = 0 do not. Here Z can represent anything that causes heterogeneity in mixing between Ys. Note that both of the assumptions underlying this example can be relaxed without affecting the conclusions derived here. For example, we can allow for different numbers of people in each of the Y|Z combinations and we can allow the degrees to vary. We have tested numerical examples under a variety of conditions . The fundamental conclusion is the same, namely that homophily on an unobserved variable that is correlated with the dependent variable will lead to a biased VHE. In contrast to the population transition matrix M, the VHE estimates sampling variance using the sample estimated 2 × 2 transition matrix C as a function of the friendship propensities and the size of each group, C ¼ 1 À i i i 1 À i " # , where i ¼ cross Y f riends total f riends ¼ H 2ðEþFÞ ¼ b 2 . A criti- cal result is that C and M have different eigenvalues. Because it is dimension two, C has only two which are 1 1 À b " # when ordered by size. By contrast, M has four eigenvalues which are 1 1 À b À a þ ffiffiffiffiffiffiffiffiffiffiffiffiffiffi a 2 þ b 2 p 1 À 2a 1 À b À a À ffiffiffiffiffiffiffiffiffiffiffiffiffiffi a 2 þ b 2 p 2 6 6 6 6 4 3 7 7 7 7 5 when ordered by size. Recalling Eqs and , for the VHE to accurately estimate the sampling variance of Y via a random walk based approach like RDS, the second largest eigenvalues of these C and M would need to be equivalent. However, they are not, which means that when a correlated unobserved variable structures the homophily on Y, the VHE will not accurately estimate sampling variance. In fact, we can make a stronger claim in this case. Because ð1 À b À a þ ffiffiffiffiffiffiffiffiffiffiffiffiffiffi a 2 þ b 2 p Þ > ð1 À bÞ for all nonzero as and bs, we can say that the VHE will always underestimate the true sampling variance with this kind of network structure. This is a general result that compliments the intuition provided in Figs 1 and2. There we showed that a random walk on a FOM network will mix more slowly than a random walk on a network with a higher order Markovian structure, but that the VHE will not be able to detect these differences. Slower mixing results in higher covariances between any two steps of a RWS or RDS sample drawn from that network, and, thus, higher sampling variance of mean estimates and larger design effects. However, the inability of the VHE to detect these differences-its biasedness-means that researchers will understate their uncertainty. --- Illustration 3: Computational Examples We now provide two concrete examples based on this illustration to demonstrate the effect of homophily on correlated unobserved variables. --- Summary of VHE Bias The model of homophily on unobserved variables presented in this section is purposively simple in order to make analytical results tractable. Nonetheless, the basic intuition should be clear: if there is clustering within categories of the observed dependent variables-such as is evident in matrix M of the second example above-then the VHE, which relies upon the observed transition matrix between categories of the variable of interest, C, will exhibit downward bias. The variance of a random walk is not just a function of dyadic homophily between different categories of the dependent variable, as both M and C have the same level of dyadic homophily but different design effects. In other words, it is network structure-not homophily on the observed, focal variable per se-that affects design effects and biases RDS sampling variance estimators downward [14]. Moreover, the examples presented here likely underestimate the role played by network structure as they focus on a simple set of networks and a limited 4 × 4 category transition matrix rather than a node level transition matrix that would be found in a real network. Indeed, the sampling variance of our computational examples could be correctly estimated by making a Second Order Markov assumption, but real world networks are unlikely to conform to that assumption as well. Below we test a modification to the VHE based on a second order assumption and show that while it frequently outperforms the classic VHE, it still does a poor job capturing the sampling variance of simulated RDS in empirical networks. The fundamental point of this section was to show that if the FOM assumption is violated, as it is by the case of homophily on an unobserved variable, then the accuracy of the VHE derived estimates of RDS sampling variance are indeterminate, and will be downwardly biased under rather general conditions. As illustrated in the these examples, unbiased RDS variance estimators are predicated on the network being well described by homophily on a single observed category, and they are of little use when there is heterogeneity in the mixing among members of the groups defined by those categories. We now consider empirical data to evaluate the generality of these problems. --- How Often Is FOM Violated in Empirical Networks? In this section, we ask whether researchers should generally expect networks to be well described by homophily on a single dimension, or, more specifically, how often they should expect the FOM assumption to be violated. The RDS literature has not explored this idea since its foundation [3]. We test the FOM assumption in 215 heterogeneous empirical networks drawn from two separate datasets. We then outline the issues faced by the two most used RDS variance estimators-the VHE and the SBE-when they are applied to networks that violate FOM, as we find that most empirical networks do. --- Data, Methods, and Measures Much prior methodological work in RDS has used simulated data [7,10,27]; however, it is challenging to accurately simulate all structural features found in empirical networks [28,29]. Because of this, we use data from the National Longitudinal Survey of Adolescent Health [30] and the Facebook 100 datasets [31][32][33]. These networks have been used in simulation based studies of network sampling performance [11,18]. In all networks, we restrict our analysis to individuals in the largest weakly connected components, and, in the Add Health data, we ignore the directionality of ties and treat all nominations as reciprocal. We use 115 networks from Add Health and 100 from the Facebook data set for a total of 215 empirical networks. In the Add Health networks, we test whether the FOM assumption holds for the following three dichotomous variables: race , gender , and sports participation . We look at the validity of the FOM assumption in two variables in the Facebook data: gender , and class year . These data are faithful to real world network patterns [34,35]. More importantly, they contain a diversity of network structures, which makes them excellent candidates for assessing the credibility of the FOM assumption in empirical networks and allows us to overcome criticisms that have plagued prior simulation work in RDS, namely that the empirical networks studied were too sparse, small, or contained "choke points". While these properties may characterize some of the Add Health networks, the Facebook networks we examine are not so limited. The best measure of choke points in a network is the average number of node independent paths. In any connected component, a set of nodes exists that, if removed, would disconnect that component. For a chain referral strategy to pass from one side of this nodal cutset to the other, it must pass through a node in this set. Menger's theorem [36] proves that the number of node independent paths in a graph is equal to the size of the smallest nodal cutset, which has been used to define the structural cohesion of a network [37]. We measured the numbers of node independent paths in the symmetrized largest connected components of the Add Health and Facebook networks used in this paper. Owing to the size of these networks and the computational complexity of calculating the number of node independent paths amongst all dyads in a network, we estimate the number of node independent paths in each network based on random samples of 10,000 dyads using maximum flow algorithms on the complete network. This provides asymptotically unbiased estimates. The Facebook networks studied had an average number of node independent paths of 30.633 , while the numbers in Add Health were smaller on average . These macro-structural features, in addition to the high average degree , suggest that we study a substantial range of networks that are not limited by the heuristic notion of choke points. Other relevant statistics are as follows. The largest connected components in the Facebook networks ranged in size from 388-16,611 with a mean of 4,701. In Add Health, these numbers were 52-1,610 with a mean of 488. The proportions female in the Facebook networks ranged from 0.24-1.00 with a mean of 0.54, while the proportions freshmen in these networks ranged from 0.14-0.46 with a mean of 0.28. In Add Health, the proportions female ranged from 0.01-0.69 with a mean of 0.54, while the proportions white ranged from 0.01-0.95 with a mean of 0.58, and the proportions participating in sports ranged from 0.28-0.95 with a mean of 0.56. In sum, the Facebook networks we examine are also quite large, orders of magnitude larger than the Add Health networks. The definition of a FOM process is as follows : [23]. Given this, a sufficient condition that satisfies that the network is not FOM is: Pr = Pr PrðY s ¼ 1 j Y sÀ1 ; Y sÀ2 Þ 6 ¼ PrðY s ¼ 1 j Y sÀ1 Þ:ð11Þ By sufficient condition, we mean that if the preceding equation is true, then the network is not FOM. However, note that because this is only a sufficient condition, a failure to satisfy the preceding equation does not guarantee that the network is FOM. This makes it a conservative test: in cases where Eq 11 does not hold , the network may still not be FOM. We test whether Eq 11 holds by estimating the following ordinary least squares regression with robust standard errors for each of the variables of interest in each of the networks from the Add Health and Facebook data sets: PrðY s Þ ¼ b 0 þ b 1 ðY sÀ1 Þ þ b 2 ðY sÀ2 Þ þ b 3 ðY sÀ1 Â Y sÀ2 Þ þ d;ð12Þ where Pr is the proportion of an ego's alter's alters with Y = 1, while and are dichotomous indicators of whether ego's or alter's Y = 1, respectively. Note that we include ego himself as one of ego's alter's alters, which suffices to retain ego's alters who lack alters in the sample, and which makes sense for the with-replacement process we study here. The resulting regression model thus contains one observation for each edge in the network . Though each ego will have several alters in the data, and we make use of even more alters' alters in our definition of the dependent variable, the use of robust standard errors reduces concerns about clustering of the data. The sufficient condition shown in Eq 11 is true if the estimated coefficients for β 2 and β 3 are not jointly equal to 0, which we evaluate with the F-test of joint significance. Our null hypothesis is that the sufficient condition shown in Eq 11 is untrue-i.e., that Pr( Y s = 1 | Y s-1 , Y s-2 ) = Pr(Y s = 1 | Y s-1 ). While this does not guarantee that the networks are FOM, in cases where this test indicates we should reject the null hypothesis it means that the network is unlikely to be FOM because the current state depends on the prior state as well as how that state was reached. In addition to testing the FOM assumption in the complete networks, we also test it in RDS samples of size 200 on those networks conducted with replacement from a single seed drawn at random from the equilibrium distribution, because this is the type of data that a researcher who had collected a single sample might possess. We allow branching to occur where the probabilities of referring 0, 1, 2, or 3 new respondents in RDS are 1/3, 1/6, 1/6, and 1/3, respectively. Because this is the approach used in an influential past study [18], we focus on these results. We also tested variants where we allow branching with the same probabilities as above, but the samples are conducted without replacement, and where we do not allow branching . We do not present these results but they did not alter our conclusions of substantial biases in the VHE and SBE. We conduct 500 simulated RDS samples in each of the 215 Add Health and Facebook schools, storing the relevant variables of interest. After testing for FOM violations in these sampled network data, we then summarize some of the problems that the VHE and SBE estimators exhibit when applied to empirical network data. In each sample, we calculate the predicted proportion of Y via the Volz and Heckathorn estimator of the mean b m RDS ¼ P s j Y j =d j P s j 1=d j , where d indicates degree [14,38]. We define the "population sampling variance" as the variance of the distribution of mean estimates obtained over R = 500 simulated samples in that network ; in other words, population sampling variance ¼ P R r¼1 ðb m RDS r À mÞ 2 R ¼ s 2 b m RDS ;ð13Þ where r indexes the simulated replication of the sample . Defined in this way, the population sampling variance is the variance of the distribution of mean estimates across repeated samples. We use the population sampling variance to define the bias for the VHE and the other popular means of estimating sampling variance in RDS, the Salganik Bootstrap Estimator , which is bias ¼ P R r¼1 ð d s 2 b m RDS À s 2 b m RDS Þ Rð14Þ where a value of zero indicates that the estimator is unbiased for that variable in that network. To gauge the potential influence of outlier RDS variance estimates on the relationship between the estimated sampling variance and the population sampling variance, we tested using the median estimate across the 500 simulations rather than the mean . This led to more severe biases and other problems than those reported in the manuscript.The next quantity of interest is the ratio of the estimated sampling variance to the population sampling variance, which helps quantify how closely the VHE and SBE approximates the population sampling variance; thus, we also examine: ratio ¼ P R r¼1 d s 2 b m RDS =R s 2 b m RDSð15Þ in each network. Finally, though it may be biased, there is the possibility that the VHE estimates of sampling variance are highly correlated with the population sampling variance and thus researchers could simply inflate the variance estimator by some factor. To examine this possibility, we consider the correlation of the mean variance estimates in each network with the population sampling variance for each variable: correlation ¼ corðð P R r¼1 d s 2 b m RDS =RÞ; s 2 b m RDS Þ:ð16Þ Finally, we note that the results about whether the network or sample is FOM pertain to whether or not researchers should expect that RDS estimators of sampling variance are underestimates and can detect those cases , as was demonstrated in the prior section. However, a different question is whether researchers can predict how large the underestimation bias in a given sample is likely to be rather than whether or not the estimators are biased toward underestimation. Building from the literature reviewed in section 2, we know that the degree of bias is determined by higher order features of a network that is not FOM. Echoing a general sentiment in the literature, it may be that homophily on the focal variable explains the degree of bias in cases where FOM is violated, so, to test this, we examine whether sample level homophily can predict levels of bias in cases where FOM is violated at the network level . To do this, we compute the sample level homophily defined as the ratio of observed cross-group ties to expected cross-group ties in a given sample. We then regress bias on this measure of homophily to determine whether there is a meaningful and consistent relationship between bias and the homophily of a sample, which, if found, would indicate that the homophily observed in a sample can alert researchers to cases where bias is especially problematic. To facilitate interpretation, we focus on XY standardized regression models, where both the independent and dependent variables are standardized to have a mean of 0 and a standard deviation of 1. In XY standardized regression, the interpretation of coefficients is natural: a one standard deviation change in X leads to a b b standard deviation change in Y . We obtained substantively equivalent results in models run without XY standardization, but we focus on the XY standardized results because of their simpler interpretation in this case. To determine whether results owe to features of the networks or estimators we study or are general, we obtain parameters from regression models with and without absorbing indicators for the network studied and for both the VHE and SBE estimators. --- Results The results of our tests of the FOM assumption on the complete networks are shown in panel A of Table 2, while results of the FOM test on individual samples are shown in panel B. Columns 1-3 show the proportion of FOM tests where we reject the null hypothesis that the network may be FOM under standard social science thresholds based on the F-test of joint significance . There are two key results. The first is that, for the complete networks , we reject the null hypothesis that the variable of interest in each network is FOM almost every single time. The one exception is a Facebook school where we could not calculate the FOM test for gender because the school is not co-educational. In other words, we find no cases where the fundamental premise of RDS variance estimation is a justifiable assumption. Worse still, the second key result in this table shows that, for the sample level tests , a near majority of the samples indicate the opposite, that the network may be FOM. This disjuncture indicates that it would be difficult for a researcher to know a posteriori whether the current methods of RDS variance estimation can be applied aptly. Though a given sample may seem appropriately characterized as FOM [3], the network from which it was drawn is highly unlikely to be. We return later to this disjuncture and its consequences for RDS variance estimation. We next consider how problematic RDS variance estimation is when it is applied to empirical data whose complete network structure violates the FOM assumption. We look at the two most commonly used RDS variance estimators. Table 3 separate these estimators into two panels, with panel A showing the VHE and panel B showing the SBE. The first column shows the mean of the bias across the networks for each dataset and variable . The key point to notice about this column is that both the VHE and the SBE estimates are negatively biased in all cases. The second column shows the mean of the ratios of average VHE and SBE estimates of sampling variance in a given dataset and school to the population sampling variance . Most of the variables understate the true variance substantially-in the Add Health schools, the VHE estimated sampling variance understates the empirical sampling variance of RDS by about 85%-but this number ranges from as low as 55% to as high as 90% in the Facebook datasets. The SBE performs better here on average, with ratios of 0.2862 to 0.4421 in the Add Health schools and 0.1494 and 0.6544 in the Facebook ones. Finally, the third column of Table 3 shows the correlations , which highlight that there are substantial deviations from direct positive correlation and that the relationships between the average RDS estimates and the population values differ substantially by variable and dataset. We argue that this variation in correlations implies that researchers cannot know a priori whether the VHE or SBE estimates of sampling variance are useful. The results in Table 3 are a conservative estimate of the problems with variance estimation in RDS. This is because they average, respectively, all VHE and SBE estimates across 500 RDS samples conducted in each school, which may paint an unrealistic picture of the practical utility of these estimators. Because researchers typically only collect one sample, we now turn to Fig 3 , i.e., within the estimated 95% confidence interval. In SRS, the coverage rate is expected to be 0.95, but, as Fig 3 shows, the coverage rate for the VHE is substantially lower, and we see substantial variability in the distributions by data set and variable. For example, the "FB Freshman" variable shows that, on average, the 95% confidence interval for the VHE estimates of the mean proportion of freshmen contained the true mean in only 36.5% of the networks under study. This is substantial failure of confidence intervals for RDS. Beyond the poor coverage seen across all of the variables, a secondary point conveyed by Fig 3 is that the SBE generally outperforms the VHE. With Table 4, we return to the disjuncture between population-level failure of the FOM test and sample-level passing of it that we noted in our discussion of Table 2. A natural question to ask is whether-a posteriori-a researcher can test her sample for FOM violations and discern whether the RDS variance estimators are likely to be biased. We split Table 4 into two panels: panel A shows the cases where the samples in Table 2 were found to not be FOM, while panel B shows those which may be FOM. The columns show the data set and variable combinations. Within each panel, we present the most relevant statistics averaged samples within that panel: the empirical design effect calculated from the population sampling variance in Eq 13 , the VHE and SBE estimated design effects , and the coverage rates from both estimators . The empirical design effects are generally smaller in the samples that may not be FOM; however, this is not true for the FB Freshman variable. However, the estimated DEs, using either the VHE or the SBE, do not appear appreciably closer to the population values . Neither do the coverage rates. Taken as a whole, these results suggest limited potential for sample-level FOM tests to be used as a diagnostic tool. Though researchers do not typically test for FOM violations, and while other, potentially yet-to-be-developed tests may be able to Lastly, we estimated the parameters of XY standardized regression models for the relationship between sample level bias in the VHE and SBE estimators of RDS sampling variance and sample level homophily. These results are shown in Table 5. The key points highlighted in this table are that the relationship between sample level homophily and bias a) are in different directions across variables, b) are generally of low magnitude and often not distinguishable from 0 despite the large sample sizes, and c) differ between the VHE and SBE estimators. The conclusion to be drawn from these tests is that sample level homophily cannot be used to characterize the degree of bias in RDS estimators of sampling variance. These results show that another feature of networks that is commonly assumed to explain biases in RDS sampling variance estimators, level of homophily, is not a reliable indicator of whether the results of a single sample are biased. In this section we found that the FOM assumption is routinely violated in empirical networks, but that researchers will not know this from the results of a single sample. Building on arguments introduced in Section 2, this finding indicates that the RDS variance estimators in common use can be expected to substantially underestimate the population sampling variance that RDS is likely to exhibit. We also examined a related question that helps contextualize the importance of our FOM results: how much do these violations of matter? More specifically, we explored the extent to which the current estimators for RDS sampling variance, the VHE and the SBE, are likely to underestimate the true sampling variance in these empirical networks. Our findings in this regard were surprising in two ways. First, Table 3 showed substantial downward biases in the VHE and SBE estimators of RDS sampling variance. It also showed little consistency across variables or data sets in the magnitude of this bias, or other properties of the relationship between the estimated values and the population parameter. This is important because it highlights that the current techniques of RDS variance estimation are wildly inaccurate, which makes sense because they are premised on a faulty assumption. In other words, this section has provided suggestive evidence that the core assumption underpinning variance estimation in RDS is violated in a large proportion of empirical cases, that RDS variance estimators are biased in such circumstances, and that researchers will have difficulty knowing when this will be the case. --- Improvements to the VHE In this section, we test the performance of two improvements to the VHE. An easily diagnosable flaw in the VHE is that it fails to account for the branching nature of RDS data. As shown in Eqs and above, the VHE uses the distance between sampled individuals i and j, which in a random walk is equal to the number of steps between their appearances in the sample. However, in RDS, because of the branching structure of recruitment, these distance calculations will be more complicated. As such, the first estimator we introduce, based off of an earlier estimator developed by Neely in a prior investigation [14], which we call the "VHE with branching correction" , explicitly tracks the network distance between individuals, so, if i recruits j who recruits both k and l, we define the distance between i and both k and l as 2. This approach should improve the VHE by more accurately calculating the covariance between cases. The second improvement we test is relaxing the FOM assumption in the VHE. The VHE assumes the network is FOM with respect to the variable of interest because it uses a 2 × 2 transition matrix populated with the observed categorical transitions in the data. However, we can relax this assumption by making, e.g., a 4 × 4 transition matrix which is populated with the observed three step transitions ; that is how often we see three-step chains with the following sequences of Y values a) 0 -0-0, b) 0-0-1, c) 0-1-0, d) 1-0-0, e) 0-1-1, f) 1-0-1, g) 1-1-0, or h) 1-1-1. where Pr 000 = p, that is the proportion of observed sequences of Y values that went 0-0-0, and Pr 001 = p, and so on . In principle, one could further relax this assumption to incorporate even high-order chains, but there is a tradeoff in terms of the number of such chains that one can observe in a single sample. As such, we test whether incorporating higher order Markov assumptions improves the validity of the VHE. In all cases, we also include the branching correction ; we call this estimator the "VHE with higher order Markov", or VHEhom for short. We test two higher order Markov assumptions: first we focus on a 2 nd order assumption then we focus on a 3 rd order assumption. We also present results for the SBE which adds another dimension of comparison with three variants of the algebraically based VHE estimator. Fig 4 presents the distribution of coverage rates across the different networks for the VHE estimates of RDS sampling variance, for the VHEwbc estimates, for the VHEhom estimates , and for the SBE estimates. In most cases, the SBE outperforms the variants of the VHE we tested in most cases. This is most clearly true for estimates of percent female in both the Add Health and Facebook networks and the percent participating in sports in the Add Health networks. The SBE's results are closer to the VHE variants for the race variable in the Add Health networks and worse than the VHEwbc estimator for the freshman variable in the Facebook networks. The proposed adjustments we consider generally improve the median coverage rate, but not substantially. Their effects also differ by variable and dataset. For the female variable in the Add Health networks, the VHEwbc improves estimates in all cases; the median, both quartiles and the outlier dots move closer to the desired 0.95. The VHEhom also improves estimates, if only marginally. For race in the Add Health networks, both the VHEwbc and the VHEhom outperform the VHE, but the VHEhom underperforms the VHEwbc. By contrast, for sports participation in the Add Health networks, the VHEhom substantially outperforms the VHEwbc. These cases illustrate that neither approach is significantly better than the other. In the Facebook 100 networks, we did not test the VHEhom owing to the size of these networks and the computational complexity of enumerating higher order chains, but suspect that the same general conclusions will hold. However, these networks are still interesting because they show just how much of a difference the VHEwbc can make. For the gender variable, there is almost no difference between the VHE and the VHEwbc. However, for percent freshmen, the difference is enormous with the 25 th percentile estimate from the VHEwbc higher than the 95 th percentile whisker from the VHE. On balance, however, the proposed corrections do not appear to substantially improve the coverage rates as none of median estimates are close to 0.95. On balance, researchers would be less likely to make inferential errors using the SBE estimator than any of the VHE variants we tested, but we note that they would still make the wrong inference frequently. As a final illustration of the potential of these adjustments, we consider Fig 5 . It plots differences between population design effects and the estimated design effects for the first , second , and third order Markov strategies for one variable across the different networks in the Add Health data. The VHEwbc estimates are primarily found in the top left, and are often about an order of magnitude lower than the population DE. The VHEhom is slightly closer to the line of parity, but still substantially different. The 3 rd Order Markov estimates are more scattered, but do not appear to be much better than the 2 nd Order estimates. Indeed, some are worse. This is because there is less and less data the higher order Markov process we estimate, and consequently additional error may be introduced by using higher order estimates. The reason for this fact is that there are fewer cases corresponding to each type of sequence the higher we go; as the cells become sparser, the precision with which they are estimated decreases. To summarize our analyses of potential corrections to the VHE, we note that the proposed corrections-more accurately accounting for the branching structure of the RDS chain rather than assuming a simple random walk and attempting to estimate higher order Markov transition patterns-do generally improve the variance estimation. However, the improvements we see are small and variable, and they do not improve coverage rates to a desirable level. Nonetheless, these procedures are a plausible first step toward improving estimates, and future work may improve on them. For instance, it may be that the eigensystem-based approach of the VHE fails with higher order Markov chains, but that a bootstrap approach more similar to the SBE would perform more desirably. We leave these questions for future work. More importantly, however, these results suggest the constraints that emerge from the typical RDS sampling methodology which focuses solely on recruiter-recruit links to the neglect of other relevant network data. We argue that a more fruitful approach may be to collect additional network data, either ego-networks [24,[39][40][41] or more complete structures [11,42]. --- Conclusions This paper has contributed to the literature on sampling hidden and hard to reach populations, and specifically Respondent Driven Sampling, by focusing on the issue of biased sampling variance estimation, which has only rarely been addressed to date [14,18]. Whereas prior work has documented biases in RDS mean estimators and the potential for RDS estimators to exhibit high sampling variance, the actual estimation of sampling variance has received considerably less attention. This is unfortunate for two reasons. First and most generally, if the RDS estimators of sampling variance are biased, then researchers cannot trust confidence intervals and hypothesis tests derived from these estimators. This is a problem for researchers and policy makers seeking to determine which populations have the highest disease prevalence, or whether observed changes in behaviors within a single population over time are due to actual changes or simply the variability that would be obtained through repeated sampling, to name two examples. Second, in the case of RDS, whose mean estimators are known to exhibit high sampling variance, an inaccurate means of estimating sampling variance will be especially problematic if it is downwardly biased. Our results suggest that the sampling variance estimators in use for RDS data are downwardly biased, indeed, massively so. Similar conclusions on a smaller scale have been highlighted in prior work [14,18]. We also found that the SBE generally outperforms the VHE or any natural extension of it, if only by a marginal amount. Further, by focusing on the exact reasons for biases in the RDS variance estimators, this paper clarifies the heuristic notions prevalent in the RDS literature about which types of networks will be "problem cases" where RDS should not be applied. Unfortunately, our results demonstrate that such "problem cases" are common. Through mathematical illustrations, computational examples, and empirical analysis of 215 observed social networks from two different data sources, we have shown that the key assumption made by current RDS variance estimators-the First Order Markov assumption-is frequently violated. In addition, our results extend those of prior work [14,18] and show that the variance estimators perform poorly in many situations, and that the VHE as well as the SBE suffers this limitation. We examined two modifications to the VHE in an effort to reduce these biases, but, though they both offered some improvement, neither fundamentally solved the problem. This paper has outlined new reasons that variance estimation in RDS needs more attention. Based on the performance of currently available estimators, a prudent researcher must wonder whether meaningful confidence intervals and hypothesis tests can be constructed. Given the results presented here, this does not appear to be the case because the variance estimators are so biased as to be effectively meaningless. However, further work may correct these issues, and other approaches to RDS estimation and diagnostic [10,13,24,39,40,43,44] and chain referral sampling [11,42,[45][46][47] are being developed. These approaches, combined with renewed attention to the issue of estimating sampling variance in RDS, should pave the way for more accurate sampling variance estimation and a renewed emphasis on collecting additional network data as part of the sampling process. ---

This paper explores bias in the estimation of sampling variance in Respondent Driven Sampling (RDS). Prior methodological work on RDS has focused on its problematic assumptions and the biases and inefficiencies of its estimators of the population mean. Nonetheless, researchers have given only slight attention to the topic of estimating sampling variance in RDS, despite the importance of variance estimation for the construction of confidence intervals and hypothesis tests. In this paper, we show that the estimators of RDS sampling variance rely on a critical assumption that the network is First Order Markov (FOM) with respect to the dependent variable of interest. We demonstrate, through intuitive examples, mathematical generalizations, and computational experiments that current RDS variance estimators will always underestimate the population sampling variance of RDS in empirical networks that do not conform to the FOM assumption. Analysis of 215 observed university and school networks from Facebook and Add Health indicates that the FOM assumption is violated in every empirical network we analyze, and that these violations lead to substantially biased RDS estimators of sampling variance. We propose and test two alternative variance estimators that show some promise for reducing biases, but which also illustrate the limits of estimating sampling variance with only partial information on the underlying population social network.

INTRODUCTION The first days and weeks of life are critical for all newborns, especially in low-resource contexts in the Global South. Approximately 75% of newborn deaths occur in the first week of life, 40% on the first day and the majority occur at home. 1 The family environment in which newborns begin life is critical to their health and survival, however, global efforts to promote newborn survival focus primarily on strengthening health services while giving limited attention to reinforcing family caregivers' roles and practices. Despite increased investment to reduce neonatal mortality, reduction has been slow. 2 Essential family-level practices have been identified, for example, thermal care, exclusive breast feeding and evacuation of newborns when danger signs appear, yet efforts to support family caregivers have been limited. All non-western cultures have strategies to promote newborn survival with advisory and caregiving roles conferred on different family members around new mothers. 3 This paper presents: the rationale for conceptualising newborn care within family systems; and examples of research from numerous contexts in Africa, Asia and Latin America that provide evidence of the role and influence of senior women, or grandmothers, on newborn care. My discussion of newborn care within family systems is framed by fundamental features of the structure and dynamics of non-western --- Summary box ► Interventions to promote newborn health and survival in the Global South primarily focus on young mothers based on a Eurocentric view of nuclear family structure. ► Evidences from studies in numerous cultural contexts across Africa, Asia and Latin America reveal the significant role and influence of experienced older women, or grandmothers, on newborn care within multi-generational family systems. ► Research reviewed from non-western collectivist cultures across the Global South suggests that grandmothers play similar core roles with newborns through advising and direct caregiving while their culturally specific practices vary greatly. ► Future newborn policies and research should be grounded in a family systems frame that reflects the structure and dynamics of families in non-western collectivist cultures. ► Grandmothers' knowledge is sometimes out of date, but given their proximity, authority and commitment to newborn survival they should be explicitly involved in newborn interventions in order to optimise this abundant local resource for newborn health and survival. BMJ Global Health collectivist cultures that have received limited attention in global efforts to promote newborn survival. CONCEPTUAL FRAMEWORK Culture: a neglected parameter in global health For many years, anthropologists have lamented the limited attention given to culture in global health. 4 Airhihenbuwa articulates this concern in his seminal book, Health and Culture: Beyond the Western Paradigm arguing that health programmes in Africa have 'consistently undermined, and in most cases ignored, the centrality of culture' often resulting in culturally inappropriate interventions. 5 From an Asian perspective, Kumar et al discuss the primacy of the sociocultural context in community health and assert that many newborn interventions are culturally blind. 3 Inadequate clarity on the notion of culture contributes to its neglect. Weidman delineates two fundamental dimensions of health culture: cognitive and normative health-related values and beliefs; and the social system, including roles, relationships and health-seeking patterns, in which healthrelated events occur. 6 Most community health research and interventions primarily focus on the first dimension, involving individual-level psychological factors, while giving limited attention to the wider social context. 7 --- Characteristics of collectivist cultures On a continuum from individualist to collectivist, approximately 88% of all societies are fundamentally collectivist. 8 The collectivist structure of the Majority World should be a central consideration in newborn research and interventions. Salient characteristics of collectivist cultures that should orient newborn programmes include: hierarchy based on age and experience; elders are teachers of younger generations; gender-specific roles; multigenerational families and childrearing; collective decision making and interdependency valued over autonomy. 9 The position of individuals in collectivist cultures, nested in family, community and cultural systems, is visualised in the onion model 9 . Asian researchers contrast the western ideal of individual agency with the non-western value of collective agency. 10 Similarly, in the African concept of Ubuntu, individuals value interdependency over autonomy. 11 And Turkish psychologist Kagitcibasi contrasts western cultures of separateness with eastern cultures of relatedness and interconnectedness. 12 For example, for a young pregnant woman in Mali or Bangladesh, embedded in the onion, it is extremely difficult for her to reject the expectations of family, community and culture. She is expected to conform while concurrently she receives valuable support from those around her. African anthropologists identify generally ignored facet of collectivist family values. 13 They explain that in African cultures, a child belongs to, and is the responsibility of the extended family. Collective responsibility for raising children has clear implications for newborn research and interventions, however, it is consistently overlooked in global health frameworks and interventions which narrowly focus on biological parents and offspring. --- Household production of health Historically, community health interventions have primarily focused on formal health systems while neglecting family health systems. 14 In the 1990s, a group of prominent social scientists asserted that determinants of child health emanate primarily from within households. They proposed the household production of health framework asserting that child health interventions should build on intra-household roles, dynamics and decision making. 15 HPH proponents emphasised the influence of culture on child health related to three aspects of non-western households: several categories of household actors participate in caregiving; roles of men and women are gender-specific and child health practices are culturally prescribed. The HPH framework implies a profound conceptual shift from a narrow focus on women and children to an inclusive family systems perspective. 15 While the ground-breaking HPH work has been largely forgotten, current interest in socioecological or systems, approaches to addressing health issues echoes the perspective of the HPH pioneers. 16 Reconsidering the mother-child dyad A key focus of maternal and child health policies, research and programmes is the mother-child dyad, rooted in the risk group orientation and nuclear family model. While this concept has been widely embraced in public health in the Global South, it is incompatible with the structure of non-western collectivist cultures where mothers and children are embedded in extended family systems. The misfit between the reductionist motherchild dyad frame and the influence of extended families --- BMJ Global Health in collectivist cultures is confirmed by research on household structure in 31 Sub-Saharan African countries. 'Most mother-child dyads are not self-contained units but are part of a larger family system in which mothers are supported by other family members and in particular by grandmothers'. 17 --- A small, but growing, body of research on collectivist cultures in the Global South looks beyond the motherchild dyad and biological parents, elucidating the roles and influence of both kin and non-kin advisors and caregivers in MCH. 18 This research on MCH topics offers increased insight into family systems structure and dynamics and consistently reveals the central role of senior women, or grandmothers, in rural contexts, for example, in India 19 and Brazil 20 and also in many urban settings for example, in Nairobi 21 and Quito. 22 From a systemic view of families in collectivist Brazil, researchers propose recentring MCH's focus on the child-mother-grandmother triad. 23 Hierarchy based on gender and experience Collectivist cultures are hierarchically structured, and authority is associated with gender, age and experience. Men's and women's roles are gender specific and with age and experience, their authority increases in their respective domains of responsibility. 'Behaviours are passed on from older, more experienced actors to younger, lesser experienced actors'. 3 In collectivist cultures, for all MCH matters, major responsibility lies with female family members whose knowledge and experience increase with age. 18 Consequently, across non-western cultures, families recognise senior women as authorities on all aspects of child development, often advising men on the support they should provide. 'In Southeast Asia, grandmothers are considered storehouses of knowledge and wisdom on a wide array of household topics'regarding MCH. 24 Grandmothers' influence often extends beyond the family. Research in 5 West African countries documents the role of influential networks of senior female kin and non-kin in newborn care collectively coaching and supervising women during pregnancy and postpartum. 25 --- Limitations of newborn frameworks Neonates are embedded in families where there are multiple caregivers. Unfortunately, global newborn policy frameworks give limited attention to the roles of various family caregivers. For example, one objective of the 2014 WHO/UNICEF Every Newborn Action Plan is to harness the power of parents, families and communities, however, no reference is made to intrafamilial roles beyond that of parents. 26 UNICEF's 2018 framework, Every Child Alive, prioritises improving health facility newborn care but makes no mention of the need to strengthen family caregiving. 2 Factors that can explain this omission include: the medicalised focus on health services for newborns; the Eurocentric assumption that mothers are autonomous caregivers 5 ; inadequate understanding of multi-generational caregiving in families; the deficit model in public health which ignores community assets 27 ; and ageist biases, particularly against older women. --- EVIDENCE OF GRANDMOTHERS' ROLE IN NEWBORN CARE In the second part of the paper, I review evidence of the role and influence of grandmothers on newborn care from published reports of qualitative research from Africa, Asia and Latin America, conducted since 2000, and published in English, Spanish and French. Studies reviewed were identified through periodic manual searches over the past 10 years in Google Scholar, Research Gate and PubMed using these key words: newborns; neonatal; MCH; and grandmother. I identified more than 70 studies from 29 countries, primarily in published articles but also in grey literature obtained mainly through personal contacts with international development organisations. Due to space limitations, I refer to only a selection of the 70 studies in the following discussion. Analysis of the identified research focused on three components of family systems proposed by Aubel and Rychtarik: core cultural values related to women and children; family and community roles related to newborn care; and cultural norms and practices with newborns. 9 Based on this family systems framework, research findings regarding seven themes related to family level newborn care are presented. --- Grandmothers' role as authoritative advisors and caregivers An earlier review presented evidence of the role of senior women, collectively referred to here as grandmothers, as caregivers and advisors on child nutrition in the nonwestern world. 18 Here, research from Africa, Asia and Latin America provides evidence of the dual-faceted role of grandmothers providing newborn care and advising young mothers, for example, from Africa: Ethiopia 28 ; Ghana [29][30][31] ; Kenya 21 ; Benin 13 ; Mauritania 25 ; Malawi 32 ; Niger 33 ; Nigeria 34 ; Egypt 35 ; Burkina Faso 36 ; Senegal 37 ; Nigeria, Tanzania and Ethiopia. 38 Mali 39 ; and Uganda 40 ; from Asia: India 27 41-43 ; Pakistan 44 ; Nepal 24 ; Indonesia 45 ; and from Latin America: Brazil 46 ; Mexico; Costa Rica 47 ; Colombia. 48 Research conducted in this wide array of sociocultural contexts reveals similar gender-specific and age-specific roles of family members in newborn care with consistent recognition of grandmothers' expertise and authority. Brazilian researchers report that young urban mothers refer to grandmothers as their eternal guides, 'They brought security and comfort justified by the experience accumulated by these women' 23 and breastfeeding research in urban Mexico concludes, 'grandmothers continue to play a leadership role in families as primary advisors and health promoters' . Two studies in Ghana conclude, 'parents consider grandmothers as the epitome of wisdom and knowledge BMJ Global Health regarding newborn care' 31 and it is the husband's mother who calls the shots on newborn care. 29 In Nepal researchers concluded that advice from revered and authoritative grandmothers cannot be refused. 24 Indian researchers report on grandmothers' influence within the web of social relationships surrounding and influencing younger mothers. 19 A similar pattern emerges across non-western cultural settings: newborn care is coordinated by older women; younger women are learners; and men are rarely directly involved. --- Inculcating cultural norms and practices As culturally designated authorities on newborn care, grandmothers are responsible for transmitting an array of culturally prescribed practices involving: massage; cord care; colostrum; prelacteals; thermal care; breast feeding; newborn illness and spiritual protection. While certain traditional practices conveyed by grandmothers are beneficial, others are not, such as giving prelacteals and withholding colostrum. In Karnataka, grandmothers initiate young mothers to the use of prelacteals, cord and thermal care. 49 Egyptian grandmothers give newborns herbal drinks to stop babies from crying at night. 35 In Nigeria grandmothers' harmful practices include giving water, prelacteals and traditional medicines 34 while in Brazil grandmothers give traditional teas and water to newborns. 50 In Northern Ghana, grandmothers' demonstrate ancestral traditions involving placenta burial and breastmilk stimulation. 29 Across cultures, grandmothers' training efforts are greater with primiparas, to instil practices that should be replicated with subsequent newborns, for example, in Uganda. 40 Young mothers are expected to defer to the advice of their recognised senior advisors. Newborn research in four African sites documents young mothers' dependency on experienced grandmothers where often 'mothers themselves only took full responsibility of newborn care after a few days or weeks'. 38 Similarly, in Nepal young mothers cannot refuse directives on traditional newborn practices from revered grandmothers. 24 Many studies, for example, in Malawi 32 Mauritania 25 and India 19 document mothers' adherence to grandmothers' traditional advice on infant care, given their trust and respect for them, even when it conflicts with health workers' advice. --- Seclusion of mothers and newborns: protecting and teaching In many cultural settings, initial newborn care, protection and feeding involve seclusion of infants and mothers, for varying lengths of time, usually under the supervision of an experienced grandmother or other senior female family member. In Karnataka, during 40 days of confinement 49 and in Nepal during 6 days of chhatiyar, grandmothers transmit their experience concerning all aspects of newborn care and feeding, to new mothers. 51 In Senegal, 7 days of seclusion protects newborns from the mystical we illness. 37 Studies in many other cultural contexts describe this protective and educational period of isolation for babies and mothers, for example, in Indonesia, 45 Niger, 33 Burkina Faso, 36 India 3 and Amazonia. 20 --- Grandmothers' role in breast feeding There is extensive evidence of grandmothers' influence on whether and how long young mothers' breastfeed. While mothers may receive occasional advice from health workers, most women benefit from in-house breastfeeding coaches, that is, grandmothers and other experienced female family members, for example, in Ghana, 52 Benin, 13 Mauritania 25 and Brazil. 50 In Bihar, mothersin-law are primary decision-makers on breastfeeding initiation, practices and problems. 42 In urban Ecuador, most new mothers identified grandmothers as their primary breastfeeding advisors and all others cited other older female kin and nonkin. In urban Mexico, grandmothers' advice was the main determinant of whether young mothers breastfed. Based on social network analysis, other Mexican researchers found that most women were primarily encouraged to breastfeed by maternal grandmothers, while less encouragement came from doctors and from paternal grandmothers . 53 Some studies report on grandmothers' advice for increasing breastmilk production, for example, from urban Bangladesh 54 and rural Ghana. 29 Grandmothers often advise certain foods and liquid drinks for lactating women, for example, in urban Bangladesh both kin and non-kin grandmothers advise specific foods to increase breastmilk. 54 In many cultures, grandmothers believe that some breastmilk is of poor quality or insufficient quantity, for example, in Ghana 55 and Egypt, 35 ideas refuted by nutritionists. In Egypt, grandmothers believe that breastmilk is poor quality when a baby cries after breast feeding and this justifies giving herbal drinks to infants. 35 As discussed above, while grandmothers' advice may be erroneous, it is often very difficult for young mothers to reject the advice of their resident senior breast feeding advisors. --- Care of sick newborns Most research on newborn illness focuses on health services and technologies. Few studies have systematically investigated household-level strategies to care for sick neonates. Evidence from various non-western contexts reveals that during neonatal illnesses, families recognise grandmothers' expertise to diagnose problems, to provide home treatment and to determine if extrafamilial support is necessary. In Ghana, researchers concluded that 'Older women and grandmothers are often the ones who decide that a child is in need of treatment and often also determine the type of treatment due to their authority within the household and the community'". 30 There are similar findings on grandmothers' central role in caring for sick newborns from Mali, 39 Indonesia 45 and India. 10 41 42 As in other contexts, in Nepal researchers found that decision making around --- BMJ Global Health newborn illness is a collective process mainly involving female family members, with husbands playing a limited role. 51 In many societies, there are mystical newborn illnesses that medical doctors do not understand but that grandmothers have experience treating, for example, in India 10 and Senegal. 37 Across Latin America the folk illness, empacho, is treated outside the formal health system by grandmothers, aunts and female neighbours. 56 --- Indigenous newborn support systems In collectivist non-western cultures, solidarity and reciprocity are primordial values reflected in the spirit of cooperative breeding 57 and embodied in ubiquitous womencentred, multigenerational indigenous support systems for women and children. 18 Unfortunately, the influence of these social support networks on pregnancy, newborn and child health has been largely unexplored in research in the Global South. South African, Steuart, was an early observer of these naturally occurring networks and referred to them as primary friendship groups. 58 In Mali, researchers identified four types of support for MCH provided by these social networks: practical, informational, emotional and material or tangible assistance. 59 In the extensive research on newborn research, these social networks has been largely ignored, reflecting the Eurocentric and reductionist focus on nuclear families and mother-child dyads. A few studies have investigated the influence of these social networks, composed primarily of experienced older women, kin and non-kin, for example, in Nepal 60 and Benin. 13 Research in Odisha describes these indigenous social support networks as webs of social influence that are 'powerful disseminators of knowledge' to young mothers on newborn care. 19 A very significant finding from this research in India is that compared with better-educated women, less-educated women's health knowledge is more influenced by their social networks, composed primarily of older women, than by formal health workers. 19 In all contexts, these social networks convey to younger women not only prescribed practices but also provide them with crucial emotional support. Most studies that discuss grandmothers' role in newborn care, focus on their knowledge and practices, while giving little attention to the emotional support they provide to young mothers. Costa Rican researchers discuss the importance to young mothers of grandmother social support networks, or redes de apoyo. 47 From two different contexts in Brazil 20 23 researchers describe the invaluable contribution of these networks in creating social bonds between women around newborn care which often provide ongoing reciprocity and support. --- Men's role in newborn care Many current MCH programmes target fathers, however, there is a dearth of in-depth research on men's roles in MCH within family systems in non-western contexts. Many studies from different sociocultural contexts conclude that: men do not play an active role either in direct caregiving nor in day-to-day decision-making concerning newborns; and that men play important supporting roles, particularly during newborn illnesses, providing logistical and financial help, for example, in Benin, 13 Ghana, 29 Burkina Faso, 36 Mali, 39 Niger, 33 Senegal, 37 Nepal 51 and Indonesia. 45 This same pattern regarding men's involvement is reported from four African contexts supporting the conclusion that 'physical and social spaces related to newborn care were closed to men, that men were perceived to lack the ability to care for the newborn and had little physical contact with the baby, but as head of the family, they were often the symbolic decision-makers and the financial provider'. 38 Families recognise that newborn care requires specialised expertise which men do not have. In many cultural settings during the first days after birth men are excluded from gendered newborn spaces. 18 It is often erroneously assumed that men, as head of household, are primary decision-makers on all family issues. The research reviewed here supports the conclusion that for newborn care men are rarely central decision-makers and more often are advisees of experienced women within the family. A present trend, reflecting nuclear family structure in the Global North, is the promotion of men's involvement in all aspects of MCH, including newborn care, based on the assumption that greater men's participation is always advantageous. Some researchers have questioned this assumption, expressing concern that increased men's involvement and authority in the newborn space can inadvertently disempower women and disrupt the women-to-women support system for mothers and neonates. Research in multiple African settings supports the conclusion that "care must be taken to ensure that interventions to involve men do not result in disempowerment of female family members". 38 --- --- CONCLUSIONS This discussion on newborn care with families in Africa, Asia and Latin America provides insight into an aspect of newborn health that heretofore has been given limited attention-the role of grandmothers. A growing body of evidence from across non-western, collectivist societies reveals the culturally designated role of these experienced, older women as newborn advisors and caregivers. Unfortunately, at the global level, newborn research, policies and interventions continue to focus primarily on medical technologies and services, 61 to a lesser extent on mothers while giving little attention to intrahousehold newborn strategies in which grandmothers are key components of families' sociocultural operating systems. 4 Research identified from 70 cultural contexts across Africa, Asia and Latin America reveals significant commonalities regarding core roles played by grandmothers with newborns, while their culture-specific BMJ Global Health practices can vary greatly. Grandmothers' core roles are summarised by researchers in Ghana 'Grandmothers play an integral and multifaceted role in maternal and infant health. Grandmothers support mothers during pregnancy, care for infants following delivery, maintain cultural traditions surrounding pregnancy and childbirth and serve as repositories of knowledge on local treatment options'. 29 These core roles of Ghanaian grandmothers during pregnancy and with newborns are virtually the same as those of grandmothers identified in studies from all three continents. Grandmothers' central role in newborn care in nonwestern societies emanates from the structure and core values of collectivist cultures where: reciprocity and solidarity are key values; the social hierarchy in families is determined by gender and age; roles are gender-specific; social norms are transmitted from older to younger women; there is collective responsibility for children, especially among female caregivers; and decision-making is collective, with experienced elder women having authority. A vital social resource for MCH in collectivist cultures, that is very beneficial to newborns are the indigenous social support networks of women, both kin and non-kin, who collaborate to ensure optimal newborn care within families. 18 These networks demonstrate what evolutionary anthropologists refer to as cooperative breeding where mothers, across the species, are supported by others, mainly female kin, and non-kin, in raising their offspring. 57 61 62 Many studies report that grandmothers' knowledge is not up to date and that some of their newborn practices are harmful, such as giving prelacteals. The research reviewed clearly reveals that while their knowledge is often inadequate, they wield significant influence over both younger mothers and other family members, based on the belief in and respect for their experience. Grandmothers' authority and power is demonstrated by the fact that respectful daughters and daughters-in-law often hesitate to contest their senior advisors' instructions when they differ from health workers' advice. This research reviewed represents a growing, but still minor, stream of research on grandmothers' role in newborn care in the Global South. Hundreds of past studies on MCH issues in non-western cultures make no mention of grandmothers' role. What can explain this oversight? In my analysis, this discrepancy points to inadequacies in the dominant conceptual frameworks used in research, of Euro-American origin, which fail to take into account the structure and dynamics of extended family systems in non-western, collectivist cultures. 7 12 Most studies on newborn care in the non-western world are conceptually grounded in a western model of the nuclear family where: care for neonates is the responsibility of biological parents; the roles of women and men overlap; grandparents often are not nearby; and young parents give greater credence to doctors' advice than to that of their elders. Conceptual frameworks for research across the non-western world, cloned from the West, consistently overlooks the distinct features of collectivist societies. 12 Research on newborn caregiving, primarily investigates mothers' knowledge and practices, and sometimes those of fathers, while giving limited attention to the wider family systems in which newborns are embedded. Future research should investigate characteristics of family systems in collectivist cultures related to: genderspecific roles in families related to MCH; multigenerational involvement in childcare; the hierarchical structure of authority among women, related to age and experience; roles of men and women of different generations; collective decision making within families; roles of men and women of different generations; and men's involvement in different facets of MCH. Regarding men's involvement in newborn care, my review of existing research presents a consistent finding that men are very rarely directly involved in newborn care. These findings question the frequently held assumption that men are central actors and decision-makers in all aspects of family life, perhaps overestimating the power of the patriarchy. The supposition regarding men's hegemonic role in women's and children's lives in nonwestern families is questionable in two respects. First, evolutionary and biological anthropologists now question long-standing assumptions regarding males' central role in child rearing in human societies in light of cooperative breeding patterns where females play a leading role. 57 61 Second, several non-western female social scientists critique the historical discourse which places men at the centre of all family life. Nigerian anthropologist, Amadiume, asserts that excessive attention has been given to the patriarchal paradigm, while at the centre of African families' social life is a matriarchal system which has gone largely unrecognised. 63 She vehemently questions widely held assumptions about the all-encompassing authority of the patriarchy in female domains of responsibility in family life and emphasises the importance of the matriarchy in those spaces. Similarly, from a South Asian perspective on gender dynamics, Mumtaz critiques the overemphasis on men's influence on women's lives and calls attention to the importance of women-to-women bonds and intergenerational female relationships wherein younger women's ties with experienced and wise older women are particularly edifying. 64 These rarely heard critiques of the notion of men's all-encompassing authority in the lives of women and children evoke the need for greater investigation into women-centred systems of collaboration and authority within family systems. Several recent experiences using a grandmotherinclusive approach to address health issues related to grandmothers' areas of responsibility and expertise provide evidence of their potential to contribute to positive health-related outcomes. Evaluations of these innovative programmes have shown positive results strengthening grandmothers' knowledge, caregiving Encouraging results of grandmother-inclusive interventions are reported from: Nepal concerning antenatal and newborn care 65 ; Uttar Pradesh addressing newborn care 3 ; Malawi, dealing both with newborn care 66 and adolescent HIV prevention 67 ; Burkina Faso addressing child nutrition 68 and Senegal addressing child nutrition 69 and adolescent health issues. 70 71 These several positive experiences, using a grandmother-inclusive approach in different cultural contexts, suggest that grandmothers may not be as intransigent to change as often assumed. These results also support the need for further experimentation with and evaluation of grandmother-inclusive interventions addressing newborn, other MCH and adolescent issues. Two major recommendations emerge from this discussion supporting the need: to reconceptualise newborn research and interventions; and to involve grandmothers in newborn interventions. First, existing evidence on the role and influence of grandmothers on newborn health highlights the incongruity between the decisive role played by these culturally designated newborn advisors and caregivers and the fact that most newborn policies and programmes virtually ignore their role in the lives of newborns and the enormous social resource they represent for community newborn interventions. To more effectively promote newborn health and survival I contend that there is a need for a major conceptual shift in newborn research to overcome the prevailing conceptual bias which reflects a Eurocentric view of families rather than the structure and values of more collectivist non-western societies. 7 16 Making this shift will require revision of prevailing conceptual frameworks and development of alternative research methodologies. This reframing should contribute to better understanding communities' emic perspective and lead to more effective strategies to promote newborn survival. Pioneering work by some scholars from the Global South exhibits a more systemic and culturally grounded frame for MCH research on intra-household dynamics indicating their greater cognizance of both the structure of collectivist cultures and of the critical role played by elders within those systems. Second, the numerous studies reviewed here provide extensive evidence of grandmothers' role as caregivers and decision-makers in newborn care across the Global South. Extant research supports the assertion that the involvement of grandmothers in newborn care is beneficial to young mothers and to newborns, however, it is also reveals that some of grandmothers' knowledge and practices are out of date, and sometimes harmful. Most studies strongly advocate for their inclusion in future newborn interventions based on their power and influence within their respective cultures and on the notion that their involvement may strengthen their knowledge and practices. While all research reviewed provides evidence of grandmothers' influence on newborn care, surprisingly, some does not explicitly recommend their inclusion in future programmes. Unfortunately, there is continued reluctance to explicitly involve grandmothers in interventions addressing newborn and other MCH issues. This appears to be related to: the often-repeated idea that grandmothers are barriers to change; a narrow perception of grandmothers focusing on their harmful traditional practices; and gender and ageist biases toward older women. Policy and programme planners should realise that the failure to explicitly include these cultural authorities can be perceived by grandmothers themselves and by their communities as an affront to their culture and engender resistance to well-intentioned programmes, thereby limiting their effectiveness. This review provides support for the use of unconventional, but culturally grounded criteria for choosing community target groups for newborn interventions, namely, cultural authority, age and experience with newborns. --- Competing interests None declared. Patient consent for publication Not required. Provenance and peer review Not commissioned; externally peer reviewed. Data availability statement This paper no data is referred to. Supplemental material This content has been supplied by the author. It has not been vetted by BMJ Publishing Group Limited and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations , and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Across the globe, the well-being of newborns is significantly influenced by the knowledge and practices of family members, yet global health policies and interventions primarily focus on strengthening health services to save newborn lives. Predominant approaches to promote newborn survival in non-western cultures across the Global South are based on a western, nuclear family model and ignore the roles of caregivers within wider family systems, whose attitudes and practices are determined by culturally prescribed strategies. In this paper, I review evidence of a neglected facet of newborn care, the role and influence of senior women or grandmothers. Based on a family systems frame, I reviewed research from numerous settings in Africa, Asia and Latin America that provides insight into family roles related to newborn care, specifically of grandmothers. I identified primarily published studies which provide evidence of grandmothers' role as culturally designated and influential newborn advisors to young mothers and direct caregivers. Research from all three continents reveals that grandmothers play similar core roles in newborn care while their culturally specific practices vary. This review supports two main conclusions. First, future newborn research should be conceptualised within a family systems framework that reflects the structure and dynamics of non-western collectivist cultures. Second, newborn interventions should aim not only to strengthen health services but also influential family caregivers, particularly grandmothers and the indigenous social support networks of which they are a part, in order to improve family-level newborn practices and save newborn lives.

INTRODUCTION Hopefulness and Frustration in Domestic Life in Manju Kapur's A Married Women Among the Indian-English women novelists, Manju Kapur has acquired a distinguished position. She has sensitively and realistically dealt with the typical Indian themes and lucidly manifested the contemporary middle-class women with a rare proficiency. In a greater sense, her themes are nothing less than human predicament and alienation. Her portrayal has become more realistic with a distinct impetus of anxiety and hostility of modern women. The women characters conceived by Kapur feel torn between the rival pulls of tradition and modernity. The predicament of the working woman in her novels gets amplified by her issues with marital compromises and her quest for her identity. Kapurs second novel, "A Married Woman" was published in 2002, and is the tale of Astha, an educated, upper-middle-class working woman. The narrative is set at a time of religious and political catastrophe. As a sensitive daughter of an open-minded father and conservative mother, Astha was brought-up in an educated middle-class family in South Delhi. Unlike most unmarried women she has her infatuated love for Bunty, a boy of a neighbor colony, and then for Rohan who goes abroad for a better job. But her real story of love and marriage began with Hemant, the son of an influential government official in Delhi. Soon after marriage, Astha gets disappointed about human nature in general and the politics of the nation in specific. She is fed up with the politicians who, in the guise of democrats and socialists try to organize different yantras for their vote banks under the pseudo-secular banner of national unity. On the other hand, Astha's family affairs are not on god terms and nothing is right with her. Hemant does not prove to be a suitable boy nor does she become a desirable daughter-inlaw in her in-law's house. http://ijlllc.org/ Astha has everything that a woman could ask for a responsible husband, children, and nice surroundings in the central part of Delhi, Indian Capital. But Astha's requirement to affirm her identity finds expression in the paintings that she makes. Art is a refuge for Astha. She strives to express herself but puts her thoughts on the canvas. Her husband was lovable and caring when Anuradha was born. He stresses giving Anuradha her Sunday bath. He has been educated abroad and has perceived how couples equally share the upbringing of the child. "He insisted on doing this, ideas about fatherhood are so antiquated in India" . But Hemant's aspiration to have a son and his emphasis that he would not stop until he has one makes Astha frustrated and anguished. She cannot understand her man. Just after marriage, Astha started teaching in a school. Her job forces her to wake up early so that she can reach her school on time. She has exercises to convert, and lessons to prepare, but because she is still very much in love with Hemant, she looks forward to spending the evenings with his company. Hemant starts coming late, accuses his manager when he fails to complete his projects, and in general expresses his disillusionment with everything. These criticisms leave her, dreary, cold, and detached from him. She has been waiting for him all day, contemplating of their being together, but nothing of this is reciprocated. He is a criminal, deteriorating her anticipation and impairing her happiness. When Astha is pregnant for the second time, everyone hopes male baby. Though she declares to the fetus that she would love the child is it a boy or a girl, she is very much anxious. Soon after, Astha's happiness receives a jolt because "between Anuradha's birth and Himanshu's, Hemant changes from being an all-American father to being an all-Indian one" . He refuses to keep with Himanshu's upbringing. He has made business trips to abroad four times a year as he started a TV manufacturing unit in Astha's name in Noida. This leaves Astha to cope singlehandedly with her dilemmas -job, children, and home. Astha dabbles with the idea of resigning but does not want to do so: "…she too had changed from being a woman who only wanted love, to a woman who valued independence. Besides, there was the pleasure of interacting with minds instead of needs" . Gradually, headaches start to attack Astha until the doctors advised an operation on her nose. For four days she is hospitalized and during this time Hemant is very lovable and caring. Returning home, she realizes that the children spend more time with the grandparents than with her. She feels lonely, but her complaint to Hemant is in vain because he too complains of loneliness stating that she spends most of her time with children and profession and not him. Astha starts writing poetry and this alleviates the pressure within her. However, Hemant understands these emotional outbursts as "positively neurotic" and fears that people might think that Astha is unhappy. She gives up writing and starts sketching and drawing, after all, "nobody could put two and two together about painting ..." . When Astha's father expires, her mother is influenced by a Swami and moves to his Ashram in Rishikesh -there she finds solace and peace. She sold her property in Delhi and gave a part of the proceeds of the sale to Heman. Astha disapproves of this and she displeasured in a couched language only to be snubbed by her husband and mother. After some time, Astha comes under the influence of Aijaz Akhatar Khan and his Street Theatre Group. It is consequential that she starts a lesbian relationship with Aijaz's wife, Pipeelika after Aijaz's death. During the summer holidays Mrs. Dubey, the principal of the school where Astha teaches, has invited Aijaz to teach the nuances of play production to the children. Astha has been put on duty much to her family's displeasure. Aijaz asks her to write a script accentuating the Babri Masjid RamJanma bhoomi issue. With the help of her daughter Anuradha, Astha learns to browse through the library books and make notes. Her efforts are satirized by Hemant, but this only serves to strengthen her effort. Aijaz is elegant, and intelligent and recognizes http://ijlllc.org/ Astha and this serves to bring them more intimate. Astha feels, "suddenly her life seemed less constricted" . Astha never wants to cross the limit of her married life: "...she does not want any other man." After the production of the play, Astha has little to do with Aijaz. But deep within the core of her heart, she cherishes the moments spent in his company. Once he touches her knee this sudden touch sends a wave of sensation in her. Later on, during a communal riot her friend Aijaz Khan dies. Astha feels it in her heart. When Aijaz dies, she cannot prevent her tears and sentiments. Kapur has admirably depicted the angst, irritation, and struggle of the Indian middle-class a woman who is at a loss to condemn social traditions. Although she tries her best to affirm her place in family and society, she reaches nowhere due to her moral disproportions and unnatural dreams of being in communion with Pipee, the widow of Aijaz Khan. After the death of Aijaz, she meets his widow-wife Pipee who becomes a source of pleasure to her. She is committed to an affair with Pipee. She takes notice of whatever Pipee tells. Astha and Pipee, both are living under the illusion of making themselves liberated from male domination but when their liberation is weighed on the scale of morality, ethics, values, and care for family peace, they stand nowhere. But the fact is, Astha is enthralled by Pipee. They understand each other, feel for each other, and stand for each other. During their pilgrimage to Ayodhya, their bond becomes stronger. Astha lays bare her lacerated breast before Pipee to which she applies the ointment of sympathy. Despite all opposition, she leaves her children and family and goes on Ekta Yatra to Kanya Kumari from Kashmir with Pipee. During their yatra, Astha gets to know of Pipee's lesbian affairs with a few other women like Sameera and Neeraj. Pipee's demanding passion and her past affairs disappoint her. She is trapped in a terrible dilemma, whether she should stay in the sheltered existence provided by family and tradition or she should run for her freedom. A Married Woman presents another form of relationship -no force, no tricks, just plain love. Both Astha and Pipee love each other and their physical intimacy is proof of that. The reader feels for them as they make each other's life joyous and trust each other with utmost love till the situations to them of Kapur's novels enable us to get an idea of the feminist struggle against male discrimination. While reading the novel, one gets the impression that a woman's life is like the life of a nation that is passing through trials and tribulations. In portraying the inner struggle of a woman's mind, Kapur reveals a mature understanding of the female psyche. Most of all, Kapur attempts to fuse the personal with the external. She speaks as someone who has lived through the tumult of the communal notes, which in some way affected the lives of all Indians. They are major historical incidents and Kapur has intertwined their understanding into the lives of Astha and Pipee. She presents an India which is relevant both to those who stayed on in their country to face the aggression of Western influences and aggravated competition, and also those who opted to perceive the nation from Western climes. Yet, occasionally Kapur's interpretation of a lesbian relationship sometimes distracts the reader from the pressures of the situation and the core emotions of the characters. However, A Married Woman is a well-balanced portrayal of a nation's inherent progress -its failures and its strengths and the angst of a woman's uproar, which is, as complicated as the political and social revolution going on around her. Kapur has restricted women with enthusiasm and appears to propose that a married woman's duty is not as a babysitter and housekeeper but to do something more. Through Astha, Kapur offers a facade challenge to patriarchal attitudes, social organization, and control mechanisms through her inherent capacity as an individual and her dream to accomplish personal recognition. In the gynocentric struggle for liberation, Astha strives for the togetherness of the family as a unit. http://ijlllc.org/ Neither as a flag waver nor as a patriot, she is wholly aware that venality, brutality, and hypocrisy are engraved on the ponderous soul of every fanatic Indian, Indian democracy has become a holy cow in a socio-political predicament, and the benefits it yields are used and abused at will by a diluted politics. Thus, A Married Woman explores the mysterious facet of a woman's passionate life as defined by comfort levels without societal sanctions. Kapur discusses logically between events, plucking out interesting emotional developments and constantly making her tale exciting and engaging. This is not a work to rival the might of such Indian masters, but there is much here to like and enjoy. Kapur's delicate writing depicts human relations and their social context with great sensibility and close attention to the details of daily life. Kapur has joined the growing number of women writers from India on whom the image of the suffering but stoic woman eventually breaks traditional boundaries. has had a significant impact. The finite dimension of the relationship between man and woman has been prescribed by man and not by woman. Her limits have been imposed on her by a man who is ruled by mastery-motive. She accepts it because of biosocial reasons. Very often this acceptance is not congruent with the reality that lies underneath. Modern women prefer to exercise their choice and break away from their traumatic experiences. Women are now portrayed as more assertive, more liberated in their view, and more articulate in their expression than the women of the past. Instead of downgrading the elements of suffering at the hands of her lover or husband or man, she has started asserting her substantive identity in action, not in words.

This paper attempts to bring new women in Manju Kapur's A Married Woman taking into account the hopefulness and frustration in domestic life. With diverse cultures, histories, and distinct forms of ethics and values, the women's question, despite essential solidarity, needs to be tackled about the socio-cultural circumstances. The lives of women who survived and struggled under the suppressive mechanism of a restricted society can be manifested in the novels of Kapur. She has written fiction focusing on the predicament of contemporary women particularly in their struggle for empowerment and independence from the male-chauvinistic social structure. Now "Woman Question" is no longer a problem limited to the condition of women within the family or their rights to equality with men in different facets of social life. It is part of the total, far broader question considering the direction of change that Indian society is making -social, political, economic, and the intellectual viewpoint and examination of that process.

triggered by body dissatisfaction . Eating disorders are prevalent among Latinas, making prevention a priority for this group . Media portrays beauty ideals that can affect self-body image, eating behaviors, and selfesteem among adolescent Latinas . As media use rapidly evolves and social media have become dominant platforms for male and female youth, mainstream beauty production and a consumption culture are available for repetitive access. More research is needed to discern the impact of social media on body image . R. J. Williams and Ricciardelli specifically highlighted the lack of insight into the influence social media has on body image and view of oneself. Latinx are the largest and fastest growing minority group in the United States however, limited research on eating disorders addresses this group. Despite the belief that Latinx cultures and communities are more accepting of curvier female body types , many U.S. Latinas seek to be thinner . Although a majority of literature on body dissatisfaction focuses on White adolescent girls, research focusing specifically on Latina adolescent girls is increasing. Latinas and White women have similar rates of body dissatisfaction . Low self-esteem is reciprocally predictive of body dissatisfaction and perceptions of negative body image, which may lead to eating disorders among adolescent girls and women . Due to multiplicative social identities of being female and a member of racial/ethnic minorities, Latina adolescents experience intersecting forms of oppression leading to poorer preventable health outcomes. Due to their marginalized status, researchers often ignore Latina adolescents and minimize their outcomes. Here, we provide a conceptual framework that can guide prevention researchers and clinicians in how to address body dissatisfaction among Latina girls. First, we discuss key indicators of eating-disorder symptomatology and address possible causes of symptoms. We then present a conceptual model, guided by objectification and intersectionality theories, which we recommend be used for intervention development in addressing body dissatisfaction among Latina adolescent girls. Objectification theory documents the links of sexual objectification with body imagery and psychological symptomatology among girls. Intersectionality highlights the racial and sexist ideologies that contribute to overly sexualized imagery of girls of color and promotes ethnic identity as a key factor in the relationship between body image and eating disorders among Latina adolescent girls and women. --- Theoretical Framework Objectification Theory Fredrickson and Roberts first proposed objectification theory as a framework to understand the experiential consequences of being female in a culture that sexually objectifies the female body. Self-objectification begins when women begin viewing themselves as objects to be looked at and evaluated on the basis of their appearance . The theory assumes that, due to women's gender-role socialization, women's unhealthy culture manifests in a variety of internalized mental health issues . Objectification theories explain mental health and body dissatisfaction issues with women of color and women from marginalized groups including African American women and Muslim American women . Although objectification theory research with Latina girls and women is limited, findings support explaining sexual objectification among this population. Research on media studies highlighted the risks of women's self-objectification because various forms of media offer unrealistic perspectives on the average person's appearance . Internalization of unrealistic perspectives can cause girls and women to judge their appearances, manifesting in anxiety, body shaming, drug use, and eating disorders . Two studies with Latina women supported the interrelationship of internalization, body surveillance, body shaming, and eating-disorder symptomatology , which did not differ significantly between Latinx and European American participants . Latina women have the same levels of body dissatisfaction and eating-disor-der symptomatology as White women, although researchers primarily focused on White women . --- Intersectionality Theory Intersectionality theory aids in understanding the multiplicative identities of Latinas that contribute to disparities in eating-disorder symptomatology. Intersectionality theory examines multiple dimensions of identities and social locations and how they intersect . Intersectionality theory specifically acknowledges people's social location that places them at risk, describing how multiple forms of oppression can affect individuals and families, leading to barriers in achieving positive health outcomes . Researchers often view salient contextual variables-race/ethnicity, gender, sexual orientation, socioeconomic status/class, education level, and ability-as separate sociocultural demographic variables that rarely influence one another. Researchers suggested that an individual's multiple identities interact and intersect to shape personal experiences and form "intersecting oppressions … that work together to produce injustice" . --- Eating Disorders Perceptions of an ideal body can vary by age, but body-image dissatisfaction among adolescents is a strong predictor of dysfunctional eating behaviors such as dieting, purging, and binge eating . Eating disorders consistently affect several million people, but mostly affect women between 12 and 35 . The three most common eating disorders are anorexia nervosa , bulimia nervosa , and binge-eating disorder. People are diagnosed with AN when they weigh at least 15% less than the normal healthy weight expected for their height . Typical behaviors associated with AN include limiting food intake, intense fear of gaining weight, problems with body image, and denial of low body weight. Individuals with BN may go unnoticed because signs are not as physically noticeable as with AN. Individuals with BN can be slightly underweight, normal weight, or overweight . Individuals with BN binge eat frequently, and the fear of weight gain prompts them to purge by vomiting or using a laxative. They tend to repeat this cycle several times a week or, in serious cases, several times a day. Unlike with BN, individuals with a binge-eating disorder tend not to purge after they eat. Binge-eating disorder involves a high frequency of overeating during a specific period . Those individuals who suffer from binge-eating disorder along with overeating also experience a lack of control, associated with three or more of the following: eating more rapidly than normal, eating until feeling uncomfortably full, eating large amounts of food when not feeling physically hungry, eating alone because of feeling embarrassed by how much one is eating, or feeling disgusted with oneself, depressed, or guilty . --- Eating Disorders Among Young Latina Adolescents Eating-related problems were lower among ethnic-minority females than Whites , though few studies examined eating disorders among diverse populations ; researchers found significant differences in types of eating disorders among ethnic minorities. A higher percentage of Latinas suffered from BN . Adult Latina women are equally as likely as White women to develop eating disorders . Furthermore, Latina adolescent girls report more diuretic and laxative use than girls from all other ethnic groups . However, little has been revealed to guide clinicians on how to identify risk and protective factors among this group. --- Body Image Body image means the physical attributes an individual possesses and their perception of those attributes in comparison of those around them . Attitudes toward and representations of physical attributes often influence adolescents' body image, central to their developing self-concept and significantly impacting adjustment . Body image is a multidimensional construct influenced by biological, psychological, and social factors . Body imagery can form perceptual attitudes among individuals, often influenced by historical, civilizing, societal, individual, and genetic factors in a population . Similar to African American girls , the media often portrays Latina female adolescents and women as overly sexual beings . Due to this misconception, stereotypes form and negatively affect the views of Latina adolescents. Latina girls and women are often expected to be voluptuous, with large breasts . Such imagery often aligns with being overly sexual, whereas the thin ideal White body frame is often view as innocent and modelesque . This conflict in media imagery can be detrimental for Latina girls who must negate imagery that views them negatively while admiring unrealistic and sometimes unattainable body imagery. Exposure to sexually objectifying media aligns with poorer body image among Latina adolescents . Because the Internet is easily accessible, young women may be consistently exposed to images that portray mainstream beauty . Because realistic and unrealistic images are fully accessible, young women may become desensitized to imagery and normalized to the notion of aspiring to obtain unrealistic images of beauty . --- Body Image and Social Media Adolescent girls are more likely to cite specific celebrities as having their ideal body image, although young women acknowledge they possess different bodies than their beauty role models . When adolescent girls make unfavorable comparisons with media images, they are less satisfied with how they look and have lower self-esteem and increased body dissatisfaction . Social media, unlike traditional media, offers an online environment filled with celebrities and pictures of contemporary and everyday peers found in TV programs and magazines . Social media is "the media of one's peers" , comprising an online environment that allows followers to experience many opportunities to compare themselves with their peers. The extent and effects of these comparisons have not yet been thoroughly examined, along with their effects on Latina identity, body image, eating disorders, and psychological outcomes. Furthermore, media representations of Latina women, family emphasis on weight, and typical foods eaten may constitute unique risk factors for body dissatisfaction and dysfunctional eating in this group . --- Conceptual Model for Latina Adolescents --- Eating-Disorder Predictors Understanding eating-disorder symptomology among Latinas is essential to ensure disparities diminish. Our conceptual framework high-lights three key risk factors for eating disorders: self-esteem, body dissatisfaction, and social media. Self-esteem.-Self-esteem is a psychological concept referencing a person's evaluation of self-worth . A person with low self-esteem may feel their life is not worthwhile or may lack pride. Someone with low self-esteem, who has a negative view of themselves, will be more likely to negatively evaluate their body and may experience body dissatisfaction. Body dissatisfaction.-Body dissatisfaction refers to negative subjective evaluations of one's physical body, such as figure, weight, stomach, and hips . Individuals with high body dissatisfaction will be at higher risk of eating disorders because they are likely to seek changes to their bodies by taking extreme measures: reducing calorie intake or participating in self-induced purging . Body dissatisfaction should not be confused with body distortion, which occurs when one views their body as significantly larger it is . Despite the belief that Latinx cultures and communities are more accepting of women's curvier body types , many young Latinas nonetheless seek to be thinner . Latina and European American women experience comparable levels of body dissatisfaction and eating-disorder symptomatology . Social media, self-esteem, and body dissatisfaction.-Among social-media users, self-esteem may directly relate to the valence of responses received about information posted on one's personal social-media page . Individuals with a negativefeedback-seeking style who received a high number of comments on Facebook were more likely to report dysfunctional eating attitudes and weight/shape concerns . During the developmental phase of adolescence, where self-esteem is integral, negative perceptions of one's body image link to lower self-esteem . Body dissatisfaction is a risk factor for low self-esteem that can lead to eating disturbances . Extreme discontentment with one's body may accompany dysfunctional eating patterns and pathways toward depression . Acculturation.-Contextualizing adolescent Latina body-image development requires an understanding of mainstream body ideals and an understanding of Latinx cultural values and the process by which Latina girls navigate between them . Acculturative stress references the strain many minority groups experience as a result of inconsistencies between expectations of their own ethnic culture and that of the dominant group. Acculturation levels span the borders between mainstream body ideals and Latinx cultural values . Acculturative stress may occur when one is attempting to learn the language and social norms of the dominant culture. Those who are not acculturated may experience greater acculturative stress because as they learn the cultural patterns of the new society, they realize how different they are from their own . Latina girls who are more acculturated into Western culture may internalize the dominant thin ideal . Because this ideal is largely unattainable for girls of all ethnicities, girls who are more acculturated may feel worse about their own bodies than those who endorse a Latinx body ideal. --- Ethnic Identity as a Protective Factor Ethnic identity.-Ethnic identity refers to self-identification with a specific ethnic group; the sense of belonging and attachment to such a group; the perceptions, behaviors, and feelings one has, due to such membership; and involvement in the cultural and social practices of the group . Latinas often use their immediate social environments to form opinions, ideas, and views about themselves when forming their own identity . Ethnic identity is the awareness and knowledge of an individual's ethnic membership that may combine with shared values and attitudes of other members of one's ethnic group . Researchers are identifying an inverse association between ethnic identity and risky behavior. Thus, strengthening ethnic identity in Latina young women may be an effective strategy to promote healthy and positive behaviors. Body dissatisfaction and self-esteem strongly correlate, especially during the adolescence developmental phase . Although findings on the effect of acculturation on body image have been mixed, findings on positive ethnic-identity development do serve as a buffer for higher self-esteem . Identity development is crucial during adolescence. Ethnic identity is a significant protective factor for minority youth. Ethnic identity and eating disorders.-Latina women and girls with high ethnic identity can be protected against eating disorders, low self-esteem, and body dissatisfaction because of their ethnic-group acceptance of larger body shapes . Latinas may experience acculturative stress as a result of discrepancies between their own ethnic body ideal and the body ideal of U.S. dominant society. Furthermore, acculturative stress increases eating-disorder predictors such as low-self-esteem and body dissatisfaction. In Latina adolescents, it is essential to view social media and its association with eatingdisorder symptomatology in the context of ethnic identity. Cultural norms and body imagery favor a thin White female . A focus on ethnic identity references how beauty production and consumption reinforce the "virtual economy of beauty privilege; an exclusive set of beauty standards that have been shaped by years of global and regional racial and national privilege culture and power" . --- Implications Future studies should explore intersectional frameworks to understand unique attributes, risks, and protective factors of specific populations. An intersectional approach and language for clinicians and researchers will avoid the notion of homogeneous treatment options and Eurocentric intervention modalities that solely focuses on eating-disorder symptomatology among White women. Eating problems are not only "gendered experiences of appearance" but are a coping strategy for various traumas including "sexual abuse, racism, classism, sexism, heterosexism and poverty" . Future researchers should study the intersectionality of race/ethnicity, gender, sexual orientation, socioeconomic status/class, education level, and ability, often viewed as separate sociocultural demographic variables. Eating-disorder researchers should consider the implications of exposure to unrealistic and unattainable in social media and how they affect young girls. As such images become more prevalent, young women may become increasingly desensitized and pay less attention to the content of such messages, causing little effect . Researchers should consider such exposure in conjunction with intersectional variables to understand body image, eating-disorder pathology, effective health promotion, help-seeking behaviors, and treatment options for the Latinx community. Quantitative and qualitative researchers can seek to understand experiences of Latina girls and women that contribute to their marginalized status and their susceptibility to eating disorders based on body distortion. Using social media as a prevention method can encourage researchers to engage in content analysis, allowing researchers to visually quantify the daily images to which Latina girls are exposed while online. Qualitative tools like PhotoVoice are useful in giving Latina girls a platform and voice to discuss their experiences and interactions with body distortions and social media. Often, eating-disorderprevention programs that exist typically target reducing eating disorders without acknowledging societal views of Latinas and how visuals can negatively affect their view. Extant prevention programs build on etiologic models that propose sociocultural pressure lead to internalization of the thin ideal, resulting in body dissatisfaction, dieting, and negative affect, thereby increasing eating-disorder risk. Through an intersectional lens, clinicians need to be aware of cultural values that are important to Latinas. Latinx families promote traditional and conservative values and specific cultural values , which emphasize the centering of family. Because Latinas and their families are culturally diverse and are not a homogeneous group, clinicians should incorporate family-based and culture-specific modes of intervention when working with young Latina women who are at risk of dysfunctional eating. Examination of diverse yet unique cultural values among Latinas and their families is essential to understand the risk factors that may contribute to the development of body dissatisfaction and eating disorders . A conceptual model of eating-disorder symptomatology outcomes for Latina female adolescents. --- Author Biographies

The purpose of this review is to propose a conceptual framework using objectification theory and intersectionality theory to examine social media's influence on body image and its effect on eating disorder predictors among Latina adolescents. To examine and explore how these effects from social media usage can result in mental health disparities that affect this group, emphasis was placed on how Latina ethnic identity mediates body image. Implications for clinicians and researchers include using strengths-based and culturally specific approaches as protective factors for Latina adolescents to strengthen ethnic identity.

in turn can alter distributions and relationships between the dimensions of socioeconomic position, thereby potentially affecting the patterning of socioeconomic health disparities over time . One important source of data on socioeconomic health disparities is the national, population-based surveys, such as the US National Health and Nutrition Examination Survey . Although several measures of multiple aspects of socioeconomic position are available for repeated cross-sectional analyses of these surveys , researchers typically analyze and report results in relation to one socioeconomic measure at a time . Education is among the more commonly used socioeconomic indicators, on account of its relative ease of measurement; applicability to persons not in the active labor force ; stability over the lifespan, regardless of changes in health status; and association with numerous health outcomes . Potential changes in the distribution and meaning of economic indicators over time, however, raise concerns about comparing socioeconomic gradients over time . In the case of education, for example, the US college wage premium since the late 1950s has increased more quickly than the high school wage premium , making the relative economic value of a high school degree in 1959 much greater than one in 2012. Thus, the "meaning" of an educational credential for an individual, in terms of socioeconomic returns in the form of income and, consequently, one's lived experience as shaped by access to material resources, has changed over time, with implications for estimates of trends in the educational gradient in health. We undertook the current investigation, motivated by recent work on trends in socioeconomic disparities in body mass index , to explore the specific question of how reliance on a single socioeconomic indicator might influence assessment of socioeconomic trends over time when relationships between observed and unobserved socioeconomic indicators in repeated cross-sectional surveys are changing. We discuss this phenomenon within the framework of measurement error and identify sources of variation that can make identification of true socioeconomic change over time difficult. Using simulations, we explore the utility of the quantile, slope index of inequality , and relative distribution approaches to minimizing bias in temporal comparisons and contrast these approaches with the use of an imputation model when validation data for the unobserved socioeconomic indicator are available. --- MATERIALS AND METHODS --- Biased estimate of association at a single time point We begin by developing simple causal directed acyclic graphs to formalize the relationships between observed and unobserved variables of interest at a single time point. Causal DAGs are a simple, flexible device for demonstrating the associations implied by a given set of assumptions about the causal structure relating variables , and they can help us answer whether a particular statistical analysis will yield an unbiased estimate of a parameter of interest. The DAGs we present are not intended to encode all possible causal structures relating multiple dimensions of socioeconomic position to health, but to focus attention on the structural relationships between observed and unobserved socioeconomic indicators and the outcome. Consider the situation where we have observed data on a socioeconomic indicator X that may have a causal effect on an outcome Y . We are interested in the potential effect on estimates of the X -Y association if there is an unobserved socioeconomic indicator Z that is correlated with X and also has a direct causal effect on Y. Figure 1 shows 4 causal structures that might account for the correlation between X and Z, Figure 1. Causal direct acyclic graphs showing causal structures that result in correlation between X and Z , both of which are associated with Y . Unobserved variables are shown in outline. In A), individual's educational attainment has a causal effect on adult income ; in B), parental income has a causal effect on individual's educational attainment ; in C), parental income is a common cause of individual's educational attainment and adult income ; in D), educational attainment and individual income are associated with mortality , and observations of Y are available only for those that survived to be included in the survey ). Because Z and W and/or S are unobserved, the observed estimate of the association between X and Y is confounded by Z. The descriptive association between X and Y reflects the net effect of the direct effect of X on Y and the "backdoor" paths through Z . We cannot control for these effects statistically , so would have to be content to interpret the observed estimate as a descriptive association rather than a causal effect. The possible exception is A), where X causes Z. Here, the descriptive association between X and Y could be interpreted as the total causal effect of X on Y, reflecting both the direct effect of X on Y and the mediated effect through Z . However, if we are interested in identifying the direct effect of X on Y or the direct effect of Z on Y, we would not be able to identify these from the data available. Temporal Comparisons of Health Disparities 871 Am J Epidemiol. 2013;177:870-881 including the possible effect of an unmeasured common cause of X and Z, W , or conditioning on a common effect, S . Under any of these scenarios, absence of observed data on Z precludes statistical adjustment for backdoor paths between X and Y, making estimates of the X -Y association only interpretable as descriptive . --- Comparing descriptive associations across time Now consider that we have observed data on X and Y from 2 cross-sectional surveys at different times, and the objective is to determine whether there has been a change in the socioeconomic gradient that is not attributable to shifting associations between the observed and unobserved socioeconomic indicators. Because a change in the strength of a relationship between variables constitutes effect modification by time, we represent this by drawing separate DAGs for each time and indicate qualitative change in effect magnitude by drawing a thicker arrow in the second DAG. Given the potential causal structures leading to correlation between X and Z in Figure 1, Figure 2 shows possible changes in the strength of relationships between variables that can lead to a change in the descriptive association of X and Y over time. We consider scenarios A and B to reflect a change in socioeconomic gradient that meets our criterion, albeit noting that scenario C arguably still reflects a change in the socioeconomic gradient captured by the total effect of X on Y . Even with no other unmeasured confounders, it remains impossible to distinguish changes in the direct effects of X on Y or Z on Y in the absence of observed data on Z. Our simulations explore how the structures depicted in Figure 2 and modeling techniques frequently used by epidemiologists can influence temporal comparisons of the association of a single observed indicator X with Y. --- Simulation framework specified as an errors-in-variables model To develop our simulations, we put a parametric form on the qualitative relationships depicted in Figures 1 and2 and associate values with the parameters governing these relationships . For continuous X, Y, and Z, we specify a series of linear regression models. Here, it is convenient to induce correlation between X and Z by specifying a data-generating model for X given Z: This has a connection to the classical errors-in-variables model when there is no direct effect of X on Y, noting that equivalent data structures can be generated by assuming a model for Z given X, or generating X and Z as bivariate normal given a vector of means and a variance-covariance matrix . We generate Z at time t as a normally distributed continuous variable with mean μ z and variance s 2 z . At a given time, the ith subject's outcome y i depends on the variables z i via the linear regression y i ¼ b 0 þ b 1 z i þ 1 i ,ð1Þ where ɛ i is a normally distributed error term with mean zero and variance s 2 1 . The observed x i at a given time depends on z i via x i ¼ a 0 þ a 1 z i þ h i ,ð2Þ where η i is a normally distributed error term with mean zero and variance s 2 h . We assume that z i , ɛ i, and η i are independent. At a given time, if the y i 's are regressed on the observed x i 's, then the estimator for the slope coefficient from this regression as n → ∞ is the following: b à ¼ b ½a 1 þ ðs 2 h =a 1 s 2 z Þ ,ð3Þ where β is the true effect of Z on Y, and β* is the observed association between the observed X and Y . Details of this derivation are presented in Web Appendix 1. As this expression clarifies, changes in any 1 of the 4 parameters will influence the association between X and Y over time. As shown in Web Appendix 2, we considered 6 different scenarios in our simulations, variously involving changes in the distribution and relationships of the variables Y, Z, and X. Values for the various parameters are summarized in Table 1. We calibrated the parameters to yield distributions of Y, Z, and X roughly similar to those observed in NHANES I and NHANES 2005-2008 for body mass index, log, and education . For each of scenarios 1-6, we simulated 1,000 data sets of 1,000 subjects at each of the 2 time points. --- Modeling methods To analyze the simulated data, we employed 6 different modeling approaches , also explicated in Web Appendix 2: These approaches include different regression models using Z and X, an imputation model, and the use of deciles, SII, and the relative distribution method, as we explain below. For each scenario and modeling approach, we estimated socioeconomic gradients in the outcome, calculated the difference in the observed estimates between time periods, d ¼ b à t¼2 À b à t¼1 , and tested this against the null hypothesis of d ¼ 0. --- Relative index of inequality, slope index of inequality, and quantile rescaling The relative index of inequality is a widely used measure for assessing trends in health disparities that is intended to address the changing composition of socioeconomic categories over time . First employed in health research by Pamuk in 1985 to analyze trends in occupational class gradients in mortality in the United Kingdom ), the RII was devised as an index "based on the mortality levels of all five social classes, and their proportionate share of the population" . It provides an estimate of the relative disparity in fitted health effects comparing the best with worst off in the socioeconomic distribution. Recent work also provides estimators for the RII that relax the assumption of linearity . Figure 4 depicts how the RII ) are methods of rescaling X as measured to the population distribution of X , analogous to taking quantiles of a continuous variable. In the context of this simulation and the additive model for Y given Z, it is the SII that we expect to be robust to changes in the scale of X, not the RII . We accordingly focus on the SII as method VI. Although the RII has been discussed primarily as a measure of "total population impact" , we note that it involves a deliberate data transformation that rescales X at each time onto the population distribution scale. The RII accordingly implies that the health effects at levels of X at the 2 different times are not comparable, and that rescaling onto the population distribution scale is required to allow comparisons that are unaffected by temporal differences in X unrelated to the true socioeconomic gradient in Y. This is equivalent to positing a latent version of X* on the population distribution scale that gives rise to observed X and assuming that only the measurement relationship between X and X* changes over time, while the "true" relationship between X* and Y is unchanged. 1. Z is an unobserved socioeconomic indicator, X is an observed socioeconomic indicator, and Y is a continuous outcome. We assume that Z is normally distributed with mean μ z and variance s 2 z . β 0 and β 1 control the relationship of Z and Y, and ɛ i is a random error term where 1 i Normalð0; s 2 1 Þ. X does not have a direct effect on Y. α 0 and α 1 control the relationship of Z and X, and η i is a random error term where h i Normalð0; s 2 h Þ. Note that parametric forms have been specified for the distribution of Z and the relationships between Z and X and Z and Y. The direction of the arrow from Z to X in this data-generating model reflects that we have specified the correlation between X and Z via an expression for f, but as we note in Figure 1 and Web Appendix 1, equivalent correlation structures can be generated under several different causal scenarios. We have represented the contribution of random error to X and Y by adding these nodes to the graph. The parametric specifications lead to the quantitative expression for bias presented in Web Appendix 1, which is not immediately apparent from the directed acyclic graphs in Figures 1 and2 because the DAG is a purely qualitative tool. Temporal Comparisons of Health Disparities 873 Am J Epidemiol. 2013;177:870-881 In spite of the widespread use of the RII/SII for comparing socioeconomic gradients over time, we are aware of no systematic investigation of how effectively the RII/SII protects against the multiple sources of bias identified in equation 3. --- Relative distribution method New work in economics and sociology on trends in income inequality has also proposed a less extreme method of rescaling, termed the relative distribution approach . Here, the distribution at one time is characterized in relation to quantile cutpoints defined by a reference distribution , thereby preserving marked differences in the shape of the second distribution relative to the first . To our knowledge, the relative distribution approach has not yet been used to study trends in socioeconomic inequalities in health and may be worth considering. --- Imputation model If we knew in advance that Z was an important socioeconomic predictor of Y, we would ideally measure it and include it in the analysis at each time point. In many settings, however, Z is not available, or if it is available, it may be subject to missingness. If data are available from an external validation source or from a subsample of the survey, an imputation model can be specified for Z given X, thereby permitting comparison of the association between Z and Y across time. We implemented a single imputation version of this as method III in our simulation by assuming that data on Z and X were available for a random sample of 30% of the population, and applied the imputation model to the rest of the population. --- RESULTS --- Simulation results We summarize the contrast in effect estimates over time under the different scenarios and models in Figures 6 and7. The boxplots show the distribution of the point estimates for the temporal contrast over 1,000 simulations, with method I showing the "true" temporal contrast, that is, the contrast that would be obtained if we observed and analyzed Z at both time points. In Table 2, we also report the mean temporal contrast over 1,000 simulations, the 95% simulation intervals, and the proportion of simulations in which the various methods detected a statistically significant temporal contrast assuming a type I error rate of 5%. In scenarios 1, 2, 3, and 4, this reflects the probability of detecting a significant temporal effect given no true change in the socioeconomic gradient, while for scenarios 5 and 6, this is the power to detect a significant change in the socioeconomic gradient given that there was a true change. In scenarios 1-4, the least biased methods are those whose median value is at or close to zero. In scenario 1 and scatterplots and fitted regression lines with Y . Distributions and fitted regression slopes at time 1 and time 2 are shown by solid and dotted lines, respectively. In A), distributions of Z at time 1 and time 2 are essentially the same, and in B), the true relationship between Z and Y at both time points is the same. In C), the change in α 0 and α 1 results in a shift in the observed distribution of X from time 1 to time 2, and in D), the observed estimate of the descriptive association of X and Y changes. In E), quantile rescaling of X restores the comparability of quantiles of X , and in F), the socioeconomic gradient is shown correctly not to change. When X is divided into categories , the proportion of the population in each category varies over time , but in H), the SII method of rescaling to the cumulative distribution of X correctly shows that the socioeconomic gradient did not change. t = 1 t = 2 t = 1 t = 2 t = 1 t = 2 t = 1 t = 2 t = 1 t = 2 t = 1 t = 2 Temporal Comparisons of Health Disparities 875 Am J Epidemiol. 2013;177:870-881 and no change in the underlying exposure-disease relationship), all methods yielded, on average, the correct characterization of no change in the socioeconomic relationship, with the exception of the relative distribution method , which detected on average a decrease in the socioeconomic gradient of and would have erroneously concluded this was significant 22% of the time. In scenario 2 , method II yielded, as expected, a biased estimate of the temporal comparison. In this case, as the strength of the relationship between Z and X decreased , this resulted in a decrease in the relationship between X and Y of 0.14, when there was no real change to the underlying relationship between Z and Y. The type I error rate was 87%. Both the quantile and SII approaches were robust to this source of bias over 1,000 simulations: Methods V and VI had type I error rates of only 5% and 4%, respectively. Method IV performed quite poorly, with a mean temporal difference of -1.83 and a type I error rate of 92%. In scenario 3 , method II also led to biased estimates of the temporal effect, with a type I error rate of 29%. Methods IV, V, and VI also produced simulation means substantially greater than zero, with type I error rates of 19%, 12%, and 10%, respectively. In scenario 4 , the bias using method II was relatively small: The simulation mean was -0.03 and the type I error rate was 9%. However, methods IV, V, and VI all perform poorly: Simulation means were substantially less than zero, and type I error rates approached 100%. In scenario 5 , all methods detected the temporal change in the socioeconomic relationship with the outcome, suggesting that all methods are sensitive, if not specific. Power to detect a significant temporal change was somewhat less with methods IV, V, and VI. In scenario 6 , methods I, II, and III did detect a significant change in the socioeconomic gradient over time. The combined effect of the multiple varying parameters on the estimated temporal change in the association of Y and X was to attenuate it relative to what would be estimated if we could regress Y on Z. However, methods IV, V, and VI all performed poorly, with simulation means going in the opposite direction of the true change in socioeconomic gradient and virtually nonexistent power for the decile and SII approaches. Over all the simulations, method III consistently yielded estimates of the temporal effect that were virtually identical to those obtained by the "true" regression of Y on Z. The slightly increased variability over 1,000 simulations shows the additional uncertainty induced by using a 30% subsample as validation data and applying the imputation model to the rest of the data set. --- DISCUSSION We have shown that temporal comparisons of socioeconomic gradients in health using a single socioeconomic indicator can be affected by an unobserved socioeconomic variable if the correlation between the observed and unobserved variables is changing. When the variables are continuous, 3 parameters potentially contribute to bias: 1) a changing systematic relationship between the observed and unobserved socioeconomic variables; 2) changing error variance ; and 3) changing distribution of the unobserved variable. Suggesting that each of these scenarios is plausible is the literature we have cited on changing Figure 5. Illustration of relative distribution method of rescaling under scenario 2 showing distributions of rescaled X at 2 timepoints and scatterplots and fitted regression lines with Y . Distributions and fitted regression slopes at time 1 and time 2 are shown by solid and dotted lines, respectively. In A), the relative distribution method rescales X at time 2 relative to quantiles of X at time 1 but is less extreme a rescaling than the quantile or SII approaches. Differences in the location and shape of the distribution of X at time 2 relative to quantiles of X at time 1 are preserved. Consequently, in B), the descriptive association between rescaled X and Y is observed to change over time, even though the true association with Z did not change. 876 Chen et al. Am J Epidemiol. 2013;177:870-881 occupational, education, and income distributions over the last 50 years . Through simulations, we confirmed that changes to each of these 3 parameters can lead to changes in the estimated association between an observed socioeconomic indicator and the outcome, even when there has been no change in the direct effects of the observed or unobserved socioeconomic indicators on the outcome. Notably, although there has been some awareness that the changing "meaning" of an observed socioeconomic indicator as reflected by its changing population distribution can lead to bias in estimating trends in socioeconomic inequalities in health , we also showed that even a simple change in the variance of the unobserved variable itself can lead to bias in temporal comparisons-a little appreciated possibility in the literature on socioeconomic disparities over time. Recent economic literature on growing income inequality suggests that changes in this latter quantity are quite plausible . Similarly, increases in s 2 h are possible if, for example, the incomes of those with high school and college degrees become more heterogeneous within those categories. We explored 3 methods of rescaling the observed socioeconomic variable to permit temporal comparisons that are robust to changes in the aforementioned parameters. One of these-the SII -is commonly regarded as a descriptive measure of "total population impact" but was explicitly developed to permit temporal comparisons when the composition of socioeconomic categories changes over time by rescaling categories of X onto the population distribution scale . Indeed, we found that the SII does effectively reduce bias due to changes in the scale of the observed relative to the unobserved variable, but it does not deal with the problem of changing error variance or changing variance of the unobserved socioeconomic distribution . Similarly, the decile approach for continuous variables was able to guard against only bias due to changing α 1 . Additionally, the relative distribution method, thus far used only as an outcome variable in the sociology and economics literature , performed poorly as a method for rescaling the socioeconomic indicator in all scenarios and should not be used as a predictor in health analyses. Although all the methods we considered were able to detect a change in the socioeconomic gradient when only β 1 changed, we found that when multiple parameters controlling the measurement model were also changing , the decile and SII approaches were unable to detect the true change. From the pattern of bias effects seen in scenario 4, it appears that the effect of increasing s 2 z effectively canceled out the true change in β 1 in simulation 6. In contrast, we found that the use of validation data to develop an imputation model recovered the true effect under all scenarios. This requires access to such data and that the unobserved variable itself be measurable with accuracy. For example, one source of external validation data on socioeconomic variables and their relationships over time in the United States is the Current Population Survey . It includes individual-level annual data on income, education, and occupation that could be integrated into analyses of health data by using Bayesian hierarchical models . Such models are more sophisticated than the simple single imputation scheme we used in our simulation and would allow uncertainty from the estimation of the imputation model to be properly reflected in the final association estimates. Because our simulations were motivated by analyses of socioeconomic gradients in body mass index, we note that among the extant studies on trends in US socioeconomic gradients in body mass index-which all reported results separately by education or income -none discussed how changes in the distribution of and relationships between these variables may have affected comparisons of the temporal trend. Nor has the literature recommending use of education as a socioeconomic indicator in health analyses addressed how analyses might be affected by the changing value of educational credentials over time . In summary, our results have 2 key implications. First, analyses of trends in socioeconomic inequalities in health should use multiple socioeconomic indicators and compare how the observed socioeconomic gradients change over time by each measure. Second, more broadly, it is essential to consider how changing relationships between unobserved and observed variables, as well as changes in the underlying distribution of the unobserved variables themselves, may affect inferences that causal relationships have changed. This holds true not only for the simple situation we simulated but also for more complex situations ). In particular, caution is needed when analyzing quantiles or using the SII/RII if there is reason to believe that the error variance of the observed variable or the variance of the unobserved variable has also changed. When possible, attempts should be made to collect data to permit estimation of an imputation model for unobserved variables; barring that, sensitivity analyses could be used to assess the potential sources and magnitude of bias . The larger lesson is that conceptual clarity about the substantive and dynamic meaning of the observed variables is thus of paramount importance, relevant to interpretation of not only trends over time but also associations observed at any one point in time. --- Conflict of interest: none declared.

Although socioeconomic position is conceptualized by social epidemiologists as a multidimensional construct, most research on socioeconomic disparities in health uses a limited set of observable indicators (e.g., educational attainment, household income, or occupational class) and typically analyzes and reports gradients in relation to one measure at a time. Societal changes in economic structures over time, however, can lead to changes in distributions of and associations between socioeconomic indicators, as has occurred with income returns to education in the United States over the last 50 years. Consequently, temporal comparisons of socioeconomic disparities from repeated cross-sectional surveys can be affected, particularly when salient dimensions of socioeconomic position are unobserved. We discuss this phenomenon within the framework of measurement error and identify sources of variation that can make identification of socioeconomic change difficult. Using simulations, we explore the utility of the quantile, slope index of inequality, and relative distribution approaches to minimizing bias in temporal comparisons and find that these methods yield correct inferences about temporal change only under limited conditions. We contrast these approaches with the use of an imputation model when validation data for the unobserved socioeconomic indicator exist. We discuss implications for analyzing changing socioeconomic health disparities over time.

Background The first 28 days of life are the most critical for a child's survival. In 2020, 2.4 million newborns died, which accounts for 47% of all under-five deaths [1]. A majority of these deaths occurred in Sub-Saharan Africa where neonatal mortality is 27 deaths per 1000 live births [1]. Preterm birth is one of the dominant risks in neonatal mortality [2], and it accounted for 35% of Page 2 of 11 Mhango and Nyondo-Mipando BMC Pediatrics 23:432 neonatal deaths worldwide in 2019 [3]. Malawi has one of the highest preterm births and neonatal mortality. Preterm birth is estimated to range from 7.9 to 29.7% [4], and neonatal mortality rate is estimated at 27 deaths per 1,000 live births [5]. Neonatal mortality rate is higher in Malawi when compared against neighboring countries, with Zambia reporting 24 deaths, and Tanzania had 20 deaths in 2021 [6]. The Malawi government developed guidelines and protocols for preterm newborn care in order to reduce neonatal mortality rates. However, Malawi still lacks a deliberate policy on male involvement in the care of preterm newborns, despite the Ministry of Health recognizing the importance of involving fathers. Fathers' involvement in the care of preterm newborns is critical to influencing the newborns' behavioural and psychological outcomes, [7] as well as overall health [8,9]. These positive outcomes are because fathers play a vital role in making decisions about the health of the children in the home [8]. Fathers who are involved in care activities become good parents [9], finding fatherhood more fulfilling and considering themselves naturally important to their children [10]. Fathers that are involved in care are more likely to have a positive feeling about their interactions with the children [11,12], better understand, and appreciate their children and develop a rich bond with them. Therefore, fathers' engagement in care practices needs to be initiated from the first days of the child's life, even in situations when the mother is recovering and able to provide care [13]. Male involvement in the care of preterm newborns can have several benefits, that could help reduce the high mortality. When fathers share caregiving responsibilities, it can lessen the burden on mothers and prevent the mothers from feeling overwhelmed [14,15]. Moreover, fathers who participate in the care of their preterm newborns are more likely to be well-informed about their baby's health, adhere to medical advice, and actively contribute to decision-making about their child's care. These factors can ultimately lead to better health outcomes for the baby. Lamb et al. argue that male involvement has three components, which include availability , engagement, and responsibility [10]. Accessibility is where a father is physically present but only plays a passive or supervisory role to ensure the child is safe. Engagement represents the more intensive, one-on-one interaction with the child, while responsibility is the ability of the father to know what is needed so that they can plan and arrange for the provision of certain aspects of child care [10]. Just like the mothers, the birth of a preterm baby in a family is equally stressful for the fathers. Fathers need targeted nursing interventions that are different from those applied to mothers [16]. Factors influencing their involvement in care need to be explored because fathering a preterm newborn is multifaceted and predisposed by personal differences, and traditional, and contextual factors that would help develop interventions that respond to fathers' needs. In Malawi, there is no policy on male involvement in the care of preterm newborns despite the numerous benefits that come with fathers' involvement. Furthermore, there is no literature on male involvement in preterm infant care despite Malawi having a high incidence of preterm births and high neonatal mortality. As such, exploring the factors that influence fathers' involvement in the care of their hospitalized preterm newborns would be key for the health care providers to professionally involve fathers in the daily care activities and build a strong foundation for continued involvement even after discharge from the hospital. There are both barriers and facilitators to fathers' involvement in the care of preterm newborns. Barriers can result in few men reported to be actively involved. Some of the barriers are social-cultural beliefs that influence gender norms [17,18], the baby's physical nature [7,13,[19][20][21], the hospital environment [13,19,22], work and other family responsibilities [19][20][21]23], and exclusion by health care staff [24]. On the other hand, factors that facilitate fathers' involvement in care are good communication from health care providers [7,22], social support [19,24,25], encouragement from male champions in the community [18,26] and their spouses [19,24,27], feedback from the infant itself when held by the father [24,28], and multiple births [19]. The study utilized the constructs of the Theory of Planned Behaviour to assess fathers' involvement based on the components of male involvement as proposed by Lamb et al. [10] . The Theory of Planned behaviour explains what encourages a person to exhibit a specific kind of behaviour [29], in light of the several social factors outside one's control that may encourage or impede that particular behaviour. The theory proposes that behaviour is not only born out of an intention but also out of one's aptitude to exhibit that behaviour [30]. According to the theory, attitudes, subjective norms, and perceived behavioural control are factors that influence the intention to perform a behaviour [31]. This means an individual is likely to have an intention to perform certain behaviours such as caring for their preterm newborns if they have a positive attitude, believe that social pressure allows the performance of that behaviour and that they believe they have what it takes to perform those behaviours correctly [29,32]. If one has these three, they are likely to have a stronger intention and more likely to perform the behaviour. However, several external factors may encourage or discourage an individual performance of that particular behaviour, despite the intention being there [32]. The aim of the study was to explore factors influencing fathers' involvement in the care of hospitalized preterm newborns. --- Methodology --- Study design A descriptive qualitative research design was used in this study. In this qualitative research, the researcher seeks to understand the participants' lived experiences in a manner that would help conceptualize the experiences and increase the researcher's knowledge and understanding of human experiences [33]. --- Study setting The study was conducted in Balaka district, Malawi. Balaka is a rural district in the southern region bordering Mangochi, Neno, Machinga, and Ntcheu districts. It has a population of 438, 379 [34]. As of 2017, the district had an estimated 91,176 women of childbearing age with approximately 18,592 annual expected pregnancies and deliveries [35]. The study was conducted at Balaka District Hospital in June 2021 where all preterm newborns in the district are referred. The district was selected due to its high infant mortality rate estimated at 51 per 1,000 live births, which is above the national average of 42 per 1,000 live births as stated in the 2017 Demographic and Health Survey [36]. --- Sample size The sample consisted of sixteen fathers who had preterm neonates hospitalized at Balaka district hospital. The number resonates with methodological literature on saturation of qualitative data [37,38], which argues that saturation could be reached with between 9 and 24 interviews, especially in research like this where the sample is relatively homogeneous and the research is aims-focused [38]. Data were collected until saturation was reached. --- Sampling The participants were chosen by convenient and purposeful sampling to get rich data. The maximum variation sampling technique was used to gain greater insights from the participants. Maximum variation was achieved by recruiting participants of different ages, several children, religions, levels of education, occupation, and ethnic background. Eligible fathers were identified by a member of the clinical staff and referred to the research assistant who obtained permission from them to take part in the study. --- Data collection Data were collected in June 2021 using in-depth interviews conducted by an experienced and trained research assistant, who has a Public Health background. The research assistant training covered objectives of the study, data collection procedures, qualitative data collection, ethics, and communication skills. The training was conducted to ensure that research assistant understands the importance of ethical considerations and is able to conduct research in a responsible and ethical manner. Interviews were conducted in a private room at the hospital and the privacy allowed participants to get engaged in a comfortable conversation making it easy for them to share experiences as compared to them having to fill out a survey questionnaire [39]. The interviews were conducted in the language of participants' choice using a semi-structured guide and were audio-recorded. Each took a maximum of 40 min. The main questions of the interview were: 1. What activities do you do with your child when you come to visit? 2. In those activities, you are yet to do; do you intend to try to do them? 3. How do you feel about the activities you are involved in as you provide care to the baby? 4. How capable do you think you are in providing the care needed for your preterm baby? 5. How do people that are considered important to you react, as you get involved in the provision of care to your preterm baby? 6. What factors enable you to be involved in the care of your preterm newborn? 7. What factors make it difficult or impossible for you to be involved in the care of your preterm newborn? --- Trustworthiness of the study To ensure the trustworthiness of the study, the researcher put into serious consideration the validity, dependability, and conformability of the results. To ensure validity, the researcher employed procedures that ensured the credibility and transferability of the data. To ensure credibility, the researcher ensured the data collection tools were checked and vetted by an expert in health-related qualitative research. The tools were also pilot-tested to see if they were gathering the data they were supposed to. During the interviews, rapport was built from the beginning to ensure participants were fully engaged and probes were used to ensure the respondent adequately addresses the questions. Adequate time was allocated to analysis to ensure interpretations are in line with the data from the participants. Transferability was accomplished by ensuring participants had varied demographic characteristics so that there was wide representation and provides a detailed description of context, data collection, and analysis approaches used to allow replication in a different place and different populations [40]. The dependability of the research process was ascertained by the clear description of procedures and processes involved during the research. These included objective recruitment of study participants and a proper description of data analysis methods employed to allow others who may wish to audit and replicate them to do so [41]. Conformability was achieved by ensuring results are objective and neutral, independent of the researcher's views [40]. Conformability enabled the findings to be consistent with those from similar related studies. Personal beliefs and opinions did not affect the research findings. --- Data analysis Data were managed using NVivo version 12. Data collection and analysis were done concurrently. Audiorecorded interviews were transcribed verbatim in Microsoft Word immediately after completion of the interviews. Rigorous review and quality control were performed by the authors to ensure accuracy and consistency between the transcripts and audio recordings. The analysis was conducted by the authors using qualitative thematic analysis approach [42]. The authors familiarized themselves with the data by reading and re-reading the transcripts while taking notes of initial codes as they read the transcripts. The coding scheme combining a priori codes and data-driven codes was developed based on key concepts from the Theory of Planned Behaviour and Lamb's model of fathers' involvement, and repeated reading of the transcripts. This approach allowed for the incorporation of existing theoretical constructs while also allowing new themes to emerge. Codes were then grouped into categories which combined similar codes and these later were organized under overarching themes based on recurring patterns, concepts, or ideas in the data and data reduction techniques were applied to identify the most illustrative examples within each themes while preserving the core meanings and experiences expressed by the participants. The authors then examined the relationships between themes, identify connections, and explore the implications of the findings to develop a coherent and comprehensive description. They also paid attention to avoid overwrapping themes and those with very minimal data to stand alone as a theme. These were combined with other themes to better illustrate the findings. The data analysis process involved iterative revisits to previous steps to achieve a better representation of the findings and encompassed multiple discussions of the results between the authors. --- Ethics The study was approved by the Malawi College of Medicine Research Ethics Committee . To protect study participants' privacy and confidentiality, a unique personal identification number was assigned to each individual in the study. This number was used instead of their names on the data collection forms and within the study databases. The only identifiable information about the participants was contained in the study's Informed Consent Forms, which were securely stored in locked filing cabinets at the Kamuzu University of Health Sciences. Access to these cabinets is restricted solely to the researchers involved in the study. All study interviews were performed in private rooms to protect participant confidentiality. Permission was also granted by the Director of Health and Social Services for Balaka district. --- Results --- Participant demographics The study had 16 participants from different demographics as per Table 1 below. Participants belonging to Lomwe and Ngoni tribes were the majority with 44% and 38% respectively. In terms of age, 38% were those aged 25-29 and 31% were those 35 years and above. The mean and median age were 29.8 years and 29 years respectively with the standard deviation 5.57 years. The majority of the participants were first-time fathers and three quarters were Christians. Slightly more than half of the participants were educated to the secondary school level and majority were not employed. --- Factors influencing fathers' involvement There were five main categories of factors that influence fathers' involvement in the care of preterm newborns. These are personal factors, interpersonal factors, infant factors, environmental factors, and economic factors as in Table 2 below. --- Fathers' personal motivational factors Fathers reported that their perception of the difficulty of activities they are involved in and the perceived benefit of being engaged in those activities influenced their involvement in care. --- Perceived difficulty of activities The fathers described that their perception of the difficulty of the tasks could not stop them from being involved in the care of the newborns and nothing was difficult for them to do. The fathers reported that despite some care activities such as bathing the infant and skinto-skin care are perceived as difficult when provided to a preterm newborn, it did not stop them from taking part in those activities. "When I was told by the nurse to put the baby in a kangaroo position, I did not find it difficult to do. It is like allowing the baby to sleep on my chest and that was not a problem. " . --- Perceived benefits of involvement Involvement in the care of the preterm infants was enabled by the fathers' belief that it was important for them to be involved because it provided them with an opportunity to show their love to the babies and enhanced the bonding. --- "The child gets to know his father because the closeness to the child gives her the familiarity of what his father is. Even if she grows and starts recognizing things she will know that this one is my father and be able to differentiate who her father is from the other men who might wish to carry her. " . Furthermore, it also helped the mothers realise that the fathers do care for the babies despite them being born preterm "It is important because my wife will also be happy that the kids are loved by their father and it makes me very happy because this is something treasured. " . --- Interpersonal factors The fathers perceived that several interpersonal factors affected their level of involvement in the care of their preterm newborns. These factors are gender roles and cultural beliefs, work and family responsibilities, provider attitude, social support, and sharing experiences with fathers who have had preterm newborns. --- Gender roles and cultural beliefs Unfriendly gender roles and cultural beliefs prevented the fathers from being involved in the care of the babies. The fathers considered some roles as belonging to women and they could only do them when the mother is sick or busy with other tasks. Fathers' inability to be involved in the roles is based on the belief that infant care was traditionally a female role while the fathers have to provide for the family's needs "That is part of the mother's care. For us men, it is to search and provide for them. " . "When the mother is busy with something, I can do like changing his diaper, but in case she is not busy, Aaah [laughing] some things are for ladies. " . --- Work and family responsibilities Various responsibilities from work and family acted as barriers to involvement in the care of preterm newborns since fathers spent their time at work and attending to other family responsibilities such as caring for the children left at home at the time the mother was in the hospital with the newborn. These multiple responsibilities reduced the time fathers could spend at the hospital and caring for the newborns. --- Provider's attitude The fathers acknowledged that the support they get from the health care providers has been a facilitator to their involvement in the care. The fathers reported that providers had been so supportive and communicative about the condition of the baby and how best the parents can provide the needed care. --- "The nurses are very good because they are always telling us what we need to do in the care of the baby and everything they are telling us here seems to be working. " . On the other hand, the behavior of other providers prevented the fathers from being involved. The fathers reported that some health providers were rude in communicating with them while other providers preferred communicating with the mothers than the fathers which impeded fathers' involvement in care. --- "This other nurse is always shouting at us and you sometimes get afraid that she will shout at you when you touch the baby. " . --- Social support The social support the fathers receive from their friends and relatives was also reported as one factor that enables them to get more involved in the provision of care. The fathers got encouragement from their social networks telling them that they need to focus on providing the care the baby needs because they were not the first to have preterm newborns. --- "They are helping me a lot because they even give me encouraging examples of people who are still alive and strong who were born preterm. That encourages me much that I have to provide good care to this child. " . Some of the family and friends do come to visit them at the hospital or call them to check how they are doing with the baby. --- "They do call me to hear how they are faring as well as encourage me to be checking on the wife and the child regularly. " . Talking to other men who have gone through the same situation was also considered a facilitator to involvement. It was reported that these had firsthand experience on how to handle preterm newborns and how their involvement helped the babies grow. Talking to such men acted as motivation to get more involved in the provision of care. --- "These people have passed through this already so their words are of great importance and we follow since we know that it is coming from someone who has experienced what he is saying. " . Furthermore, one participant reported that he is involved in the care so that he could be a role model to other men, that the mother needs to be supported in the care of the babies. --- "What makes me happy in doing this is that I want other fathers to take it as a lesson that we need to be always available to our children because caring for the children is for both father and mother. " . --- Infant factors Fathers' involvement in care was also influenced by infant factors. These range from the babies' physical appearance and nature, the health status of the infant, feedback from the baby, and multiple births. --- Baby's physical appearance or nature The physical nature of the baby prevented fathers from getting more involved in the care. The fathers were afraid because the baby was fragile-looking and too small and thought they will harm the baby when they try to get involved in the activities. --- "I might not know how to touch or hold him since there might be ways how to touch such a child when bathing him. The child is too small so it might be difficult to ably do it. " . --- Baby's health status The health status of the baby was another infant factor that influences fathers' involvement in care. Fathers found it difficult to get involved when the baby was sick or not in good health. However, the positive changes they noticed acted as an encouragement for them to get more involved. Others had seen their babies transferred from the isolette to an open crib, and were able to finally hold them. --- "The improvement encourages me because it shows a difference from the time the child was born to now --- so when I look at those positive changes it encourages me that my child will be like me one day a fully grown person so I have to do more for him. " . Furthermore, babies' growth helped remove the fears that they might lose the baby. Dealing with these fears helped them accept the situation and motivated them to get more involved in the care. "It is encouraging to me that the child is growing fast and is healthy which motivates me to do more. " . --- Feedback from the baby The fathers described that the feedback they get from the babies while providing care acted as a facilitator to their involvement. They reported that the positive feedback given by the babies such as stretching and smiling made the fathers feel acknowledged and made them want to do more with and for the babies. --- "The first day I carried the child, she was not comfortable or open up as how a child is supposed to do when the child is born. However, for the past three days when I carry the child, I see that the child tends to stretch its arms, smile, and do other things. So, this tends to make me happy and encouraged. " . --- Multiple births All the fathers who had twins acknowledged that having two babies was an enabler for them to get involved because twin birth made it difficult for the mother to handle both babies alone. Inability to handle both babies at the same time allowed fathers to come in and help. For example, when the mother is feeding or holding one baby, the fathers attend to the other baby. --- "When they were giving milk to one I had to carry the other one who was also waiting to be given after the other one. So I just carried one and after giving the other one milk I then carried the other one. " . --- Environmental factors The physical environment in the hospital is another factor that affected involvement. Fathers reported that they were unable to take on some tasks because the ward is just an open space where they could not privately take on some roles such as skin-to-skin care. After all, they were in sight both men and women in the ward. --- "The problem is that you find that in the ward the baby is in, there are several women and each can see what every other person is doing. As such, it becomes difficult to help with other things such as kangaroo care. " . Furthermore, in situations where the newborn was in the isolette, the fathers were unable to interact with the baby because there was limitation on who is allowed to touch the baby. Care guidelines allow only the mother and health care provider to touch the baby in the isolette. "She is in the incubator so I cannot even touch her. " . --- Economic factors Fathers acknowledged that their inability to raise enough funds for the provision of care affects their involvement in the care. They reported that all that the baby needs at the moment requires money which most of them do not have. As such, they spent much of their time away from the hospital searching for resources that would enable them to provide for the baby's needs, which ultimately reduced the time the fathers would be at the facility providing care to the baby. --- "The most challenging thing is money because I may not manage since the work I do is not that so much paying considering it is in the village and what I get there may not be enough to provide for the baby's needs. " . In addition, the advent of COVID-19 has worsened the situation on the fathers' ability to generate income and provide the basic needs for the babies. --- "Due to the COVID-19 pandemic, a lot of things have changed that even the way income is being generated these days is a bit a challenge so this might be the thing that may hinder the work of the provision of care that involves the provision of all the necessity a child needs and the availability in the life of this child. " . On the other hand, fathers that were formally employed and had a steady source of income reported that being financially stable helps them spend more time at the facility with their babies and gives them more opportunities to get involved in the provision of care. --- "Having enough money to provide for my child would help me be at home most of the time and spend more time with the mother and the child. " . --- Discussion The findings from the current study suggest that several factors act as barriers as well as facilitators to fathers' involvement in the care of hospitalized preterm newborns. Guided by the Theory of Planned Behaviour and Lamb's model of fathers' involvement, it is evident that fathers' personal motivational factors, interpersonal factors, infant factors, environmental factors, and economic factors affect fathers' involvement in care. The findings suggest that fathers did not find tasks such as skin-toskin care difficult and believed their involvement was an opportunity to express love to the newborn and develop a bond with the child. These findings are consistent with other studies that reported that fathers considered child care provision as responsibility for both the father and the mother, and that fathers' involvement enhances the bonding between the father and the baby [43]. This suggests that more involvement in caregiving would be a motivating factor for fathers to frequently visit the baby at the hospital and create more chances for them to get involved in care as it would help them feel more responsible for their newborns. Interpersonal factors such as unfriendly gender roles and cultural beliefs made fathers consider some roles as feminine hence not taking part in care. Earlier studies [44][45][46] found that childcare is culturally considered a feminine role. This necessitates a shift of mentality to help fathers realise that they can also take part in caregiving activities. Work and other responsibilities have been reported in the current study as impeding fathers' involvement in the care of hospitalized preterm newborns. This is consistent with findings from other studies [19,[47][48][49][50], which reported that multiple responsibilities prevented the fathers from being engaged in caregiving activities. It is worth noting that at the time of the study, the Malawi labour laws were not allowing for paternal leave for those formally employed, and also the majority of the participants were not in formal employment which meant they still needed to work to earn something for the families. Fathers felt the communication and support from the health providers helped them get better engaged in care. These findings are similar to other studies which acknowledged that inclusive interaction about childcare with the providers helped the fathers feel recognized and get more involved in the care of their preterm newborns [13,43,51]. These findings, however, differ from other studies [7,52] where fathers felt sidelined by the health providers in the provision of care to the newborns which mean providers' attributes and approach to work may affect fathers' involvement in care. Additionally, the findings show that getting help from different individuals, sharing experiences, or talking to fellow fathers who are having or have ever had preterm babies helped them cope with the situation and encouraged them to get more involved in care activities. The findings are consistent with those from other studies conducted in Malawi where parents received social support from family and community members [44,53]. The experiences shared with fellow fathers of preterm newborns helped fathers deal with feelings of being isolated [54], and helped them realise that they are not alone [55]. These findings imply that talking to fellow fathers of preterm newborns or being able to interact with those that have ever gone through a similar situation helps the fathers get to learn from each other on how best to handle the situation; hence the need to provide safe spaces where they could be having such interaction. A case could also be made from this on how important male champions could be in reproductive health. Infant factors such as physical appearance, health status, and feedback from the baby influence fathers' involvement in care. The findings from the current study suggest that physical appearance and fragilelooking skin impeded fathers' involvement since they feared they could harm the baby. These findings are similar to another study conducted in Malawi, which reported that caregivers including fathers, were afraid of holding the preterm newborns because they looked small and fragile, and the caregivers feared harming them [44], which is consistent with findings from studies conducted in other countries [51,56]. This fear made fathers unwilling to touch and hold the babies [7,19,20,57]. On the other hand, fathers felt encouraged to get more involved when they heard and saw that the babies' health status was improving. This was further enhanced by the feedback they were getting from the baby whenever they hold or talk to the baby. These findings are similar to other studies where fathers reported feeling encouraged by babies' developmental improvements [58], and the feedback helped them deal with their fears and get more involved in care [19,20,59]. On the other hand, the absence of improvement in health status and a lack of feedback from the newborn may lead to withdrawal from care by the father [19]. Providers supporting fathers in the hospitals, therefore, have a duty to help fathers understand the different non-verbal cues the baby makes so that lack of them may not act as a deterrent from involvement in care. The presence of multiple births had a positive influence on the fathers' involvement in the care of preterm newborns because fathers felt the mother on her own was lacking since there were two babies to be attended to and the fathers had to come in to help with caregiving activities. These findings are consistent with a Swedish study which found that twin birth meant there was a need for more people to help with the caregiving activities and enabled the fathers to get more involved [19]. For instance, when the mother was done breastfeeding one, the father took the responsibility of burping that one while the mother breastfed the other twin. The physical environment at the hospitals where the babies were admitted was reported as another factor impeding the fathers from being involved in care activities. Fathers in the current study needed more private space and not an open ward where everyone could see them. Other studies have also reported that lack of privacy in open-spaced hospital wards prevented the fathers from getting involved in care [60]. However, these findings do not support previous research, which found that the open-spaced beds allowed the fathers an opportunity to see their colleagues hold the babies and get involved in care activities which helped them realise that it was possible to get involved in care [19]. As such, there is a need for the healthcare staff attending to the preterm newborns to frequently engage fathers and get their preferences, and if resources are available, provide more privacy to those that may need it so that they get more engaged in care. Economic factors influence fathers' involvement in care. As per the model of male involvement proposed by Lamb et al. [10], responsibility entails the father being able to provide for what the newborn may need. The current study suggests that fathers that had formal employment considered their financial stability as a facilitator to involvement in care while those financially unstable would be spending more time running errands and doing piece works to earn some money to provide for the baby thereby limiting the time they could take part in care activities. These findings are consistent with other studies, which reported that the ability to provide for basic needs meant the father had time to be involved in care [23,61] and further reduced parents' psychological distress which usually affects their involvement in care activities [61]. In Malawi, it was reported that due to financial challenges, fathers were unable to provide for what is needed for the baby [26,44]. Some of the limitations of the study are the fathers were reporting about their own experiences thereby making it prone to self-reporting bias, and interviews were conducted by a female research assistant which may have prevented the fathers from further opening up as compared to if the interviewer was male. The study was also conducted during the time Coronavirus disease was at its peak which resulted in there being limited visitations to the health facilities as a preventive measure and affected recruitment of participants. Despite these limitations, the participants in the study were from varied cultural, ethnic, educational, and economic backgrounds which allowed for varied experiences in preterm care. Furthermore, the concurrent data collection and analysis helped determine saturation and also allowed the researcher to get in-depth data from the participants especially when it was noticed certain data gaps in the preliminary analysis. The use of the theory of Planned Behaviour and the Model of fathers' involvement as engagement, accessibility, and responsibility helped the study to get the experiences that were more focused and also measure the chances of the fathers having the intention to be involved in care. --- Conclusions The study findings suggest that several factors impede or facilitate fathers' level of involvement in care. These factors range from those considered personal such as perceived benefit to involvement, those related to the newborn, financial, facility environment, gender roles, sociocultural norms, support from health providers and relatives, and other responsibilities the fathers have. As such, health care providers and policymakers have a role to play in ensuring more fathers are taking part in the care of their preterm newborns since its impact is well known to surpass the cost. Fathers could be supported through education programs, training sessions, and support groups, which can help fathers overcome barriers and promote better outcomes for preterm newborns and families. Education programs can provide fathers with knowledge and skills related to neonatal care and the benefits of their involvement. Training sessions can help fathers develop practical skills and receive feedback from healthcare providers. Support groups can provide emotional support, a sense of community, and opportunities to share experiences and learn from other fathers. These interventions can also help promote gender equality and shared responsibility for caregiving within families. --- --- Abbreviations --- --- --- --- Competing interests The authors declare no competing interests. Mhango and Nyondo-Mipando BMC Pediatrics 23:432 • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research ? Choose BMC and benefit from: ? Choose BMC and benefit from: ---

Background Malawi has one of the highest incidences of premature birth, with twice the mortality compared to fullterm. Excluding fathers from preterm newborn care has negative consequences, including father feeling powerless, missed bonding opportunities with the newborn, additional strain on the mother, and negative family dynamics such as breakdown in communication, reduced trust, and strained relationships. In Malawi, there is no deliberate policy to have fathers involved in preterm care despite having high incidence of preterm birth and neonatal mortality. There is also limited literature on the factors that influence fathers' involvement in the care. The aim of the study was to explore factors influencing fathers' involvement in the care of hospitalized preterm newborns. Methods A descriptive qualitative study design was used, guided by Theory of planned behaviour and the model proposed by Lamb on male involvement. Sixteen in-depth interviews were conducted with fathers of preterm infants purposively and conveniently sampled in June 2021. Interviews were digitally recorded and transcribed verbatim. Data were organized and analyzed using Nvivo software and thematic analysis approach was used because the approach allows deeper understanding of the data, identification of patterns and themes, and provides rich insights into participants' experiences and perspectives.The barriers and facilitators that influence a father's involvement in the care of preterm newborn babies include: perceived difficulty with care activities and benefits of involvement, gender roles and socio-cultural beliefs, work and other family responsibilities, social support, baby's physical appearance/nature and health status, feedback from the baby, multiple births, and hospital's physical environment and provision of basic needs.The study found that fathers value their involvement in caring for hospitalized preterm newborns but face barriers. Evidence-based interventions like education programs, training sessions, and support groups can help fathers overcome barriers and promote better outcomes for infants and families.

INTRODUCTION Festivals and special events are known to play important roles in destination development as attractions, image makers, animators of static attractions, and catalysts for other developments 1 . In addition to the economic benefit that is derived from the increased number of tourists, festivals and special events also expand the tourist season of the destination 2 , provide cultural and educational opportunities, foster a feeling of community pride 3 , help conserve sensitive natural, social and cultural environments, and contribute to sustainable development 4 . Event and festival tourism is one of the fastest growing forms of tourism. It is becoming increasingly popular in rural areas as a means to revitalize local economies. Advancing technologies and changing weather patterns have brought about the need for diversification of industry in areas previously reliant on traditional industries such as farming, or narrowly focused tourism industries. Events and festivals have been used with great success in many such areas as a means to draw people to regions that traditionally have a seasonal tourist appeal 5 . According to a survey conducted by the International Festival and Event Association , the special events industry is estimated to include some 4 to 5 million regularly reoccurring events and has a significant economic impact globally 6 .The advantages of festivals and special events are demonstrated more explicitly in rural settings, particularly in boosting local economies, continuing employment, and rural destination promotion 7 . Every year, festivals and events generate more than 22,000 jobs in Ontario. Tourism spending contributes more than $22 billion annually to the economy, and supports approximately 300,000 direct and indirect jobs in Ontario8 . --- Background --- What is a Festival? A festival can be defined as -A series of performances of music, plays, films/movies, etc., usually organized in the same place once a year; a series of public events connected with a particular activity or idea‖ 9 . --- Definition of Events The term ‗event' is used to describe a wide range of activities many of which have quite different characteristics. Such events range from the Olympic Games at the mega-event end of the scale to small regional festivals. An event can be defined as ‗a onetime or infrequently occurring event of limited duration that provides the consumer with a leisure and social opportunity beyond everyday experience' 10 . --- Categorization of Events Events vary from local community based events to major events 11 . The nature of events can also differ due to the number of venues used during the event. Events can be all inclusive at one venue or held at many venues throughout a region. Multiple venue events are more difficult to evaluate or assess as boundaries become blurred, the sample and study area gets larger, and there is often a chance of counting the same visitor many times. A framework for the categorization of events is useful for determining strategic gaps through the identification of different events, their timing, location and themes. It is also useful to classify events in order to assess economic performance against criteria such as the timing of the event , its location, or its themes . --- Factors that make an event successful 6 The thematic nature of events can be used to either justify or disqualify the benefits or advantages of events based on their importance to the host community 12 . An event that spreads the seasonality of tourism in a region also distributes the flow of money into a region, increasing the opportunities for full time employment. Other factors that may contribute to a successful event could include13 : • Increasing visitor length of stay in region/town; These factors are beneficial to a region and its tourism industry, although the cost at which they are obtained must be justified. Furthermore, these impacts differ due to the size and scale of the event ranging from a local community based event such as an art show to a mega-event, such as the Olympics. For instance, little real estate speculation could occur as a direct result of the hosting of a local event such as an art show, compared with the hosting of a mega-event such as the Olympic Games. Festivals attract tourists who might otherwise never visit the area, and the benefits of festivals extend well beyond generating tourism dollars and include strengthening rural communities and enriching the quality of small-town life 14 . In addition to the social positives, holding festivals also improves the environment of the host community. --- The importance of festivals Events or festivals are, according to the South Australia Tourism Commission, celebrations of something the community wishes to share; their objective is to attract the maximum number of people to participate 15 or, to provide 16 the opportunity for -a public, themed celebration.‖ Goldblatt suggests the increasing number of festivals is due to small and large communities seeing the opportunity to obtain tourism dollars 17 . Dwyer et al. further suggest that events are seen as creating income and jobs in the short term and generating increased visitation and related investment in the longer term 18 . All levels of government see them as a way to promote tourism 19 , while regional tourism operators use them as part of their marketing strategy. Festivals can be organized for reasons that include the preservation of culture and history and the provision of recreation and leisure 20 or for their ability to give an area a competitive advantage by targeting special interest markets that may lead to the meeting of social, economic, and environmental goals 21 . Festivals and community events have the ability to revitalize, reimage, and expand existing markets and in most cases bring economic benefit to the destination that stages them 22 Festivals have direct and indirect impacts on their communities through the opportunities they provide for; participation, skills development and volunteering. They can also have environmental and political impacts. Benefits of festivals to communities can be extensive, particularly when the draw is from outside of the region. Local vendors, artisans, craftspeople, restaurateurs, hoteliers and innkeepers are kept busy and may indeed make a large portion of their annual income during the course of a weekend. Mainstreet revitalization can also be a positive economic spin-off from festivals and 1997;Walo, Bull, and Breen 1996;Hall;1989b22 Goldblatt 1997;;Hall 1992;Getz 1991 special events. Appearance improvements made for an annual celebration have year-round benefits and can create stimulus for improved downtown retailing and service industries. --- Importance of economic impact studies The economic impacts of events on the macro-economy of a region are categorized into primary and secondary economic impacts. The primary impact is the direct economic impact of the event measured by surveying participants or measuring business sales. Secondary impacts are those that result from the introduction of new money into the economy. Types of Events Economic impact studies provide a good tool to estimate the influence of particular events on the economy. The results of economic impacts are useful in making decisions and may assist developers and local politicians to justify public funds by citing the economic benefits events have for the whole community23 . The studies of economic impact focus on how a project can impact the surrounding community by creating employment, income, and helping territory spatial organization24 . Economic impact is not just about direct or directly-induced expenditures. In a brief recently released by the National Governors Association on the Role of the Arts in Economic Development, it was noted that "the non-profit arts industry is a potent force in economic development nationwide. States and communities have integrated the arts into their economic development arsenal to achieve a wide range of direct and indirect economic goals. Arts and festivals assist municipalities in the following ways: They leverage human capital and cultural resources to generate economic vitality in underperforming regions through tourism, crafts, and cultural attractions; They restore and revitalize communities by serving as a centerpiece for downtown redevelopment and cultural renewal; They improve quality of life, expand the business and tax revenue base, and create a positive community image; and They make communities more attractive to highly desirable, knowledge-based employees thus stimulating new forms of knowledge-intensive production to flourish." Events are increasingly used in place marketing and can offer the following advantages25 : They can attract tourists and visitors, both national and international. The tourism influx generated is especially welcome when it occurs in a medium or low-tourism season since it reduces seasonal ebbs. Events help to capture attention and promote attractions and infrastructures. They make it possible to maximize and rationalize the use of certain spaces. Preservation of these spaces may result in financial benefits and dissemination of artistic and cultural heritages. Through cultural events, places may attract not only tourists but also investments. The tourism industry that supports events generates employment and economic diversification. Other possible impacts of community events and festivals can be26 : Beaverhill Lake, located in central Alberta, Canada, attracts thousands of Snow Geese and a wide variety of other birds on their spring migration. In turn, these birds attract thousands of bird watchers to the Lake and to Tofield, the nearest town . The annual Beaverhill Lake Snow Goose Festival, begun in1993 and held in the third or fourth week of April, attracts 3000-7000 people each year27 . During the two-day festival, participants can go on a guided bus tour or hike, take a free shuttle bus to the lake, visit the nature centre, or take part in a wildlife trade show, art show, or nature's market. Registered participants spent an average of $32.68 per person, of which $22.48 was spent in Tofield and $10.20 outside of Tofield. Nonregistered participants spent slightly less. In total, festival participants spent $146,145, of which $98,050 was spent in Tofield. By expenditure category, the percent spent in Tofield was, from highest to lowest, restaurants , souvenirs , miscellaneous costs , other costs , registration , groceries , accommodation , and travel . Festival organizers also spent $12,000, of which $7,200 was spent in Tofield. During the festival, 58% of businesses reported that they had attracted new customers. Of responding businesses, 30% of sales during the week of the festival were attributed to the festival. More than a third reported a moderate increase in income during the festival season. During the festival, 32% of businesses reported expenditures on advertising and facility development. Most of the businesses indicated moderate/partial support and 11% of the businesses indicated an overwhelming support for the festival. --- Sylvan Lake, Alberta The Sylvan Lake supports year-round recreation suitable for all visitors. The summer months promote swimming, sailing, boating, and fishing, while the winter months offer hockey, crosscountry skiing, snowmobiling, and ice fishing. Boasting a beautiful natural ice surface, the town is able to host activities such as the Sylvan Lake Pond Hockey Tournament and outdoor speed skating competitions, Jazz at the Lake Festival and Shake the Lake festivals. These events and festivals contribute to the town's economy28 . The Town of Sylvan Lake hosts approximately 906,744 visitors annually. Visitor spending creates an economic impact of $48.4 million each year. Food, beverage, accommodations, and recreational spending account for over 80 per cent of all visitors spending. --- Arts and Culture, Edmonton, Alberta The Edmonton Arts Council and Economic Development Edmonton collaborated on two studies of the economic impact of Edmonton's non-profit arts and festival organizations on the economies of the Edmonton region and on the Province of Alberta 29 . The study was limited to 101 arts and festival organizations that applied for an operating grant from the City of Edmonton in 2000 and which contributed detailed information on their operations and attendance during their 1999 season. During the 1999 season the 101 participating organizations entertained an estimated total of 2,737,268 patrons including approximately 590,264 patrons visiting from outside Edmonton. The results show that each local resident attends approximately 3 performances or venues per year. Patrons spent an estimated $57.7 million in Edmonton during their trips to participate in Edmonton's "Arts and Culture‖ activities. The "Arts and Culture‖ organizations themselves spend approximately $40.3 million annually on local goods and services. When spin-offs to other parts of Alberta are included the total impact that Edmonton's "Arts and Culture‖ has on the province of Alberta grows to approximately $116.6 million. --- Prince Edward County, Ontario Prince Edward County is a rural community in Eastern Ontario. It is located in the heart of Canada's Creative Corridor between Toronto, Ottawa, and Montreala -mega region‖ that represents 50% of Canada's GDP 30 .This location provides PEC with a unique opportunity for growth in the creative economy. In 2006, the population was 25,496; however, the median age for PEC was 47.7 compared to 39.0 for Ontario 31 . PEC has a higher percentage of its population between 25 and 64 with a university degree when compared to other areas in the province that are outside a major urban region 32 . Agriculture has a long history in PEC. The dominant agricultural sectors include dairy, beef, and grains and oilseeds 33 . Prince Edward County is at the forefront of Ontario's new creative food movement 34 . In fact, PEC is quickly becoming one of Canada's fastest growing wine regions and it has been officially designated as Ontario's 4th Designated Viticultural Area . Tourism also plays a significant role in the PEC economy, drawing traditionally from its natural amenities including Sandbanks Provincial Park 35 . Over the last decade, PEC has led an aggressive strategy to attract educated and creative workers to the area. PEC has attracted many ‗Escape Artists', creative professionals who desire to leave the city to work and live in rural areas 36 .PEC is already home to over a hundred independent artists and galleries complemented by artistic institutions like the Regent Theatre and an annual Jazz Festival. The County also uses an aesthetically-pleasing website, www.buildanewlife.ca, to attract creative investment. PEC has further recognized the importance of industry clustering in the creative economy through the Taste Trail. This involves strategic partnerships and co-marketing between farms, cheese producers, wineries, breweries and restaurants in the County. --- Economic Impacts of 97 Festivals and Events of Ontario Ontario's festivals and events provide significant economic benefits for the province through spending by the festivals themselves and by their visitors 37 . The report, -Economic Impacts of 97 Festivals and Events of Ontario‖ prepared for the Ontario Trillium Foundation, examines the economic impacts of various sizes of festivals and events in the province. The overall impact of 97 culture, sports, recreation and community festivals and events amounts to nearly $80 million in economic return to the province's Gross Domestic Product . The 97 festivals and events also generate over $30 million in taxes for all levels of government and help create 2,600 jobs and over $50 million in wages and salaries. This fact sheet highlights the breakdown of these key economic impact figures for small, medium and large festivals and events, including 39 small festivals and events , 37 mediumsized festivals and events , and 21 large festivals and events . There are significant differences between the 97 festivals and events in terms of type of event, budget, attendance and location. The festivals and events include non-profit organizations involved in culture, sports, recreation and community-based initiatives. Their expenditures range from a few thousand dollars to nearly $2 million, and attendance figures vary between 150 and 700,000. In addition to these differences, the festivals and events vary in their staffing and use of volunteers. These differences mean that the results presented here should not be used to calculate the economic impacts of any one festival or event. --- Economic Impact Analysis of ECBC's Festival and Events Initiative Enterprise Cape Breton Corporation launched its Festival and Events Program in fiscal year 2002/2003 to assist incorporated, non-profit organizations in Cape Breton interested in hosting festivals and events 38 . Since that time ECBC has provided approximately $729,000 in funding assistance to two hundred and fourteen festivals and events throughout Cape Breton and Mulgrave area. These festivals and events were attended by over 1.12 million people, of which 27% were off-Island visitors. These approximately 300,000 off-Island visitors generated a positive economic impact of $76.2 million on the Gross Domestic Product of Cape Breton, created 2,621 Based on the net economic impact or Gross Domestic Product the total contribution to the Cape Breton economy from every $1 in Festival and Events Program funding is $104. Obviously the Festival and Events Program has made a very significant contribution to the Cape Breton economy over the past four years. --- EXAMPLES FROM OTHER JURISDICTIONS --- UK a) The creative economy The arts are an important part of the rural economy. Across the country, arts venues create jobs, support local businesses and attract visitors39 . Cumbria, for example, has 21 theatres, each contributing to local cultural, social and economic vitality. In Keswick, the lottery-funded Theatre by the Lake is the town's second-largest employer, creating 80 jobs and bringing £2 million each year to the local economy. Many artists run businesses in rural areas, working in fine art, craft, design, advertising, music and digital media. Research in the South Westwhere 47 per cent of the population lives in rural areasidentified at least 5,000 craftspeople, most of them working from home in villages and small towns. Crucially, this research also showed that, between 1998 and 2002, regional employment in visual arts and crafts rose by 37 per cent, while income increased by 11 per cent. In the South West and other rural areas, the creative industries are a cornerstone of a strong rural development strategy. --- b) Art Connections Art Connections was developed to help North Yorkshire artists market their work, and provide business support. Established with support from local authorities, Yorkshire Forward, Arts Council England, the Countryside Agency and the Small Business Service, the project offers an information service, mentoring scheme and a website promoting the work of selected artists. A pilot business network with 12 members has generated over £150,000 of new business and £100,000 of orders for suppliers. --- c) Rural festivals The movement restrictions imposed during the FMD crisis highlighted the importance of tourism and leisure to the rural economy. The arts bring visitors to the countryside. Annual festivals, from international events such as the Three Choirs to local ones like Rye Festival , bring thousands of artists and audiences, giving a valuable boost to the local economy. Three quarters of the people who come to Buxton Festival live outside the town. The event contributes £1.3 million to the town's economy each yearand only 10 per cent of the festival's annual turnover comes from public sources. --- d) Cumbria festivals The Cumbria Rural Regeneration Company is working with the Arts Council England, the Cumbria Tourist Board and the county council to support over 60 festivals with a substantial arts element. The program includes new education work, research, publishing information resources and a £450,000 Festivals Fund, through the North West Regional Development Agency. One early outcome is planning a Cumbria Biennale that will build on the county's strengths in visual arts. --- e) Farm diversification Agricultural diversification has grown rapidly as farmers move away from mass food production, and the arts provide new opportunities to add value, for instance in marketing organic foods or in environmental initiatives. Artists' studios are being developed in redundant farm buildings while farms themselves are becoming sites for events and exhibitions. At Middle Rocombe, in Devon, a dairy farm and ice cream factory has become an ‗art farm', where, each September, up to 50 artists show their work. In 2003, the event attracted over 3,200 visitors and produced sales of £13,000. Elsewhere, farms are used as sites for painting holidays, learning about traditional crafts and galleries. --- USA a) Economic Impact of America's Nonprofit Arts & Culture Industry Nationally, the nonprofit arts and culture industry generates $166.2 billion in economic activity every year-$63.1 billion in spending by organizations and an additional $103.1 billion in eventrelated spending by their audiences 40 . The impact of this activity is significant, supporting 5.7 million U.S. jobs and generating $29.6 billion in government revenue. Arts & Economic Prosperity III is the most comprehensive study of the nonprofit arts and culture industry ever conducted. It documents the economic impact of the nonprofit arts and culture industry in 156 communities and regions , and represents all 50 states and the District of Columbia. The diverse communities range in population and type . Researchers collected detailed expenditure and attendance data from 6,080 nonprofit arts and culture organizations and 94,478 of their attendees to measure total industry spending. Project economists customized input/output analysis models for each study region to provide specific and reliable economic impact data. The study demonstrates that the nonprofit arts and culture industry is an economic driver in communities-a growth industry that supports jobs, generates government revenue, and is the cornerstone of tourism. Nationally, the nonprofit arts and culture industry generates $166.2 billion in economic activity every year-$63.1 billion in spending by organizations and an additional $103.1 billion in event-related spending by their audiences. The study is the most comprehensive study of the nonprofit arts and culture industry ever conducted. It documents the economic impact of the nonprofit arts and culture industry in 156 communities The annual Elvis Revival Festival in the small town of Parkes, in rural Australia explores the way in which a remote place with few economic prospects has created a tourism product. The festival subsequently captured national publicity, through a festival based around commemoration of the birthday of Elvis Presley, a performer who had never visited Australia, and certainly not Parkes. The Festival began in the early 1990s, when a keen Elvis fan rallied promoters around the idea of bringing Elvis impersonators to the town for an annual celebration. Since then, the Festival has grown in size, with notable economic impact. The economic impact of the festival has been considerable, and visitor surveys in 2004 indicated the extent of the direct economic effects on Parkes. In 2004, visitors spent an average of A$440 per person over the festival weekend, translating to an injection of over A$1.1 million into the local economy. Accommodation , food and drink and entertainment were the most common forms of expenditure, with smaller amounts spent on souvenirs and other services such as fuel . The festival improved employment multiplier impacts by generating extra work in those activities that, in turn, are most closely embedded in the local economy rather than others that rely on goods and services imported from Australia's large urban centres. --- PORTUGAL a) A Cherry Festival A Cherry Festival contributes positively in the economy of a rural municipality of Funda˜o, in the Beira Interior of Portugal 41 . Funda˜o has 31.482 inhabitants, a geographic area of 700.4 square kilometers and is characterized as being a region that is losing population, for having an extremely old population, for having a high number of companies from the primary sector and for having a purchasing power per capita lower than the national average 42 . This festival is organized by Funda˜o's Town Council. During the festival ; some residents open the ground floors of their houses to visitors in what is called -small taverns‖. In these taverns, visitors can buy and experience cherries and their 41 Impacts of small tourism events on rural places by Helena Maria Baptista Alves 42 INE, 2001 derivatives along with other regional products such as wines, cheeses, and handicrafts, especially rattan baskets. The Cherry Festival through local tourism operators also enables tourists to experience a taste of the natural and cultural environment of the region, including cherry harvesting, mountain biking, walking, and garden tours. During the festival visitors and tourists also experience exhibitions, concerts, street fairs, and professional performances. With this festival, the municipality of Funda˜o hopes to attract some tourism that helps to boost the region's economy and also promote the image of the cherry both nationally and internationally. It was observed that in economic terms the Cherry Festival of 2008 had a direct and indirect effect of €267.55for the local and regional community. Although this number is not in itself significant, its importance can be considered by increases in occupancy rates of hotels during the festival , as well as increases in the number of meals served by restaurants . That is, both the hotels and restaurants on average doubled their capacity during the days of the festival. In economic terms, the effect of the induced impact, an impact that although roughly estimated, is extremely important and should also be considered. Furthermore, these effects can also be seen in the future, as visitors and tourist's experience with the local hotels and restaurants may lead to future returns. Another very important component of this event is its social impact on the community. In this study, it was found that this impact is perceived to be higher than the economic impacts, as both residents and suppliers agree that the Cherry Festival has a greater social than economic impact in the local community. The residents and the suppliers agreed that the greater impact was on -building community pride‖ and -enhanced community image‖ than on economic impacts. --- FINAL THOUGHTS Event tourism can contribute to economic diversification and profitability by increasing employment, improving basic services, and increasing economic equity between urban and rural populations. Events and festivals can help people be more aware of their capabilities, perceive new job and business opportunities and improve quality of life. Community events and festivals can attract local, national and international tourists and visitors. Events help to capture attention and promote attractions and infrastructures. They make it possible to maximize and rationalize the use of certain spaces. Preservation of these spaces may result in financial benefits and dissemination of artistic and cultural heritages. Events and festivals play a role in putting a region on the map and distinguishing their basket of attractions from other similar regions. Events can also result in social benefit such as -increased community pride‖ and -enhanced community image.‖ Research shows that the more involved local people and suppliers are in terms of provision of services, food, beverages, and attractions, the greater the economic benefits to the region. Local vendors, artisans, craftspeople, restaurateurs, hoteliers and innkeepers may make a large portion of their annual income during the course of an event. Organizers should build partnerships and involve the greatest number of possible local operators. Organizers should think about all the potential benefits, and consider the community events to be an investment in community sustainability. Community events and festivals can make a significant contribution for rural development with strategic planning, well-defined goals, local level partnerships and funding. However, rural communities often organize tourism events to inject cash into their economies. This cash is not enough to achieve comprehensive rural economic development. Events assist the tourism businesses of the region but are not enough to revive the entire economy. The Association of Festival Organizers and the Arts Council in England determined the total expenditures arising from 350 folk festivals in England 57 . In the United States, the International Festivals and Events Association assessed the combined economic impact of festivals and events produced by IFEA members 58 . 57 Association of Festival Organizers 58 Zoltak --- Appendix A --- Some other examples Some other examples of economic impact analyses are as follows 43 : The tourist spending of 90,000 visitors to Louisbourg in 1995 was assessed in terms of its impact on GDP 44 . Two spending estimates, one -conservative‖ and one -optimistic‖ were developed for the Cabot '97 events in Cape Breton 45 . An assessment was carried out of the total economic impact of the Louisbourg -Encampment ‗99‖ event 46 . An analysis of the Tall Ships 2000 event determined the incremental visitor spending which resulted in Nova Scotia 47 . A -social cost-benefit analysis‖ of the Stan Rogers Folk Festival in Canso calculated the net social benefit 48 . A study of Festivals and Events in Nova Scotia used data collected from a sample of F&E to calculate a -province-wide economic impact assessment‖ 49 . The Atlantic Theatre Festival in Wolfville assessed the economic impact of their event after three years of operation 50 . A 2002 study of 19 major events in Québec for REMI evaluated the contribution to the economy and estimated that for each dollar invested in the industry, another $5.36 was generated 51 . A 1997 study of festivals in Ottawa determined the spending by visitors from outside the municipality , the gross economic impact resulting from this spending , and the federal, provincial, and municipal taxes collected 52 . A study of Ontario's festivals and events industry calculated the total direct revenue 53 . The Edmonton Arts Council's study of festivals and the arts calculated the total net impact on the Edmonton region , and, with spin-offs included, the impact on the province of Alberta 54 . The City of Langley, British Columbia calculated overall visitor spending for -cultural tourism‖ , and, using the provincial government's multiplier formula, determined the economic spin-off 55 . The economic impact of the Springfest event in Ocean City, Maryland was calculated , together with the economic impact on personal income , which the authors feel to be the most meaningful measure 56 .

This research provides a literature-review-based overview of community events and festivals. This report also considers the event categories, their importance and the associated impacts on rural communities. The report reviews examples from Canada and other jurisdictions. Events and festivals tourism is one of the fastest growing forms of tourism. They are becoming increasingly popular in rural areas as a means to revitalize local economies. Festivals and special events play important roles in destination development, image makers, animators of static attractions, and catalysts for other developments. The term ‗event' is used to describe a wide range of activities many of which have quite different characteristics. Events vary from local community based events (e.g. Folk Festivals) to major events (e.g. The Olympics). Events have direct and indirect impacts on communities. They provide opportunities for participation, skills development, volunteering and social, cultural economic and environmental developments. Community events and festivals can attract tourists and visitors at regional, national and international level. Events help to capture attention and promote attractions and infrastructures. They make it possible to maximize and rationalize the use of certain spaces. Preservation of these spaces may result in financial benefits and dissemination of artistic and cultural heritages. Benefits of festivals to communities can be extensive, particularly when the draw is from outside the region. Research shows that the more involved local people and suppliers are in terms of provision of services, food, beverages, and attractions, the greater the economic benefits to the region. Local vendors, artisans, craftspeople, restaurateurs, hoteliers and innkeepers may make a large portion of their annual income during the course of an event. Festivals and events have impacts that go well beyond what can be measured in economic terms. They contribute to the quality of life across Canada by strengthening communities, providing unique activities and events, building awareness of diverse cultures and identities, and acting as a source of community pride. Community festivals and events are also revolutionizing businesses and prompting economic development across the globe. Community events, festivals and arts venues create jobs, support local businesses and attract visitors across the UK, USA and Australia. Governments offer a wide range of products and services to help promote community events in rural areas. Community events and festivals can make a significant contribution for rural development with strategic planning, well-defined goals, local level partnerships and funding. However rural communities often organize tourism events to inject cash into their economies. This cash is not enough to achieve comprehensive rural economic development. Events assist the tourism businesses of the region but are not enough to revive the entire economy.

Introduction Women have traditionally married and had their first child at a young age in many parts of the world 1 . Recently, however, teenage sexuality and pregnancy have been framed as a social problem, whose negative outcomes for the lives of adolescent women have been widely used to justify the need for public intervention and expansion of family planning programs to target the issue 2,3,4,5,6 . The assumption that teenage pregnancy is frequently unwanted or unplanned suggests that one of the causes may be the insufficiency or inadequacy of family planning programs 7,8 . However, despite the availability and free distribution of contraceptive methods in some regions, the reluctance to use them 9,10 , and persistent age-specific fertility rates point at flaws in a model that analyzes fertility intentions as a direct result of contraceptive use, furthermore ignoring cultural norms and ideals of motherhood and family size. Brazil has presented declining fertility levels since the early 1970s, but adolescent fertility is still considered to be an increasing phenomenon, especially among the poor strata 11 . In a culture where motherhood is desirable and central to womanhood 8,12 , this study aims to ARTIGO ARTICLE Cad. Saúde Pública, Rio de Janeiro, 28:655-664, abr, 2012 investigate whether adolescent females consider pregnancy to be a problem, analyzing their discourse in the light of their socioeconomic scenario and consequent aspirations while investigating the role of different actors in their decision-making process. Guided by the hypothesis that motherhood can present itself as a milestone and transition into womanhood, this research takes socio-cultural aspects into consideration to understand local views in teenage pregnancy and how these views might affect girls' attitudes towards contraception and childbearing. In order to attempt to answer that question I chose to investigate what appears to happen in a shantytown within the metropolitan area of the city of Belo Horizonte, Minas Gerais State, Brazil, by making use of qualitative methods for data collection such as participant observation, in-depth interviews, and focus group discussions. --- Methodology The research site chosen was Vila Novo São Lucas, located within the Metropolitan area of Belo Horizonte, with a population of around 4,000 inhabitants spread over 151.326m 2 . Vila Novo São Lucas is one of the six slums that together form the "Aglomerado da Serra" -Belo Horizonte's biggest settlement, currently housing over 50,000 people 13 . For the purposes of this study qualitative methods were used for data collection, which consisted of interviews with key informants , in-depth interviews with young mothers and mothers of teenage mothers, participant observation and field notes, as well as focus group discussions. Reflective analysis was an on-going process taking place throughout the data collection. And although interview guides were developed beforehand, new ways of approaching unforeseen topics were developed and the interview guides were refined accordingly. In-depth interviews were conducted as follows: with key informants who are local community figures, 10 young mothers who were under 18 at childbirth, 10 young mothers who had their first child after age 18, and 5 mothers to teenage mothers. The interviews were made in various locations where privacy was possible, recorded then transcribed, and the profiles were later translated to English. Consent of the participants was requested and given. This research followed all the necessary ethical principles and was approved by the University College London Research Ethics Committee. The open-ended interviews were conducted using a general guide approach intended to ensure that the same general areas of information were collected from each interviewee, while still allowing a degree of freedom and adaptability in obtaining the information. Having had case-oriented and comparative perspectives, this research made use of content analysis and grounded theory 14 , the latter being used to analyze the corpus of data and identify different variables and their interrelationships. Coding started with full transcription of the interviews in an effort to recognize key words or sentences, incorporated by field notes to identify emerging patterns, allow comparisons, and provide guidelines for collecting additional data. The next step was to organize the responses into labeled clusters, furthermore breaking the discourse of the participants and creating categories, or codes, which could be compared and contrasted. Through systematic analysis and comparison of data, the number of codes were organized in a way that showed the relationship among them, thus giving rise to a core category representing hypothetical relationships between the different categories. The core category in this study is discussed in the next section, but the model constructed details some specific conditions in the lives of the studied population that seems to give rise to the phenomenon of teenage pregnancy, revolving around one main theme: the belief that childbearing brings about a sense of achievement and self-fulfillment and that, far from being an unwanted occurrence of negative consequences as often cited in the literature, it is closely linked to local ideals of femininity, maturity, and small family sizes. --- Results In their own words: young mothers and their discourses about pregnancy How does pregnancy "occur"? • Motivations and unplanned consequences Retrospective accounts of life histories were, in this research, the only way of getting an insight into young women's intentions during their teens. Additionally, as Williams et al. 15 and Trussell et al. 16 note, when asked retrospectively about their fertility intentions, some women may report their feelings according to changes they have undergone in their lives, and declarations of wanted or unwantedness may correlate more closely with the outcomes of their pregnancies and current circumstances. In the sample, only two girls said their pregnancies were planned as the result of a decision made jointly with their partners. However, from data collected in the focus group meetings and field notes, this decision was expressed as natural and expected from couples who were in longterm relationships, regardless of their age. Data gathered shows the search for more freedom at home, i.e. from their patents, as a reason why one might want to get pregnant early in life. In addition to that, girls as young as 12-13 years old are allegedly trying to become pregnant from drug lords or drug dealers, in an attempt to obtain financial help and enjoy high status among their peers. This is consistent with the findings in Cameron 17 , Kenya 18 , and South Africa 19 , according to which the exchange of sex for gifts or financial support help shape young girls' sexual behavior, also favoring the non-use of condoms. Most girls who became pregnant before age 18, however, had not planned the event, which happened soon after their first intercourse. According to them, the absence of contraceptive methods allied to their "naivety" at the time contributed to their early pregnancies. Such occurrence could also indicate insufficient knowledge on sexuality and reproduction. --- Knowledge and attitudes towards contraception Despite their broader knowledge of contraception and relative experience of using different types of methods, women who were 18 or older at the time of their first pregnancies surprisingly also reported them as unplanned. Such women had been in and out of relationships for years and had used different types of contraceptive methods, the pill being the most popular one. As in Brazil one does not need a prescription to purchase contraceptives over the counter, this is considered easy and preferred to queuing at the health center with other patients. Half of the girls who got pregnant in their early 20s also mentioned they had been on the contraceptive pill when they became pregnant, thus being unable to understand how that happened. Possibly ill-use or simultaneous consumption of other medicines, such as antibiotics, added to the fact that over-the-counter drugs do not come with the doctor's recommendations. Perhaps, as Ajayi et al. 20 and Juarez 21 point out, sporadic sex may help determine contraceptive choices. Young women might choose to interrupt the use of long-lasting contraception, such as the pill, in the absence of a stable relationship, thus becoming more exposed to unplanned pregnancies in the event of unprotected sex. For the younger ones the lack of knowledge about reproduction could be seen in some cases and reinforced through contradictions that appeared in their discourses. This was true for most girls who got pregnant very early, showing insufficient or inadequate knowledge on the matter, including formal knowledge gained at school 20,22,23 . --- Embarrassment, moral barriers, and • contraceptive choice Adapting the educational system to meet adolescents' needs could certainly improve their understanding of their own body and sexuality, but the question as to whether their real doubts are being clarified will remain unanswered for as long as teenage sexuality remains a taboo. None of the girls interviewed reported being able to talk about sex with their parents or older relatives, and 90% had never seen a gynecologist or talked to a reproductive health professional until they found out they were pregnant. Reasons cited include embarrassment to see a gynecologist and lack of anonymity in the local health center. The center has a good varied team of health personnel and offers condoms, injectables, and pills as contraceptive methods. However, it does lack privacy, as everyone has to be screened by the social worker and explain their reasons for visiting. Additionally, health workers are normally people from that area, therefore acquainted with the girls who, in turn, fear the disclosure of their sex lives. Furthermore, some health clinics demand under-aged girls to come accompanied by their parents to prescribe free contraceptives, which could understandably be a barrier. Despite the different approach taken by the local health center, figures pointed out by the senior nurse show a timid demand proceeding from adolescents: for every one hundred patients she screens every day, only 2 or 3 adolescent girls would come every week in search of contraceptive methods. She states that the highest demand for pills comes from girls around the age of 23, who had already had a child and wanted to space births. Even so, she was looking at around 30 prescriptions per month, as opposed to the 900 condoms distributed monthly. Table 1 shows the mothers' profiles and their chosen contraceptive methods before and after pregnancy: demand for hormonal contraceptives is slightly higher for girls over 21 despite their similar educational levels, showing perhaps not so much discrepancy in terms of formal knowledge gained at school, but probably less embarrassment to discuss their issues and gain access to certain methods. Perhaps the most distinctive finding in this research was the fact that, unlike results found by some scholars, e.g. Manlove et al. 24 and Khan et al. 25 , rapid subsequent births or the non-use of contraceptive methods after birth for teenagers who had not used a contraceptive method during their first sexual intercourse were not a visible concern in this sample. Thus the reluctance to use contraceptives is not associated with their unavailability or with the girl's lack of experience with different methods, since 90% of them opted for the IUD to space births and reduce the risk of another unplanned pregnancy once they had their first child. Perhaps this would help explain the low demand for other hormonal methods at the health center and raise the question as to whether or not the implementation of more family planning programs is the answer to tackle teenage pregnancy. --- Education and life plans Hoffert et al. 26 refer to teenage pregnancy as one of the main contributors to young girls' discontinued education, obstructing their career paths and limiting their job prospects. Among the girls interviewed, truncated education is a common occurrence, but before making a premature association with early pregnancy, other factors regarding their background should be analyzed carefully. From the sample, the girls who got pregnant during their teens show almost the same educational level as the girls who got pregnant at a later stage in life, and despite their expressed desire to finish school and go to university, no one really tried to do so, even before they became pregnant. Almeida et al. 27 found results that may partly resemble this reality when analyzing school trajectory and teenage pregnancy in three Brazilian state capitals, showing that despite the fact that young mothers had interrupted their studies, for many of them this decision was made before the onset of pregnancy. Similar results were found by Hof & Richters 28 in Zimbabwe associating school interruption with existent financial problems prior to pregnancy. Moreover economic hardship, especially in female-headed households, reinforces the need of getting a job at early ages in order to complement their family's income. As a result they have to juggle long working hours and evening classes, a combination that often proves unsuccessful. Some of the girls interviewed, for instance, had already been failing school for years before eventually dropping out. In their research carried out in Brazil, Almeida & Aquino 29 suggest that the risk of interruptions in education and dropping out of school under unfavorable family and social contexts may contribute to the acceptance of motherhood as the only possible successful path for impoverished girls. Saying they had interrupted their schooling because they got pregnant and had to provide for their babies only tells part of the story. However a child in this scenario can legitimate their choice and a decision that was imminent even before pregnancy occurred. Jonhson-Hanks 30 draws attention to the need of understanding the place of schooling in girls' aspirations and analyzing it in conjunction with other sources of influence which are present in their social environment. It seems that in Vila Novo São Lucas education, however important, gives way to experiences that give the young mothers an immediate sense of fulfillment and achievement . Cad. Saúde Pública, Rio de Janeiro, 28:655-664, abr, 2012 Education and the job market • Jobs that require little formal education are a common source of income for many women in Vila Novo São Lucas. Data on their working histories shows a preexistent financial need impelling girls to an early start in the job market, creating a feeling of economic stability, thus making education a laborious if unnecessary exercise, especially in the event of childbirth: "I got pregnant when I was 14, but I've always worked, since the age of 11 I thought that since I had a profession I could make money wherever I went, and my kids wouldn't starve. That was when I got pregnant again…" . --- Motherhood • Description of motherhood as an inevitable and desired occurrence was mentioned by 95% of the young mothers. Negative views on parenthood were absent and, differently from young girls' perception in South Africa 31 , were not viewed as compromising personal, professional, or financial aspirations. Whereas plans of educational attainment can be vague and dependent upon their financial situation, childbearing is an event that gives them a sense of purpose in life and a valid reason for their hard work that goes beyond money. Responsibility is listed as the first and most important change motherhood had brought to their lives. However young those girls might appear to start raising a family, their discourses invariably highlight the fact that they went from "going out and having fun" to "being responsible and hard working" after childbirth, even though they all used to work to help their families prior to pregnancy. --- The social environment and the roles of different actors The family's reaction and their support • The reaction of the family in face of unexpected pregnancies was one of surprise and sorrow. Mothers to teenage mothers reported being very upset when their daughters became pregnant and such reaction was also recollected by young women who got pregnant at age 23, for example. Ability to work and support themselves was reported by many girls as the reason why their parents were supportive of their pregnancies. Since most of them had already been working before their pregnancies and continued to do so, financial security ceased to be a concern for the parents, giving space to gradual demonstrations of fondness towards their pregnant daughters. The family's support is extremely important for the young mothers, especially as many continue to live with their parents after birth. Young girls who do not have children mentioned their parents as their prime source of support in the event of an unplanned pregnancy, and every girl interviewed, despite their hesitation in telling their families at first, received unconditional support from them. Their help goes from emotional to financial support, with women helping look after their grandchildren, giving the young mothers time to make a living and acquire the highly desired sense of responsibility. --- The father's role and the decision to have a baby In some cases pregnancy was planned, and since it was a joint decision, the couple stayed together to bear their child. Interpretation of the data gathered shows that such event often occurs in long-term relationships -nevertheless not very common in my sample -resulting in most participants continuing to live in their parents' house. Pregnancies resulting from casual relationships were frequent, and girls' decision to have a baby was not strongly linked to the father's support, as they relied on their respective families. --- A social phenomenon rather than • isolated incidents The incident of teenage pregnancy in Vila Novo São Lucas is not considered a rare phenomenon by local people; likewise the naturalness of their speech regarding the subject may help to socially validate an event that became indisputably frequent. Despite the initial shock described by the mothers, the collectiveness of the event seems to lift the weight of having a teenage daughter who is pregnant. Perhaps a collective understanding of teenage pregnancy as recurrent, nonetheless familiar, could help bring its occurrence to a level of social acceptance and support as it acquires a status of common phenomena. Cad. Saúde Pública, Rio de Janeiro, 28:655-664, abr, 2012 --- Discussion Despite the fact that most teenage pregnancies were reported as unplanned, none was said to be unwanted, this being the most important finding and aim of this research, thus contradicting many studies on teenage pregnancy that fail to look at the girls' perspectives and subsequent influence on their behavior. The desire to become mothers acts as a drive to young girls and their decision not to use contraception is based on beliefs that motherhood will bring fulfillment and take them to a different level of maturity. For those in longterm relationships who actually decided to have a baby, their decision was based on their ideals of motherhood and life goals, and the fact that they were minors at the time did not weigh much in their decision. The fact that very young girls were not using any contraception at the time they conceived was more related to lack of information about sexuality and reproduction, than to unavailability of contraceptive methods, since their local health center offers contraception for free. Teenagers reported lack of privacy as the major reason for not using their services, pointing to the inadequacy of family planning programs to adolescents' needs, and how this can be a major obstacle to their access in spite of their availability. Gupta & Leite 7 note that insufficient family planning services may contribute to high fertility rates amongst adolescents, especially if the services available are inadequate to their needs. In Vila Novo São Lucas some of the health providers believe they are offering good enough services to teenagers -despite being undistinguished from the ones offered to adult women -even though they do not seem to understand the low demand coming from adolescents. This study found that despite the presence of a local health center and free distribution of different types of contraceptive methods, the lack of anonymity and fear of disclosure of their sexual lives was an impediment to very young girls, as similarly found by Tangmunkongvorakul et al. 32 in Thailand. The fact that health providers lived in the neighborhood was seen as an obstacle to getting contraceptives from the center, reinforcing the paradox existent in contemporary Brazilian culture as proposed by Heilborn 8 , according to which, teenage sexuality is evident and socially tolerated, but in the form of "blaming the victim". Even though the girls showed knowledge about contraception, only the older ones who were already mothers would go to the health center to obtain it. Blanc & Way 33 found similar results elsewhere in Latin America. No negative health outcomes were related or associated with pregnancy at early ages, including those who had their first child at age 13. However, like Luke's 18 findings in Kenya, age and economic asymmetries between sex partners can play a role in exposing young girls to the consequences of unprotected sex i.e. STDs and HIV/ AIDS, in addition to unplanned pregnancies, since their partners are generally older, further contributing to power imbalances and girls' inability to negotiate contraceptive use 17,19 . Focus group meetings also confirmed girls' apprehension to say "no" to their partners if asked to have unprotected sex, under fears of being accused of unfaithfulness. Reluctance to use hormonal contraceptives was not based on assumptions of adverse side effects 9 , as emotional immaturity 34 was initially used to justify very young girls' pregnancies. However, throughout the process of data collection, it became clear that the lack of information about their sexuality and reproductive system was a major determinant of unplanned pregnancies, especially for 13-year-olds and those who became pregnant soon after their first intercourse. Dialogues about sex and contraception are inexistent within the family domain, and the formal knowledge acquired at school doubtfully covers all their queries 20,22,23,35 . Polit & Kahn 36 , Manlove et al. 24 , and Raneri & Wiemman 37 mention subsequent fertility and high parity as a problem related to early motherhood, inhibiting career prospects for the mothers and helping perpetuate poverty. Such subsequent fertility was not present in the analyzed sample. On the contrary, only two girls deliberately decided to have one more child after giving birth in their teens, agreeing with their desired family size. For all the others, the decision to have the IUD implanted was made soon after childbirth However absent in the beginning of their sex lives, contraception was widely used to space births once the first child was born, showing no apparent reluctance to the use of modern contraceptive methods. Although the problem of unwanted pregnancies is resolved by using the IUD, exposure to STDs and HIV/AIDS is not. Most of the participants got pregnant from casual partners, and another unplanned pregnancy seemed to be their biggest concern in not using condoms. High incidence of STDs, as portrayed by the local doctor, and the continuous practice of unprotected sex can pose a serious threat to sexually active adolescents and young adults 5 in Vila Novo São Lucas. Obviously, there are other aspects of teenage pregnancy that were not covered by the scope of this study, such as sex work performed by street children in Brazil 38 . These marginalized teenagers possibly represent a significant proportion of unwanted or unplanned pregnancies, as well as STDs. Such cases, although distinguished from the ones covered by this study, also deserve further investigation. One aspect worth mentioning is the fact that abortion is illegal in Brazil and can still be a major cause of death for young girls who do not have access to adequate treatment and attempt to abort by themselves, as shown by Alubo 39 and Almeida & Aquino 29 , noting that in Brazil clandestine abortion clinics are not an option for economically underprivileged girls who, in turn, often have no option but to have the baby, though keeping fertility rates higher for poor teenagers as compared to teenagers of all social classes. Such discrepancy also deserves further investigation and could possibly point towards new teenage pregnancy rates across different social strata. Early childbearing is seen as a major contributor to truncated education, and is believed to reduce career prospects for young women 26 . However, I found it hard to establish a direct causal relation between school dropout and pregnancy. Their financial reality and the very fact that girls had to start working quite early in life to help in the household income seems to have triggered the high levels of school failure previous to pregnancy, thus contributing to their gradual demotivation towards schooling. Heilborn et al. 8 found that, in another shantytown in Belo Horizonte, a high percentage of young females who experienced early pregnancy had already left school when the event happened. Although such occurrence was not clear in the sample, it is often ignored in studies on early childbearing. Pregnancy might not be the ultimate reason for school interruption, as other possible reasons can also discourage girls from attending classes, e.g. long working hours, and motherhood representing a more immediate successful activity when school does not. Qualitative data collection and in-depth analysis of their circumstances prior to and after childbirth can help explain the ulterior motives for quitting or not trying to return to school. Finally, the ulterior motive for not using contraception and 'allowing' themselves to get pregnant seems to be the longstanding desire to become a mother, mentioned by 95% of the young mothers, and the sense of purpose and achievement associated with motherhood which corroborates Fustenberg Jr. 40 , Almeida & Aquino 29 , and Bessa 12 . In people's statements it becomes clear that working for one's children is more dignifying than working for oneself and having fun. Equally, the family's ambivalence in being initially upset over a pregnancy out of wedlock, but gradually supporting their daughters both financially and emotionally, indicates that, despite being recriminated, teenage pregnancy is related to Brazilian women's ideal of motherhood 8 and that, so far, these young parents have been following the traditional pattern of womanhood set out for them according to their local culture. --- Considerations The study shows that, for young Brazilian women, the ideal of motherhood and the sense of achievement it brings are a desired event in any woman's life, regardless of age. Despite their young age when giving birth for the first time, girls often mentioned the positive side of childbearing as a milestone in their lives, bringing responsibility and representing the beginning of adulthood. Pregnancy is thus seen as a desirable event that gives young mothers a purpose in life, and its social acceptance makes it legitimate -whether in or out of marriage -through parental support and the young woman's new role as mother. Pregnancy and motherhood are far from being a problem for girls in Vila Novo São Lucas. For them the benefits outweigh the problems potentially associated with early childbearing, which is seen as life following its natural course. Despite the great sense of achievement, pregnancy should not be the only avenue through which such girls can succeed. Financial strain pushes them to an early start in the job market, performing activities that require little education, making schooling an arduous and often unsuccessful task. Low educational attainment may result from poor living conditions for these girls, combined with limited prospects for pursuing further education and getting a better job. The girls' knowledge about reproduction and sexuality could be improved if parent-child communication were more frequent and unrestricted. However, the intergenerational inability to approach these issues in the domestic sphere can make this a difficult target. Alternatively, schools, mass media, and health workers could reach such girls in different spaces if programs were targeted to their specific needs and issues. Improving the entire community's economic well-being seems to be the first necessary step by the government in order to deal with poor career prospects and subsequent poverty. Viewing teenage pregnancy as merely related to lack of contraception, resulting in truncated education and poor economic prospects, is a shortsighted perspective that overlooks the social environment legitimating motherhood. It also ignores the girls' own ambitions, which may not go much further than what they know their realities can provide. As their economic reality makes working more necessary than schooling, their horizons remain limited and are likely not to extend beyond motherhood. --- Resumo Este artigo apresenta os resultados de um estudo socioantropológico com mulheres de uma comunidade de baixa renda em Belo Horizonte, Minas Gerais, Brasil. Usando métodos qualitativos, analisa a gravidez pela perspectiva das jovens mães e a contribuição de seu ambiente socioeconômico. Mostra o papel de diferentes atores nas decisões reprodutivas e identifica normas culturais contraditórias que recriminam práticas sexuais na adolescência, mas vê a maternidade como ritual de passagem para a vida adulta. Com base em críticas sobre a insuficiência de programas de planejamento familiar e das consequências para sua saúde e futuro econômico, este artigo argumenta que a maternidade é uma atividade bem sucedida por meio da qual essas jovens preenchem o ideal coletivamente reconhecido de se tornar mulher, além de carregar um senso de realização num ambiente onde a falta de oportunidades prevalece antes mesmo da ocorrência de uma gravidez. Adequação na instrução formal e melhores perspectivas econômicas são possíveis soluções para provê-las com metas que vão além da maternidade, porém dentro dos limites de sua realidade econômica. --- Gravidez na Adolescência; Anticoncepção; Planejamento Familiar

This paper presents the results of a socio-anthropological study with women from a low-income community in Belo Horizonte, Minas Gerais State, Brazil. Through the use of qualitative methods it looks at teenage pregnancy from the young mothers' perspective and the contribution of their socioeconomic environment. It shows the importance of different actors in their fertility decision and identifies contradictory cultural norms that recriminate teenage sexual activity while seeing motherhood as a ritual of passage to adulthood. Following criticisms of insufficient family planning programs and negative health and economic outcomes for the lives of youngsters, the paper argues that motherhood stands as a successful activity through which they fulfill the collectively recognized ideal of womanhood, also carrying a sense of achievement in an environment where lack of opportunities prevail long before pregnancy occurs. Improving formal knowledge and economic well-being are possible solutions to provide these girls with goals that go beyond parenthood while within reach of their economic reality.

Introduction A recent article mentions that rural Americans have limited access to health care 1 . The authors mention two reasons for this; the first being many Americans, especially in rural areas, are without health insurance, and the second being only 11.4% of the country's doctors practiced in rural areas in 2005, serving about 20% of the country's population . Family physicians distribute themselves in proportion to the population in both urban and rural areas but doctors of other specialties are much more likely to settle in urban areas. A study published in 2009 has shown that international medical graduates filled the gaps in requirement of primary care doctors in many rural areas but there were wide differences between various states in America 2 . The authors concluded that policies to address the mal-distribution of physicians in the United States should also consider the role of IMGs. Critical access hospitals are a federal Medicare category for isolated rural facilities with 15 or fewer beds and IMG physicians play an important and growing role in staffing these hospitals 3 . Canadian trained physicians migrate both nationally and internationally and IMGs play an important role in providing healthcare in physician losing locations in Canada 4 . The author concludes that IMGs will be needed in underserved locations for years to come. A study published in 2013 showed Caribbean IMGs had the highest proportion of physicians practicing in primary care specialties compared to all other categories of graduates 5 . They make an important contribution to the US primary care workforce. A study in Canada found that around 20% of Canadian IMGs obtained their medical degree from the Caribbean 6 . A study which examined US citizens who graduated from medical schools outside the US and Canada found that the Central American and Caribbean region graduated more physicians than any other region 7 . Thus Caribbean medical schools may play an important role in producing primary care physicians for the US and Canada. Xavier University School of Medicine is an offshore Caribbean medical school located in Aruba, kingdom of the Netherlands, admitting students from the US, Canada and other countries to the undergraduate medical course. Students complete the basic sciences in Aruba and do their clinical rotations in the US. The number of students in XUSOM at present ranges from eighty to a hundred students a semester with around twenty new students being admitted each semester, which is less than the number enrolled in the bigger Caribbean medical schools. In addition, the major shareholders of the school are a group of Indian Americans and a large proportion of the student body is of South Asian and Indian origin. There are also students originating from other countries in Asia. Most students in XUSOM are from middle and upper middle class socioeconomic backgrounds. We do not have information on the financial and social background of students in other offshore Caribbean medical schools but believe most of them are likely to be from a similar background. The school also offers a premedical program for high school students to prepare them for admission to the MD program. The premedical program runs for four semesters, while the preclinical MD program is of six semesters. Each semester is of 15 weeks duration. The studies mentioned in this article from the US and Canada have defined a Caribbean IMG as either a citizen from the US/Canada who had studied in an offshore Caribbean medical school or students from other countries who studied at these schools and eventually plan to practice in the US/Canada. For the purpose of this study 'IMGs from Caribbean medical schools' may be considered as undergraduate medical and premedical students from XUSOM who intend to practice in the US/Canada after graduation. We make the assumption that all our students intend to practice in the US/ Canada after graduation which may be an overestimation as not all students may do so. Student perception about working in rural US/Canada has not been previously studied in the institution. Hence the present study was carried out. The study was conducted among premedical and basic science undergraduate medical students to obtain basic demographic information, understand their perceptions regarding why doctors are reluctant to work in rural US/Canada after graduation and conditions to be fulfilled before they would consider working in a rural area, identify possible problems of working in rural areas and obtain possible solutions. --- Methods The study was conducted among Premedical 1 to 4 semester and undergraduate medical students at the Xavier University School of Medicine , Aruba during the month of October 2014. Students during the clinical years of study in the US and Canada were not included. Students were informed about the aims and objectives of the study and invited to participate. They were informed that participation in the study was voluntary --- Amendments from Version 1 In the revised version of the manuscript we have tried to modify the manuscript as per the reviewers' comments. We have tried to define what exactly is meant by a Caribbean IMG, have explained why no significant differences with regard to perception about working in the rural US/Canada according to respondents' demographics was noted, and clarified that students during the clinical years of the MD course were not included in the study. As requested we have provided more background information about the school, mentioned possible differences between Xavier University School of Medicine and other Caribbean offshore medical schools, highlighted possible initiatives offshore Caribbean medical schools may undertake to meet rural practice needs, and expanded on the limitations of the manuscript. We have also briefly explained certain areas of the analysis procedure as requested by the reviewer. We thank the reviewers for their comments. --- See referee reports REVISED a semirural area. The respondents in our study were diverse but no important difference was noted with regard to perspective about rural practice among different subgroups. The number of students was low and this may have been partly responsible. Also the study was not specifically designed to examine for differences in perception about working in rural areas among different subgroups of respondents. Table 2 shows important conditions which may need to be fulfilled before respondents would consider working in a rural area. and they were free not to participate. Written informed consent was obtained from each participant. The study was approved by the Institutional Review Board of XUSOM vide notification XUSOM/IRB/2014/05. The questionnaire used was modified from that used in a previous study on student perceptions about working in rural Nepal after graduation 8 . The questionnaire was modified to the US/Canadian context by the authors. Inputs were also obtained from selected faculty members of the institution. The questionnaire used in the study is shown as Supplementary File 1. Information about the semester of study, gender, nationality, place of family residence and occupation of parents were noted. Students' responses to different questions were noted. Common responses were tabulated. Among the topics studied was whether the strident had lived in a rural area before joining XUSOM, whether they plan to work in rural US/Canada after graduation, and if yes whether they plan to do so immediately after graduation/residency or later in their career. They were also requested to state three important reasons why doctors are reluctant to work in rural areas, and also their opinions regarding how governments can encourage doctors to work in rural US and Canada. Students were also asked whether they felt their medical curriculum prepared them adequately for rural practice and if they felt this was not the case, what modifications they felt were necessary in the medical curriculum. The data was analyzed using descriptive statistics. The percentage and number of respondents mentioning a particular response to a specific question was noted. Their opinion regarding whether students were aware of the problems of life in rural areas, problems which they anticipate in dealing with the local population, their knowledge of initiatives to address the shortage of doctors in rural US/Canada and in other countries and whether poor return of investment was a factor hindering doctors from working in rural areas was noted. The data was analyzed manually by the authors. The demographic information was analyzed using Statistical Package for Social Sciences version 20 for windows. --- Results Ninety-nine of the 108 enrolled students participated in the study. Table 1 shows the demographic characteristics of the respondents. Twenty-eight of the 34 premedical students and 71 of the 74 MD students participated. The number of male and female respondents was approximately equal and 28.3% of respondents were from rural areas. The majority of the student's parents worked in non health related professions. Forty respondents were in favor of working in rural US/Canada after graduation. Ten respondents were not sure about whether they planned to work in rural US/Canada at this initial stage. With regard to the duration for which they planned to work in a rural area, 12 stated for most of their career or indefinitely while nine respondents stated for 5 to 10 years while four respondents stated between 1 to 5 years. Eighteen respondents preferred to work in a semi urban or semirural area while eight preferred a rural area. However neither the authors nor the respondents defined what exactly was meant by a rural or Respondents were concerned about housing, regular supply of electricity, internet and telecommunication facilities, and good quality water supply. Family-related factors were mentioned as important and were related to whether the doctor would be able to bring their family to the location, the place should be not very far from his/her hometown, and there should be facilities for education of children. When respondents were referring to proper working conditions they were mainly talking about adequate diagnostic and treatment facilities in the hospital or clinic where they would be working. Respondents also mentioned availability of treatment facilities for themselves and their family if they fall ill is important, and that the practice area should not be extremely cold, the community should welcome individuals from different backgrounds, absence of racism, provide opportunities for personal growth and help with repayment of student loans as necessary conditions. A respondent [Participant 83] mentioned, "I have lived in a rural area for almost my whole life. I have seen how we have to travel to cities for basic medical services and to help make a change in this situation I would like to work in a rural area". Forty-four respondents mentioned that they plan to work in a rural area immediately after graduation while 23 respondents mentioned that they would prefer to do so later in their careers. Table 3 shows important reasons why doctors are reluctant to serve in rural areas according to the respondents. Among other reasons mentioned were issues of safety, isolation, slow pace of rural life, racism and greater potential for career and personal growth in an urban area. A participant stated, "Rural practice is mostly a GP job and most future doctors want to be specialists . Away from home . Looks 'boring'". With regard to how the government can encourage doctors to work in rural US/Canada important initiatives/measures mentioned were providing greater benefits and salary to rural doctors as compared to doctors practicing in urban areas , loan forgiveness or help with repayment of student loans , improving the facilities at rural and community hospitals , provide opportunities for professional growth and development for rural doctors , and emphasize 'educating doctors for rural areas' . Four respondents mentioned making certain duration of rural service mandatory as has been done in many developing nations. A respondent [participant 98 ] stated, "Make it mandatory for graduates to serve in rural areas and have them serve at least a year or two". Seventy-one respondents felt their medical training adequately prepared them for a career as a rural physician while 19 respondents were of the opinion that it did not do so. Many respondents did not answer the question regarding suggestions for re-focusing medical education for preparing doctors for rural practice. Only fourteen respondents answered this question. Among the suggestions mentioned were educating students on how to treat and manage medical conditions in a resource constrained setting, having preceptors with rural experience, initiating clinical exposure for students in rural areas and conducting sessions on cultural diversity and cultural issues in rural areas. When asked whether medical students in the institution were aware of the problems of life in rural areas 43 students were of the opinion that they were aware while 39 respondents stated that students were not aware. Regarding how the awareness can be created 19 respondents mentioned increasing rural exposure of students, while three respondents stated sessions by rural doctors should be conducted. Among other suggestions were sessions on this topic during the orientation program, medical humanities sessions and talks by rural community leaders. Regarding potential problems which students anticipate while dealing with the rural population fourteen respondents mentioned difficulty in being accepted by the rural community, while eight respondents mentioned that the rural population may be less educated which may lead to problems in their acceptance of treatments and advice provided by the doctor. Eight respondents opined that they did not have much awareness of and exposure to rural life which may lead to problems in adjusting to rural practice. Culture gap, fewer available resources and differences in lifestyle were also mentioned. Racism and language barriers were also mentioned. Twenty-four respondents mentioned they were aware of initiatives to address the shortage of rural doctors in the US and Canada but specific initiatives were not mentioned. Fifty-seven respondents mentioned they were not aware of these initiatives. Among the initiatives mentioned were debt forgiveness, return of service obligations, flying doctors and encouraging IMGs to practice in rural areas. Fifty-six respondents felt poor return on investment was a factor preventing doctors from working in rural areas while 17 mentioned that this was not a factor. Respondents mentioned recovery from medical school debt requires a job/position which pays well, the world is money oriented, doctors have to study for a long time before they can make money, there are fewer opportunities for making money in rural areas and urban areas provide greater opportunities. A respondent stated, "Doctors are always in debt after med school. They do not want a poor return. They want progress --- Discussion Only about 30% of respondents were from rural areas. Around 41% of respondents were in favor of working in rural US/Canada after graduation. Many respondents however preferred to work in a semirural area. Good living conditions, family related factors, welcoming community were among the conditions necessary to be fulfilled. Issues of safety, isolation, lesser potential for personal growth were among the reasons mentioned why doctors were reluctant to serve in rural areas. Difficulty in being accepted by the community, lack of exposure to rural life, cultural issues were mentioned as potential problems for rural practice. Like in the Nepalese study 8 , the majority of the respondents were from urban areas. Students from countries other than the US or Canada planned to work in US/Canada after graduation. Caribbean offshore medical schools offer a similar curriculum to US medical schools and students complete their clinical rotations in the US. Most offshore medical schools admit three groups of students a year and provide an increasing contribution to the US health workforce 9 . Like in Nepal, most respondents willing to work in a rural area preferred a semi urban/semirural area as their work location. The conditions that need to be fulfilled were also similar to the Nepalese study. In Tanzania, increased salaries, hardship allowance, decent housing, good infrastructure, and offering continuing education after a period of service were mentioned as factors making rural jobs more attractive to health workers 10 . In a survey conducted among healthcare professionals in the Scottish highlands it was found that a rural background, perceived ease of access to children's education, access to job for spouse and healthcare were important factors influencing rural location 11 . Professional isolation was an important issue for those working in rural areas. In a study conducted among occupational therapy students in Australia it was seen that rural fieldwork experience and the influence of fieldwork supervisors were important positive factors for rural practice while the desire to be a specialist, and lack of professional development opportunities in a rural area were negative factors 12 . The desire for specialization was also mentioned as a negative factor in the present study. In the state of North Carolina in the US incentives such as loan repayment programs, salary guarantees and practice assistance for rural physicians were important factors in attracting primary care physicians to rural areas 13 . Having grown up in an area with a population of less than 11000 was highly predictive of rural practice. A study conducted in the US found that debt forgiveness, financial incentives and wage guarantees may increase participants' interest in rural practice 14 . This was similar to that reported in our study. In the US loan repayment and direct financial incentive programs were successful in recruiting and retaining doctors in needy local communities 15 . Family-related factors were mentioned by respondents as influencing their decision regarding working in rural areas. In the US study 14 it was noted that medical students with a rural background and with spouses and significant others having a rural background were more likely to practice in a rural area. In a Turkish study it was noted that respondents who were born in an underdeveloped region of the country or had lived for a significant amount of time in an underdeveloped region and with a lower background family income were more likely to work in a rural area 16 . Higher salaries was the most influential factor for working in these areas. In XUSOM, a large percentage of students, although they are US or Canadian citizens, are of Asian origin. A large percentage of them are from cities and have minimal exposure to rural life and they anticipate problems in being accepted by rural communities. Medical schools in developed nations with a mandate to create rural doctors are providing a greater proportion of decentralized teaching-learning to students. Students stay in small groups in rural communities and lessons are delivered using the internet and other media. At the University of Calgary in Canada, two separate preclinical programs were taught at two separate rural sites and no significant difference in examination scores were noted between students at the rural sites and students learning at the main university 17 . At the University of Missouri School of Medicine in the US the summer community program was developed in which second year medical students work alongside rural, community-based physician preceptors 18 . The program positively influenced student perspective of rural practice and lifestyle and increased their interest in rural practice. In the US state of Kansas, the Scholars in Rural Health program is designed to attract and retain young rural Kansans with a high probability of successful careers in rural communities. Scholars accepted into and satisfactorily completing this program are admitted automatically to the School of Medicine. The program has shown success in maintaining a pipeline of doctors for rural Kansas 19 . At the University of Alabama in the US, fifteen years of operation of the rural health leaders pipeline was studied 21 . Rural students are targeted at multiple levels ranging from elementary school through residency. Puppet shows highlighting different health professions are conducted in elementary schools, a rural health schools program is offered to 11 th grade students, a minority rural health pipeline program is offered to college students and a rural medical scholars program and assured admissions to family medicine residency are offered to medical students. A recent article analyzing rural medical education in the US, Canada and Australia identified three types of medical schools, mixed urban/ rural schools, defacto rural schools and stand alone rural schools 21 . These schools all adopted a pipeline approach to meeting the need for rural doctors focusing on: early recruitment; admissions; locating clinical education in rural settings; rural health focus to curriculum; and support for rural practice. In Australia, the Broken Hill University Department of Rural Health was established to improve health care in far-western New South Wales and offers an Enhanced Rural Inter-professional Cultural Health program to health science students designed to deepen the medical student experience 22 . The program uses community resources, professional, cultural and artistic to provide stimulating educational resources. Offshore Caribbean medical schools are privately run for-profit institutions. However, the cost of education at many Caribbean medical schools may be lower compared to their US counterparts. Most Caribbean medical schools at present to the best of our knowledge do not have specific incentives and programs to create physicians for rural areas. A large percentage of our students were from urban areas which may also be true for other Caribbean schools. The US and Canadian governments can offer student loans to their citizens who study in Caribbean medical schools in return for rural practice obligations. At present certain accredited schools are eligible for student loan support. They can also offer residency and other rights to foreign citizens studying in these schools in return for their willingness to practice in rural US/Canada. The Caribbean medical schools can develop closer links with the country's health system to offer their students increased opportunity for early clinical exposure. Caribbean offshore schools may not have significant financial incentives to produce rural doctors unless specific financial incentives are provided by the US and Canadian governments. We have come across very few studies regarding the rural practice intentions of Caribbean medical students. Further studies in this area are required. We have not been able to obtain studies regarding the career intentions of graduates of offshore Caribbean medical graduates. A study conducted at the University of West Indies in Jamaica had shown that two-thirds of those who planned to work in a rural health facility planned to do so immediately after graduation but most did not plan to work in these facilities for more than five years 23 . A study conducted in California had shown that IMG holders of temporary visas had the highest obligation to serve in health professional shortage areas 24 . A similar phenomenon was noted in the present study with many students from countries other than the US and Canada expressing willingness to serve in areas of physician shortage in return for residency rights. XUSOM also offers scholarships to students from Caribbean community countries . Providing rural exposure on the island of Aruba with well developed infrastructure and health system is challenging. Among the affiliated US hospitals where students do their clinical training most are in urban areas but community hospitals are also included in the list. Students graduating from XUSOM may be interested in serving in rural US/Canada if appropriate incentives and opportunities for professional growth are offered. The high response rate was the strength of the study. Also a variety of responses were obtained in response to the questions. The study also had limitations. Student perception about working in rural US/ Canada was obtained using a questionnaire which had been used in a previous study in Nepal. The questionnaire was modified to suit the present context, and feedback about the modified questionnaire was obtained. The findings of the present study were not triangulated with information obtained using other modalities. Respondents' awareness about possible changes in medical education and curricula to produce doctors for rural areas and about specific initiatives underway in the US and Canada to address the shortage of rural physicians was low. Though the overall response was high the response to certain questions was low. The study was limited to premedical and preclinical students and was not conducted among students during the clinical years of the course due to logistical difficulties. Seventy-one students from the basic science years participated in the study and the school had around 65 students in the clinical years of study. Their responses were not studied which could affect the generalizability of the study findings. --- Conclusions Most respondents were from urban areas. About 40% of respondents were in favor of working in rural US/Canada after graduation. Most were however considering working in a small town or a semi-rural area not far from a big city. Like in previous studies good living conditions, family related factors, incentives were among conditions necessary to be fulfilled before students would consider working in rural areas. Most respondents considered reduced opportunities for growth, their lack of exposure to rural life, and cultural issues as hindering factors. Students from offshore Caribbean schools could play a role in partly addressing the physician shortage in rural US/ Canada provided the right incentives and growth opportunities are offered. Similar studies are required in other offshore Caribbean medical schools. The study's respondents are diverse, as shown in Table 1 -including "premedical" and MD students, different nationalities, etc. It would be informative to see more comparison in the analysis, to see if demographic characteristics are related to perspectives on rural practice. There is some mention of this in the Discussion but it is not detailed in the Results. It would be helpful if the authors would clarify whether all enrolled premedical and MD students were invited to participate in this study. The description in the Methods suggests that this is the case but it is not explicitly stated how many students were enrolled at the school at the time of the study, including a breakdown by premedical and MD. This information would be helpful, as well as confirmation that the study respondents are representative of the overall student population. Also, some more background on the school, including how it is comparable to and how it differs from other Caribbean schools would be helpful for judging the generalizability of the study results. Some of this information is in the Discussion. The authors note that "Many respondents did not answer the question regarding suggestions for re-focusing medical education for preparing doctors for rural practice." It would be helpful to know exactly how many respondents there were to this question, particularly given its importance. It is understandable that the authors would want to be cautious about stating results from this question given the limited response. However, while stating this caution, the authors may want to offer in the Discussion possible actions that Caribbean medical schools might take to begin to address rural practice needs, including direction for future research. The low question response rate might also be noted as a study limitation. --- Competing Interests: The lead author is a graduate of and has been program faculty for the FAIMER fellowship programs. 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. their clinical postings in geographically diverse locations across the US and Canada. This has been mentioned as a limitation of the study on page 10, last paragraph. Comment: Also, some more background on the school, including how it is comparable to and how it differs from other Caribbean schools would be helpful for judging the generalizability of the study results. Some of this information is in the Discussion. We have provided this information on page 3, paragraph 3. As the reviewer has mentioned some of this information is also in the discussion. XUSOM has lesser number of students and a greater proportion of students of South Asian and Asian origin compared to other bigger and older schools. --- Comment: The authors note that "Many respondents did not answer the question regarding suggestions for re-focusing medical education for preparing doctors for rural practice." It would be helpful to know exactly how many respondents there were to this question, particularly given its importance. Only 14 respondents answered this question. We have mentioned this in page 7, paragraph 1. Comment: It is understandable that the authors would want to be cautious about stating results from this question given the limited response. However, while stating this caution, the authors may want to offer in the Discussion possible actions that Caribbean medical schools might take to begin to address rural practice needs, including direction for future research. We have briefly mentioned some of these initiatives on page 10, paragraph 2 of the manuscript. --- Comment: The low question response rate might also be noted as a study limitation. We have mentioned this in the limitations of the manuscript on page 10, last paragraph. --- Competing Interests --- Anthony David 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] --- Data availability F1000Research: Dataset 1, 2 and 3. Demographic characteristics and responses of student respondents, 10.5256/f1000research.5927. d40411 25 --- Student perception about working in rural US/Canada after graduation questionnaire. Modified questionnaire used in the study "Student perception about working in rural US/Canada after graduation". Click here to access the File. Author contributions PRS, AKD, and AN conceived the study. BLH and BWL provided inputs during the design of the study. PRS, AKD and AN conducted the study. PRS was involved in analyzing the data. AKD, AN, BLH and BWL helped in interpretation of the data. PRS wrote the manuscript with inputs from the other authors. All authors were involved in the revision of the draft manuscript and have agreed to the final content. --- Competing interests No competing interests were disclosed. --- Grant information The author declared that no grants were involved in supporting this work. --- Open Peer Review Author Response 13 Apr 2015 Pathiyil Ravi Shankar, International Medical University, Kuala Lumpur, Malaysia We are submitting the revised version of the manuscript titled 'Student perception about working in rural United States/Canada after graduation'. The manuscript has been revised in the light of the reviewers' comments. Response to specific comments is as follows: S Friedman: We thank the reviewer for her comments. We have explained what we meant by 'IMGs from Caribbean medical schools' on page 3, paragraph 4. The studies we have mentioned from the United States and Canada had defined a Caribbean IMG as either a citizen of the US or Canada who had studied in an offshore Caribbean medical school or students from other countries who had studied at these schools and eventually plan to practice in the US/Canada. Comment: The study's respondents are diverse, as shown in Table 1 -including "premedical" and MD students, different nationalities, etc. It would be informative to see more comparison in the analysis, to see if demographic characteristics are related to perspectives on rural practice. There is some mention of this in the Discussion but it is not detailed in the Results. No important difference with regard to perspective about rural practice was noted among different subgroups of respondents. The number of students was small which may have been partly responsible for not detecting differences. Also the study was not designed to statistically examine for differences among subgroups. This has been mentioned on page 5, paragraph 1 . Comment: It would be helpful if the authors would clarify whether all enrolled premedical and MD students were invited to participate in this study. The description in the Methods suggests that this is the case but it is not explicitly stated how many students were enrolled at the school at the time of the study, including a breakdown by premedical and MD. This information would be helpful, as well as confirmation that the study respondents are representative of the overall student population. In the Methods section we have clarified that only students enrolled during the basic science years of the MD program and in the Premedical program were included in the study. In the first paragraph of the Results section we have provided the number and percentage breakdown. Students during the clinical years were not included as they do Department of Physiology, Deccan College of Medical Sciences, Hyderabad, India The title could be changed a little. The characteristics of the students, rural or urban also was studied and that is not reflected in the title. The study as such is well researched and clear. The findings have been well analyzed and presented in the paper. The abstract reflects the paper well. The explanation of the study design & methods is clear. A little more explanation on the analysis would be useful. The conclusions are well justified by the results. The data is useful for replication. Overall this paper can be indexed with a few minor modifications as suggested. 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. --- Author Response 13 Apr 2015 Pathiyil Ravi Shankar, International Medical University, Kuala Lumpur, Malaysia Dr David We thank the reviewer for the comments. Even though we have studied the demographic characteristics of the student respondents the study was not designed with the objective of comparing perception about working in rural areas among different groups of students. Hence we feel a change of title may not be required as with a change in title readers may expect us to study differences in perception among subgroups of respondents. We have explained the analysis towards the end of the 2 nd paragraph of page 4. We used mainly descriptive statistics and numbers and percentages. We thank the reviewer for the other comments Competing Interests: None declared

Introduction: Rural residents of the United States (US) and Canada face problems in accessing healthcare. International medical graduates (IMGs) play an important role in delivering rural healthcare. IMGs from Caribbean medical schools have the highest proportion of physicians in primary care. Xavier University School of Medicines admits students from the US, Canada and other countries to the undergraduate medical (MD) course and also offers a premedical program. The present study was conducted to obtain student perception about working in rural US/Canada after graduation. Methods: The study was conducted among premedical and preclinical undergraduate medical (MD) students during October 2014. The questionnaire used was modified from a previous study. Semester of study, gender, nationality, place of residence and occupation of parents were noted. Information about whether students plan to work in rural US/Canada after graduation, possible reasons why doctors are reluctant to work in rural areas, how the government can encourage rural practice, possible problems respondents anticipate while working in rural areas were among the topics studied. Results: Ninety nine of the 108 students (91.7%) participated. Forty respondents were in favor of working in rural US/Canada after graduation. Respondents mentioned good housing, regular electricity, water supply, telecommunication facilities, and schools for education of children as important conditions to be fulfilled. The government should provide higher salaries to rural doctors, help with loan repayment, and provide opportunities for professional growth. Potential problems mentioned were difficulty in being accepted by the rural community, problems in convincing patients to follow medical advice, lack of exposure to rural life among the respondents, and cultural issues. Conclusions: About 40% of respondents would consider working in rural US/Canada. Conditions required to be fulfilled have been mentioned above. Graduates from Caribbean medical schools have aApproval Status 1 2 version 2 (revision)

Introduction The role of privacy in information systems design has been well established in numerous domains, including the one that is the focus of this paper, namely the blockchain . Information management and governance, ensuring data is tamper proof and managing business risks related to privacy have created challenges for many companies, because information is seen as a strategic asset. A blockchain is a tamper-proof and distributed data repository that supports information management and governance through automating trust, thereby reducing transaction costs; for these reasons, blockchain technology is of advantage in business . However, blockchains suggest challenges for privacy and the aim of this paper is to explore some of these challenges. Privacy has been conflated with security and confidentiality and this ontological vagueness obscures all three concepts and curtails clarity in research. In this paper, a clear definition of privacy enables us to establish a stronger position on privacy with respect to the blockchain, utilising the example of Holochain. Privacy is important because it supports freedom , dignity , autonomy , justice , and democracy . Westin considers that "… privacy is a quality of life topic worth the best scholarship, thoughtful advocacy, and continuing attention of us all" . The literature on privacy encompasses diverse themes : control of data and self-determination restricting access to self and data , privacy and data as commodities that may be traded , privacy as a social good differing from context to context , and the view that privacy takes shape according to the technologies forming the infosphere . Data privacy has been seen as distinct from physical privacy and while the distinction affords analysis , it obscures the foundation of privacy: social context. As Parent put it, "a lonely man isolated on a desert island could hardly be expected to cherish his privacy. So, we serve no useful or constructive purpose in ascribing it to him" . It is social context that gives rise to meaningful privacy, regardless of whether that context is the blockchain or in a grocery store. Here, privacy is understood to be an intrinsic and pliant feature of social contexts, shaped by social contexts and to a lesser extent, shaping of social contexts and therefore we define privacy as Privacy is a highly contextual social good which is instrumental to the perpetuation of other social goods. People exercise privacy through personal information preferences and practices which are specific to social contexts and changing over time. This view is consistent with those held by others. Gavison states "My purpose is to point out the many contexts in which privacy may operate ..." . One of three elements in Solove's work on conceptualising privacy is "a recognition of context and contingency" and in a later paper, Solove considers the question "How can privacy be addressed in a manner that is non-reductive and contextual, yet simultaneously useful in deciding cases and making sense of the multitude of privacy problems we face?" . Finally, Nissenbaum details a framework for revealing the integrity of privacy within specific contexts. According to Panichas , "Privacy has undeniable implications for persons' abilities to define themselves and to command the requisite degree of respect for their dignity and autonomy." The right to be forgotten is a human right established when the European Court of Justice ruled against data controllers in Google v Spain1 . Today, the right to be forgotten exists in Europe and Argentina. It acknowledges that one's conduct in one social context ought not to impact on one's success in other social contexts. For example, youthful drunkenness ought not to damage future access to employment opportunities. The right to be forgotten may require data controllers to remove data from indexes. Therefore, the right to be forgotten supports privacy in online contexts. Similarly to privacy, socio-technical IS has features of social contexts, perhaps not as pliant as privacy, but shaped by social contexts and shaping of social contexts. The bitcoin blockchain is an example of technology that was shaped by social contexts in which economic transactions occur. --- Limiting privacy through a Blockchain Blockchain is considered as a data structure that can provide the capabilities of a ledger with three main properties : 1. Consensus: All participants must be agreed on what data should reside on blockchain 2. Persistence: once data has been added to the blockchain it can never be removed and 3. Liveness: anyone can add data to the blockchain and it is not possible to prevent anyone from writing new data so system can never get stuck. Blockchain was originally intended to digitally timestamp documents to make them tamperproof. Blockchain is considered a transparent, time-stamped and decentralized system as it provides digital trust by recording information in a public space and is tamper-proof . In a blockchain, a block is a cryptographic hash of the previous block's identifier, a timestamp, and transaction data . When a new block is added, it encapsulates the previous block and is itself later encapsulated by subsequent blocks . This ensures all blocks in a blockchain are in accord; blocks may be added to the ends of a blockchain, but once added, blocks are immutable . In other words, a blockchain can be added to and read, but it cannot be amended. If it were possible to change a block, the entire chain would fail because the block's encapsulation would no longer accurately represent the previous and subsequent blocks in the chain. Therefore, data cannot be changed within or removed from blockchains. One of the major issues with blockchains is the lack of privacy of transaction data. One of the primitive requirements is to prevent double-spending attacks . To do that, a user is required to reveal some information for authentication. Each computer on the network receives a record of every single transaction and update, and each computer validates these transactions. The transactions contain information like sending account, receiving account, amount, and any other details that are required for validation. In some applications, it is not acceptable for all transactions to be revealed to all participants in real time. These problems are motivating privacy as an emerging research topic in the study and development of blockchain technology. Let us review privacy concerns in the context of bitcoin financial transactions in between two business parties and review how solutions to such concerns are evolving in with respect to blockchain technology. So, if we look at initial way of bitcoin transactions, we can see the source addresses are written on the blockchain as well as the all the amounts, and information about payer and payee. Even though due to cryptology, it is directly difficult to figure out the physical identify associated with these addresses, there are companies that analyse the blockchain and map these addresses to physical entities. It is a problem for business payers, payee and amounts are revealed, making the blockchain less trusted for business payments. For instance, if costs in supply chain are known to everyone, a business cannot gain a competitive advantage. One idea to deal with such privacy concerns is confidential transactions . The idea is rather than writing amounts in the clear, system can replace amounts with commitments to the amount. A commitment is a way to represent data in a way that no one knows what the data is but the person who wrote the data no longer can modify it. So, they are committed to a particular value. Even though no one knows what the value is: Like Pederson commitment : Commit=gv.hr where, r is random. We put value v as an exponent and we blind it by random r. Unfortunately, we lose the opportunity to verify the transaction. Normally, when we verify, we check whether the number of inputs is equal to the number of outputs in a transaction plus the fee for the miner. However, a solution named zero knowledge proof is used to counter it. It means that payee attaches zero knowledge proof to the transaction. Some blockchain systems like zcash and Monero have implemented transactions where not only the amount is hidden but also payee and payer information is hidden. Such systems use techniques such as ZnSNARK . The right to be forgotten enshrines a legitimate right to request that personal information be erased from the Internet, so that it cannot be found by search engines. In contemporary digital history, it is associated to the case of Google Spain SL, Google Inc for request by Mario Costeja González. In the 1990s, González had financial debts that were reported by an online newspaper and later resolved. Years later, González requested his past to be forgotten, but the internet would not forget. On May 2014, the European Court of Justice ruled that Google had an obligation to remove links to González's personal data that was no longer correct . It led to the European Union law referred to as the right to be forgotten. However, it has serious implications for corporate burden and access to information. After this decision, Google received 41,000 requests for data to be forgotten and in 2015, requests were reported to be around 1,000 a year . In response, Google has removed various URLs from its search results. The right to be forgotten is a perpetual burden on corporate resources and is considered very hard in implementation terms. It also suggests questions regarding the scope of European law at global scale. Finally, it highlights the enduring tension between privacy and utility, as it invites consideration of whether the benefits of privacy outweigh the costs to web indexing organisations. While the outcome of such consideration is a clear 'yes' in the deliberations of the European Court of Justice, the question remains open in other jurisdictions. Blockchain enables the decentralised web by relying on a network of computers for distributing data . Each computer can act as a node, with power and memory on a distributed storage network system. The data is not stored in any one storage and therefore there is no central point to hack and control. This peer-to-peer infrastructure model of nodes is similar to a blockchain's distributed ledger and therefore it could be the answer to create a decentralized web. If we reimagine the right to be forgotten, it becomes highly infeasible from the implementation point of view. One reason is the highly decentred nature of the next generation of the Internet as no one would have the central authority on data. As a data controller makes personal data public, exercise of the right will also place responsibility upon a node to take reasonable steps, to inform other controllers of any erasure request. To comply with this responsibility, data controllers will have to take all technical measures based on the available technology and the cost of implementation. In similar terms, other world states are contrary to the right to be forgotten like United States and Australia where access to the information is considered linked to the free speech and freedom . --- Holochain privacy There have been various suggestions for supporting privacy , including attempts to leverage blockchain technology to provide privacy . One such attempt is Holochain, which provides "Individual authority over data sharing, access, and storage" . Holochain is a platform for which apps are developed . hApps provide for various distributed and transparent economic exchanges of value and examples include energy trading, farm produce trading, village resource-sharing, agile project management, publishing, heterogeneous IoT, social media, and event and disaster management2 . In other words, Holochain enables distributed, server-less apps running on devices with localized data storage and management. For this reason, Holochain is disruptive of client/server platforms and cloud technologies. A hApp "… consists of a network of agents maintaining a unique source chain of their transactions, paired with a shared space implemented as a validating, monotonic, sharded, distributed hash table where every node enforces validation rules on that data in the DHT as well as providing provenance of data from the source chains where it originated" . In less compact terms, in a hApp each node creates its own blockchain for holding verified data which is shared to a DHT . The DHT is sharded and distributed back to the nodes in the hApp , which validate the shards against their source blockchains . An inherent limitation of any blockchain is its computation and communication cost, which makes a blockchain less efficient as it gets bigger. A hApp is more scalable than a blockchain because validating shards of DHT at nodes is a more efficient validation process than validating an entire blockchain. As a blockchain gets bigger, it gets less efficient and more resilient; but as a hApp gets bigger, it gets more efficient and more resilient. Nodes in a hApp may be controlled by agents, and an agent may be a person. In this way, Holochain does indeed support "individual authority of data sharing, access, and storage." Also, Holochain transfers power from centralised data controllers such as Facebook to individual agents. However, authority over data sharing, access, and storage is not always consistent with privacy. According to the definition of privacy adopted in this paper, privacy may take different forms in different contexts. For example, in one such context, authority over data sharing may be a privacy preference and practice for many people, but not others. In another context, authority over the sharing of sensitive data may not be practicable; for example, when a patient loses consciousness they can no longer exert authority over the sharing of sensitive data related to their health, yet it may be in the patient's best interest that the data is shared. Finally, there are many social contexts in which someone's privacy preferences and practices dictate that their authority over data sharing be waived . As it is impossible to specify the privacy preferences and practices for every social context, so it is impossible to precisely state that Holochain provides privacy. However, it is clear that Holochain supports privacy for a subset of social contexts. The contexts in this subset are to be determined by Holochain's users as they develop and use hApps. For this reason, we suggest that it will be one of many privacy tools deployed by those with an interest in exercising their privacy preferences and practices. 4 Holochain privacy and the right to be forgotten One form of privacy which is likely to be consistent with Holochain's features is the right to be forgotten. This right applies to data indexed by search engines, that is, it applies to the web. If a search engine indexed Holochain nodes, then the data returned to web search queries would be under the control of the Holochain agents. In cases where the agent is an individual person who wants to exercise the right to be forgotten, they may re-create the small blockchain at their own node making sure to prevent the sharing of the data which is to be forgotten. Then, the agent's small blockchain is re-shared with the Holochain's DHT, which is resharded, and re-distributed to the nodes for verification. If such a Holochain web search engine existed, its computational and communication costs are likely far less than a blockchain deployed for this purpose. However, these costs would be greater than we observe in today's web search engines. Our point remains that a Holochain platform is consistent with the right to be forgotten. --- Conclusion Privacy needs to be studied in an ontologically robust way, in relation to blockchain and in relation to any other technology or social context. It is important because it is mutually shaping of freedom, dignity, autonomy, justice, and democracy. It encompasses diverse themes including the control of data and self-determination, restricting access to self and data, privacy and data as commodities that may be traded, privacy as a social good differing from context to context, and the view that privacy takes shape according to the technologies forming the infosphere. Blockchains are technologies that were shaped by, and are now shaping of, social contexts in which economic transactions occur. As noted above, Holochain is disruptive of client/server platforms and cloud technologies. Whether Holochain's data control and computational advantages are sufficient to disrupt data is the new oil business models is a matter of conjecture, but for privacy, Holochain hApps offer significant advantages over blockchains. Privacy and data protection laws around the world represent a real compliance challenge for public and private distributed implementations of blockchain technology; and Holochain platforms have the potential to meet these challenges.

Privacy is important because it supports freedom, dignity, autonomy, justice, and democracy, and therefore it is important that privacy is studied in ontologically robust ways. A form of privacy is implemented in the right to be forgotten, which is a human right established by the European Court of Justice. Blockchain and Holochain are examples of recently emerged technologies that were shaped by, and are now shaping of, social contexts in which economic transactions may occur. The right to be forgotten represents a compliance challenge for public and private implementations of blockchain technology. This paper describes a few of these challenges.

Introduction Adolescents and young adults remain disproportionately burdened by STI/HIV, accounting for over half of new sexually transmitted infections , 1 and over one in four new HIV infections. 2 Young women in particular are highly affected; those ages 15-24 demonstrate the highest rates of Chlamydia and gonorrhea, 1 with over 750,000 cases of Chlamydia reported for U.S. women ages 15-24 in 2011 alone. 3 Young women also experience the highest rates of physical and sexual intimate partner violence , [4][5][6] , with recent national data confirming that over two thirds of those exposed to partner violence report that the first such incident occurred before the age of 24 years. 4 IPV has emerged as a consistent STI/HIV risk factor for both adolescent and adult women; [7][8][9] prospective research strongly indicates a causal linkage of IPV with incident STI/HIV infection. [10][11][12] Associations of IPV with sexual risk behavior, including unprotected sex and multiple and concurrent partnering, are, in part, responsible for the links of violence with STI/HIV. For example, qualitative and quantitative data alike demonstrate that IPV and the threat of abuse can undermine women's ability to ensure successful condom use and refuse unwanted sex. 8,[13][14][15] While relatively understudied as a correlate of violence, heterosexual anal intercourse has emerged as a prevalent HIV risk factor; 16 its transmission efficiency 17 affirms the need to similarly evaluate potential links with IPV. To date, gendered power imbalances, coercive and violent dynamics have been implicated in heterosexual anal intercourse. 13,1415,18,19 Violence is also closely associated with substance use, including injection drug use, [20][21][22][23] which both directly causes HIV infection and increases risk through unsafe sex that is more likely in the context of substance use. Violence can be particularly influential for drug involved women, as abuse and the threat of violence undermine both sexual and drug-related HIV prevention. 24,25 So too, women may turn to substance use, including injection drug use, as a maladaptive coping mechanism in the face of violence. 25 Much of this past work has focused on the HIV-related risks imparted by a history of abuse, with evidence that violence can impart a trajectory of subsequent sexual risk-taking, 8,26,27 lower women's perceived control in sexual relationships, prompt fear of subsequent condom negotiation, and heighten vulnerability to coercive sexual risk. 15,28 Yetqualitative evidence also illustrates the potential for STI/HIV risk behavior and IPV to co-occur within relationships, with perpetrators using violence and threats of abuse to garner women's compliance with unwanted and unprotected sex. [29][30][31][32][33] Clarifying the extent to which women's IPV experiences are concurrent with sexual and drug-related STI/HIV risk is critical to understanding the interwoven epidemics of IPV and STI/HIV. Such research is particularly important in health care settings most likely to encounter those affected. The high burden of IPV among patient populations that seek sexual and reproductive health services [34][35][36] renders family planning clinics one such critical research and intervention setting. Sexual and reproductive health providers also stand to serve a crucial role in intervention, through patient risk assessment and risk reduction at the intersection of violence and sexual health. To inform these gaps, we sought to evaluate the prevalence of IPV in the past 3 months and its associations with sexual and drug-related STI/HIV risk and related care-seeking over this same time period in a sample of adolescent and young adult female family planning patients. These data allow a new understanding of the concurrence of IPV with sexual and drugrelated STI/HIV risk. --- Methods --- Data A cross-sectional survey that served as baseline data for a prospective study was conducted among English-and Spanish-speaking females ages 16 to 29 years seeking care at one of 24 free-standing Title X family planning clinics in Western Pennsylvania. Data were collected between October 2011 and November 2012. Upon arrival, all women were screened for age eligibility by trained research staff. Eligible women interested in participating were escorted to a private area in the clinic for consent and survey administration. As participants were receiving confidential services, parental consent for participation was waived for minors. Data were collected via Audio Computer Assisted Survey Instrument, a self-administered program that allows participants to complete surveys on a laptop computer with questions read aloud through headphones. Materials were in English or Spanish based on patient preference; materials were developed in English, professionally translated to Spanish, and back translated to English to check for accuracy. At the conclusion of the survey, participants were offered a resource sheet of local social services and received a $15 prepaid debit card to thank them for their time. Further details have been described elsewhere. 37 All study procedures were reviewed and approved by Human Subjects Research Committees at the University of Pittsburgh. Data were protected with a federal Certificate of Confidentiality. These procedures generated a total of 3,682 participants; for the current analyses the sample was further restricted, with women reporting never having sex and women reporting primarily same-sex sexual partnerships removed, as well as those providing incomplete data on IPV , resulting in a total analytic sample of 3,504. --- Measures All measures were self-reported. Single items assessed demographic characteristics, including age, race, education, and relationship status. Intimate relationship partners were defined as persons the respondent reported "dating or going out with." All exposures and outcomes were assessed using a referent time period of the past 3 months. The primary exposure of interest, physical or sexual IPV was measured via three items modified from the Conflict Tactics Scale-2 and the Sexual Experiences Survey, 38,39 specifically, "hit, pushed, slapped, choked or otherwise physically hurt", "used force or threats to make you have sex when you didn't want to?", and "made you have sex when you didn't want to, but didn't use force or threats?" Outcomes of interest were assessed via single items. Standard assessments were used for unprotected vaginal sex, anal sex, and unprotected anal sex, as well as self and partner use of injection drugs. Other sexual risk behaviors included condom non-use against her will , fear of condom request , fear of refusing sex . Participants were asked to self-report their history of STI over the past 3 months , syphilis, herpes, genital warts, Hepatitis B, or crabs"). Participants also reported their primary reason for attending the clinic on the day of the survey; those indicating STI testing or treatment were classified as having sought STIrelated care. --- Analysis Prevalence of recent IPV was calculated and differences across demographic characteristics were tested via Wald Log-Linear Chi Square tests for clustered survey data with significance set at P<0.05. Estimates of the prevalence of recent sexual risk and drug injection behaviors are reported for the total sample, and by exposure to recent IPV. Logistic regression models for clustered survey data were specified to account for clinic-level clustering and statistically adjusted for demographic characteristics, including age, race, education, and relationship status. To preserve analytic power, small amounts of missing data on demographic characteristics were recoded to the mean value; and the effective sample size was allowed to fluctuate in cases of missing data for outcomes of interest . All statistical analyses were conducted in SAS, Version 9.3. --- Results In this sample of family planning clinic patients, over one third were ages 16-20 , 80.8% were white, and slightly more than half had at least some college education . One third of the sample were single, while 59.6% reported being in a serious relationship. Recent IPV was reported by 11.0%, with significant demographic differences observed. This violence was most prevalent among those ages 16-20 , and among multi-racial women . Recent IPV prevalence was highest among women with less than a high school education and women who reported their relationship status as single or dating . More than two-thirds of the sample reported engaging in unprotected vaginal sex in the past 3 months . One in ten women reported recent anal sex and 8.4% of women had recent unprotected anal sex, i.e., the majority of recent anal sex was unprotected. Women with histories of recent IPV were more likely than those who did not experience IPV in the past 3 months to engage in unprotected vaginal sex , anal sex , and unprotected anal sex . Women with recent histories of abuse were also more likely to report being forced to have sex without a condom (AOR 1.87, 95% CI 1. 51 --- Discussion Findings confirm high, recent, interrelated and concurrent burdens of IPV, STI/HIV risk, including coercive sexual and condom-related risk, among adolescent and young adult female family planning clinic patients. Recent IPV was significantly associated with sexual and drug-related risk, including unprotected vaginal and anal sex, coerced unprotected sex, as well as injection drug use. Likely as a result, IPV also related significantly to STI careseeking. Findings build on prior reports, 10,15,[40][41][42] including prospective studies illustrating a longitudinal association of violence with STI/HIV risk 27 , i.e., a trajectory pathway, by confirming an additional pathway of concurrence of IPV with sexual and drug-related risk. Specifically, adolescent and young adult women's experiences of IPV are not only related, but are often simultaneous with STI/HIV risk behavior. These data further affirm that mitigating the disproportionate burden of STI among young women 1 requires addressing the potential for physical harm as well as coercive sexual risk in STI/HIV prevention and intervention efforts. Recent intervention evidence illustrates the value of clinical screening for violence and related reproductive coercion in the context of harm reduction and referral to violence support services; 43 current findings suggest the value of expanding this type of intervention approach to address STI/HIV risk and coercion that specifically inhibits STI/HIV risk reduction. That those exposed to IPV were more likely to present for STI-related clinical services further suggests the likely utility of such an intervention in family planning clinics as well other sexual health services. Referral to violence support services are critical components in the clinical response to IPV. While leaving an abusive relationship, even temporarily through shelter services, is often the desired outcome for patient safety, providers should maintain empathetic and hold realistic expectations in supporting patients to make the right decisions for themselves, as the separation process is often lengthy with innumerable barriers 44 and heightened homicide risk 45 Current evidence of a two-fold increased risk for recent unprotected anal intercourse based on recent IPV is particularly relevant, and echos past evidence linking lifetime IPV and sexual coercion with anal intercourse. 15,46 Growing awareness of the prevalence and STI/HIV transmission efficiency related to heterosexual anal intercourse 16,17 has prompted calls for routine assessment for anal intercourse. 47 These findings affirm the need to broaden sexual risk assessment and risk reduction counseling to include heterosexual anal intercourse, particularly in family planning clinics which have historically focused exclusively on sexual behavior that relates to pregnancy risk. These findings also suggest that, within such sexual risk assessment, it is critical to address the potential for violence and coercion in the context of unprotected anal intercourse. Findings also illustrate the need to address young women's agency or control in sexual decision-making in abusive relationships in the context of STI/HIV risk reduction. One in five participants experienced condom non-use against their will in the past three months, with significantly elevated risk for this exposure among those in abusive relationships. Partner violence also elevated risk for women's reports of fearing both condom requests and refusing sex. These data highlight the challenges of STI/HIV risk reduction through condom promotion among young women; provision of condoms and condom promotion should be accompanied with an acknowledgement that condom use is not always within women's control, and probes for related patient concerns and for current partner abuse. As condom promotion alone may not be sufficient in reducing young women's STI/HIV risk, it is essential to ensure access and connection to STI/HIV testing and treatment across multiple care settings, including non-clinical services for abused women, to fully support the health needs of women in abusive relationships. Where possible, harm reduction strategies such as supporting women to negotiate lower-risk sexual activity may be valuable. Providers working with abused women in crisis-based and long-term counseling should be aware of the heightened STI/HIV risk among their clients, including unprotected anal intercourse, coercive sexual risk and condom practices, and injection drug use of both abused women and their partners. Given these data, it is essential to support strong referral networks among STI/HIV clinics, needle exchange services, family planning clinics, and violence-related crisis/counseling programs to create a system of care that is responsive to STI/HIV risk, and the counseling and support needs that stem from IPV. Finally we note that the impact of the elevated STI/HIV risk identified among those in abusive relationships is likely amplified by the STI/HIV risk behaviors enacted by their partners; prior reports illustrate that abusive men engage in greater sexual risk behavior and are more likely to be STI/HIV infected. [48][49][50][51] Current findings illustrating elevated STI/HIV risk to women in abusive relationships are best considered in the context of the broader literature, including evidence that violence can undermine safe and successful STI/HIV partner notification, 52,53 and that violence and other life stressors can compromise treatment uptake and adherence, and advance disease progression. [54][55][56][57] Taken together, these data argue for a fuller consideration of the impact of IPV across the HIV treatment cascade. Findings should be considered in light of several limitations. All data were self-reported; despite the use of self-administered computerized data collection, responses may be subject to social desirability biases, recall biases and other sources of inaccuracy. While our use of a clinical sample may enhance the accuracy of self-reported STI data in that participants are receiving routine medical care, reliance on self-reported STI is a particular limitation. The extent to which the anal sex reported herein was subject to force or coercion is unclear. Further research is needed to clarify elements of force, pressure and coercion in anal sex assessments. Findings illustrate the need to extend and modify standard violence assessment tools, such as the Conflict Tactics Scale-2, 39 to clarify the nature of unwanted or forced sexual activity and, in turn, the likely STI/HIV risk introduced. Together, findings demonstrate a high burden of recent IPV in a large, clinic-based sample of adolescent and young adult women, with significant and co-occuring sexual and drugrelated STI/HIV risk. Mitigating the burden of STI/HIV among the youth most affected requires addressing IPV and the gendered power imbalances that impart risk. It is critical to engage sexual and reproductive health clinics in addressing violence and related sexual and drug-related STI/HIV risk among patients. --- Key messages • Recent physical or sexual violence from intimate partners was common among adolescent and young adult female patients of family planning clinics • Intimate partner violence was significantly associated with both sexual and drug-related STI/HIV risk during this same time period • Findings affirm the concurrence of IPV with sexual and drug-related risk behaviors, and highlight the need for safety promotion and harm reduction for such patients.

Background/Objectives-Adolescent and young adult women are at high risk for both STI/HIV and intimate partner violence (IPV). We evaluate the prevalence of IPV in the past three months and its associations with STI/HIV risk, STI, and related care-seeking over the same time period. Methods-Female family planning clinic patients ages 16-29 (n=3,504) participated in a crosssectional survey in 2011-2012 as a baseline assessment for an intervention study. We examined associations of recent IPV with sexual and drug-related STI/HIV risk behavior, self-reported STI, and STI-related clinical care seeking via logistic regression. Results-Recent physical or sexual IPV (prevalence 11%) was associated with recent sexual and drug-related STI/HIV risk, specifically unprotected vaginal sex (AOR 1.93, 95% CI 1.52, 2.44),

Introduction Globally, health worker shortages represent a serious constraint for health systems. 'Task-shifting' and mobilization of community health workers are increasingly being promoted to extend primary health care to underserved populations in resourcepoor settings . Community health workers are a cadre of local community members who receive short training to conduct health education and sometimes treat illness; they may or may not be remunerated . The use of community health workers holds tremendous potential to extend access to very basic health care in Sub-Saharan Africa where the population is mainly rural, health indicators are poor, and health worker shortages and financing constraints are immense. Evidence for community health worker effectiveness in improving child health is growing. Extensive community health worker programmes are being scaled-up throughout Sub-Saharan Africa , including in Uganda where the Ministry of Health has begun rolling out its national village health team strategy . The challenge of motivating, and in turn, retaining community health workers is of paramount importance as poor motivation and high drop-out rates can impact programme effectiveness, cost and sustainability . Some authors suggest attrition is higher, especially in largescale programmes, when workers are not paid , while others question the ethics of using volunteer labour in very poor communities . However, volunteerism in developing countries remains poorly understood . While the World Health Organization is encouraging greater financial remuneration of community health workers as a means of improving motivation and retention , the resources required, even for minimal remuneration, may not be available in resource-constrained countries. Despite the ongoing debate on financial remuneration, retention rates and incentives in volunteer and non-volunteer programmes have not been well documented . Recently though, some countries with large community health worker programmes have begun to undertake systematic reviews . While it is difficult to assess programmes by comparing numbers alone, retention rates-in those studies that have published results-have varied greatly from 23 to 97% . Published rates include 66% community health worker retention after 11 months in an HIV/AIDS home care programme in western Kenya ; 43% in a government programme in Kalabo District, Zambia ; 97% in a volunteer programme managed by BRAC during the 1990s , but only 55% over 3 years in a different maternal and newborn health trial in the Sylhet District of Bangladesh that paid comparatively high salaries ; 70% over 9 months among unpaid volunteers in Sine Saloum, Senegal and 50% over 2 years in a national volunteer community health worker programme in Niger . Understanding the apparently large variance in retention across programmes is constrained by the limited number of studies that document retention data in general and the virtual absence of studies that present data disaggregated by attrition type or systematically track retention against different packages of incentives . The main objectives of this study are to examine retention data in detail for volunteer community health workers trained by Healthy Child Uganda and assess factors related to community health worker selection and motivation. HCU was established in 2004 as a Ugandan-Canadian university partnership aiming to promote child health in rural southwestern Uganda through education and training of volunteer community health workers. A recent 3-year impact evaluation of HCU demonstrated significant reductions in reported child deaths, as well as decreases in malaria, diarrhoea and underweight status of children . --- Methods --- Study setting and population From 2004 to 2009, HCU trained 404 volunteer community health workers to conduct child health promotion in Mbarara and Bushenyi Districts in southwest Uganda. The project reached 175 villages in 18 parishes, covering a total population of 95 000 people. Villages are clustered together into administrative groupings known as 'parishes'. An average village has a population of 450, of whom 20% are under 5 years of age. Most communities are remote, with poor access to health care and very poor baseline child health indicators . --- Intervention Community health workers were selected and trained in two main cohorts: Phase I was initiated in 6 parishes ; Phase II involved 12 additional parishes . Each community health worker served on average 25 households and 45 children under 5 years of age. --- Selection Community health worker selection is intended to encourage broad community participation. Community leaders organized village meetings where project representatives provided a clear description of community health worker volunteer roles and expectations. Building on suggested criteria , community members subsequently developed their own criteria and determined the process for selecting individuals to be trained. --- Training and supervision Government health staff trained by HCU led 5-day introductory training workshops and subsequent 2-day annual refresher workshops on basic child health promotion, prevention and assessment of sick children, as well as community facilitation and counselling skills. Health centre trainers met teams of community health workers monthly for reporting and refresher training. --- Community activities Community health workers registered children, conducted home visits and health education sessions, assessed ill children and mobilized communities for health outreach activities such as national immunization days. --- Incentives Community health workers were not paid a salary, but received various forms of financial and non-financial incentives such as: Financial incentives: transport stipends to attend monthly meetings and 2-5 day-long annual training workshops . Material incentives: job aids , token gifts and small prizes during training. Recognition awards: certificates and awards for various competitions . Social networking opportunities: exchange visits (i.e. between community health worker groups), inter-village health promotion competitions, annual holiday party and engagement with local council. Income-generating opportunities: community health worker-initiated income-generating activities and savings/ loans groups; HCU-supported training for specific incomegenerating activities . A detailed description of the HCU intervention and its impact can be found in our report, The HCU Community-Owned Resource Persons Model 2010 , available at www.healthychilduganda.org/resources. --- Study design In November 2008, HCU undertook a study to examine community health worker retention over the course of the project and elicit factors supporting retention and motivation of its volunteers. Study approval was obtained from the Mbarara University of Science and Technology Institutional Ethical Review Committee in Uganda and the Carleton University Research Ethics Committee in Canada. All survey participants provided informed written consent. --- Community health worker registry Basic demographic information for all community health workers trained by HCU between February 2004 and July 2009 was entered in an MS Excel database and updated at least quarterly. Deaths, resignations and new replacements were recorded. Periodically, additional community health worker demographic data including age, level of education and number of children were collected and recorded in the registry. --- Community health worker survey Two separate Questionnaires were constructed de novo in English to address selection and motivation concepts. Questionnaire A collected both quantitative and qualitative data and applied a participatory rural appraisal technique called 'pairwise ranking' to compare a series of potential motivating factors. The results of an earlier brainstorming exercise were used to identify the motivational factors. Questionnaire B consisted of open-ended questions to elicit more in-depth responses to those given in Questionnaire A. Both questionnaires were translated into the local dialect, back translated and piloted with four community health workers from non-study parishes for clarity. No major revisions were required. A stratified sampling frame was designed to survey community health workers from half of the parishes where HCU operates. The nine participating parishes were selected based on the date of their next scheduled community health worker meeting . From these nine parishes, the names of 25% of active community health workers were drawn randomly in advance. Any pre-selected participant not present on the day of the survey was replaced through a second random draw conducted on site. The number of community health workers in each parish varies according to the population size of each parish. Time and resource constraints limited a larger sample size. Questionnaire A was initially conducted in six parishes . Preliminary analysis revealed a strong consensus among respondents, suggesting data saturation. The survey was then modified to elicit more indepth information and build on responses from Questionnaire A. Twelve respondents completed Questionnaire B. This sampling process is summarized in Figure 1. Surveys were constructed de novo in English, translated into local dialect, back translated for accuracy, then verbally administered in the local dialect by research assistants who did not have a role in HCU programming. Interviews were conducted on an individual basis to avoid peer and group influence on individual responses. --- Data analysis Using 2009 data from the registry , descriptive statistics for community health worker demographics and retention were calculated using means and frequencies for numerical data and proportions for categorical data. The retention analysis considered aggregated results for the duration of the project, as well as disaggregated results by training date . For the survey, descriptive statistics were used to calculate proportions for quantitative questions. Final pair-wise ranking scores were calculated by averaging the score for each motivator across all respondents. For closed-ended, multiple-choice questions, deductive coding was used to calculate frequency. Open-ended answers were grouped thematically and multiplecoded manually by the lead researcher. --- Results --- Community health worker demographics Between February 2004 and July 2009, HCU trained 404 community health workers in 175 villages . Demographic data were collected from 302 community health workers from December 2009. Community health workers ranged in age from 20 to 73 years with an average age of 36.7 years . Ninety-eight per cent of community health workers had at least one child, with an average of 4.9 children. All community health workers had completed at least primary school grade 3: 30% completed primary 3-6; 33% completed primary 7 and 38% had gone beyond primary school. Male community health workers reported a higher average level of education than did women , [t P < 0.05]. --- Community health worker retention The study analysed the retention of community health workers using data from the project registry. As of July 2009 , 350 of all community health workers were still active. Fifty-four community health workers had resigned with no statistically significant difference between genders [ 2 ¼ 1.84, P > 0.05 ]. In the Phase I cohort , the 5-year retention rate was 86.3%. However, once reasons unlikely to be related to poor motivation-changed village of residency, death, schooling and health problems-were excluded, the 5-year community health worker retention for the Phase 1 --- HCU-Trained Community Health Worker Coverage --- Convenience Sample of Parishes --- Random Sample of Community Health Workers Questionnaires A&B • Population: 95,000 • Children <5 years: 19,000 cohort was close to 95%. Almost all villages had at least one active worker for every month during the 5-year period. --- Length of time until resignation or death For the 54 community health workers who were no longer active, the average length of time from initial training until resignation or death was 20 months with a range of 1-60 months . Of the total 404 community health workers trained between 2004 and 2009, 389 were still active 1 year after initial training . By 2 years post-initial training, 36 community health workers had died or resigned, leaving 91% retained. In summary 96% of CHWs were active for at least 1 year. 91% of CHWs were active for at least 2 years. 86% of CHWs were active for 5 years. --- Reasons for community health worker attrition The most common self-reported reasons for dropout were due to being too busy, moving to another village or needing to focus on business or other paid activities . Six community health workers died, while another six resigned due to divorce. In cases of divorce, the woman typically returns to her parents' village. --- Participation in the questionnaire studies In total, 58 community health workers participated in the verbally administered survey, representing 17% of all 350 community health workers active at the time. Forty-six completed Questionnaire A and 12 completed Questionnaire B. In each of the nine parishes, one or two of the pre-selected community health workers did not attend their scheduled monthly meeting; reasons for non-attendance were not documented. No community health workers at the meeting refused to participate in the survey. The demographics of those participating in the survey were similar to the overall group in terms of age and gender , but the average level of education was slightly higher with 48% having at least some secondary school education. --- Community health worker selection In Questionnaire A, most respondents reported having been chosen through a community 'election'. The remainder was selected either by the local council or by the local women's council without wider input from the community. Of those 'elected', 89% recalled at least half of the community being present. Eighty-seven per cent estimated that women comprised at least half of those in attendance at the election. Respondents described methods used to select the community health workers in their village: each side of the village chose one community health worker; interested individuals self-nominated themselves and were confirmed by community members; nominations from peers that were confirmed by consensus or by vote. Voting was often carried out by show of hands or by standing behind the candidate of choice. Community health workers almost universally reported that community election was the most desirable means of selecting candidates for the post. Community selection was preferred because community members have intimate knowledge of who would be most suitable and communities should have the 'right' to select their own representative . --- Community health worker motivation Community health workers were asked to compare seven motivators using a pair-wise ranking tool . As seven factors were presented, the highest possible score was 6, indicating that the respondent ranked that motivator above Improved child health When asked 'Why do you volunteer?' respondents explained that the training increased their awareness of how they could take action to improve the health of children in their community and in their own families : 'I realised I will benefit personally as well as my neighbours in a way that diseases would be fought and our children would grow healthy, strong, and have a right future as opposed to how it was before'. Respondents gave concrete examples of how their families and others in their villages had adopted new health practices and observed improved child health. 'In our village child mortality was very high, but after we got involved through the project, it has now decreased very much. Almost every home if not all have vegetable gardens. This has improved greatly the health of our children. The neighbouring village which is not part of Healthy Child Uganda . . . their children die almost daily'. Seventy-six per cent of respondents reported that since initial training, the number of hours they dedicate to carrying out volunteer, community health worker-related activities have increased. --- Education and training Furthermore, half of all respondents cited refresher training/learning opportunities as something they 'like about being a community health worker': 'The re-fresher courses they give us keep opening our eyes./ . . . I realized knowing these things could greatly change my family and the whole community at large'. --- Requests for advice/assistance Surveyed community health workers indicated that they gained pride and respect when village members consult them: '. . . people learn from me; they come to me for advice and I feel proud and useful in my community. I am even appointed to represent them elsewhere'. Though respect was not rated as highly as some of the other motivators in the ranking exercise, it emerged more clearly in the open-ended responses. --- Discussion Critics have flagged low retention as a serious weakness of community health worker programmes. However, poor retention does not have to be the rule. Our study is one of few to report medium-term retention from East Africa and present a detailed, disaggregated analysis of retention data for volunteers. The study provides insight into reasons for attrition, differentiating those which may be motivation-related from those due to other factors . An impressively high retention rate was demonstrated even after 5 years. Although many programmes do struggle to retain community health workers , our findings alongside other successful examples-such as the low 3% dropout reported by the BRAC programme in Bangladesh -suggest that high community health worker retention can be achieved. Moreover, the high retention in HCU-trained community health workers challenges the assertion that non-financial incentives are insufficient to retain community health workers , since HCU does not provide significant salaries. Debate in the literature provides competing views on whether non-financial motivators such as 'a sense of achievement' and 'recognition' are more important than salary in low income countries . Our study, based on the HCU context found that volunteer community health worker motivation stems primarily from perceived improvements in health outcomes, training opportunities and being a valued asset in the community. We hypothesize that a number of HCU programme design factors may have helped promote community health worker retention and served to motivate volunteers, most of which have previously been cited as retention promoting: an inclusive, community-centred approach to selection ; recruitment of individuals with a sense of community spirit and social responsibility ; frequent, participatory training sessions and regular and supportive supervision from health centre staff . Furthermore, the study findings suggest our programme focus on 'child health' and subsequent notable child health improvements in intervention areasdocumented and attributed to the community health worker intervention -were key motivators for volunteers, whose activities, in turn, further improved the health of children in their communities. We believe that HCU's process for selecting community health workers may have played a key role in recruiting candidates who have a strong likelihood of staying on the job. The selection process emphasized clear communication of volunteer roles and expectations, promoted broad community attendance and community input into selection criteria and encouraged community selection of their own representatives. Perhaps community members are best positioned to select those individuals who have a demonstrated track record of volunteer spirit and commitment to the community. Furthermore, community-led selection may also help select individuals whose demographic profile will most likely lead to longevity in that community and on the job . Finally, those volunteers selected by their own peers may experience the most accountability and responsibility towards their own communities. Evidence from the literature and this study suggest that a transparent, participatory community selection process is a worthwhile investment for community health worker programmes of all sizes. In addition, HCU's use of interactive learning pedagogy and emphasis on community development and facilitation skills combined with regular contact and training by supervisors likely served as retention promoting, non-financial motivators. Other key HCU programme elements not cited in the literature but which may have played a role in promoting retention include the role of a 'uniting cause' and organization of community health workers into teams. Both of these create a platform for peer support and an opportunity to be a 'champion for change'. We acknowledge important limitations of this study. Surveying only community health workers in attendance at monthly meetings may over represent more 'motivated' volunteers. Additionally, community health workers may have responded so as to satisfy interviewers, including underrating the significance of the small transport allowance. Our motivation and selection questionnaires lacked input from supervisors, those who had dropped out, and community members served by community health workers-all of which may have highlighted different perspectives on selection and motivation. Although we consider HCU community health workers to be 'volunteers', they did receive a very modest transport allowance to attend monthly meetings which may have acted as a financial incentive. Of reassurance, these allowances were discontinued completely 1 year after conducting this survey, without substantial impact on meeting attendance, reporting or retention during the following 12 months. As countries and donors increasingly invest in large-scale community health worker programmes, more studies examining retention and motivation across various programme designs are needed. While HCU has demonstrated positive results, more evaluation is needed to assess whether high retention is possible within larger programmes, such as the roll-out of Uganda's Village Health Teams and other country-level programmes. Ensuring quality of selection, non-financial incentives, supervision and refresher training may be more challenging as programme size increases. Future studies should explore how different 'packages' of incentives may impact motivation and retention, how retention varies over time and how often and through what mechanism community health workers should be replaced or re-affirmed. Further investigation is also needed to better understand how assigned tasks , responsibilities and workload impact retention and motivation. --- Conclusion Our study demonstrates that under the right conditions it is possible to retain volunteer community health workers over the medium term in a rural, sub-Saharan Africa context where financing is extremely limited. Community health worker programmes should not be viewed simplistically as a quick fix, but as one component of a broader primary health care approach involving diverse human resources for health. Community health worker programmes require investment in high-quality selection, training, supportive supervision, incentives and evaluation. Policy-makers and programmers should consider the following among key elements in designing and implementing community health worker programmes: community involvement in community health worker selection, recurrent training, incentives, peer and supervisory support, and focus on specific, easily understood, high priority health issues. Narrowly debating the merits of paid vs unpaid community health workers may not be the most productive way forward. Rather, researching incentive and behavioural models could strengthen understanding of motivation and retention, and support the implementation of more effective, sustainable programmes. This will be particularly important as we move into the post-2015, post-MDG era where the dialogue on how to achieve sustainable health gains and address continuing shortages of human resources for health is of increasing importance. --- Other contributors We thank the many community health workers, trainers, health centre staff, community members and leaders whose input shaped the HCU model and who participated in the study including HCU staff , academic advisors , research assistants , analysis and editing advice . --- Funding This study was funded by the Association of Universities and Colleges of Canada through its Students for Development program. The AUCC played no role in the study design, data collection, analysis, or report writing. We are also grateful for support which has enabled community health worker training and programming since 2004 from the Canadian International Development Agency, Healthy Generations Foundation, Buy-A-Net, Mayfield Rotary Club and the University of Calgary Annual Giving Program.

Globally, health worker shortages continue to plague developing countries. Community health workers are increasingly being promoted to extend primary health care to underserved populations. Since 2004, Healthy Child Uganda (HCU) has trained volunteer community health workers in child health promotion in rural southwest Uganda. This study analyses the retention and motivation of volunteer community health workers trained by HCU. It presents retention rates over a 5-year period and provides insight into volunteer motivation. The findings are based on a 2010 retrospective review of the community health worker registry and the results of a survey on selection and motivation. The survey was comprised of qualitative and quantitative questions and verbally administered to a convenience sample of project participants. Between February 2004 and July 2009, HCU trained 404 community health workers (69% female) in 175 villages. Volunteers had an average age of 36.7 years, 4.9 children and some primary school education. Ninety-six per cent of volunteer community health workers were retained after 1 year (389/404), 91% after 2 years (386/404) and 86% after 5 years (101/117). Of the 54 'dropouts', main reasons cited for discontinuation included 'too busy' (12), moved (11), business/employment (8), death (6) and separation/divorce (6). Of 58 questionnaire respondents, most (87%) reported having been selected at an inclusive community meeting. Pair-wise ranking was used to assess the importance of seven 'motivational factors' among respondents. Those highest ranked were 'improved child health', 'education/training' and 'being asked for advice/assistance by peers', while the modest 'transport allowance' ranked lowest. Our findings suggest that in our rural, African setting, volunteer community health workers can be retained over the medium term. Community health worker programmes should invest in community involvement in selection, quality training, supportive supervision and incentives, which may promote improved retention.

Background A growing body of literature and health research policies has, in recent decades, emphasized the importance of more participatory approaches in health research that include views, knowledge, experiences and actions from those who are in focus of the research. Similarly, it has been highlighted that children and young people should be involved in co-creating new knowledge [1][2][3][4][5][6]. Although participatory approaches have user involvement in common, they differ in terms of the explicit guidance on how to actually involve and engage children and young people in health research [7,8]. Many factors can challenge children's and young people's involvement, such as an underestimation of children's competence to participate in research [1], attitudes that child involvement might adversely affect the quality of the research [9][10][11] or fear that participation might harm the child [2,[9][10][11]. As a result, children are usually merely involved as subjects in research and not as active research partners [3,5]. Most health research for children and young people is thus primarily based on the involvement of parents, caregivers, and other stakeholders. However, such stakeholder perspectives cannot replace the qualities that come with genuine participation by the children and young people themselves [3,5,12,13]. Children as social actors were emphasized by the Convention on the Rights of Children in 1989 [14]. The principles of the convention are relevant for how researchers relate to children's participation and serve as a standard for how integration and assessment of children's participation in research should be planned and assessed. Increased emphasis in health research on the need to develop solutions to improve the health of children and young people's thus also requires that children are viewed as important partners in co-creating such solutions [15], providing both relevant knowledge for the design of solutions and guidance and planning for evaluation and implementation in practice [16]. When co-production approaches are used in health research, they increase the likelihood of developing more effective and efficient interventions that more precisely targets health and well-being issues among children and young people. This, however, requires a shift in the way research is designed, communicated and performed [17] and necessitates that children and young people are recognized as experts with the capability of contributing with unique experiences and knowledge [8]. However, it is difficult to classify and evaluate health research in ways that shows both the level of children and young people's participation during the research process and its relation to the impact of the research and translation into practice [18]. In the application of participatory research it is important to distinguish three overall stages of participation [6], that differ both in principle and in the way they can be translated into practical implementation in health research. These stages are; nonparticipation, consultative participation, and collaborative participation. In the stage of nonparticipation, children are either not involved at all or are involved in ways that have no real impact on the research or that give a false semblance of partnership and sharing of power. At the stage of consultative participation, adults acknowledge the expertise of children and involve them in sharing their views and experiences, primarily through interviews or questionnaires. However, such approaches are neither giving the children control over the focus of the research or influence over the analysis or interpretation of data. At the collaborative stage, children are not only involved as experts but also take part at various degrees in initiation, planning, analysis and dissemination of the research. The partnership between the researcher and the children at this level is based on both trust and shared decisions [6]. Models that describe how participation can occur at different levels of intensity and quality are valuable as benchmarks in the planning of research and as guidance for evaluation of participation. Hart's Ladder [19] adapted from Arnstein's work [20] and Shier's Pathways to Participation [21] are commonly used models when developing and evaluating children and young people's participation in projects. Both models can be suitable when evaluating children and young people's level of participation in research. In this scoping literature review, however, we decided to use Shier's model [21] based on this model providing a more practical framework for planning and evaluating children and young people's participation in practice. Shier's model combines five levels of participation and three stages of commitment at each level describing the child's transition from a passive actor towards having a partnership position where the child and the adult have equal positions [21]. Despite the increasing expectations that children and young people are involved as partners in health research, and despite that models for discriminating between different levels of participation and how to tackle challenges to participation are available, there is still considerable uncertainty among researchers about how to optimally provide opportunities for involving children and young people in health research. The aim of this scoping review was thus to systematically map recent research involving children and young people in the development of interventions targeting issues of health and well-being. The specific objectives were to; a) identify the extent of recent research using participatory approaches to involve children and young people in the development of interventions targeting health and wellbeing, b) grade the level of participation in such participatory approaches, and c) identify areas for further research. --- Methods --- Study design An interpretative scoping literature review, based on the framework of Arksey and O'Malley [22], was chosen for the study design. A scoping review contributes a systematic knowledge synthesis in a defined area based on an explorative research question with the aim of mapping key concepts, available evidence, and gaps in the research [23]. Scoping reviews are useful for summarizing and describing data from a wider range of fields and disciplines and for identifying gaps in the literature, and thus the quality of the studies are not the focus of the evaluation. To enable replication and strengthen methodological rigour, this study follows the five-stage methodological framework of Arksey and O'Malley [22]; identifying the research question, identifying relevant studies, study selection, charting the data, and collating, summarising and reporting the results. --- Identifying the research question The core question in this scoping review was: In which areas and to what extent were children and young people involved in the development of interventions targeting children and young people in health and well-being. The definition of children, in this scoping review, was a person under 18 years old, which is in line with The United Nations Conventions of the Rights of the Child [14], while young people are referred to a person between 15 and 24 years old, in accordance with The United Nations [24]. --- Identifying relevant studies The evidence was searched by way of electronic databases, reference lists, hand-searches of key journals and contact with existing networks. With the help of experienced librarians an electronic database search incorporating Academic Search Elite, CINAHL, ERIC, Medline, PsycInfo, Sociological Abstracts and SportDiscus was conducted in December 2014, updated in April 2015 and finally updated in December 2017. Different techniques and terms were used for expanding and narrowing searches, including search tools such as medical subject headings , Boolean operators and Truncation. Single and combined search terms included the key words: "child", "adolescent", "participate", "participation", "collaboration", "involve", "involvement" and "intervention". Relevant publications were defined as any empirical peer-reviewed paper. No limitations were set in terms of the publications date, as no previous review focusing on children and young people's participation in the development of interventions had been performed. English was chosen as the language for the database search, as it is the most commonly used language in scientific journals. Inclusion criteria were: articles with children and young people under 25 years old; articles in which children and young people participate in one or more levels of the development of an intervention aimed at children and young people in health and well-being. Exclusion criteria were: articles in which children and young people were only participants in an intervention or evaluated an intervention. --- Study selection The searches identified 4458 articles, which were catalogued in EndNote®. Duplicates were removed by automation, supplemented by manual checking. An initial scan of title and abstracts identified large numbers of irrelevant studies. A total of 383 articles were retrieved and read in full text by the authors. Seventy four studies were identified as potentially relevant after an elimination process based on the inclusion and exclusion criteria. These articles were reviewed by three of the authors and by consensus a final decision was made on which articles to include. The development of a particular intervention was sometimes described in a number of articles and in these cases only one of these articles describing the intervention was included in this scoping review. In some articles, when the target group for the intervention was children and young people, the development of the interventions was poorly described, but it appeared that the researchers developed the intervention together with parents, health professionals or other adults. Articles were excluded if children and young people participated in the implementation of an intervention or in the evaluation of interventions with the specific aim of modifying an intervention designed by the researchers. A total of 41 articles were finally included in the scoping review . --- Data charting and collation The authors created a data charting form which included: reference , country, key aims, number and age of children and young people, type of intervention, aim of participation, and finally in which parts of the developmental process the included participants were involved. The authors first extracted data independently and then met to determine whether the data extractions were consistent with the aim of the study and the research questions . --- Collating, Summarising and reporting findings In order to grade the level of participation in the analysis, we summarised and categorised the data in accordance with Shier's Pathways to Participation Model [21]. This model includes the following five levels of participation: Level 1, Children and young people are listened to, requires that researchers listen to the children and young people when the children and young people take the initiative to express their views. In this stage the children and young people can thus be seen as responsible for taking active initiatives in order to participate. At level 2, where children and young people are supported when expressing their views, the initiative and responsibility is moved to the researcher who has to find and facilitate ways for the children and young people to express their views and by this enable the children and young people's participation. At level 3, the degree of participation is extended and children and young people's views are taken into account. This level is distinguished from level two in that children and young people's views are not only listened to or asked for but are also seriously taken into account with the aim of having influence. At level 4, children and young people are involved in decision-making processes. This level and level 5, where children and young people share power and responsibility in decision-making are characterised by a successive transition from children and young people seen as consultants to a stage where they obtain a position of power. These are characterised by the willingness of researcher to share or give up their power in favour of children and young people's contribution. In moving up the levels the model describes the children or young people moving from being a passive informant to an active agent towards a partnership position where researchers and children or young people are in an equal position. In order to categorise the level of children and young people's participation, the included studies were judged overall based on a) quantitative aspects of participation i.e. number of activities or stages in the developmental process that included children and young people and b) qualitative aspects i.e. to what extent such involvement was based on reciprocity concerning influence, power and decision-making in the developmental process. In order to strengthen the validity in this process the authors , who were multidisciplinary, independently analysed the children and young people's level of participation in each intervention in accordance with Shier's five-stage framework [21]. The authors discussed and compared their analysis and level of participation in the articles until consensus was reached. --- Results Our overall aim with this scoping review was to grade children and young people's participation in the development of interventions. To enable this goal and to gain an overall view of the field we needed to make sense of our complex findings. Therefore, in addition to displaying the number of articles, we also found it useful to map the research fields according to general characteristics of the included articles such as the settings and countries were the research took place. It was also considered of interest to focus on the interventions made and the methodological characteristics of the included articles, i.e. the actions agreed as the result of the research and the types of practices that were needed to promote children and young people to participate in the research field. Furthermore, we used Shier's [21] model to analyse and grade our findings according to the children and young people's level of committed participation in the research. To allow this view of the field to be seen, the findings are illustrated in three areas; general characteristics of the included articles, methodological characteristics of the included articles and children and young people's level of participation in the development of interventions in the included articles, with a combination of texts, tables, and figures. --- General characteristics of the included articles There was a total of 41 articles included in this scoping review. These were published between 2000 and 2017 and were from different parts of the world. A majority of the studies, 21, were conducted in North and Central America and 15 in Europe [46][47][48][49][50][51][52][53][54][55][56][57][58][59][60]. A small minority of the studies were thus undertaken in the other continents, Asia [61][62][63], Africa [64] and Australia [65]. The participants in the included articles varied in age between 3 and 25. Four of the included studies did not, however, explicitly declare the participants' age and instead used terms such as teenager or teens [44,53], adolescents [41] and students [36]. Another way of expressing the participants' "age" was to refer to which grade or school system the participants were in when the research took place [33,45]. Furthermore, the included articles in this scoping review varied according to research settings. We categorised three broad settings, 13 were conducted in community settings [26, 29-32, 35, 37, 38, 41, 43, 44, 61, 65], a further 11 in healthcare settings [25,28,42,46,49,52,53,55,[57][58][59], and 17 in school settings [27,33,34,36,39,40,45,47,48,50,51,54,56,60,[62][63][64]. The reviewed articles had two different foci for the developed interventions, support for lifestyle changes and support in managing illness and disease . However, both of these foci were supporting interventions concerning health and well-being. The main areas in 30 articles were supportive lifestyle interventions and concerned: a healthy diet and obesity [27,36,39,41,47,54,56,60], physical activity [26,36,38,39,41,45,48,54], substance abuse such as, alcohol, tobacco and drug use [29,34,40,43,44,51,61,62,65], sexual and reproductive health [29,31,42,43,50,64], violence [29,37], stress [33], social skills [63], health beliefs [30] and mental health promotion [35]. In addition to lifestyle issues, eleven of the articles considered issues that supported children and young people with their illness or disease. These were supportive interventions to manage healthcare situations [58], support children and young people with cancer [25,46,57,59], diabetes [28,52,53], mental illness [32,49], and asthma [52,55]. --- Methodological characteristics of the included articles Most of the studies included in the scoping review have used interviews as a data collection method to give a voice to the participants. Focus group interviews was the most common data collection method [26, 27, 29-34, 37, 39, 41, 44, 45, 47-56, 59-64] while other studies have used individual interviews [25,28,30,34,36,42,46,48,49,57,58,64] with children and young people. However, seven of the studies reported including both interviews and surveys to involve the children and young people's views in the development of the intervention [27,30,42,44,45,48,65]. As a supplement to these more traditional data collection methods there were also research designs that included a range of innovative methods, such as video recordings [40,49,57] photographs [38,40,54] drawings and texts [46,58,59] advisory boards [33], e-mail and a social networking site [55], observations [57,64], script-making [40,43], storyboarding [63], active and spontaneous role play [43], and videoconferences and in-person meetings [35]. The interventions were sometimes tested for feasibility and usability and a wide range of methods were used, including face validity [25] think-aloud methods [43,46,47,57], interviews [31,32,34,65] and observations [33,57]. Written feedback [27,31,34,54], workshops [52,59,65] and a mobile survey [65] were also used. --- Children and young people's level of participation in the development of interventions The articles were also graded according to the children and young people's level of participation in the development of the interventions. We used Shier's model, which contains five levels of participation [21], as a framework to map our findings . All studies regardless of design were graded from level one to level five. However, due to the inclusion criteria in this scoping review, where children and young people had to be involved in the development of an intervention, all the included articles met the criteria for the second level according to Shier's model [21]. Level one is thus not further described in the findings section. There was, however, a variation in how clearly and detailed the authors described the research process as to how the voice of the children and young people influenced the development of the interventions. The findings showed a variation in the articles in terms of the children and young people's level of participation. Furthermore, this participation varied both in quantitative and qualitative terms from just being an active informant to an active agent taking part in more steps in the research process as a co-researcher. When the research process was shaped by views of a higher level of mutuality the participants were enabled to share power and responsibility in the research process. --- Level 2. Children and young people are supported to express their views in the development of interventions This was the lowest level for participation found in this scoping review. The children and young people simply had a participatory role as informants in the research process at this level. Only three articles [50,51,60] in this scoping review were analysed as only meeting the criteria for level two. The low rating for these articles was due to the researchers only describing that they had supported the children and young people to express their views and that they had facilitated ways for listening to the participants. The researchers did not, however, describe how the information was used. Furthermore, these articles only stated that the performed focus groups were to inform the development of an intervention but the researchers did not explicitly describe the ways in which the information affects the intervention [50,51,60]. It is unclear if the participants' voices were taken into account or not thus leaving the reader with unanswered questions. --- Level 3. Children and young people's views are taken into account in the development of interventions The vast majority, in total 28, of the included scoping-review articles, met the criteria for level 3 [25-29, 32-34, 36-41, 43-45, 47-49, 53-56, 61, 62, 64, 65]. The distinction between this third level and the previous one was that the children and young people's views were not only asked for and listened to, but their voice was also seriously taken into account with the aim of influencing the further process of the development of the intervention, and that this was explicitly expressed in the article. Expressions such as; "helped to develop the intervention", "guided" or "informed" were used to describe the ways in which the children and young people's views were taken into account at this level. We can therefore conclude that the children and young people were listened to and that they also had an impact on the development of the intervention. However, the words that explain the participatory part of the development are relatively vague leaving the reader uncertain as to how much of the information was used by the researchers. --- Level 4. Children and young people are involved in the decision-making processes in the development of interventions Seven of the 41 articles in this scoping review were considered to fulfil the criteria for the fourth level of participation [30,35,42,46,57,58,63]. To be able to attain this assessment the article had to explicitly describe that the researchers involved the participants in the decision-making process when the intervention was developed. This implies that the children and young people's voice has an even greater level of importance and is taken into consideration. It is not sufficient for this level of participation that the researchers themselves choose which part of the information is to be used in the development. The children and young people's voices need to instead be taken more seriously and there needs to be a successive transition, from just seeing the children and young people as consultants to a position where they have a much more important and extended position in the development of the intervention. In the articles at this level, the researchers explicitly expressed that the participants; "developed the design", "determined the development" or "accounted and contributed to the design" as well as being "central collaborators" and "co-designers". --- Level 5. Children and young people share power and responsibility in the development of interventions Only three of the included articles managed to reach the highest level in Shier's model of participation [31,52,59]. Level 4, children and young people are involved in decision-making processes, and level 5, where children and young people share power and responsibility in decision-making are characterised by a successive transition from children and young people as consultants to a stage where they obtain a position of power. These are also characterised by the willingness of researchers to share or give up their power in favour of the children and young people's contribution. In moving up the levels the model describes the child or young person moving from a passive informant to an active agent towards a partnership position where researchers and children hold an equal position. The three studies that achieved this level were more characterised by a willingness from the researchers to share the power to influence the developing process with the participants. The research idea in the study by Garafolo et al. [31] originated from the participants themselves. Young transgender women created the intervention with assistance of the research team. The feasibility of the intervention in this study was then tested and evaluated with the participants and their further suggestions were used to refine the intervention. The researchers and the participants in the study by Kime et al. [52], worked "side by side" at different stages of the research process. Finally, in the study by Wärnestål et al. [59], the researchers explicitly named the participants as "designers" and "innovators" and described the co-creation in developing the intervention during the whole process together with the children. This demonstrated that the children and young people had an extended role where they could influence the development. The designation as a designer also proved that the researchers were willing to share a certain amount of power over the developmental process. In addition, the participants were included in several of the research steps as informers, testers, evaluators and finally as informants about the usability of the intervention. --- Discussion The need for and operationalization of more explicit participatory approaches in research where participants actively take part throughout the research process has increasingly been discussed. This review provides an overview of research during 2000-2017 with a focus on children and young people's level of participation in research aiming at developing health and well-being interventions. Results from research from both the health and social sciences were included, which to our knowledge has not previously been performed. The main findings showed that the explicit level of participation in the included studies varied greatly, both in quantitative and qualitative terms, and that the designs and the methodologies used in order to increase the level of participation were generally poorly described. Even though studies reported an ambition to increase children and young people's participation in the research process, actually doing so was more seldom supported by the results. The participants only took part as informants in a majority of the studies [25-29, 32-34, 36-41, 43-45, 47-51, 53-56, 60-62, 64, 65] and surprisingly few reached the higher levels of participation [30,31,35,42,46,52,[57][58][59]63] in Shier's model [21]. These findings were somewhat unexpected since approaches, in which representatives from the target group formulate the needs and goals of an intervention, increase the level of sustainability of the outcomes [5,66]. In line with the work of Reich et al. [67], the studies that reached the fourth [30,35,42,46,57,58,63] and the fifth [31,52,59] levels appeared to be less expert-driven. Furthermore, a specific feature of the studies on the fifth level was that the children and young people were co-researchers in all parts of the development process. For example, participants were described as designers or that they were working alongside the researchers. In one of the studies [59] working pairs of participants and researchers or professional designers were formed. These types of descriptions indicate a higher level of partnership mutuality and that a transfer of power, from researchers to participants, was facilitated and had possibly taken place. Broström maintains that research needs to include the participants at all stages of the process if the participants are to be recognised as co-researchers [68]. Consequently, participants also need to have an influence on the research questions and development of the interventions from the start to ensure that the intervention responds to their needs and takes their specific contextual situation into account [69]. It was only in the study by Garafolo, in this review, where it was reported that the participants had an opportunity to influence the intervention from the conceptual stage [31]. In spite of having participatory approaches as a common denominator, it was notable that the descriptions of the methods varied considerably as to how these approaches were carried out throughout the research process. Several studies had inadequate descriptions of the methods used and were therefore excluded at an early stage. Moreover, information concerning design and methodology were lacking or incomplete in some of the studies that met the criteria for inclusion. This is an important result since a more accurate and thorough methodological description could help revealing which factors, conceptions and assessments influenced the participatory approach in the reviewed studies. Moreover, the information could have been significant for the understanding of how different mechanisms in the research process were influenced by contextual factors as well as how they in turn informed the results of the developed intervention [70]. Levels of possible and expected participation will by nature vary with the age of the children as well as with their knowledge and experience of a certain issue or situation. Since previous studies have pointed out that children's capabilities are often underestimated [1,4] it was particularly interesting, that one of the studies reaching level four, involved children as young as three to 5 years of age [58]. Appropriate and mixed data collection strategies are necessary for successful inclusion of children and young people as active research participants [71] and indeed those studies that reached the highest level of participation used a variety of data collection strategies. In addition to interviews or surveys, the most common methods of data collection for studies at higher levels of participation, were methods where children could be active and express themselves in other ways than verbal ones [30,31,35,42,46,52,[57][58][59]63]. Such methods could for example include drawing, painting, writing [72], theatre [73], photography, filming [74] workshops, storytelling using software and digital tools, and mapmaking [73]. Using such methods at different stages in the research process does not in itself guarantee a participatory approach [74,75], but they can promote the participants' sense of control and thereby enable them to take a more active part in the research process [73]. Methodology in itself can thus push a traditional top-down research paradigm in a more egalitarian direction [73] and increase credibility and reduce potential biases by triangulating different data sources [76]. The scoping review showed that participatory approaches in the development of interventions were most often used in school settings and aimed at supporting lifestyle issues or managing illness or disease [27,33,34,36,39,40,45,47,48,50,51,54,56,60,[62][63][64]. None of the included studies focusing on school settings reached the highest level of participation and only one study reached level four [63]. It is possible that the school context, with its traditional structure of power, is preserving a top-down approach in the development of school interventions. In addition, researchers, based on their own experience, might be confident in having sufficient knowledge about the school environment and children and young people's daily lives and how interventions could best be put forward [67]. However, such attitudes can have a negative impact on utilizing the potential of the school as an arena for health interventions that are co-developed in partnership with the children, strategically to prevent the occurrence of ill health and tactically to promote the health of vulnerable groups or individuals with special needs. In conclusion, researchers often advocate a participatory approach in which children and young people are seen as credible informants on interventions aimed at improving their health and well-being. However, this scoping review has showed a somewhat ambiguous result concerning implementation of this understanding in research practice. In order to recognize children and young people's capabilities as active contributors instead of passive recipients of researcher-driven interventions, a consistency between the theoretical understanding of a participatory approach and what is undertaken in research practice in developing health interventions together with the target group is needed. --- Strengths and limitations Despite a focus on a broad research field and a fairly wide range of publication years, only 41 articles met the inclusion criteria. In an effort to provide clarity and focus for the inclusion and exclusion criteria the research group attached great initial importance to discussing concepts linked to the aim, such as "what action is required to be called an intervention" [77,78]. Furthermore, additional methodological rigour in the inclusion process included two of the authors independently reading the full text articles. If uncertainty occurred related to article selection, the third author also read the article and a final inclusion decision was then reached in consensus [77]. As an aid to transparency and reproducibility a flowchart and additional text from the search process were included in the method section. Efforts were made to do more than just map and describe the characteristics of included studies by using Shier's model [21] as an analysis of the levels of participation. The model has been used in enhancing children and young people's participation in decision-making in society but to our knowledge the model has not been used before in grading of children and young people's participation levels in the development of interventions in health and well-being. However, we consider the model to be useful in order to map and clearly illustrate at which level children and young people are involved in these research studies and thus pinpoint the state of knowledge. The search process in our study is, however, subjected to certain limitations since our strategy was restricted to health, social science and educational databases. Although experienced librarians performed extensive literature searches in several research databases resulting in a large number of articles, it is possible that relevant studies were missed. We have selected original research articles, which thus did not include reviews or grey literature that may have contained studies that could contribute understanding to this topic. Despite only original research articles being included in this scoping review, there is a limitation since scoping reviews do not typically include a quality assessment of included studies [77][78][79]. This review may also have been limited by restricting the search to English articles as it is the most commonly used language in scientific journals [78]. Moreover, we may lack full knowledge of existing international networks in the research area. A further limitation could be said to exist in the potential risk for the validity of the grading levels of participation. Even though, the study's research team was multidisciplinary, they all shared the same geographical and academic context. Moreover, based on the variations in journal requirements, it should be acknowledged that these requirements might be lead to publication bias in terms of how the concept of participation was presented. It was not possible to further discuss country-specific differences in relation to participatory research approaches as the relatively small total of 41 studies were from a total of 13 different countries. Questions regarding differences between countries and contexts in attitudes concerning to what extent children and young people can be, and are involved in the development of interventions, were not part of the aim for this study but are recommended to be included in future research. --- Conclusions This scoping review showed that work remains in enabling children and young people to influence the development of health interventions. Many studies were keen to discuss the importance of taking children and young people's views, experiences and suggestions into account when developing interventions. However, relatively few invited these children and young people to share power in the design, implementation and analysis of the research or as partners in the process of developing the intervention. Participatory approaches aiming for a higher level of reciprocity where children and young people work together with the researchers in partnerships are thus warranted. Such studies also need to carefully describe the methods used in the collaboration with children and to use models, such as the one described by Shier, in their description of how and to what extent the children were actually involved. Only then will it be possible to proceed from describing children's participation from a theoretical point of view to confronting such argumentation based on previous research. Examples and experiences from further research is a necessity for being able to elaborate on what participation actually means and which level of participation that is most appropriate given a certain context or target group. Further research is also needed to investigate to what extent there is a more beneficial outcome from the interventions, depending on whether the children are involved in the development of the intervention or if the intervention is developed solely by the researchers. --- Abbreviations CINAHL: Cumulative index to nursing and allied health literature; ERIC: Education resources information center; MESH: Medical subject headings; PsycInfo: Psychological information database; UN: United Nations Authors' contributions All authors made significant contributions in the scoping review, by together identifying the research question and design the scoping review. identified and selected relevant studies, charted the data, and collated, summarised and reported the results while have contributed with support throughout the process. All authors have read and approved the final manuscript. --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Greater interest is being shown in participatory approaches, especially in research on interventions that concern children and young people's health and well-being. Although participatory approaches have user involvement in common, they differ in terms of the explicit guidance on how to actually involve and engage children and young people in health research. The aim of this scoping review was to systematically map recent research involving children and young people in the development of interventions targeting issues of health and well-being. Methods: An interpretative scoping literature review based on: a scientific literature search in (health and social science) databases, reference lists, a manual search in key journals and contact with existing networks was conducted. A total of 4458 references were identified through the literature search, of which 41 studies published between 2000 and 2017 were included in the review. The target population was children and young people under 25 years old. Level of participation was categorized according to Shier's Pathways to Participation Model. Results: The review showed that participatory approaches were most often used in the development of interventions in school settings and in community and healthcare settings and on issues concerning support in lifestyle or in managing illness or disease. The level of participation varied from children and young people taking part just as active informants, through stages of greater participation both in quantitative and qualitative terms, to children and young people becoming an active agent involved as a co-researcher where the research process was shaped by views of a higher level of mutuality. Most of the studies were categorised at a medium level and only three studies were judged to involve the children and young people at the highest level. Conclusions: This scoping review showed that work remains in enabling children and young people to influence the development of interventions targeting health and well-being. In relation to level of sustainability in the interventions, it is relevant that goals, strategies and processes are formulated by those who can gain from the interventions. Participatory approaches aiming for a higher level of participation where children and young people work together with the researchers in partnerships are thus warranted.

Background Obesity has physical, psychosocial, and economic consequences during childhood and across the lifespan. Effects at an early age include type 2 diabetes, hypertension, dyslipidemia, sleep apnea, negative body image, low self-esteem, and depressive symptoms [1]. Obesity during childhood also tends to track into adulthood [1] during which it is associated with cardiovascular disease, diabetes, several types of cancer, sleep disorders, and all-cause mortality [2]. As the United States population ages, the long-term effects of youth obesity are likely to become a greater public health burden. Using data on adolescent obesity in 2000 and historical trends of obesity tracking into adulthood, Bibbins-Domingo et al. [3] projected that coronary heart disease prevalence in the US will increase 5-16% by 2035 as a result of obesity among today's adolescents, resulting in an increase of 100,000 prevalent cases. The prevalence of obesity among 12-19 year-olds increased from 5% in 1976-1980 to 16% in 1999-2000 [4], but in recent years there has been conflicting evidence of whether it has reached a "plateau." Ogden et al. [5] reported no linear trend across 2-year periods from 1999 to 2008, based on data from the National Health and Nutrition Examination Survey . In contrast, Bethell et al. [6] and Singh et al. [7] both reported that the prevalence of obesity increased among 10-17 year-olds from 2003 to 2007, based on data from the National Survey of Children's Health . Singh et al. [7] also reported substantial between-state variation in obesity trends, as changes in the prevalence ranged from -32.1% to 45.8% . Such variation raises the question of whether that US is achieving one goal of Healthy People 2010 at the expense of another [8]. The discrepancies between national obesity trends estimated in NHANES and NSCH, along with the state variation reported by Singh et al. [7], highlight the need to continue to investigate national and state trends using different data sources. In a systematic review, Rokholm et al. [9] recommended that future studies explore: non-linear trends in obesity prevalence and behavioral changes that may account for the stabilization. As the authors discussed, investigating the association between trends in weight-related behaviors and changes in obesity trends is an opportunity to explore causes of the obesity epidemic [9]. This opportunity is salient when examining both national obesity trends and between-state variation in trends. Many states have introduced legislation designed to reduce obesity by targeting different weight-related behaviors, but legislative activity has varied considerably by state [10]. Identifying behavioral trends that distinguish states that reversed or attenuated youth obesity trends from states that did not can provide insight as to how certain states have been successful in preventing further increases in obesity. Behavioral differences are just one of many potential source of state variation in obesity trends. Geographic disparities in obesity are a common finding worldwide [11][12][13][14][15][16][17][18] and can result from a complex interplay of sociodemographic, behavioral, and environmental differences. Du et al. [19] argued that many longitudinal studies ignore population shifts that occur during follow-up, and such demographic changes must be considered particularly when studying larger geographic areas such as states that are more dynamic. Singh et al. [17] identified many behavioral, demographic, and contextual sources of cross-sectional geographic disparities in youth obesity, but did not explore whether these variables were associated with state variation in obesity trends. The objectives of this study were to build upon previous studies of obesity trends and address the gaps noted by Rokholm et al. by: estimating between-state variation in adolescent time trends in body mass index percentile and weight-related behaviors from 2001 to 2007, using Youth Risk Behavior Survey data, testing for nonlinear trends, and identifying behavioral, demographic, and contextual factors that distinguish states with disparate trends. --- Methods --- Sample The time-series design combined cross-sectional samples from the 2001, 2003, 2005, and 2007 YRBS, a biennial survey of 9th-12th grade students, administered on the national, state, and local level. Data used in this study came from the state-level YRBS. Participation by states was voluntary in each year and individual students were not followed over time. Students were sampled by the state using a two-stage cluster sampling design, and data were weighted according to school and student response rates to produce estimates that are representative of the state jurisdiction [20]. Data were weighted only in states that provide appropriate documentation and have an overall response rate ≥60%. Years in which states did not fit these criteria were excluded from analyses because their data were not considered representative. To allow estimation of non-linear trends, we limited analyses to states that provided height and weight data in --- Variables Anthropometric and Behavioral Data-Data were collected using a written questionnaire administered to students in classrooms. Our outcome of interest was BMI percentile, calculated from self-reported height and weight. BMI percentile accounts for developmental differences between boys and girls of different ages by measuring each student's BMI relative to a reference population composed of children of the same age and sex in the US from 1963 to 1994 [21]. Brener et al. [22] studied the validity of self-reported height and weight data in YRBS and found that students over-reported their height by an average of 2.7 inches and underreported their weight by 3.5 lb, but the correlations between measured and self-reported BMI was 0.89. We focused on trends in BMI percentile rather than obesity for several reasons. The 95th percentile of BMI is commonly used to define obesity among youth [23], but the ability of this cut-point to identify youth of excess fatness can vary by race/ethnicity, particularly among girls [24]. Weight gains in childhood have also been associated with cardiovascular risk factors independent of weight classification [25,26]. Finally, Ogden et al. [5] noted changes in the BMI distribution that may not be reflected in changes in obesity prevalence . To facilitate comparisons of our results with other studies, we also repeated analyses of time trends using obesity as the outcome. Our analyses included all nutrition and physical activity behaviors that were measured in the majority of states in each survey year from 2001 to 2007: sports played in the past 12 months, days of physical education per school week, hours of television watched per school day, and daily consumption of 100% fruit juice, fruit, salad, potatoes, carrots, other vegetables, and milk. Daily servings of fruit, salad, potatoes, carrots, and other vegetables were summed to create a measure of fruit and vegetables per day. All behaviors were hypothesized to be negatively associated with BMI percentile except TV viewing, which was hypothesized to be positively associated with BMI percentile. Demographic Variables-The demographic variables of interest were self-reported age, gender, and a 4-category measure of race and ethnicity: non-Hispanic White, non-Hispanic Black, Hispanic, and non-Hispanic Other. Contextual Variables-Singh et al. [17] found that state-level poverty rate, income equality, and violent crime rate statistically accounted for 17, 16, and 7% of the state variance in youth obesity respectively, independent of individual-level behavioral and demographic variables. We obtained data on these variables for each survey year to determine if they were similarly associated with adolescent BMI percentile trends. Statelevel data on household income inequality, as measured by the Gini coefficient, were obtained from the US Census, as were poverty rate and violent crime rate [27]. Multiple studies have examined the impact of cigarette prices and taxes on adult BMI trends, with two studies finding a positive association [28,29] and one study documenting a negative association [30]. This topic has not been explored in adolescents to our knowledge. Given that youth are responsive to changes in cigarette prices [31], we examined whether state-level changes in cigarette taxes were associated with state BMI percentile trends, using data from the Tax Foundation [32]. --- Statistical Analyses Overall Time Trends-All analyses accounted for the complex sample design in YRBS [20]. As a preliminary analysis, we estimated the time trend in BMI percentile and all weight-related behaviors in the sample overall. We used a linear model with a robust standard error to account for state clustering in this part of the analysis [33]. In the model of BMI percentile trends, we tested for an interaction between gender and year and found that it was significant. Subsequently, all analyses were stratified by gender. We chose not to test for an interaction between race/ethnicity and year because the primary purpose of this study was to estimate state disparities, and several states did not have sufficient racial/ ethnic diversity to accurately estimate within-state time trends by race/ethnicity. After stratifying by gender, we tested for non-linear time trends by including a quadratic term for time in both gender groups . --- Associations Between BMI Percentile and Weight-Related Behaviors- Subsequently, we used a linear model with robust standard errors to estimate the associations between behavioral variables and BMI percentile. Gender-specific models adjusted for race/ethnicity, age, and all contextual variables. Juice consumption and TV viewing were modeled as continuous variables, and the following behaviors were dichotomized: sports , PE attendance , fruit/vegetable consumption , and milk consumption . The cut-point for milk consumption was chosen after exploratory analyses revealed that the bivariate association between milk consumption and BMI percentile in the sample was decidedly nonlinear, with an obvious change in the direction of association at 4 glasses per day. Between-State Variance in Time Trends-Linear mixed models were used to estimate variance in the time trends of BMI percentile and weight-related behaviors across states. The notation for the mixed models was adapted from Murray [34]. BMI percentile is used as an example, but identical models were used to estimate time trends in each weight-related behavior. For this part of the analysis, all variables were modeled as continuous outcomes: BMI i:jk represents the BMI percentile of the ith individual nested within state j in year k. Mixed models account for state clustering by estimating an overall intercept, β 0 , as well as a random intercept by state, . The overall intercept can be interpreted as the mean state-level BMI percentile in 2001. T t k represents the linear time trend, or the average change in mean BMI percentile per year across states. S j T t k , , represents a random slope for time by state. When trends were found to be non-linear, we added another random component to estimate the variance of the quadratic component of the estimated time trend. We calculated the intraclass correlation for the intercept and time trend . The ICC represents the proportion of variance in BMI percentile due to state differences in each parameter. All random parameters were allowed to covary, allowing us to determine if, for example, states with a higher mean BMI percentile in 2001 had a larger or smaller increase in BMI percentile from 2001 to 2007. Mixed model analyses were conducted with Mplus Version 5.21 to incorporate the sampling weights in the YRBS [35]. All other analyses were conducted with Stata 11 [36]. State-Level Case-Control Analysis-After estimating between-state variance, we explored differences between states in which BMI percentile increased between survey years and those in which it decreased. We created a data set in which state was the unit of analysis and each 2-year survey interval was a separate observation, meaning states would appear in the data set 3 times if data were collected in each survey year. Each observation was categorized according to whether the mean BMI percentile increased or decreased during that interval in that state. We used t-tests to compare state/interval observations in which BMI percentile increased to those in which it decreased-analogous to a case-control analysis-with respect to changes in demographic, behavioral, and contextual variables during the same interval. --- Results --- Descriptive Statistics Table 1 provides the sample size, mean BMI percentile, obesity prevalence, and distribution of demographic variables overall, and descriptive statistics of state participation and withinstate sample size and student response rates. BMI percentile and prevalence of obesity, as defined by the 95th percentile of BMI [23], both increased across years. Across states, the mean BMI percentile in 2001 ranged from 52.7 in Utah to 65.5 in Washington D.C. BMI percentile clustered by region, as most states in the Southeast had a mean percentile above the sample median, while all states in the Rocky Mountains had a mean percentile below the median. --- Estimated Trends in BMI Percentile and Weight-Related Behaviors Table 2 displays the estimated 2001 mean and time trends in BMI percentile and weightrelated behaviors, by gender. The mean BMI percentile in 2001 was 57.67 and 63.57 among girls and boys, respectively. Among boys, the mean increased modestly over time, in a linear fashion, by 0.18 units per year : 0.07, 0.30). Trends among girls were non-linear, and the linear and quadratic coefficients for time suggested that the mean BMI percentile increased throughout the study time period but the magnitude of increase declined from 1.02 units in 2001-2002 to 0.23 units in 2006-2007 . Analyses of obesity trends showed a linear increase over time among girls , but no change among boys. TV viewing, milk consumption, and 100% fruit juice consumption decreased across years in both genders, but the change per year was <0.05 units. Milk consumption, for example, decreased each year by 0.02 glasses per day among girls, or 1.7% of the 2001 mean. PE attendance, sports participation, and fruit and vegetable consumption did not change over time in either group. --- Associations Between BMI Percentile and Weight-Related Behaviors Cross-sectional associations between each behavior and mean BMI percentile on the student level are displayed in Table 3. TV viewing was the only behavior that was associated with mean BMI percentile in the hypothesized directions among both boys and girls. Playing at least one sport was negatively associated among girls, but positively associated among boys. Greater juice consumption was also associated with lower mean BMI percentile among girls. Among boys, higher mean BMI percentile was associated with drinking four or more glasses of milk per day or having at least 1 day of PE per week. Other estimates were close to the null. Table 3 also displays the association between sociodemographic and contextual variables and BMI percentile. Mean BMI percentile was substantially higher among non-Hispanic Blacks and Hispanics, relative to non-Hispanic Whites, among both boys and girls. Poverty status was positively associated with mean BMI percentile among boys , while income inequality was positively associated with mean BMI percentile among girls . --- Between-State Variance in Trends Though the behavioral changes over time were small overall, Fig. 1 shows that time trends in most behaviors varied by state in magnitude and even in direction. Each line represents the time trend among girls for a different state, as estimated by the mixed model. Graphs of state trends among boys were qualitatively similar , and the intraclass correlations for both genders are displayed in Table 4. Even though the proportion of variance attributable to state differences in time trends was low for each behavior , Fig. 1 shows a heterogeneous combination of state trends in fruit and vegetable consumption, sports participation, and PE attendance. For each behavior, some states experienced an increase over time while others experienced a decrease. All states experienced a decrease in milk and fruit juice consumption among girls, but the range of decline was small State trends were negatively correlated with their 2001 mean for each behavior, indicating that states with a lower mean in 2001 had a greater increase over time. ICC 0 estimates indicate that state differences accounted for 1-2% of the variance in sports participation, fruit and vegetable consumption, and fruit juice consumption in both genders, as well as PE attendance in boys, and 5-7% of the variance in TV viewing and milk consumption, and PE attendance in girls. In contrast to Fig. 1, the state variance in BMI percentile time trends was very low , particularly among boys . The ICC T was higher for girls, but the shape of the trend was very similar across states. BMI percentile increased from 2001 to 2007 among girls in all states, but the rate of increase generally declined over time. --- Case-Control Analysis Table 5 displays the results of the state-level case-control analysis. TV viewing was the only variable that distinguished states in which BMI percentile increased from those in which it decreased in both boys and girls. For example, during 2-year YRBS intervals in which states experienced a decrease in BMI percentile among girls , average TV viewing declined by 0.08 h per day, compared to a decline of only 0.01 h per day in intervals in which states experienced an increase in BMI percentile among girls . Changes in cigarette taxes, juice consumption, and the distribution of race/ethnicity also differed across categories of states among girls. During intervals when BMI percentile increased, there was a greater decrease in the proportion of non-Hispanic Whites ; conversely, the proportion of non-Hispanic Blacks increased by 0.51 percentage points during intervals when BMI percentile increased and decreased by 0.23 percentage points during intervals when BMI percentile decreased . --- Discussion Among 9th-12th grade students, states experienced different trends in weight-related behaviors from 2001 to 2007, but BMI percentile changes over time were similar across states. These results are not contradictory given that both the behavioral trends and the association between the behaviors and BMI percentile were modest in size. Sports participation among girls, for example, had a relatively high association with BMI percentile, but its prevalence was stable from 2001 to 2007, and the change in prevalence did not differ between states with different BMI percentile trends. Among girls, differences in BMI percentile between states may be attributable to demographic shifts in the racial distribution more than behavioral changes. It should be noted that our estimates of behavioral trends and associations between behaviors and BMI percentile may have been attenuated by measurement error or unmeasured confounding. Overall, however, results suggest that the measured behaviors do not account for recent trends in adolescent BMI percentile, which increased from 2001 to 2007 particularly among girls. Among the behaviors that we examined, TV viewing had the strongest evidence of being associated with BMI percentile on the student level and BMI percentile trends on the state level. The positive association on the student level, coupled with the decline in TV viewing, might seem to contradict the overall increase in BMI percentile. However, TV viewing declined primarily in states in which BMI percentile decreased. This finding suggests that the decline may partially account for the decelerating BMI percentile trend overall. TV viewing also had the highest overall ICC of any behavior, implying that it may be influenced by state-level factors. Future research could explore state-level determinants of TV viewing and determine if adolescents have replaced TV viewing with alternative pursuits that are more active or sedentary . The lack of state variance in BMI percentile trends is interesting given that legislative activity to reduce childhood obesity has varied considerably across states [10]. Arkansas, for example, has taken aggressive action to remove vending machines from schools and require that students' BMI be measured and results confidentially mailed to parents. After the legislation was passed in 2003, researchers reported that obesity did not increase in Arkansas between the 2003-2004 and 2004-2005 school years [37]. Arkansas' results are encouraging, but they must be considered in light of the fact that the state had the 2ndhighest BMI percentile among states in our sample in 2001. We found that populations with a higher BMI percentile at baseline generally experienced less BMI percentile gain, and thus Arkansas' stability may simply be due to a nationwide pattern of heavier populations reaching a plateau. States with similar BMI percentile distributions at baseline would need to be compared to support the causal effect of legislative changes. Similar to the studies of obesity time trends conducted with NSCH data [6,7], our analyses of YRBS data suggested an overall increase in obesity from 2001 to 2007 that was particularly strong among girls. The discrepancies in these results and those reported by Ogden et al. [5] may be due to the fact that NHANES measures height and weight directly while YRBS and NSCH rely on student and parental self-report, respectively. YRBS height and weight data are prone to measurement error [22], and we likely underestimated adolescent BMI percentile as a result. This would not create a bias toward detecting a time trend, however, unless measurement error decreased over time. In addition to being limited to self-reported height and weight, our analyses were limited to self-reported behaviors that are also prone to measurement error [38,39]. Non-differential error may have caused us to underestimate the association between these behaviors and BMI percentile. Even when associations were detected, we were unable to assess causality between weight-related behaviors and BMI percentile because the analyses were based on a series of cross-sectional surveys and did not measure within-student changes over time. Our analyses were also limited to the 29 states that provided representative data in enough survey years during the study period. Results cannot be generalized to other states, and we were unable to estimate state variation in time trends by race/ethnicity because of the limited sample. BMI is also an imperfect measure of adiposity because it does not distinguish between fat mass and fat-free mass, and this limitation may explain the positive associations that we found between BMI percentile and physical activity behaviors among boys. Boys who play more sports and attend PE class more often may build up more muscle mass, resulting in a higher BMI even if their body fat distribution improves. Boys who consumed four or more glasses of milk per day were also found to have a higher mean BMI percentile than those who consume less. This finding is similarly difficult to interpret because the YRBS does not distinguish between types of milk; additional research would be needed to determine if the association was due to the quantity of consumption per se or the specific types of milk that boys consumed. Similar to the behavioral variables, contextual variables that we examined did not account for state trends in BMI percentile even when they were associated with BMI percentile on the student level. Of all the variables that we examined, race/ethnicity had the strongest evidence of association with BMI percentile on both levels. The large differences in BMI percentile between both non-Hispanic Blacks and Hispanics, compared to non-Hispanic Whites, is not surprising given the consistent racial/ethnic disparities in obesity [40]. Many societal trends that we did not examine have been hypothesized to contribute to the contemporary "obesogenic" environment, such as increased portion sizes [41], urban sprawl [42], and growth of the processed food sector [43]. These trends may impact other behaviors such as soda consumption, which the YRBS did not measure until 2007, or measurement units that are not captured by YRBS . Future research should explore if such trends have contributed to recent increases in youth adiposity. Our study was strengthened by using multiple years of data designed to be representative of individual states. The sampling design of YRBS and sample size of our analysis provide evidence that adolescent BMI percentile has not peaked in many states even if youth obesity prevalence is no longer increasing in the US overall [5]. Moreover, the fact that BMI percentile trends differed by gender in both slope and shape, and obesity trends differed by state in NSCH [7], indicates that national trends are complex and dynamic, and must be carefully deconstructed in future research. National, state, and local health organizations should continue to monitor adiposity patterns to determine if subgroups with a lower BMI percentile are continuing to gain weight. The lack of increase in obesity in individual states, such as Arkansas, is encouraging. However, rigorous studies are needed to determine if legislative action caused the stability, so that states in which obesity continues to rise can apply effective methods to alleviate the trend.

Evidence is conflicting as to whether youth obesity prevalence has reached a plateau in the United States overall. Trends vary by state, and experts recommend exploring whether trends in weightrelated behaviors are associated with changes in weight status trends. Thus, our objective was to estimate between-state variation in time trends of adolescent body mass index (BMI) percentile and weight-related behaviors from 2001 to 2007. A time series design combined cross-sectional Youth Risk Behavior Survey data from 272,044 adolescents in 29 states from 2001 to 2007. Selfreported height, weight, sports participation, physical education, television viewing, and daily consumption of 100% fruit juice, milk, and fruits and vegetables were collected. Linear mixed models estimated state variance in time trends of behaviors and BMI percentile. Across states, BMI percentile trends were consistent despite differences in behavioral trends. Boys experienced a modest linear increase in BMI percentile (β = 0.18, 95% CI: 0.07, 0.30); girls experienced a nonlinear increase, as the rate of increase declined over time from 1.02 units in 2001-2002 (95% CI: 0.68, 1.36) to 0.23 units in 2006-2007 (95% CI: -0.09, 0.56). States in which BMI percentile decreased experienced a greater decrease in TV viewing than states where BMI percentile

Introduction We know, so far, that the idea of an integrated disaster risk management has been around for at least 3 decades. Starting from the 1990s, conversations on integration and disaster risk management have been intertwined with concepts such as sustainability and climate change. Nevertheless, conceptualizing IDRM has been elusive partly because it has never taken a central place in the disaster discourse and partly because "integration" tends to mean a lot of things to a lot of people and fields, from system research to sociology and anthropology. In this work, we investigate IDRM from an international perspective and analysis in national contexts, adopting a politico-economic and social constructionist approach. By conducting a literature review and analyzing key definitions from selected works, we dig into how the concept of integration has emerged, and we ask what elements of a DRM system are necessary to consider it "integrated." Some of these questions, such as where did the idea of IDRM come from, what does IDRM really mean, and how can we assess or evaluate "integration" in a national context, have guided this research. The starting point is a basic yet significant interpretation of DRM. According to widely used international definitions, DRM "is the application of disaster risk reduction policies and strategies to prevent new risk, reduce existing risk and manage residual risk, contributing to the strengthening of resilience and reduction of disaster losses" . Although useful for some contexts, this definition leaves aside key elements that later will be fundamental to better understand "integration" in the context of DRM. This is the case of understanding 1 3 DRM as a social product intrinsically tied to the way that different groups define risk and their means to "manage" it or reduce it at a specific time in history. This relational approach entails that DRM is a dynamic sociocultural process subjected to the historical process of social formation, routinization of everyday life, and institutionalization, implying that it is embedded in and is the result of societal relations and processes that are historically defined . A systemic view of disaster risk formation/definition, management, and reduction considers that DRM is largely affected by societal everyday life experiences and definitions, including processes of power relations, division of labor, and class, among others. This relational approach is also linked to the systemic nature of risk where disaster risk is associated with cascading impacts that spread within and across systems and sectors through the movements of capital, goods, information, and people across regions and countries . The impacts of the COVID-19 pandemic, climate change, and more recently the war in Ukraine clearly show how the challenges of addressing risk in an interconnected and interdependent world go beyond traditional notions of DRM and risk governance . These crises have also demonstrated the need to better understand "compound risk"amplified by underlying vulnerabilities-and "cascading impacts," as well as the political and societal responses to disasters. Consequently, addressing these complexities from a systemic viewpoint will also require integrating different cultural and politico-economic perspectives and fostering system thinking while implementing key intergovernmental agendas, such as the New Urban Agenda, the Paris Agreement, the Sendai Framework, and the Sustainable Development Goals . Along the literature review, we analyzed different IDRM ideas and experiences-especially from China, Japan, and Mexico-to later explore potential indicator candidates or "proto-indicators" as proposed by Czúcz et al. . These proto-indicators are considered entry points for further discussions on IDRM. In total, 29 proto-indicators were found, grouped in three meta-categories that relate to different kinds of integrations detected in our study: sectoral, spatial/ hierarchical, and temporal. We consider that a prospective analysis may help to guide the discussion and support strategic planning for IDRM implementation in national and international contexts. Finally, the reflections collected in this study may help to guide upcoming discussions and implementations of IDRM, as expressed in the Sendai Framework for Disaster Risk Reduction 2015-2030 , and especially in a context where integration between complementary global agendas is crucial. --- Methodology In order to examine different approaches on IDRM at an international level, we adopted a literature review strategy following a Preferred Reporting Items for Systematic Reviews and Meta-Analyses approach to existing published material . We identified a list of databases to find articles and publications about the integrated management of disasters and risks. The databases selected were Web of Science, Scopus, and Google Scholar; the Google search engine was used to look for other relevant websites. The second step entailed the definition of inclusion search criteria, screening titles and abstracts with related terms or related concepts such as "integrated management" and "disaster risk." Articles were excluded if they were editorials, opinions, or commentaries without any substantial evidence independent of the study design and if the content was unrelated to the topics or not available on the Internet . Additionally, we reviewed the databases of several international organizations, such as the United Nations Office for Disaster Risk Reduction , the German Agency for International Cooperation , and the Integrated Disaster Risk Management Society, among others, as well as the reference lists of the articles judged to be relevant to the topic of interest. As the content structure of the selected documents tends to differ , we adopted a qualitative approach to content analysis . Documents were assessed on MaxQDA 22.0.1, looking at potential definitions and description of empirical cases related to IDRM while considering a historical perspective . We developed content categories to help us define IDRM at an international level in terms of what elements are necessary to consider a DRM system integrated, for example, multi-hazards approach to risk management and consideration of inter-and transdisciplinary research on risk, among others. Based on this, we briefly explored some results of the literature review with documented experiences from China, Japan, and Mexico. Finally, we refined these content categories into "ideas for indicators" using Czúcz et al.'s approach as they may work to explore and analyze IDRM in specific national contexts. In the 1994 Yokohama Strategy, the concept of integrated management of disaster risks pointed to a better strategy to achieve goals and objectives. It assumed that disaster response approaches alone are not sufficient to reduce disasters and risks; on the contrary, it presented disaster prevention as a fundamental element: "it contributes to lasting improvement in safety and is essential to integrated disaster management" (UN-IDNDR 1994, p. 2). Interestingly, the idea of "integration" that influenced the Yokohama Strategy can be tracked back to the 1992 Rio Declaration , where the integration of environmental and development concerns was the paramount goal. Figure 2 illustrates landmarks in the historical development of IDRM ideas. Between the 1960s and 1980s, the international debate tended to concern how to provide coordinated and efficient humanitarian assistance after disasters. This period started with the international efforts to provide humanitarian assistance to Iran after the Buin Zahra disaster in 1962, and it culminated with the creation of the United Nations Disaster Relief Office in 1971. This period is characterized by a "reactive" approach to disaster as well as a lack of awareness about the underlying causes ). During the 1990s, the foundations for IDRM were laid by incorporating the idea of sustainability and development into the discourse of disaster and risk reduction. Although global objectives still pointed to a reactive approach in 1994, there was a growing awareness of DRC and the necessity to "integrate" different actors and development processes. This type of integration is defined by Wisner as "integration by adding," and it is characterized by a growing institutionalization of DRM with a multi-disciplinary approach. By 2002, the Johannesburg Plan on sustainable development requested actions under the chapeau: "An integrated, multi-hazard, inclusive approach to address vulnerability, risk, assessment and disaster management, including prevention, mitigation, preparedness, response and recovery, is an essential element of a safer world in the twentyfirst century" . This supports a holistic approach that integrates DRM and DRR efforts and principles into sustainable development processes. However, it was the Hyogo Framework that prioritized the multi-sectoral integration of DRR among relevant actors beyond the conventional domains of DRM and stressed the need to integrate DRR into development policies at all levels of government. For Zhang et al. , IDRM means the management of all types of disasters and all phases of disaster management, focusing on hazard and vulnerability, including the underlying causes of risks, and emphasizing a multi-level, multi-sectoral, multi-disciplinary coordination among stakeholders. In Zhang et al.'s words, IDRM "represents a kind of philosophy and concept, a kind of basic system arrangement, a kind of integrated flow, a kind of mode and scientific method and a kind of future trend of disaster management" . Likewise, the Sendai Framework deepens these notions of integration by focusing on reduced vulnerabilities and exposure and strengthening resilience. For example, recent mid-term reviews point out "risk-informed sustainable development" to stress the linkages between DRR and development , such as the underlying causes of inequality, poverty, and other forms of vulnerabilities. The Sendai Framework asserts that the realization of DRR requires "the implementation of integrated and inclusive economic, structural, legal, social, health, cultural, educational, environmental, technological, political and institutional measures that prevent and reduce hazard exposure and vulnerability to disaster, increase preparedness for response and recovery, and thus strengthen resilience" . This aspirational goal highlights the relevance of vertical and horizontal integration through DRR and DRM policies articulated from international to local levels and among different actors and cross-cutting societal issues, such as the Sustainable Development Goals and climate change. Unfortunately, some studies have pointed out the Sendai Framework's lack of concrete connections with these societal issues and processes: "despite the rhetoric of vulnerability, the measurement of progress towards DRR remains event/hazard-centric" . Wisner refers to this overall integrative approach as "doing things differently" and calls for development in a "risk aware manner." The latter means "integrating" the awareness of disaster risk creation and reduction into other aspects of development. For instance, the World Bank's Global Facility for Disaster Reduction and Recovery has shown how sectors such as water, agriculture, and housing have been made risk-sensitive in a number of countries. In this sense, integrated DRM can be seen as a process that involves the whole society and its activities and relations on all scales. A relational approach-in Holling's ecological sense-is particularly useful here. Complex systems where nature and society are interconnected and interdependent do not behave deterministically, predictably, and systematically, but "systemically," with complex feedbacks that cannot be conclusively described and that "cross" multiple scales, temporally and spatially . Then, a systemic understanding of integrated DRM should start by considering the general conditions of the interplay of stabilizing and destabilizing variables that lead to a system maintaining a certain coherence and structure despite rapid and radical variation of internal elements and the surrounding environment . In other words, "change" on multiple scales should be considered the norm, with periods of slow, faster, and suddenly radical changes, whereby different spatial and temporal scales interact . This is the context for IDRM. It is with this background that Shi emphasized the importance of governance in IDRM and of mapping the roles of the various actors to understand the complexity of integration: "The range of actors and the design of the processes vary a great deal depending on the specifics of the political systems as well as the socioeconomic and cultural contexts." However, examining IDRM actors and their relationships, as well as structures and functionality of IDRM processes is methodologically challenging. Analyzing IDRM in a specific country will require a spatiotemporal analysis of actors, processes, and relationships that govern how disaster and risk are created/avoided, managed, and reduced in each society. According to Alcántara-Ayala , there is a paradigm shift toward an integrative change in DRM when governance is involved, "as overarching efforts need to be strengthened first and foremost within the scope of regulatory frameworks and mechanisms of implementation." In 2019, Alcántara-Ayala and colleagues pointed out the systematic nature of IDRM and the importance of addressing the root causes of disaster vulnerability and exposure, defining it as: A complex systematic process consisting of a series of decisions, actions and activities, as well as a transver-sal coordination between the different institutional and social actors, to understand and transform the needs and weaknesses expressed in the different aspects of vulnerability and exposure, in specific responses and collective solutions, whose main objective is the deconstruction of risk. Similarly, Blümel et al. analyzed different sectors-that is, DRR, healthcare, humanitarian assistance,business management, and education-aimingat establishing a working definition of the term "integratedmanagement." The authors define IDRM as: A complex and dynamic societal process in which all aspects perceived as relevant by horizontally and vertically as well as internally and externally cooperating actors are understood in their context, and corresponding effective and efficient measures are taken in a coordinated manner by all relevant actors to prevent crises or disasters and, in case of their occurrence, to avert harm and ensure the well-being of the people at risk under dynamically changing conditions. Like Alcántara-Ayala and Wisner , Blümel et al. underlined the "coordination" of a multiplicity of actors working on different geographical scales and at all DRM phases with a greater focus on prevention and the underlying factors of risk. One novel aspect refers to "horizontal" and "vertical" integration in contrast to "internal" and "external." For example, in the field of science, "horizontal integration" refers to the coordination of different disciplines such as in multi-hazards research and also in transdisciplinarity. On the other hand, "vertical integration" in the public sector refers to the cooperation of different government levels-national, regional, local, and international. Besides these, we encountered diverse cases where IDRM ideas are used in more specific domains, namely "integrated landslide risk management," "integrated flood risk management" , and "integrated urban risk management," among others. Such approaches are indeed practical examples illustrating a variety of actors working on or influencing the DRM process. For example, Wang et al. studied IFRM as an effective method to reduce damage from floods in Beijing, China, while Mercado et al. studied the governance aspects of IFRM in the case of Metro Manila, Philippines. During the analysis we also detected several peer-reviewed papers and institutional reports linked to specific countries: China, Japan, and Mexico. As the literature addressing the term IDRM is in fact limited, we decided to explore these outputs. Some of the following examples reveal that IDRM ideas have moved from academic discourses to practical approaches and actual policies. In China, according to Shi , IDRM is seen as a "basic feature of sustainable development." Shi explored past disasters in China to point out key aspects of the Chinese government's role in IDRM: overall leadership, engaging civil society, and international cooperation. The World Bank highlighted Shi's observations, especially about the government's role in overall leadership; however, it also criticized its "vertical leadership" as interests between local, provincial, and national levels in DRR rarely coincide. Shi and Zhang et al. are some of the few scholars found in our research who specifically look at IDRM in China. Unfortunately, their studies lack comprehensive reflections on how, for instance, China's unique politico-economic system and culture influence IDRM approaches and practice. In Japan, according to Ikeda et al. , an integrated framework for DRM has been developing since the Great Hanshin-Awaji Earthquake in 1995 . Okada et al. illustrated these changes in Japan toward IDRM by contrasting two main DRM approaches: the Jiishubosai-soshiki and Machizukuri. Historically, community organizations known as Jiishu-bosai-soshiki were common and active after disasters. These organizations were oriented toward rescue and relief as well as self-evacuation. After the 1995 Kobe earthquake, Jiishu-bosai-soshiki were influenced by local governments to focus on prevention and preparedness too. In a study on Jiishu-bosai-soshiki, Bajek et al. concluded that they tended to be guided and mobilized by local governments, regularly supplementing government actions on reducing disaster risks in residential areas. In contrast, after the 2011 Great East Japan Earthquake disaster, the Machizukuri approach includes a wide range of local initiatives aimed at reducing disaster risks or mitigating disaster effects. From the viewpoint of IDRM, the difference between Jiishu-bosai-soshiki and the Machizukuri approach is that the latter is holistic and multi-focused; it is therefore not limited to disaster concerns only. Machizukuri is citizen-led , involves multiple stakeholders, and takes account of day-to-day issues instead of focusing on one-time problems. Although community-led DRM has proven to be an effective way to integrate DRM at local levels in Japan, the necessary structural changes to prevent the creation of disaster risks require the involvement of actors and stakeholders at all levels, such as national and regional institutions. In Mexico, according to Alcántara-Ayala et al. , the 2012 General Law on Civil Protection established the basis for an integrated management of disaster risks, defining IDRM as: The set of actions aimed at the identification, analysis, evaluation, control, and reduction of disaster risks, considering them from their multifactorial origin and in a permanent process of construction. It involves the three levels of government, as well as all societal sectors, to facilitate the creation and implementation of public policies and strategies which are integrated to sustainable development policies and principles. These take on the structural causes of disasters and strengthen the resilience capacities of society. Integrated actions cover the stages of risk identification and/or its process of formation, forecasting, prevention, mitigation, preparation, response, recovery, and reconstruction. Despite this holistic statement, some studies showed that in Mexico traditional views persist in DRM that rationalize the current implementation of this law . On the other hand, Alcántara-Ayala et al. analyzed "integration" beyond DRM institutions by looking at urban development and environmental policies, budgeting, and governability processes, such as accountability. This novel angle tries to capture the comprehensiveness of DRM and DRR processes beyond the traditional and often compartmentalized DRM and civil protection domains. Interestingly, while this approach pays close attention to the interconnections and interdependencies of different public systems, the research leaves aside other relevant dimensions and sectors detected in our literature review, such as the role of academia and research institutions, civil society, and the private sector. These sectors play important roles in integrating different phases of DRM . Based on Puente and Oliver-Smith et al. , Alcántara-Ayala et al. proposed that the transition from traditional DRM to IDRM could be based on at least five normative principles: efficiency and equity, integrality, transversality, co-responsibility, and accountability. In the case of Mexico, these principles intend to guide policy design and the implementation of IDRM. Unfortunately, no recent evidence on how this approach is being taken up by Mexican authorities or recent examples of policy design were found. The short examples from China, Japan, and Mexico show that an integrated management of disaster risks may be moving from a conceptual debate to an actual implementation debate. The examples reveal, nonetheless, that there is not a one-size-fits-all way to approach IDRM, and a mechanism or framework to assess the integration of DRM at country and city levels does not exist. The following section deepens these issues and proposes key elements to advance the study and assessment of IDRM. --- Discussion and Proposals: Proto-Indicators for Advancing Integrated Disaster Risk Management The need for an integrated approach to DRM lies in the fact that the root causes of disaster risks and vulnerability are embedded in a complex web of societal processes and actors that goes far beyond the traditional domain of civil protection and risk management. This means that this integration requires looking at development processes and societal and individual relations . Thus, IDRM should not be seen only as a "part" of sustainable development pathways but as a "transversal" element without which sustainability cannot be reached. On the importance of integrating DRR and DRM into other international agendas , there are three identified problems. First, if we use a systemic perspective on DRR/DRM where historically defined processes of social formation are taken into account, "sustainable development," "disaster," and "risk" need to be contextualized and defined in the first place, that is, from within the respective sociocultural orders one is aiming at . This is extremely complex as it requires the development of sociocultural methodological frameworks that orient the navigation within this complexity in different contexts. Voss et al. are developing the conceptual Culture-and-Catastrophe Framework for this purpose. The second problem, as pointed out by Wisner , refers to the relationship of disaster risk with other urgent societal issues, such as climate change, poverty reduction, gender equality, and development in general. As far as disaster risks continue being perceived as a sectoral problem, IDRM will necessarily tend to compete on different scales with other developmental issues, for both attention and resources. The third problem relates to a multi-dimensional integration. As IDRM looks to integrate with other societal agendas, it does not question if, for example, the SDGs and Paris Agreement are internally integrated themselves. This leads to the conclusion that an IDRM can exist at two levels: it can be "internally integrated" when directly related processes of disaster and risk and related stakeholders articulated at different dimensions, including spatially, temporally throughout the DRM phases, and sectorally; and it can be "externally integrated" when DRR are intertwined with other societal processes, such as gender equality, poverty reduction, and sustainable development in general. This type of integration means that DRR and DRM are coupled with multiple systems, namely, climate systems, ecosystems, and human systems . Thus, avoiding, managing, and reducing such risks entail a systemic task that goes from addressing the interactions of different coupled systems to considering the singularities of societies and territories. There are also several challenges in implementing IDRM from a systemic point of view. As pointed out by Alcántara-Ayala , little attention has been paid to integrative approaches to DRM, in part due to the resistance of compartmentalized styles of public management that have encapsulated it into rigid and often instrumentalized institutions. But beyond the changes needed for IDRM, it is necessary first to point out basic ideas that define IDRM in a general sense. On this point, our study identified in the literature 29 elements that could lead towards an identification and assessment of IDRM . We have decided to call these elements "proto-indicators" as they are presented as "ideas" or "candidates" for indicators . The 29 proto-indicators reflect the multidimensionality of IDRM in relation to at least four important dimensions or directions . First is the integration of different kinds of actors, either within or among them -for example, units of reference such as classes, groups, institutions, and social sectors. Regardless of the proposed actors in Fig. 3, the "red axe" can be expanded with other actors and institutions, depending on the specific characteristics of the context where IDRM is being analyzed. Second is integration at different geographical scales or levels, that is, actors or mechanisms of DRM that operate from the national to the local simultaneously and hierarchically. This spatial dimension must also include the integration of states into international and/or upper-regional levels, such as the EU Civil Protection Mechanism in Europe. Third is integration in the temporal dimension , that is, actors or mechanism of DRM that work throughout the main DRM phases , with special focus on prevention and transformative resilience . Fourth is integration of and in different relevant societal processes, such as climate change, development, and urbanization. This type of integration means that DRR/DRM thinking participates or is embedded in systemic operations of other societal processes and that it goes beyond the typical civil protection or risk management organizations. This dimension relates to the division of labor and contrasts with the traditional organization of societal sectors that have tended to work in "silos"-something recently highlighted by the Bali Agenda for Resilience in the 2022 Global Platform for Disaster Risk Reduction . Figure 3 uses a cube-shaped figure to illustrate the integration of DRM across different sectors, scales/hierarchies, and temporally across the DRM phases and resilience pathways . Importantly, integration of DRM requires synergies with other societal challenges and processes, such as the SDGs. --- Conclusion This study analyzed the key literature at the international level with the aim of offering a closer look at the concept of IDRM while exploring potential indicators and conceptual elements that may help to advance IDRM in national and international contexts. One of our first observations is that there is not a concrete or identifiable IDRM approach in the literature but rather a set of ideas and experiences related to what IDRM is and how it should be operationalized. Such ideas and experiences are linked to scientifically digested content and contextualized events, such as "integrated landslide disaster risk management in Mexico" or IFRM in China , but they are rarely based on international and comparable cases. Thus, it is possible to conclude that IDRM research still requires adopting inter-and transdisciplinary approaches that adequately incorporate other forms of knowledge and practices. In this sense, this is also a limitation of the study: analyzing experiences and elements of IDRM that are not systematically documented is a challenge for comparative and international studies, although this could be sorted out in the future with more systematization and data collection . Conceptually speaking, Blümel et al.'s approach to integrated management could count as an exception. They considered IDRM multi-dimensionally and highlighted the differences between internal and external integration as well as horizontal and vertical integration. Nevertheless, Blümel et al. did not reflect on cross-cutting societal challenges, discussed in this work, in relation to the international agreements: the SDGs, the Paris Agreement, and the New Urban Agenda. Importantly, we are not here talking about "external integration" with other sectors but rather with a wide range of sectors , such as those working on climate change and sustainable development issues. In this study we have also proposed that IDRM can be better understood from at least four dimensions: sectoral, spatial/hierarchical, temporal, and externally with other cross-cutting societal challenges. These dimensions are interconnected in a system , and all are necessary to achieve an effective and efficient IDRM. To that end, we proposed a series of 29 proto-indicators to guide the exploration and analysis of IDRM in specific national contexts. Nonetheless, we have Recognition of distant root causes of disaster risks Acknowledgment of the causal chain of disasters and risks that begins with distant "root causes," spatially and temporally. This includes how disaster risks are transmitted through "dynamic pressures" such as weak government, unplanned urbanization, and so on, so shaping "unsafe conditions." Wisner et al. ; Wisner ; Alcántara-Ayala Reduction of multi-dimensional vulnerabilities also detected the need for more comprehensive theories and methodologies to further advance IDRM in this regard. In sum, IDRM encompasses different kinds and levels of "integrations" that go from internal to external , including temporal and spatial integrations. In other words, we are talking about a dynamic and systemic integration of DRM. This makes IDRM something complex to study and especially implement, as it requires a profound understanding of the cultural and social conditions that shape DRR and DRM domains in each context. If these cultural settings are not considered or addressed in the first place, any attempt to advance IDRM may tend to fail. Presumably, inside-out and bottomup approaches that consider co-design and co-production of solutions and governance approaches may have greater chances of success. After all, advances in DRR and DRM are widely accessible, and knowledge is openly shared, but their usage, impacts, and effects will be diverse depending on the institutions and cultures of each society. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. --- References

This study analyzed the international key literature on integrated disaster risk management (IDRM), considering it a dynamic sociocultural process subjected to the historical process of social formation, offering a closer look at the concept while exploring conceptual elements and ideas to advance IDRM in both national and international contexts. Methodologically, the study adopted a literature review strategy, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) approach, combined with qualitative content analysis. This article examines the history of IDRM, discusses current challenges for implementation, looks at some experiences, and proposes avenues for further research. Some findings point out the lack of an overarching IDRM approach, which is characterized by a rather disperse set of ideas and experiences concerning what IDRM is and how it should be operationalized, thereby revealing the need for a more comprehensive theory and methodologies to further advance it. Other findings highlight that IDRM encompasses different kinds and levels of "integrations" that go from internal (that is, disaster risk reduction and management domains) to external (that is, all societal processes such as sustainable development), including temporal and spatial integrations. Hence, we are talking about a multidimensional integration of disaster risk management. In this regard, the article proposes four dimensions for integration: sectoral, spatial/hierarchical, temporal, and externally with other cross-cutting societal challenges, especially climate change and sustainable development. These dimensions cover 29 ideas for indicators or "proto-indicators" to guide the discussion, exploration, and analysis of IDRM in specific contexts.

Background Unmet need for family planning defined by the World Health Organization as "those who are fecund and sexually active but are not using any method of contraception, and report not wanting any more children or wanting to delay the next child" [1]. It is one of the indicators used to determine the achievement of universal reproductive health coverage [2]. Family planning is one aspect of women's rights and constitutes the major feature of their reproductive health [3]. Globally, an estimated 214 million women have an unmet need for family planning services [4]. In developing countries, more than two hundred million women did not receive family planning services in 2012 [5]. In Ethiopia, the 2016 Ethiopia Demographic and Health Survey report reveals that 36% of currently married reproductive aged women years use contraception and more than one in five married women have an unmet need for FP. With a total fertility rate of 4.6 [6]. The same report showed that the Somali, Afar, and Oromia regions of Ethiopia remained the highest total fertility rates with the fertility rate of 7.2, 5.5, and 5.4, respectively [6]. Studies elsewhere in Ethiopia have also revealed that the unmet need for family planning ranged from 16.2 to 37.4% [7][8][9]. As a result of unmet need for FP, high fertility, unintended pregnancy, and unsafe abortion are the main outcomes [10]. Approximately 80 million unintended pregnancies occur each year in the world. This in turn, has an effect on the mothers, children and the society at large [11]. For example, in low and middle income countries an estimated 18 million unsafe abortion takes place each year [12]. Although several interventions have been used to prevent unintended pregnancies, women, of low and middle-income countries are still affected by unplanned pregnancies [13,14]. Failure to fulfill family planning also leads to short birth intervals this which results in both the newborn and the preceding child being at high risk of morbidity and mortality [15]. Religion, women's age, number of living children, place of residence, respondent's education, knowledge of family planning, respondent's work status, being visited by a family planning worker, parity, and number of children at first use of contraceptives were factors associated with unmet need for family planning [7-9, 16, 17]. Existing studies also revealed that inaccessibility of family planning methods, lack of knowledge, religious belief, fear of side effects of contraceptives, and opposition by the husband were identified as reasons for not using contraceptives [18,19]. Understanding the prevalence and predictors of unmet need for family planning helps each country to overcome the problem [20]. It is possible to prevent more than one million infant deaths and 54 million unwanted pregnancies by reducing the unmet need for FP [5]. Besides, it benefits a woman's well-being, autonomy, and educational attainment. In Ethiopia, this study contributes to the field of family planning by identifying different predictors of unmet need for FP at the individual and community levels. The finding is also important for policymakers and program planners to realize women's need for family planning in developing countries. Therefore, this research aimed to assess the magnitude and predictors of unmet needs in selected high fertility regions of Ethiopia. It is also important because it provides rational policy recommendations to help women meet their FP need. --- Methods --- Study design, setting, and period The study was a cross-sectional analysis of data from recent EDHS, which was conducted by the Central Statistical Agency in collaboration with the Federal Ministry of Health and the Ethiopian Public Health Institute , which was a national representative sample conducted from January 18 to June 27, 2016. There are nine regional states in Ethiopia , Harari, and Somali), and two administrative cities , 611 Districts, and 15,000 Kebeles. Afar, Somali and Oromia regions were included in this study. These regions were selected because they are the high fertility rate regions in Ethiopia with fertility rates above 5.0, a value that is higher than the rate of 4.6 in Ethiopia and 2.47 worldwide [6,21]. The health care system in Ethiopia is structured in a three-tier system: primary, secondary, and tertiary levels of care. The primary level of care including primary hospitals, health centers, and health posts), The secondary level of care is delivered by general hospitals and the tertiary level of health care is given by specialized hospitals [22]. high fertility regions of Ethiopia. Any interventional strategies that reduce the unmet need for family planning should consider these factors to overcome the problems in the regions. --- Keywords Unmet need for family planning, High fertility regions, Ethiopia --- Data source, sampling procedure, and study population The study was conducted using the EDHS 2016 data by accessing the official DHS program database at www. measuredhs.com after obtaining permission via an online request by explaining the purpose of the study. A twostage stratified cluster sampling technique was used to select populations using the 2007 Population and Housing Census as a sampling frame. It has been stratified by separating the nine regional states and the two city administrations of Ethiopia, into urban and rural areas. At the first stage, a total of 645 Enumeration Areas proportional to EA size were selected, at the second stage, 28 households per cluster were selected with an equal probability of systematic selection [6]. For this study, we have used the women's individual data set and the source population was currently married reproductive age women during the survey in the study regions . While the study population was all currently married reproductive age women who were in the selected EAs included in the analysis. Currently married reproductive age women who had never sex, and were not sexually active were excluded from this analysis. Data were collected by trained data collectors using pretested, structured, interviewer-administered questionnaires. A weighted sample of 4312 currently married reproductive age women was included in this study . --- Measurements of variables The dependent variable for the current study was unmet need for FP . It is the sum of unmet need for spacing and limiting [23,24]. Reproductive age women who were currently married, and or sexually active have unmet needs for family planning if they desire to space the next pregnancy or limit future pregnancy but not use any methods of family planning. Amenorrheic or pregnant women with unwanted or mistimed pregnancies or births who did not use family planning were also considered as unmet need for family planning [7,24,25]. Of the community level variables, residences were directly accessed from DHS data sets. However, community level poverty and community-level media exposure were constructed by aggregating individual-level characteristics at the cluster level [26]. They were categorized as high or low based on the distribution of the proportion values generated for each community after checking the distribution by using the histogram. The aggregate variable was not normally distributed and the median value was used as a cut-off point for the categorization [26,27]. --- Data processing and analysis Data were extracted from EDHS 2016 and cleaned, recoded, and analyzed using STATA version 14 Statistical Software. Throughout the analysis, sample weights were done to adjust for non-proportional allocation of the sample to strata and regions during the survey process and to restore the representative. Descriptive statistics were described using frequencies, percentages, median, and interquartile range, and were presented using tables, figures, and narratives. A multi-level logistic regression analysis was used after checking the model eligibility. It was assessed by calculating the intra-class correlation coefficient , and a model with an ICC greater than 10% was eligible for multilevel analysis. In this study, the ICC was 49%. Hence, the data were hierarchical . First bi-variable multilevel logistic regression analysis was performed, and those variables with a p-value of < 0.20 were considered in the multivariable analysis. After selecting variables for multivariable multilevel analysis, four models were fitted; the null model , mode I , mode II , and model III . Deviance was used to assess the model fitness, and the model with the lower deviance was the best-fitted model. --- Results --- Socio-demographic characteristics of study participants A total of 4312 weighted samples of reproductive age women in high fertility regions of Ethiopia were included. The median age of the study participants was 29 years old. Nearly half of the study participants were under the age group of 25-34 years, 2806 of women had no formal education. Moreover, the majority of women were rural dwellers and 1868 of women fell in the poor wealth categories . --- Obstetric history of the study participants Of the total study participants, 3436 were multipara. More than half of women had ANC follow-up, and eight hundred and twenty-eight of participants decided jointly with their partners on contraceptives . --- The magnitude of unmet need for family planning The magnitude of unmet need for FP in high fertility regions of Ethiopia was 29.78% , with Oromia recording the highest prevalence of 30.21% . --- Random effects and model fitness The intra-class correlation in the null model indicated that 49% of the overall variability of unmet need for FP was attributed to cluster variability. The median odds ratio for unmet need for FP was 4.52 in the null model, which indicates that there was a variation in unmet need for FP between clusters. This means that if we randomly selected individuals from different clusters, those in the cluster with the highest unmet need for FP had a 4.52 3). --- Predictors of unmet need for family planning In the final model , after adjusting for individual and community level factors, individual-level variables such as education, wealth quintile, media exposure, and parity, sex of household head as well as community-level variables such as residency, were identified as Accordingly, the odds of unmet need for FP among women with no formal education were 1.65 times higher than those women with formal education . Women who had no media exposure had 1.32 times more odds of having an unmet need for FP than those who had media exposure . Multiparous women were found to be 1.57 times more likely to have had an unmet need for family planning compared with those women with primipara . The odds of unmet need for FP among women who live in rural areas were 2.45 times more likely than among women who live in urban areas. The odds of unmet need for FP from households with poor wealth status were 1.67 times higher than women from rich households' wealth status . Those households lead by male were 1.39 times more likely experienced unmet need for family planning as compare to counterpart. --- Discussion This study was conducted to examine the magnitude and predictors of the unmet need for family planning among currently married reproductive age women in high fertility regions of Ethiopia. In the current study, the unmet need for FP was found to be 29.78% in high fertility regions, with Oromia having the highest prevalence of 30.21%. The result of the study showed that level of education, wealth quintile, media exposure, parity, sex of household head, and residency were identified as the predictive factors for unmet need for FP among currently married reproductive age women in high fertility region of Ethiopia. The magnitude of unmet need for FP in the current study is higher than the national magnitude, which was 22% [6]. And significantly higher than previous regional studies in Ethiopia [8,23,28]. It is higher from the national target of reducing the level of unmet need for FP to 10% by 2020 [29]. The magnitude was higher compared with the United Nations sphere standard of unmet need for FP, which is considered high if greater than 25%, and the global estimate of unmet need for FP reproductive age women [27,30]. The finding was also higher than studies conducted in Zambia [31], and Bangladesh [24]. The higher magnitude of unmet need for FP in this study might be due to the fact that the current study was conducted in high fertility regions of Ethiopia; those regions are developing regions; the pastoralist community of Ethiopia; has limited access to health facilities; low access to contraceptives; low awareness of modern contraceptive methods; poor infrastructure; and husbands are strongly opposed their wife to use family planning [32][33][34]. Furthermore, in the study areas where most of them are uneducated, Muslim religion followers and the culture in those regions are not allowed to use family planning even if the women wanted to use it. Members of the Muslim religion are strongly opposed to contraceptive use [35,36]. However, the magnitude of unmet need for FP in the present study was lower than in the study conducted in India [37]. This variation could be attributed to the difference in the target population. The current study was done among reproductive age women while the study conducted in India was done among young women. The evidences showed that usually, young women are far from adequately meeting the needs of family planning and lack the knowledge, agency, or resources to make decisions regarding their reproduction [38,39]. It suggests that even though there was an improvement in contraceptive prevalence among currently married reproductive age women in the regions; achieving the target and maximizing the benefits of contraception requires dedication to provide contraceptives to those women with the identified unmet need. In this study, women with no formal education were more likely to have an unmet need for FP as compared to women with formal education. The finding of this study is in agreement with those of studies conducted in Ethiopia [7,32,33], Nigeria [40], and Malawi [41]. The possible reason might be that women with no formal education are less likely to be exposed to FP through the media and other ways of exposure, which compromise access to contraceptives and have no easy to understand the health benefits of the contraceptive in reducing fertility, unintended pregnancy, unsafe abortion, and other maternal and child problems [42,43]. In addition, non-educated women are less likely to be empowered, which subsequently increases their unmet need [42]. This suggests that education will be one way to reduce the unmet need for contraceptives in these high fertility regions. The likelihood of the unmet need for FP among women from households with poor wealth quintile was higher than those from households with rich wealth quintile. This finding is supported by studies done in Ethiopia [25,44], Nigeria [45], and developing countries [46]. The reason might be that women from poor households cannot be able to deal with the cost barrier associated with access to contraceptive use as compared to those from rich households since they cannot be able to overcome both the direct and indirect costs associated with contraceptive uptake [47]. Another possible reason could be that as income decreases, exposure to different types information and the financial accessibility of services will be compromised [48]. of contraception. The reason for this may be that women with no media exposure might not have a better understanding of contraception, which cannot have a positive change in their attitude toward contraception [50,51]. The study indicates that media exposure will reduce the barriers to access and use of health care services, including contraception. According to this study, the odds of having an unmet need for FP among reproductive age women living in rural areas were 6.16 times higher than their counterparts. This is similar to studies conducted in Ethiopia [52], Nigeria [53], and Bangladesh [54]. This might be due to a variety of reasons; in Ethiopia; rural residents have poor health service accessibility and low awareness of contraceptives due to the fact that rural women are less educated, have limited access to mass media, have insufficient income, and poor infrastructure, which has a positive impact on modern contraceptive use [6,44]. Moreover, evidence revealed a high concentration of sexual and reproductive health services delivery in an urban area in Ethiopia [23,27,55]. The odds of unmet need for family planning were higher among male-headed households as compare to female-headed house. This finding is in line with study conducted in Nigeria [45] and East Africa [56]. This might be due to the female headed household may increase resource gain and control [57]. In this study, parity was found to be one of the predictors of unmet need for FP. Multipara women were 1.77 times more likely to have unmet needs for family planning when compared to their counterparts. This is congruent with the results of other studies [7,23], which revealed multipara women were more likely to use family planning compared to primipara women. The possible explanation might be that the more children the woman is having; the more likely she wants to space or limit the number of children she will have [7,33]. This indicates that the more she uses family planning, the more she has an unmet need. The main strength of this study was that it used nationally representative survey data and focused on high fertility regions in Ethiopia. In addition, the DHS uses validated instruments in its appraisals of datasets along with the large sample size and well-designed procedures, such as training field enumerators and employing welltested methods for data collection. However, since the surveys are cross-sectional design, causality cannot be established for the findings. Additionally, due to the use of secondary data, essential factors like attitude toward family planning methods, husband perspective on FP, and socio-cultural factors were not available in the EDHS data set. Hence, it was not possible to incorporate these variables. --- Conclusion In the current study, the magnitude of unmet need for family planning among currently married reproductive-age women in high fertility regions of Ethiopia was high when compared to the national average and the United Nations sphere standard of unmet need for family planning. Education, wealth index, mass media, parity, sex of household head and residence were independent predictors of unmet need for family planning among reproductive-age women in high fertility regions of Ethiopia. Any interventional strategies that reduce the unmet need for family planning should consider these factors to overcome the problems in the regions. Future researchers interested in this area should also consider qualitative variables such as socio-cultural factors, which might have a great effect on the unmet need for family planning. --- Data Availability This study used data from the 2016 Ethiopian Demographic and Health Survey, which is freely available online at . --- Another factor that predicts the unmet need for FP found in our study was media exposure; A woman who had no media exposure was 1.53 times more likely to have an unmet need for family planning as compared to those who had media exposure. The finding is consistent with studies done in Mali [49], and Nigeria [50], where exposure to mass media has a substantial positive effect on contraceptive use and intended future use --- Authors' Contributions Both authors conceived the study, reviewed the literature, carried out the statistical analysis, interpreted the result and wrote the manuscript. Gave final approval of the manuscript to be published, and agree to be accountable for all aspects of the work. --- Declarations Ethics approval was not required for this study since the data is secondary and is available in the public domain. To conduct our study, we registered and requested the dataset from DHS online archive and received approval to access and download the data files. According to the 2016 EDHS report, all respondents' data were anonymized during the collection of the data [6]. --- --- Competing interests The authors declare that they have no competing interests. ---

Background Unmet need for family planning refers to fertile women who want to limit or space their delivery but are not using contraceptive methods. Despite multiple studies were conducted to address family planning in Ethiopia, there is limited information on unmet need in high fertility regions. Knowing the magnitude and predictors of unmet need in the study area helps as an impute for interventions. Therefore, this study aims to assess the magnitude and predictors of unmet need for family planning among reproductive age women in high fertility regions of Ethiopia. Methods A secondary data analysis was performed using the Ethiopian Demographic and Health Survey 2016. A total sample weight of 4312 currently married reproductive age women were included in this study. A multilevel mixed-effect binary logistic regression model was fitted. Finally, the odds ratios along with the 95% confidence interval were generated to determine the individual and community level factors of unmet need for family planning. A p-value less than 0.05 was declared as statistical significance.The overall unmet need for family planning among currently married reproductive-age women in high fertility regions of Ethiopia was 29.

Introduction Understanding and supporting HIV-infected women's awareness of their own status and their options regarding pregnancy requires an understanding of their backgrounds and the potential barriers to care and safe pregnancy. Studies have shown that both social pressures and concerns for vertical transmission play a large role in HIV-infected women's choices about pregnancy [1][2][3]. Information from around the world supports the fact that since the introduction of reliable methods for the prevention of mother-tochild transmission , fewer HIV-infected women are choosing terminations and many are choosing to become pregnant, even to have multiple pregnancies, after their HIV diagnosis [2,[4][5][6]. Options for safe conception, especially in serodiscordant couples, now include artificial insemination and "sperm washing" [7]. These methods, even where they are available, however, remain relatively unknown and infrequently recommended by many healthcare providers, even in developed countries [8] and may be prohibitively expensive or inaccessible for some couples. In addition to these planned pregnancies, a large number of pregnancies remain unplanned [9,10]. Women who live in developing countries, where safer methods of conception are relatively unknown or unavailable, are nevertheless choosing to become pregnant after their HIV diagnosis. A survey of 459 HIV-infected men and women in Cape Town, South Africa found that twothirds of women, who became pregnant after commencing highly active antiretroviral therapy , had intended to become pregnant [11]. Furthermore, 50% of HIV-infected men and women were open to the possibility of conceiving children after their HIV diagnosis; being on HAART had no significant impact on whether or not men intended to pursue pregnancy, but did make women more likely to consider pregnancy [11]. It is unclear what impact moving from a developing to a developed country may have on childbearing intentions, but it is possible that this change in their socioeconomic environment and perceived options for care and support may play a role in women's childbearing intentions. Women in the USA, who are HIV-infected and foreignborn, may have additional issues to consider during pregnancy. They are more likely to have language, educational, and economic barriers to care and to exercising their options, but may be even more likely to choose to keep a pregnancy because of cultural emphasis on childbearing and negative views of termination. Women who are born in areas of the world where fertility rates are between 5 and 7 children per woman, especially Africa and parts of South and Southeast Asia [12], may experience particular spousal and familial pressure to become pregnant, even if they have disclosed their HIV status [13,14]. A cross-sectional study done in Canada revealed that 69% of 490 HIV-infected women desired future pregnancy and African ethnicity was significantly correlated with intention to become pregnant [15]. Although foreign-born persons with HIV living in the US appear to be a growing proportion of the HIV-infected population [16], current national sociodemographic data do not accurately reflect this demographic, as they are generally listed under their ethnicity without distinction about place of birth [17,18]. Thus, persons who are originally from Sub-Saharan Africa, where the highest number of HIV-infected individuals exists, are listed as black/African-American in data used for determining funding and resource allocation. African-born persons often live within their own communities and may not be reached by programs that target African-Americans [19,20]. In Washington state, for example, the HIV diagnosis rate among blacks in the state is five times higher than the rate among whites, and 40% of all HIV diagnoses among blacks in Washington state have been among foreign-born persons [21]. Similar data collected from five different states and high-prevalence areas show that across all areas, up to 41% of diagnoses in women and up to 50% of diagnoses in blacks occurred among Africanborn individuals [16]. Additionally, we should not presume all foreign-born persons were necessarily infected with HIV prior to arrival in the USA. Data suggest that in some parts of the country, especially those parts with high numbers of Hispanic immigrants, patients are more likely to have been infected after arrival [22], which may indicate increased vulnerability among foreign-born persons. Unfortunately, the complex social issues that contribute to women becoming HIV-infected often present barriers to adherence with prescribed medication regimens and prenatal followup, as well as with ensuring adequate testing and followup of the HIV-exposed infant. Studies have shown that 45% of mothers of HIV-infected infants had missed opportunities for perinatal HIV prevention [23], indicating that although appropriate protocols are in place, additional factors contribute to transmission. Rhode Island, the smallest state in the United States, had an HIV prevalence of 209 per 100,000 population by the end of 2009 [24]. Among the general population, 81% of persons identified themselves as White, 12% identified themselves as Hispanic, and 6% identified themselves as black or African-American [25]. Amongst the 3,080 HIV-infected persons who have been diagnosed in Rhode Island since 1982, 54% identified themselves as White, 26% identified themselves as African-American, and 19% identified themselves as Hispanic [25]. Thus, 45% of HIV cases in the state have occurred in the 18% of the population identified as Hispanic or African-American. No data is available on the percentage of HIV-infected persons in the state who are immigrants or foreign-born. Following the lead of several other states, Rhode Island adopted a law mandating testing of pregnant women during pregnancy or their children immediately after birth. This was done in order to ensure that children receive medication to prevent mother-to-child transmission in a timely manner, and that women are appropriately identified if they need HIV-related services [26]. This law has already led to increased rates of HIV testing during pregnancy, from 52.8% in 2005-2006 to greater than 95% after the law changed in 2007 [27], but it is unclear if it has led to an increased number of HIV diagnoses. We seek to describe the experiences of HIV-infected pregnant women and their children followed at a large HIV clinic in Rhode Island. --- Materials and Methods The Immunology Center at the Miriam Hospital in Providence, Rhode Island, is an urban HIV clinic with 1400 active patients. In 2009, 25% of HIV patients followed at the Miriam Immunology Center were uninsured or covered only by the hospital's free care program and 32% were foreign-born. All HIV-infected patients in Rhode Island have access to antiretrovirals either through health insurance or through the Ryan White Program, provided they have been registered as Rhode Island residents. The Immunology Center has been caring for HIV-infected women before, during, and after their pregnancies since its establishment in 1986. The Hasbro Children's Hospital Pediatric Infectious Diseases II Clinic in Providence generally sees all the HIVexposed infants delivered to these women. These clinics follow DHHS guidelines for perinatal prevention of motherto-child transmission and follow-up testing for HIV-exposed infants. The two clinics are the largest providers for HIVinfected women and their children in Rhode Island. This study aims to characterize and understand the recent trends in the HIV-infected pregnant population and HIV-exposed children in Rhode Island. Social and clinical data from HIV-infected pregnant women collected regularly by the Immunology Center staff for care and research purposes were analyzed to identify and characterize trends in demographics, viral control, terminations, miscarriages, timing of diagnosis, and adherence to followup. Data on the adherence with follow-up appointments and testing for the HIV-exposed infants were obtained from paper and online charts from both Hasbro Children's Hospital's primary care clinics and Pediatric Infectious Diseases II Clinic. Data were analyzed for means of continuous variables , percentages of descriptive variables and significant correlations between variables using Statistical Package for the Social Sciences Version 17.0 . Statistical significance was assessed using a chisquare test. This project was reviewed and approved by the Miriam and Rhode Island Hospital Institutional Review Boards. --- Results Between January 2004 and December 2009, 321 HIVinfected women between 18 and 45 years old were seen at the Immunology Center. Seventy-six HIV-infected women became pregnant, with a total of 95 pregnancies. Seventeen women had more than one pregnancy, and two women had three pregnancies. Fifty-five percent of these women were foreign-born. Sixteen of the women were diagnosed with HIV during their pregnancy while 79% knew their HIV status prior to becoming pregnant. Sixteen of the women chose to terminate their pregnancies ; two of these women were diagnosed during this pregnancy and both were US-born. Seven women suffered miscarriages. The trends by year and breakdown of specifics between foreign-born and US-born women are presented in Tables 1 and2. The breakdown by continent/region of origin of the women is shown in Figure 1. Among the thirty women from Africa, sixteen were from Liberia, three from Burundi, two from Guinea, Kenya, and Senegal and one each from Angola, Ghana, Ivory Coast, Mali, and Sierra Leone. Among the seven women from Latin America, there were three women from the Dominican Republic, two from Guatemala, and one each from Venezuela and Argentina. While the number of HIV-infected women becoming pregnant increased only slightly over the years of the study, the proportion of women that were foreign-born rose from 41% between 2004 and 2005 to 57.5% between 2006 and 2009. There was no significant difference between USborn and foreign-born women in terms of their likelihood of being diagnosed with HIV during pregnancy, choosing termination or choosing to have a second or third pregnancy. US-born women were more likely to have fewer than two visits during the pregnancy or Department of Children, Youth & Families involvement with their children. Of the 72 women who had pregnancies that did not end in miscarriage or termination, mode of delivery was known for 49. Seventeen delivered vaginally, twenty-five underwent elective caesarian section, and seven required emergency caesarian sections. Sixty-three women had CD4 counts and plasma viral loads measured during pregnancy. Fifteen women had detectable plasma viral loads on their last test prior to delivery, and all but one of them underwent caesarian section . Thirty women had both information on viral load prior to delivery and mode of delivery available. Of these, sixteen had undetectable viral loads and six had viral loads less than 1000. Only eight of the women who underwent caesarian section had viral loads above 1000, and five of these women had either repeat caesarian section or underwent the procedure due to emergency indications unrelated to viral load. Nine women were considered to have AIDS at the time of delivery, based on a last CD4 count prior to delivery less than 200 cells/µL. Seven of these women were foreign-born. The mean CD4 count prior to delivery was 480 cells/µL. Three children in this cohort were confirmed to be perinatally-infected. All of them were born to mothers with psychological and/or social issues that had a significant impact on their ability to adhere to appropriate followup and preventive measures. One mother was a young perinatallyinfected teenager at the time of her first pregnancy. She received care in both Rhode Island and Massachusetts, but was unable to adhere adequately to care and treatment at either site. The second child was born to a woman from West Africa who had been diagnosed with HIV three years before the pregnancy upon immigration to the USA. She did not adhere to her medication regimen during pregnancy and DCYF was involved soon after the child was born. The third child was born to a US-born woman who had been diagnosed with HIV during a prior pregnancy. She had delivered an HIV-negative child at that time, who was subsequently removed from her custody. She had been out of care and off HIV medications for many months when she presented to the emergency department with complaint of abdominal pain. She was quickly determined to be in active labor. She denied knowing that she had been pregnant and had received no prenatal care. --- Discussion As has been seen around the world, our data suggest that HIV-infected women are increasingly choosing to become pregnant and to keep pregnancies that may have been unplanned. The number of terminations has remained low. This is consistent with other studies that have shown that fewer HIV-infected women are choosing termination now that the efficacy of methods for prevention of motherto-child transmission are well-known and wellvalidated [2,[4][5][6]. Additionally, a growing number of women are having second and even third pregnancies after their HIV diagnosis. This trend likely reflects both the strength of the desire to have children and the perception of HIV as a controllable illness that does not preclude the creation of a family. Our data indicate that Rhode Island's relatively new optout law for HIV testing among pregnant women, and the resultant increased rate of testing, has not resulted in a higher number of HIV diagnoses; however, its effects may not yet be reflected in our data as this law was only passed in July 2007. In addition, Rhode Island is known to have a low prevalence of HIV, and the increased testing may not result in a significantly higher number of diagnoses. Furthermore, laws to encourage testing during physician encounters outside of pregnancy may negate any potential increase in diagnoses during pregnancy. While the number of HIV-infected women becoming pregnant in our clinic has increased only slightly, the proportion that are foreign-born has been steadily rising, from an average of 41% between 2004 and 2005 to 57.5% between 2006 and 2009. Now that the Obama administration has lifted the travel ban on HIV-infected persons entering the USA , this trend may increase as the number of foreign-born HIV-infected women entering the USA continues to grow. However, the numbers seen in Rhode Island may also decline, as the freedom to enter any state may lead to some choosing to settle elsewhere. Previously, Rhode Island was one of the few states that was allowed to accept HIV-infected individuals emigrating to the USA under the travel ban. Available virologic control data for our clinic indicate that most women have favorable CD4 counts and undetectable viral loads prior to delivery. All the women who had detectable viral loads underwent caesarian section, which is associated with a lower risk of viral transmission for women with HIV plasma viral loads >1000 copies/mL near the time of delivery. The three cases of perinatal transmission are not necessarily representative of the overall outcomes of pregnancy among HIV-infected women. However, they are indicative of the fact that while PMTCT protocols are now able to reduce the risk of HIV transmission to less than 2%, they cannot do so if patients are unable to adhere to them. In our small state, during the time period of this study, there was essentially one clinic and one HIV-trained obstetrician/gynecologist who saw the women during pregnancy, one hospital where they delivered, and one pediatric infectious disease clinic where the children were followed. Each of these settings has robust medical, social work and support staff that fully supported each HIV-infected pregnant woman in order to optimize her care and adherence to therapy. These data highlight the fact that despite this level of coordination and support, nonadherence does occur and perinatal HIV transmission may ensue. Further study to identify "at-risk" women may be warranted, in order to target further medical and psychosocial interventions in order to improve outcomes and decrease maternal-to-child transmission. These study results support prior trends noted from US and international studies that HIV-infected women are increasingly choosing to become pregnant or to continue a pregnancy. Studies are needed to determine the knowledge and attitudes of foreign-born and US-born HIV-infected women regarding childbearing and prevention of maternalto-child transmission. Additionally, comparative studies among women in high HIV-prevalence regions and women who have moved from those regions to resource-rich settings would help healthcare providers and systems to anticipate and meet the needs of a growing population of HIV-infected women who may have different understandings of, and barriers to, exercising options for safe pregnancy. --- Conclusion The face of HIV in pregnancy is changing. HIV-infected women in Rhode Island are more likely to be foreign-born, to know their HIV status prior to pregnancy, and to actively choose to become pregnant or to continue pregnancies, whether planned or unplanned. Despite increasing rates of HIV testing in pregnancy, an increased number of HIV diagnoses has not been seen. When HIV-infected pregnant women engage with appropriate prenatal care, virologic control is generally excellent and outcomes are good for mother and child. However, risks for HIV transmission beyond lack of awareness of HIV status are emerging, especially among women with psychological or social issues, which may preclude adherence to PMTCT protocols and lead to unnecessary infection of infants. Foreign-born women are a growing percentage of the population of HIV-infected women becoming pregnant in the USA and may benefit from further study and targeted interventions. Identifying "at-risk" women during their pregnancy and providing extra social, psychological, and other support to them may be an important additional focus for PMTCT programs in order to continue to prevent perinatal HIV transmission.

Meeting the needs of HIV-infected pregnant women requires understanding their backgrounds and potential barriers to care and safe pregnancy. Foreign-born women are more likely to have language, educational, and economic barriers to care, but may be even more likely to choose to keep a pregnancy. Data from HIV-infected pregnant women and their children in Rhode Island were analyzed to identify trends in demographics, viral control, terminations, miscarriages, timing of diagnosis, and adherence to followup. Between January 2004 and December 2009, 76 HIV-infected women became pregnant, with a total of 95 pregnancies. Seventy-nine percent of the women knew their HIV status prior to becoming pregnant. Fifty-four percent of the women were foreign-born and 38 percent of the 16 women who chose to terminate their pregnancies were foreign-born. While the number of HIV-infected women becoming pregnant has increased only slightly, the proportion that are foreign-born has been rising, from 41 percent between 2004 and 2005 to 57.5 percent between 2006 and 2009. A growing number of women are having multiple pregnancies after their HIV diagnosis, due to the strength of their desire for childbearing and the perception that HIV is a controllable illness that does not preclude the creation of a family.

Introduction Th e labour market, the welfare state and the family are three main societal pillars, although their relative signifi cance diff ers considerably in diff erent historical contexts, countries and life phases of individuals [see e.g., Goodin 2000;Furlong & Cartmel 2007]. For example, the transition from adolescence to adulthood and fi nancial independence takes longer nowadays and varies more than it did a couple of decades ago [Furlong & Cartmel 2007;Buchmann & Kriesi 2011]. Th e signifi cant role of education is one main reason for keeping young adults fi nancially more dependent on the generosity of the welfare state and/or own parents, whereas without qualifi cation transition to the labour market is diffi cult nowadays. Furthermore, those who are already on the employment ladder have to face overall insecurity in the labour market in the form of temporary jobs and turbulence [Furlong & Cartmel 2007;Settersten 2007;Swartz & O'Brien 2009]. Young adults in the Nordic countries have tended to be more independent of their family of origin than their peers in other regimes [Attias-Donfut, Ogg & Wolff 2005;Settersten 2007;Buchmann & Kriesi 2011]. Th ey move out of the parental home on average at a younger age than their European counterparts, for example [Eurostat 2008]. Individually based benefi ts, such as the student grant, housing and unemployment allowances as well as the existence of aff ordable housing are at least partly behind this early home-leaving pattern in the Nordic countries [see e.g. Isoniemi 2006]. Th e level of state support has been diminishing, however, and for example the fi nancial situation of young Finnish adults has been deteriorated during and aft er the recession of the 1990's [Moisio 2008;van Gerven 2008]. It seems, however, that while the welfare state has withdrawn from some of its responsibilities, increasingly more parents are supporting their adult children [Majamaa 2011]. Brandt [2013] has noticed that collaboration between state and family leads to broader and feasible private support, which is usually of a voluntary nature. Overall, the family's role is mainly complementary, at least in the Nordic countries [se e.g. Berndtson 2004;Björnberg & Latta 2007]. Broad research on intergenerational help and support has revealed that usually need is behind the parental help and support, but also resources, ability, possibility and willingness to give help eff ects this. Parental help is not axiomatic, however, and the focus of this study is on the lack of parental help in Finland. Parental help is seen here more or less as a positive sign of the existence of a safety net and an assistant to everyday life not as a vital phenome-non. Lack of parental help can be seen to be more problematic when it relates to scarce resources of parents and accumulates to certain families than when it relates to the fact that there is no immediate need for help. In other words, if the welfare state has less resources to support young adults in their rocky road to independent life nowadays, and if also parents, especially with fewer resources are not capable of supporting their off spring when needed, opportunities in life diff er greatly depending on the family background. Th is article comprises fi ve sections. Following this introduction it fi rst gives a short review of postponed adulthood and previous research results related to parental help. Also an illustrative fi gure of the socio-demographic variables related to parental help follows this review. Parental help and support is known to be common in the Nordic countries [see e.g. Brandt 2013], however, given the gap in research related to intergenerational help and support, the paper addresses the question of why not all adult children receive help? Th e second section introduces the data, variables and methods used in the study, whereas the third section takes a closer look at Finnish baby boomers who do not give any practical help or fi nancial support to their adult children. Th e lack of parental help and the possible reasons for this is also considered amongst adult children in the fourth section. Th e fi ndings suggest that rare communication, in particular, predicts a lack of parental help and that the ability to give help as well as a diminishing need are strong determinants. Th e last section concludes the paper and suggests that the lack of parental help can be seen, at least partly, as one link in the chain of educational inheritance. --- Postponed adulthood and parental help Young people gradually move into an adult role and make a wide range of choices about their studies, way of life, work, relationships and children. Nowadays these transitions are more varied and tend to take longer than a couple of decades ago and the order of events is not rigid [see e.g., Furlong & Cartmel 2007;Buchmann & Kriesi 2011]. Th ese transitions vary amongst European countries, however, and perhaps the most striking diff erences are in the patterns related to leaving home. Co-residence is a typical way for parents to support adult children in Southern European countries and direct money transfers are less common, whereas co-residence is less of a normative parental-support strategy in the Nordic countries. Adult children in Central European countries can expect fi nancial support when they are residentially autonomous but co-residence is more common than in the Nordic countries [see e.g., Albertini & Kohli 2012]. Th e lack of aff ordable housing is one reason for the delayed departure from the parental home [Eurostat 2009, 30] but institutional factors also have an eff ect. State benefi ts and allowances tend to be allocated on an individual basis in the Nordic countries, for example, which encourages early home leaving, whereas young adults in Southern European countries have to lean more on family support [Buchmann & Kriesi 2011]. Even if young adults in the Nordic countries have tended to be more independent of their family of origin than their peers in other welfare regimes [Buchmann and Kriesi 2011] parental help and support is more or less obvious also here [Haavio-Mannila et al. 2009]. Previous studies have shown that many socio-demographic factors interplay with parental help. For example, the age of both the parent and the adult child is associated with helping: the younger the adult child is, the more likely he or she is to receive help and support and the younger the parent, the more likely it is that he or she will help and support an adult child [see e.g. Albertini, Kohli & Vogel 2007]. In particular the age of the adult child is negatively associated with received fi nancial support [Kohli 1999;Attias-Donfut & Wolff 2000a;Majamaa 2013], as well as with childcare help [Attias-Donfut & Wolff 2000b;Vandell et al. 2003;Majamaa 2012], whereas age has less of an eff ect if practical help is considered in general [see e.g., Hillcoat-Nallétamby & Dharmalingam 2003]. Moreover, relatively young grandparents aged 50 to 59 are most likely to provide childcare help, but giving regular help seems to be more common among 60-to-69-year-olds old [Hank & Buber 2009]. Attias-Donfut and Wolff [2000a] argue that gendered helping is related to the implicit gender contract: care giving is more for women than for men. Th is gender pattern applies not only to the care of elderly, but also to care related to grandchildren [Hank & Buber 2009;Majamaa 2012]. Th e association is less clear with regard to fi nancial support [see e.g., Fritzell & Lennartsson 2005;Majamaa 2013], even if Björnberg and Latta found [2007] that men were more likely to give fi nancial support than women because they have more assets. Gender and receiving parental help have been found to have also some interplay: Fritzell and Lennartsson [2005] showed that women are more likely than men to receive fi nancial support and Attias-Donfut and Wolff [2000a] found a similar relation in respect of childcare help received. Resources and the ability to give help and support have a strong eff ect. Parents with few children are more likely to give and adult children with few siblings to receive fi nancial support from their parents. A similar association has been found with regard to practical help [see e.g., Attias-Donfut & Wolff 2000a;Fingerman et al. 2010;Majamaa 2012]. Having a partner also seems to aff ect resources: parents who are married or cohabiting are more likely to give practical help and fi nancial support to their adult children than parents living alone [Lennartsson, Silverstein & Fritzell 2010]. Forming a relationship, in turn, appears to decrease dependence on the family of origin amongst adult children, who seem to rely on their parents when they are alone but turn to their partner aft er establishing a relationship [Sarksian & Gerstel 2008;Swartz et al. 2011]. Furthermore, the need for extra help usually increases when children come along. Lehto and Sutela [2008] found, for example, that over 60 per cent of working Finnish parents received extra help from relatives and friends to balance their work and family commitments. Th e need for childcare help is one reason but love, aff ection and a desire to spend time with grandchildren also seem to be strong motives [Fingerman et al. 2010;Settles et al. 2009;Majamaa 2012]. Adult children from families with more available resources and stronger emotional commitment seem to have a smoother transition to adulthood [Swartz & O'Brien 2009;Swartz et al. 2011]. According to various studies, parents with a higher socio-economic position are more likely to give fi nancial support than those with a lower socio-economic position [see e.g., Kohli 1999;Fritzell & Lennartsson 2005]. From the receivers' point of view parental financial support helps to prevent poverty amongst young adults [Fingerman, Silverstein & Fritzell 2009;Björnberg & Latta 2007], although parents seem to be especially keen to help and support their off spring if they are progressing towards independent adult life [Swartz et al. 2011]. Students, regardless of age, tend to receive fi nancial support from their parents [Kohli 1999;Swarzt et al. 2011, Majamaa 2013]. Some of the research fi ndings indicate, however, that private support is oriented to advantaged adult children, in other words those with an established position in adult life [Fritzell & Lennartsson 2005;Lennartsson, Silverstein & Fritzell 2010], whereas personal disadvantages are found to go hand-in-hand with the absence of a personal safety net [Harknett & Hartnett 2011]. Health status and the amount of available spare time also appear to be associated with the giving of practical help [Attias-Donfut & Wolff 2000a;Tan et al. 2010]. Th e likelihood of looking aft er grandchildren has been found to be negatively associated with the poor health of grandparents [Hank & Buber 2009;Tan et al. 2010]. Furthermore, grandparents who are still active on the labour market have less time to give practical help in the form of looking after grandchildren [Albertini, Kohli & Vogel 2007]. On the other hand, some studies [see e.g., Guzman 2004] report that retired and unemployed grandparents provide childcare help less frequently than those who are employed, whereas others report no association between their labour-market position and childcare help given [see e.g. Majamaa 2012]. As Attias-Donfut and Wolff [2000a] put it: "Grandmothers in general arrange to spend time with their grandchildren whether they work or not. " External factors such as long geographical distances also aff ect the ability to help. Hank and Buber [2009], for example, found a negative association between an increasing geographical distance separating the younger and older generations and the likelihood of looking aft er grandchildren [see also Attias-Donfut & Wolff 2000a;Tan et al. 2010]. Nevertheless, the eff ect of geographical distance seems to be much weaker if less frequent and more frequent childcare help are studied separately [Majamaa 2012]. Even if resources, ability and needs are associated with intergenerational help, more or less conscious motives for giving may determine intergenerational support in the family [Kohli & Künemund 2003]. Doty [1986], for example, distinguishes three broad categories that infl uence transfer behaviour: aff ection and norms of both responsibility and reciprocity. Tan and her colleagues [2010], in turn, found that emotional closeness and more regular contact had a strong impact in predicting grandparental involvement in childcare, whereas Grundy [2005] noted a strong reciprocal element related to intergenerational help and support. Kohli and Künemund [2003] constructed a motivational space with fi ve basic orientations: altruism; delayed, indirect or generalized reciprocity; a sense of duty; separation and direct exchange. Th ey argue that when intergenerational help is based on concern for the wellbeing of others the motive could be considered altruistic. Reciprocity in helping may be indirect, delayed or generalized depending on from whom and when the help is received and to whom and when it is given. A sense of duty is related to normative obligations, whereas maintaining autonomy and distance relate to not-giving and to the idea that the grown-up child should stand on his/her own feet, for example. If the giver wants to receive some help and support in return, the motive is based on direct exchange. As we can notice, parental help and support is a widely studied fi eld of research. Figure 1 represents an attempt to systemize and summarize the research framework with regard to parental helping patterns. Generally speaking background variables tend to measure parental resources and the ability to help, whereas factors related to adult children measure the variety of needs. Motives, attitudes and cultural norms also have an eff ect but in this study they cannot be measured directly. Furthermore, families live in a specifi c histori-cal context and in a certain society and aggregate-level factors such as the generosity of the welfare state aff ect parental helping [see e.g., Kohli 1999]. Overall, families live in a specifi c society and at a specifi c historical time with specifi c legislation and institutions. Th e welfare state generosity changes over time, however, and aff ects the extent of parental help and support. In some countries private fi nancial transfers are the only source of income amongst young adults nowadays, whereas in others these transfers complement state assistance and services [see also Kohli 1999;Attias-Donfut & Wolff 2000b]. Neither should one forget the cultural factors as well as the social norms and expectations which shape parental help and support in a certain family context. --- More precise research questions Parental help and support is common but it is not axiomatic. Parents may not have the ability, the resources or the will to support their adult children and if children do not receive help and support during diffi cult times or for further education they may fi nd themselves at a disadvantage [see e.g., Swartz Various previous studies are related to intergenerational help but none could be found which referred directly to the lack of parental help and support. In focusing on the fl ip side of the coin, in other words, those who lack parental help and support this study off ers a new perspective and sheds further light on the concept of intergenerational help. It also highlights new research results in recognizing the heterogeneity of adult children. Firstly, the article considers parental helping and non-helping from the perspectives of the givers and the receivers as a general phenomenon. Aft er that the focus turns on those who lack parental help and the paper seeks answers to the following questions: Who are these adult children who lack practical parental help and fi nancial support? Why do they lack help and is the lack of help related to the sparse resources of the givers, the age of the adult children or a certain phase of life or is it simply that there is no need for help? --- Data Th is study is based on information obtained from questionnaires distributed to families spanning two generations. Th e fi rst generation comprised socalled Finnish baby boomers, born between 1945 and 1950, and the second off spring generation included their adult children who were born between 1962 and 1988. At the time of survey they were 18-44 years old. Statistics Finland2 drew up and collected the data in 2007 and with the permission of the respondents merged some individual-level administrative register data into the survey data such us information about employment, the level of education and occupational status in 2007 and in previous years. Because of the low response rates among the baby boomers and their adult children a non-response analysis was conducted [Majamaa 2009a]. According to the results more women than men returned both questionnaires and amongst the adult children age was associated positively with the response rate. Non-response bias in the children's data related mostly to education: the higher the educational level, the higher was the response rate . Income level did not separate the respondents from the non-respondents. Amongst the parents the response rate was higher among women in the higher income deciles than for those in the lower deciles although this tendency was not evident amongst the men. Moreover, those with a higher educational level responded more actively than those with only a basic level [Majamaa 2009b]. Th e fi rst phase of the study included baby boomers with at least one adult child living outside the parental home . Amongst the potential help receivers all adult children with at least one living parent and who did not live in the parental home were included in the descriptive part . In the fi nal analysis , however, the data was reduced by a third because the adult-child-parent dyads were possible to form only if both the child and a parent returned the questionnaire. Information on the parent was linked with the data on their adult child by means of the family identifi cation number. --- Lack of help -two dependent variables Practical help 3 which includes childcare help 4 was considered separately from fi nancial support 5 . Financial support means giving money or covering specifi c costs such as for schooling or certain purchases. Altogether both questionnaires included 10 ways of helping. Not helping was considered amongst both the parents and their adult children . Amongst the parents help and support given to all their adult children was taken account. Two groups were distinguished: those who did not give any practical help to their own adult children and was distinguished from those who gave at least some help . Also in the case of fi nancial support, parents who did not give any support were distinguished from those who gave at least some help . Correspondingly the same distinction was made amongst the potential receivers . In the descriptive part lack of help and support was considered in cases of both parents but in the explanatory part the focus was on lack of help from a particular parent, in line with the adult-child -parent dyads. Th is child-parent data included 141 adult children who did not receive any practical help and 639 who received at least some practical help from one parent. Th e same distinction applied to fi nancial support: 441 adult children did not receive any support from their own parent, whereas 339 received at least some. Th e questionnaires did not include any variables about direct social support. Th ere was information about contact frequencies between parent and child, however, which was used as one of the control variables. --- Independent variables Th e review highlighted the importance of socio-demographic characteristics, especially age, socio-economic variables and emotional aspects in explaining parental helping. Th e choice of control variables was based on these fi ndings. Eight variables covering the socio-demographic characteristics of parents were included in the fi rst analysis related to not helping . Th e fi rst three of these were gender, living with a spouse and number of children . Th e next three variables represent the parental socioeconomic position: educational level , working status and disposable income per month . Information about disposable income per month was based on taxable income less paid taxes in 2007 divided by 12. Th e last two variables were self-reported health and having grandchildren . Th e analyses related to adult children covered seven socio-demographic characteristics. Furthermore, two variables concerned geographical and relationship closeness with a parent and two were socio-economic indicators concerning the adult child's parents . Th e fi rst four control variables were gender, age group , number of siblings and family type . Th ree socio-economic indicators were included: educational level , working status and monthly disposable income . Geographical distance to the parental home was categorized in three groups: less than 15 kilometres, 15 to 149 kilometres and at least 150 kilometres. Th e questionnaire of the adult children contained also the question about the frequency of contact with their parents: daily or more oft en than weekly, weekly and less oft en than weekly. Finally, two parental socio-economic variables were included: educational level and working status. --- Methods In the explanatory part binominal logistic regression was used, in which the response variable is dichotomous and comparisons are made between these two categories. Firstly the focus is on the parents. Th e parents who did not give any practical help were compared to those who gave at least some. Similar comparisons were made with regard to fi nancial support. Th e focus then moved to adult children and the lack of received help and support. In other words the adult children who did not receive any practical help from their own parents were compared to those who received at least some help and a similar comparison was made related to fi nancial support. Th e fi rst category in each explanatory variable was the reference group with an odds ratio of 1. Tables 1 and2 include two models. Model 1 indicates how each predictor aff ected non-helping / lack of help, whereas Model 2 presents the estimates for non-helping / lack of help adjusted for all the variables. Odds ratio of not helping / lack of help for each category versus the reference category were obtained from the estimated logit coeffi cient by means of transformation e b . Stata's statistical soft ware cluster option was used to compute the standard errors because of the clustered data on adult children. --- Downward-fl owing help -a lack of resources among parents? Almost all the parents gave at least some kind of help to their adult children: only 13 per cent did not give any practical help or fi nancial support . Most parents gave practical help , whereas fi nancial support was less prevalent . Approximately 40 per cent of parents gave both practical help and fi nancial support. Th e strongest tendency was to give two types of help, usually fi nancial support and childcare help . Approximately one in ten parents helped in at least six diff erent ways the mean score being 2.8 = 0.07), whereas the biggest tendency was to off er three types of help. Binominal logistic regression was used in the next explanatory part. Table 1 shows the odds ratios for the eight control variables and not giving practical help, as well as not giving fi nancial support separately . Model 2 shows the estimates adjusted for all the chosen variables at once. Surprisingly, neither gender nor working status was associated with not giving any practical help amongst the parents whereas the other factors were associated . According to the results, living without a spouse, having a low number of children, a low educational level and low disposable income and poor self-reported health predicted a lack of practical help to adult children. Non-helping was also more common amongst parents who did not have grandchildren. When all the socio-demographic variables were controlled the likelihood of not giving practical help was signifi cantly higher amongst those with no grandchildren than amongst those who were grandparents. Th is result was not surprising as previous research related to childcare help given by grandparents shows that approximately 90 per cent of Finnish grandparents give at least some childcare help to their adult children [see e.g., Lammi-Taskula, Suhonen & Salmi 2004; Majamaa 2012]. Furthermore, living without a spouse indicated a higher propensity of not helping. Previous studies [Lennartsson, Silverstein & Fritzell 2010] also report that living with a spouse increases the probability of giving help. However parents with only one child were almost three times as likely not to help their adult children than those with at least three children. Th is fi gure implies that the fewer children parents have the more likely they are not to help their own children. In other words, parents with a number of children give help to at least one of them [see also Majamaa 2012]. A positive association between parental educational level and fi nancial support has been reported [see e.g., Kohli 1999;Swartz et al. 2011] but in the case of childcare help the association is less clear [Majamaa 2012]. Interestingly the likelihood of not giving practical help was twice as strong amongst parents with a low level of education compared to those with a high educational level when all the variables were included . It may be that a low level of parental education indicates low overall resources [see also Harknett & Hartnett 2011] and that a helping culture is less prominent than amongst the more highly educated. Table 1 shows also the odds ratios for parents who did not give any fi nancial support to their adult children variable by variable . In general a low economic position was strongly linked to not giving fi nancial sup- port, whereas those in a better economic position were more likely to support their adult children. Th ere seem to be also other factors involved in terms of not giving support such as a low number of children, poor health and having grandchildren. Th e full model gives only a little new information related to not giving fi nancial support: the strong association with a low educational level weakened only slightly aft er all the variables were added and the likelihood of not giving support was still almost three times as high amongst those with a low versus a tertiary level of education [see also Swartz et al. 2011;Majamaa 2013]. Moreover those who were not working carried excess risk in terms of not helping their off spring fi nancially. Th e association between both disposable income and self-reported health and not giving support disappeared when the educational level was included in the model. Th is seemed to be attributable to the fact that those with less education have lower disposable income and worse self-reported health than the more highly educated . Th e results show that parents' scarce resources predict non-helping whereas a lack of help and support is more evident amongst parents who lack financial resources or experience health problems, for example. Furthermore the accumulation of educational capital seems to go hand in hand with helping -the likelihood of not helping, in terms of both practical help and fi nancial support, was higher amongst parents with a lower educational level than amongst the higher educated. Spousal support also seems to be important on the level of practical help whereas living without a spouse was associated with not giving help. However aff ection and the desire to spend time with grandchildren in particular, more oft en than duty or need, seemed to be behind childcare help given by grandparents [Settles et al. 2009;Fingerman et al. 2010;Majamaa 2012]. Overall if variables related to values and motives could be included the analysis would benefi t. --- Lack of need amongst adult children? Based on results retrieved from the data on adult children a total lack of parental help was quite rare: only eight per cent did not receive any practical help or fi nancial support from their own parents . Almost half received fi nancial support, 88% received some kind of practical help and 44% received both types of help. Most typically they received two types of help, usually fi nancial support and childcare help . Approximately one in ten of the adult children received help in at least six diff erent forms, the mean being 3.1 = 0.05) and three being the most common. Figure 2 gives more details about the associations between the age of the adult child and the lack of both practical help and fi nancial support. Approximately 15% of those under the age of 25 did not receive any fi nancial support whereas in the oldest age group the fi gure was 78%. Th ere was no clear evidence of a growing trend in not receiving practical help however. Up to age of 35 the proportion of adult children not receiving practical help remained at approximately 10%, whereas in the oldest age group it increased to a quarter. Furthermore overlap in intergenerational help was common, especially amongst younger people. Approximately 75% of adult children in the youngest age group received various forms help from their own parents as opposed to 20% in the oldest age group . Only three per cent of adult children who did not live in the parental home and who were under the age of 25 did not receive any help from their own parent whereas in the oldest age group the proportion of those lacking both practical help and fi nancial support increased to 25% . Overall a lack of fi nancial support was approximately four times more common amongst adult children than a lack of practical help and there was a clear, positive association with the age of the adult child especially in the case of financial support. But is it really the age of the adult child or is it more a question of life phase and the need for help that explains the increasing trend? So far the results have indicated that parents need some resources and reasons to give help although the adult child's age, certain life phases, frequency of contact and geographical distance seem to be associated more or less with a lack of parental help and support . Table 2 shows the results related to a lack of help when all these factors are considered, fi rst variable by variable and then following adjustment for all the variables at once . Age-adjusted odds ratios were also counted but the results are not shown. As model 1 in Table 2 shows the adult child's gender, family type, educational level, distance to the parental home and frequency of contact were associated with a lack of practical parental help, whereas, and perhaps surprisingly, there was no association with either the adult child's or the parental working status. As expected all the socio-economic variables related to the adult child and his or her parents were strongly associated with a lack of financial support. Also the child's gender, age, number of siblings, family type and frequency of contact predicted a lack of parental fi nancial support. As Figure 2 shows, age in particular is positively associated with a lack of fi nancial support but does age really explain the diminishing parental help? According to the results adjusting for age either eliminated or weakened the associations between the studied variables and a lack of help. As expected adjusting for age had hardly any eff ect when a lack of practical help was considered. Th e odds ratios increased only amongst those who had no children following adjustment for age implying that having children would decrease the probability of the lack of practical help even more if the age structure was the same as amongst those with and without children. Moreover the response rate was higher amongst the older groups [Majamaa 2009a] and given that having children is more common in the older age groups adjusting for age increased the odds related to a lack of help. Th e associations between gender and a lack of fi nancial support and family type disappeared following adjustment for age. Th is relates to the fact that the response rate was higher amongst women [ibid.] and that having children is more common in the older age groups. Furthermore age adjustment weakened the associations related to educational level, working status, disposable income and the parent's educational level and working status. Th ese results refl ect the fact that, fi rst of all, older adult children are more likely to have an established fi nancial position than younger ones thus need for fi nancial support is less evident. Secondly, younger adult children are more likely to have a parent with a higher level of education, and their parents are more likely still to be actively employed . Following the inclusion of all the variables in model 2 the association was strongest between the frequency of contact and a lack of practical help: the likelihood of a lack of help was almost eight times stronger if the adult child and the parent were in contact less oft en than weekly than if they contacted one another several times a week. Interestingly the likelihood of a lack of practical help was also high amongst those with only a basic level of education, in other words with no qualifi cations and only compulsory education than amongst students. Educational inheritance is strong in Finland [see e.g., Kivinen, Hedman & Kaipainen 2007;Myrskylä 2009] but it may be that the adult child's educational level predicted a lack of help only because there was no need for help. On the other hand adult children could be unmotivated to educate themselves if they lack parental help and support , which could be one determinant related to educational inheritance. Th is causality relationship could not be tested directly here because of the cross-sectional nature of the data. However some evidence of the latter alternative emerged: adult children whose parents had a lower level of education were more likely not to receive fi nancial support than those whose parents were more highly educated . Not-giving amongst lower educated parents may relate to the idea that a grown-up child should stand on his/her own feet earlier [see Kohli & Künemund 2003], or on the contrary the lack of parental help and support may be one link in the chain of educational inheritance. Furthermore a lack of practical help was more prevalent amongst those who were living alone than amongst those who were cohabiting or married and had children. Looking aft er grandchildren is an important form of grandparental help and a major reason for care seems to relate to a desire to be with grandchildren [see e.g., Fingerman et al. 2010]. Brandt [2013] also notes that if intergenerational help is based on voluntariness rather than necessity, it is usually more forthcoming. Th e results show that a long geographical distance between generations reduces the likelihood of practical help given and received which other studies have also noted [Hank & Buber 2009, Tan et al. 2010]. However following simultaneous adjustment for all the variables no association could be found between the gender, age, number of siblings, working status, disposable income of the adult child or the parental working status and the lack of practical help . Becker [1991, p. 369] argues that in supporting education parents wish to make a sensible investment in the future of their children. Th e present fi ndings support this claim, in that students in particular received fi nancial support . Overall socio-economic factors amongst both receivers and givers seemed to interact strongly with the lack of parental fi nancial support. Having educational qualifi cations and a high disposable income seemed to predict a lack of fi nancial support, in particular6 , but parental socio-economic characteristics also mattered. Adult children whose parents had a lower level of education were more likely to lack fi nancial support than those with highly educated parents . Previous studies also report an association between parental high education level and fi nancial support given [see Kohli 1999;Björnberg & Latta 2007;Majamaa 2013]. Parental working status also predicted a lack of fi nancial support: the children of unemployed parents were more likely not to receive support. Overall it seems that the parental socioeconomic background is highly predictive of received and not received financial support. Furthermore a weak independent association with age and a lack of fi nancial support remained between the oldest and the youngest age groups . Previous studies [Attias-Donfut & Wolff 2000b;Hillcoat-Nallétamby & Dharmalingam 2003] also report that younger adult children are more likely than older ones to receive fi nancial support from their own parents. Results indicated that larger number of siblings increase the risk of lack of parental fi nancial support. Th ose with at least two siblings had almost twice the risk of lack parental fi nancial support than those with one sibling . Association was not statistically signifi cant amongst those who did not have any siblings. In previous studies those who had no or only a small number of siblings were more likely to receive fi nancial support [see e.g., Attias-Donfut & Wolff 2000a;Fritzell & Lennartsson 2005;Majamaa 2012]. Interestingly the association between the frequency of contact and a lack of fi nancial support remained or even strengthened in the full model: children who were in touch with their parents less oft en than weekly had approximately three times higher odds ratios than those who were in touch daily or almost daily when a lack of support was considered [see also Tan et al. 2010]. It seems that close parent-child relationships in particular foster, not only in practical terms but also on the fi nancial level, helping. Gender, family type and distance to the parental home showed no association with a lack of fi nancial --- Lack of practical help support following adjustment for age and there was hardly any change when all the variables were included in the model. Overall it seems that rare contact between an adult child and his or her parents in particular forecasts a lack of practical help and also a lack of fi nancial support. In terms of practical help diff erences emerged already between those who were in touch more oft en than weekly and those in weekly contact. Interestingly adult children with only a basic level of education were approximately four times more likely to lack practical help than students. A long geographical distance to the parental home and not having any children also predicted a lack of parental practical help. Th ere was some interplay between the age of adult children and a lack of practical help , but the diff erences were not statistically signifi cant . Th e age of the adult child did not fully explain the lack of fi nancial support: the strongest determinant was apparently the diminishing need associated with certain life phases. According to the results and as expected, students were most likely to receive fi nancial support from their own parents whereas the likelihood of not receiving support was highest among those with a tertiary level of education. A high level of disposable income also increased the odds related to a lack of fi nancial support in line with the need hypothesis: the lower the level of disposable income, the greater is the need and the higher the likelihood that parents will help fi nancially. Parental socio-economic variables were also associated with the lack of fi nancial support: the lower the parental educational level, especially if the parent was unemployed, the more likely was the adult child not to receive any such support from a parent. Th ese fi ndings refl ect the results obtained from the parental data . It seems that need for fi nancial support accumulates amongst young adults who are studying and have low income and most parents respond accordingly. In sum the present Finnish data gives further evidence that the helping behaviour of parents is related both to their resources, ability and willingness to give help and support, as well as to the needs of their adult children. Firstly, parents geographically removed from their children could not easily give practical help, whereas having grandchildren clearly increased their willingness to provide such help. A low socio-economic position and being retired, as well as living without a spouse, lowered the ability and perhaps also the readiness to give fi nancial support. Th ese associations were evident even aft er controlling the child's socio-demographic characteristics. On the other hand from the receivers' point of view adult children in a better socioeconomic position evidently had no need for fi nancial support and they were also more likely not to receive any than those with a lower socio-economic position. With regard to practical help socio-economic position determined a lack of help only amongst those with a basic level of education. Th e interplay between the age of the adult child and a lack of fi nancial support was limited. Finally, a low level of communication between adult child and parent, measured here in terms of frequency of contact, predicted strongly a lack of parental practical and fi nancial support. --- Conclusions Th is article analyses the lack of parental practical help and fi nancial support from both the parents' and the adult children's points of view. Even if help and support are available from many sources, such as volunteer organizations, relatives, friends and neighbours, familial bonds seem to be strong and parental bonds in particular [see Haavio-Mannila et al. 2009;Bengtson 2001;Dykstra & Fokkema 2011]. Finnish baby boomers follow with this trend. A lack of parental help and support was quite rare: approximately 10% of parents did not give any help to their adult children, and approximately 10% of adult children did not receive help from their own parents. Th e results also indicate that practical help extends beyond the early years of adulthood, whereas financial support relates more strongly to those early years . Children received parental fi nancial support especially in their early 20's but in quite small amounts. According to Brandt [2013], the more prominent role of public transfers and the social services is at least partly behind this sporadic and less time-consuming intergenerational help and support in the Nordic countries. Individuals' expectations and thoughts have changed over the course of time [Settersten 2007] and many young adults increasingly tend to feel that parents have a duty to support them fi nancially past the age of 18 [Social Issues Research Centre 2009]. Indeed most parents help and support their children in some way aft er the age of maturity and one main goal for all parents appears to be that their children achieve independence and getting a good job is one way to establish fi nancial independence. Finnish baby boomers in particular have a good enough economic position and the resources to help their own adult children [Majamaa 2013]. Present results indicate that adult children receive parental fi nancial support especially during their studies when their economic circumstances are bad but their willingness to invest in their own future is high. Aft er gradua-tion and with a higher income the lack of support is more evident. Parental educational level, however, is a strong positive determinant of their children's educational level in Finland, too [see e.g. Kivinen, Hedman & Kaipainen 2007;Myrskylä 2009]. Current fi ndings show that a low level of parental education predicts a lack of parental fi nancial support potentially aff ecting the child's future prospects and opportunities. Th e lack of parental fi nancial support may be one of the factors that weaken the will to invest in education and in postcompulsory studies. Furthermore if scarce resources of parent and adult child interplay with lack of help the importance of welfare state support stands out. Th e role of the Finnish welfare state has been deteriorating, however, and there is some evidence that families are nowadays playing a bigger role in the welfare of young adults [Majamaa 2011]. Policy makers should understand that with fewer resources and fewer options from which to make life choices the inequality in terms of opportunities will increase amongst those from less advantaged backgrounds. It is vital, also in the future, to invest in education and to maintain equality in educational opportunity by maintaining a certain level of welfare-state benefi ts , which constitute one primary form of economic welfare amongst young adults in Finland. Even if the socio-economic characteristics of parents and adult children showed no clear association with the lack of practical help, those adult children with only a basic level of education had a higher risk of the lack of parental help. Unemployment makes people feel an outsider in society and passes on from one generation to another, also in Finland [Myrskylä 2011], and this result suggests that there could be some interplay between a disadvantaged position and a lack of parental support. At least fi nancially poor mothers and mothers with personal disadvantages seem to be more likely to lack a personal safety net than their more advantaged counterparts [Harknett & Hartnett 2011]. Th ere are also factors related to parental help other than resources and ability. As Fingerman and her colleagues [2010] state, "need and aff ection appeared to be at the heart of off spring support" [see also Doty 1986;Kohli & Künemund 2003;Tan et al. 2010]. As reported in this study the frequency of contact between adult child and parent is a prime factor related to the lack of parental help: adult children in more sporadic contact were at a substantially higher risk of lacking practical help and this also relates to the lack of fi nancial support. However the lack of parental help may weaken the child's will to maintain contact with the parent. Overall it seems that parents need to have the resources to help and support their needy adult children, but sporadic communication between generations in particular predicted a lack of help. Finally longer life expectancy and better health status, falling numbers of siblings and increasing marital instability seem to reinforce vertical bonds amongst family members and parents in particular seem to be a signifi cant source of help and support to young adults in contemporary society [Bengtson 2001;Brannen, Moss & Mooney 2004;Schoeni & Ross 2005;Swartz et al. 2011]. Studying parental help from the reverse perspective, namely focusing on factors that relate to not giving help to children, and on not receiving help from parents, sharpens the overall picture. Admittedly more fi ne-tuning is needed but the underlying mechanism and the implications related to the lack of parental help and support deserve closer attention.

Th e main focus of this paper is on adult children who do not receive any practical help or fi nancial support from own parents. Th e perspective is two-folded: the lack of help is considered both from the potential givers' (parents') and from the potential receivers' (adult children's) points of view. Th e results indicate that the lack of fi nancial support is more common than the lack of practical help, whereas received practical help extends beyond the early years of adulthood. According to the binominal logistic regression analyses, the lack of parental help is most evident when there is no need, in the other words, fi nancial support is less probably received aft er graduation and with good incomes, whereas practical help is received especially aft er having children. Lack of regular contacts between parent and adult child and long geographical distance seem also be detrimental to parental helping. Furthermore, there was some interplay between an adult child's disadvantaged position and a lack of parental support. If scarce resources of both parent and adult child interplay with lack of help, social problems may arise more easily, in other words, the lack of help accumulates more or less to certain families.

Background Depression affects mood, quality of life, and physical health [1], and is an independent risk factor for increased mortality [2,3]. The prevalence of depression among Chinese adults was 20% from 2011 to 2019 [4]. Depression is one of the most common mental health problems among the middle-aged and elderly and has become an important public health problem in China [5]. It not only poses a significant threat to physical and mental health and ability to live, but also brings a heavy burden to families and society. With the aging and the popularity of healthy aging, the prevention and treatment of depression in the middle-aged and elderly is becoming increasingly important. With the high popularity of the Internet, social media has become a factor that affects mental health. The use of social media could reduce anxiety and loneliness and promote physical health [6], as well as reduce depression levels and increase life satisfaction [7]. Social media can provide many convenient services such as information retrieval and online communication for the middle-aged and elderly with mobility problems, so it is gradually winning their affection. Online socialization could overcome geographical barriers [8], and keep older adults in close contact with family and friends [9], therefore enhancing their social support [10,11]. WeChat is a popular social media in China and has become an inseparable part of the work and life [12]. According to a report by the Chinese Academy of Social Sciences, WeChat has become the most commonly used online social tool among the elderly [13]. WeChat usage has some positive effects on both physical and mental health of middle-aged and older adults. WeChat usage could reduce the risk of depression in the elderly [14] and had a significant positive impact on subjective health status [15]. A study suggested that WeChat usage may also boost memory in older adults by reducing risk of depression [16]. However, some studies have noted that excessive use of social media, such as WeChat addiction, can have negative effects on the physical, psychological, and social health of users [17][18][19]. We attempted to clarify the correlation between WeChat usage and depression among Chinese middle-aged and older adults, and to explore whether there were mediating variables in the correlation. Social participation is an important factor, among influences on depression. The definition of social participation is not entirely uniform. In general, social participation refers to activities in which individuals engage in interactions with others in society or the community [20]. In China, whatever way the elderly keep in touch with society is considered to be a form of social participation [21]. Studies showed that social participation was an important part of healthy aging [22], and great for reducing loneliness, relieving depressive symptoms [23,24], increasing life satisfaction [25], and improving mental health [26]. The higher the frequency and levels of social participation, the lower the risk of depression in older adults [27]. Social participation had a positive effect on health by increasing their social capital and contact between friends and family [28]. Older adults who engaged in social participation tend to have better cognitive behaviors [29]. Many types of socially participating activities had effects on mental Fig. 1 Data inclusion process health, such as volunteering and healthy exercise [30], paid work [31], religious activity [32]. Different types of social participation could positively affect physical and mental health as well as life satisfaction of older adults through different mechanisms [33]. In the digital era, it has been suggested that social media usage could contribute to social participation among the middle-aged and elderly [34]. Does social participation play a mediating role between WeChat usage and depression in middle-aged and older populations? Do different types of social participation have different mediating effects? In this study, we aimed to 1) clarify the relationship between WeChat usage, social participation, and depression in the middle-aged and elderly in China; and 2) explore the mediating role of different types of social participation activities in the relationship. --- Methods --- Study design and study sample We used data from the China Health and Retirement Longitudinal Study in wave 4 of 2018 for cross-sectional analysis. CHARLS aims to collect highquality data on households and individuals aged 45 and older in China to analyze aging and promote research on healthy aging. CHARLS surveyed participants for basic information, health status, health insurance and health care, and retirement. CHARLS was approved by the Ethics Review Committee of Peking University, and all participants signed an informed consent form before the investigation and voluntarily participated in the survey [35]. The target population selected was people aged 45 and above. The missing rate of the dependent variable values was 18.18%, and 15,636 participants were included in the study after excluding those with missing key variable values. The total missing rate for the remaining values of all variables was 1.33%, so we directly excluded participants with missing variable values, and the final number of samples included in the study was 15,428. Data inclusion process are shown in Fig. 1. --- Depression The dependent variable in this study was depression, which was set as a dichotomous variable based on the Table 1 Coding of variables CHARLS questionnaire. The 10-item Center for Epidemiologic Studies Depression Scale , which is a simplified version of The Center for Epidemiologic Studies Depression Scale, is included in the questionnaire of the CHARLS. Each question was scored on a scale of 0-3 . The final score was calculated cumulatively, with a total score of 0-30. When the scores were ≥ 10, the respondent was considered to have depressive symptoms, and below 10, the respondent was considered normal. The depression score also reflected the levels of depression in the sample; therefore, the depression was also examined as a continuous variable in this study. The higher the depression score, the higher the levels of depression . --- WeChat usage The independent variable of this study was WeChat usage status. In the questionnaire, participants were asked "Do you use WeChat?" and the answers included "yes" and "no". When participants answered "yes", they were considered to use WeChat, otherwise they were considered not to use WeChat. --- Social participation The mediating variable in this study was social participation. Participants were asked in the questionnaire whether they had engaged in 10 activities in the past month. Participants were considered to be socially engaged if they participated in any of the activities. We also measured the levels of social participation. Participants earn one point for engaging in an activity. Points were accumulated, and the total score ranged from 0 to 10 points. The higher the score, the higher the levels of social participation. We also tried to explore which type or types of social participation have a mediating role. Based on the experience of previous papers [29], we categorized the 10 activities mentioned in the questionnaire into voluntary activities, recreational activities, cultural activities, and other activities. Voluntary activities included: providing assistance to family, friends or neighbors who do not live --- Covariates We selected a series of control variables that may be associated with depressive symptoms, including demographic characteristics [36][37][38] , health status and health behaviors [39][40][41][42] , smoking, drinking, sleep duration, chronic disease status), and protective factors [31,[43][44][45][46] . For age, we selected people aged 45 and above; for marital status, we reclassified them according to the answers of the questionnaire, and considered married and living with spouse, married but not living with spouse temporarily as married; separated and no longer living with spouse, divorced, widowed, and never married as unmarried. Educational attainment was classified into five categories: no education, elementary school, middle school, high school, and college and above. Since the sleep time showed a skewed distribution, we logarithmically processed the sleep time. For chronic disease prevalence, we divided the population into five categories: no disease, one chronic disease, two chronic diseases, three chronic diseases, and four or more chronic diseases. The detailed coding of the variables is shown in Table 1. --- Data analysis We would use t-tests for continuous variables and chisquare tests for categorical variables to compare the sample characteristics between the WeChat users and non-WeChat users. Given that WeChat usage was not randomly distributed in the study, we used propensity score matching to match the subjects. We used Psmatch2 to identify covariates and perform propensity score analysis. A 1:2 matching group was eventually constructed by nearest-neighbor matching, and the whole set of control variables was included in PSM. Then, we used binary logistic regression and multiple linear regression on the matched groups to analyze the correlation between WeChat usage and depression. Since one of the dependent and mediating variables were both dichotomous, we used stepwise regression to examine the presence of a mediating effect. We also used the KHB method to check the mediating effect again and to analyze whether the mediating effect differed in different populations. The data were processed and analyzed using stata16.0. --- Results After data cleaning, we included 15,428 participants in the study. After PSM, we finally matched 4545 participants for the final analysis, including 1906 participants who used WeChat and 2639 participants who did not use WeChat. The PSM results showed that the t-value of ATT was -4.72, and the p-values of t-tests for all control variables were greater than 0.1, and the bias was less than 10%, which implied that the balance test was passed. According to the kernel density plot , we found that the differences between the treated and control groups were large before matching, and the differences decreased significantly after matching, which indicated a better matching effect. After PSM, the significance of differences in gender, marital status, sleep duration, ADL, smoking, employment, pension insurance, health insurance and chronic disease prevalence status no longer existed Regression results for WeChat usage, social participation and depression are presented in Tables 3 and4. Table 3 used the occurrence of depression as the dependent variable. Model 1 measured WeChat usage and demographic characteristics, Model 2 included all control variables. Models 3 and 4 incorporated the presence and levels of social participation respectively. We could find that in models 1-4, WeChat usage was consistently and significantly associated with a lower risk of depression, as well as the presence and levels of social participation were significantly associated with the lower risk of depression. At the same time, we found that the effect of WeChat on depression was correspondingly reduced after including the variables related to social participation, which suggested social participation might have a partial mediating effect. We could also find that older age, being male, living in a city, having health insurance, better self-reported health and longer sleep duration were significantly associated with not having depression. In contrast, impaired ADL and having more chronic diseases were risk factors for depression. In Table 4, we replaced the dependent variable with the levels of depression, and WeChat usage was consistently and significantly associated with lower levels of depression. In Models 7 and 8, the presence and levels of social participation were significantly associated with lower levels of depression. The effect of WeChat on the levels of depression was also reduced with the inclusion of variables related to social participation, suggesting that social participation may also have a partially mediating role. Compared to Table 3, the significance of the association between education levels and depression levels increased, and the association of being married and employed with lower depression levels also showed significance. After matching, the most popular activity was recreational activities , followed by voluntary activities . While cultural activities and other activities had lower participation rates . Table 5 reports the results of the binary logistic regression of the four types of social activities and depression after the inclusion of all control variables. The results showed that participating in recreational activities and participating in two recreational activities were significantly associated with non-depression. Also, in Table 5, WeChat usage continued to be significantly * p < 0.05, ** p < 0.01, *** p < 0.001 associated with non-having depression, and the effect of WeChat usage on depression decreased in all models compared to model 3. In Table 6, we replaced the dependent variable with the levels of depression. We found that participating in recreational activities and levels of recreational activities were significantly associated with lower levels of depression. Compared to Model 6, the effect of WeChat usage on depression levels decreased both in Model 18 and Model 22. Thus, there might be a partial mediating effect of recreational activities. Then we used binary logistic regression, ordered logistic regression and linear regression to test the significance of the correlation between the independent variables and the mediating variables. The results in Table 7 showed that the use of WeChat was significantly correlated with all four mediating variables. We therefore suggested that social participation and recreational activities might have a partially mediating role. Combining the results in Tables 3,4, 5, 6, 7, we could calculate the percentage of mediating effects. When using being depressed as the ending variable, the mediating effect of performing social participation accounted for 49.86% of the total effect , the mediating effect of the levels of social participation accounted for 12.68% , the mediating effect of participating in recreational activities accounted for 42.82% . When the levels of depression were used as the ending variable, the mediating effect of performing social participation accounted for 40.28% of the total effect , the mediating effect of the levels of social participation accounted for 12.07% , the mediating effect of participating in recreational activities accounted for 28.97% . Since the level of recreational activities was an ordered categorical variable, we could not directly calculate the percentages. To verify the accuracy of the mediating effect of social participation, we verified the mediating effect of social participation again using the KHB method, which extended the decomposability of linear models to nonlinear probability models for a variety of situations [47]. The results in Table 8 showed that the mediating effects of social participation and recreational activities were significant for the dependent variable whether it was occurrence of depression or levels of depression. We could assume that all four variables had mediating effects. Finally, we divided the population into two groups respectively according to age and gender: the middle-aged , the elderly , the female and the male. We tried to analyze whether the correlation between WeChat usage and depression as well as the mediating role of social engagement were heterogeneous because of age and gender. WeChat usage could reduce the levels and risk of depression in all groups and reduced the risk of depression better in the elderly than in the middle-aged. Apart from among women, social participation played the mediating role in almost all groups. However, the mediating effect of recreational activities was only present in the middle-aged. --- Discussion This study concluded that WeChat usage was significantly associated with lower risk of depression and lower levels of depression in middle-aged and older adults, while social participation had mediating effects. Moreover, there was heterogeneity in the mediating effect across groups. After we used PSM and adjusted for a range of control variables, the study found that WeChat usage was significantly associated with lower rates of depression and lower levels of depression. Social media usage was generally significantly associated with fewer depressive symptoms [48,49], which is generally consistent with previous studies. There is an explanation that increasing age decreases physical functioning and impedes mobility in older adults, which increases the vulnerability to social isolation and loneliness dilemmas, causing mental health problems [50]. According to a survey, almost half of the elderly in China lived alone or only with their spouses [51]. Due to physical and distance reasons, it is difficult for them to meet with their relatives or friends frequently. Besides the convenience and speed of social media, its real-time updates and interactivity can facilitate the middle-aged and elderly to communicate with others and share their emotions, which may alleviate the problem of social isolation. For example, the use of social media such as WeChat and Facebook could expand the social network of the middle-aged and elderly, enhance their interaction with others, and promote communication and emotional cohesion with family members [52,53]. So, we think that WeChat, as a comprehensive social media, offers online communication technology that may benefit people who cannot or rarely interact with others face-to-face, and its mental health benefits for middleaged and older adults should be considered. Social participation has a great impact on the occurrence of depression. We found that middle-aged and older adults who do not engage in social participation have a higher risk of depression than those who do, and among the four types of social participation activities proposed in this study, recreation could significantly reduce the probability and levels of depression. In line with the results of previous studies, active social participation could significantly reduce the risk of depression [22][23][24]. Because positive social participation may increase communication and interaction between middle-aged and older adults and help them understand the aging process as well as mitigate the negative effects of aging on mental health [54]. Recreational activities chosen according to their interests can also help them to relieve negative emotions effectively [21]. In this study, we found that recreation was the main social activity of Chinese middle-aged and older adults. Competitive activities, such as playing chess and mahjong, can fully exercise their thinking skills and bring joy as well as a sense of accomplishment when they achieve victory. Leisure activities, such as dancing and qigong, can effectively relax their bodies. These activities are usually participated by many people and they can have sufficient conversations while engaging in recreational activities, which can allow them to relax and release the negative emotions caused by stressful events. Therefore, we hold the opinion that it may be effective to reduce the risk of depression by organizing more forms of recreational activities to meet the needs of the middle-aged and elderly, and encouraging them to try more forms of social participation. This study found that WeChat usage can promote social participation in middle-aged and older adults, which is consistent with previous findings [34,43,55]. We further found that WeChat usage can influence depression through social participation. A possible reason is that online socializing can alleviate the shrinking social circle of older adults [15,56]. Especially for those who have fewer social activities, online communication technologies allow them to gain more opportunities to communicate with others [57]. We think that the photo sharing, voice messaging and video calling technologies of social media have greatly increased the fun of communication for the middle-aged and elderly. It also fulfills their need for emotional expression and sharing their lives. Social media enhances their resistance to depression by increasing their emotional support and social support. On the other hand, social media can have an impact on different groups of people [58]. We think the online social function of social media is applicable not only to individuals, but also to groups. For example, different family WeChat groups can be established among different family members. The middle-aged and elderly can also form WeChat groups with like-minded people to promote further communication among them and facilitate them to organize more offline activities. Both online interaction and offline communication can alleviate the loneliness of the middle-aged and elderly and have a positive effect on mental health. * p < 0.05, ** p < 0.01, *** p < 0.001 Our study found that WeChat usage promotes social participation among middle-aged and older adults, but only the recreational activities among them have a mediating effect. In previous studies, voluntary and cultural activities were also beneficial to physical and mental health [30,46,59], but in this study, we did not find that they could have a mediating effect. Probably the reason is that the sample of this study had a high participation rate in recreation and a low participation rate in other types of activities, especially cultural activities, which social activity participation status had an impact on the final results. Another possibility is that WeChat usage does not have an impact through these types of activities, which needs to be further investigated. We also found some differences in the levels of effect of WeChat usage on depression and in the mediating effect of social participation because of differences in age and gender. The positive effect of WeChat usage on depression was better in the elderly than in the middle-aged. Although social participation had mediating effects in most groups, WeChat usage reduced the prevalence of depression through recreational activities only in middleaged adults and men. Possible explanations are that communication is the most pressing need for WeChat usage [60]. The use of social media such as WeChat can help older adults communicate more easily with family and friends, share their lives and communicate emotionally [15], which can have a more positive effect on mental health. Compared with the elderly, middle-aged people are less likely to face obstacles in action and communication because of better physical health. On the other hand, older adults still face some difficulties in using WeChat, but the middle-aged group has better learning and receptive abilities, so they are more likely to fully utilize social media such as WeChat for recreational activities and thus gain mental health benefits. We propose that the accessibility of social media for older adults should be taken into account to fully utilize the positive effects of social media on mental health. This study still has some limitations. First, we were using cross-sectional data as the CHARLS question item on WeChat first appeared in 2018, so we were unable to predict the causal relationship between WeChat usage and social participation and depression. In future studies, we will consider using longitudinal data to study this aspect; Second, the effect of WeChat usage on depression may also arise through other mediating variables, but this study only explored the mediating role of social participation. We will look for the possibility of other mediating variables in future studies. Our classification of activities may not be sufficiently objective because we did not find a unified classification standard for social participation activities. Finally, we have adjusted for covariates as much as possible and used propensity score matching to make the results as robust as possible, but the factors affecting depression are complex and there are still some unobserved or unaccounted for biases. Despite some limitations, this study also has some practical implications. First, we verified that WeChat usage and social participation were protective factors for depression in Chinese middle-aged and elderly people. Second, we found that WeChat usage could reduce the risk and levels of depression through social participation. This would imply that we can further promote social participation rates through social media usage to reduce the risk of depression among the middle-aged and elderly. --- Conclusions Our findings suggested that WeChat usage was significantly associated with lower risk of depression and lower levels of depression in a Chinese middle-aged and elderly population, with a partial mediating effect of social participation and recreational activities. These findings may have important implications for the prevention and treatment of mental health in middle-aged and older populations. Encouraging middle-aged and older adults to actively use social media and participate in social activities may be considered for the prevention and treatment of depression in the future to enhance mental health. --- --- Abbreviations --- CHARLS --- --- --- --- --- Competing interests The authors 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: ---

Background We aimed to explore the association between WeChat usage and depression in the Chinese middleaged and elderly and the role of social participation. Methods Data were obtained from China Health and Retirement Longitudinal Study (CHARLS) of 2018. The dependent variable was depressive symptoms, measured with the 10-item Center for Epidemiologic Studies Depression Scale (CES-D-10). We used the propensity score matching (PSM) to match the WeChat users with the non-WeChat users. Correlations between WeChat usage and depressive symptoms were verified by using logistic regression and linear regression, and the mediating role of social participation was verified by using stepwise regression and KHB method.Four thousand five hundred forty-five samples were ultimately matched for analysis in this study. After including all control variables, results of logistic regression showed that WeChat usage was significantly associated with a lower prevalence of depression (aOR:0.701,95% CI: 0.605-0.812). And the results of linear regression showed that WeChat usage was associated with lower levels of depression which was significant (p < 0.001). The results of the stepwise regression and the KHB method showed a mediating role of social participation in WeChat usage and depressive symptoms. Among the four types of social participation, the mediating effect of recreational activities was significant, while the mediating effects of voluntary activities, cultural activities, and other activities were not significant. Meanwhile, the effect of WeChat usage on depression and the mediating effect of social participation were heterogeneous because of differences in age and gender.Social participation partly mediated the effect between WeChat usage and depression in middle-aged and older adults. Among the four types of social participation, only recreational activities had a mediating effect. Encouraging more active social participation and other types of social activities should be considered to improve the mental health of the middle-aged and older adults in China through social media usage.

Background Substance use is a well-known risk factor for disease, e.g. cardiovascular diseases and cancers, and premature mortality worldwide [1][2][3][4]. Recent evidence from the Global Burden of Diseases, Injuries, and Risk Factors Study 2016 showed that smoking was globally the leading risk factor in terms of attributable disability-adjusted life years in men, while alcohol use ranked fourth [5]. For women smoking and alcohol were among the 20 leading risk factors for disease and premature mortality [5]. Similarly, a range of adversities has been associated with frequent cannabis use, particularly with an early onset in adolescence [6,7]. Despite its significant burden on individuals and society, research on substance use has remained a low priority in particular concerning migrant populations [8]. Little information is available on the substance use of adult migrant populations compared with the general population in the European context, especially regarding the association between substance use and socioeconomic and migration-related factors. Traditional cultural patterns of substance use in migrant populations may evolve due to acculturation and other factors after immigration to a new country [9]. Alternatively, maintaining the substance use habits of the country of origin can become an important part of the individual's ethnic identity. Migrants are exposed to vulnerabilities, such as stressful life situations, lack of social support and socioeconomic disadvantage, more often than the general population [10]. These vulnerabilities are often associated with substance use, and it has been suggested that especially young migrants of ethnic minorities are more prone to develop risky behaviours due to the acculturative stress and encountering discrimination and prejudices [11]. Population-based research on migrant's substance use in European context is scarce, although population-based research design allows for taking the diversities between migrant populations into account. In addition, to our knowledge the prevalence of alcohol use and illicit drug use among Kurdish migrants has not been reported before. According to previous literature, alcohol use among migrants in Europe seems to depend on the region of origin, and alcohol consumption and binge drinking among non-European migrants seems to be lower than in the general population [12][13][14][15][16][17]. The smoking rates of migrant populations in comparison with the general populations in European studies seem to vary according to migrant groups studied, and depend on the region of origin, gender and age of the migrants [12,15,16,18]. The findings of European studies reporting the prevalence of cannabis use among migrant adolescents of non-European origin compared with the general population adolescents are contradictory [12,13,19]. Information on substance use among Kurdish migrants is limited. To our knowledge, the prevalence of smoking and excess alcohol consumption among Kurdish migrants has only been reported in a Finnish study examining cardiovascular risk factors among older migrants in Finland [20,21]. However, different methodologies used in previous research, hinder comparing the results between different studies. Based on previous research, it has been proposed that migration-related phenomena, such as acculturation, the role of native cultural and social context, education, and generational differences, influence the substance use of migrant populations [14,17,19,22]. In contrast to other European countries, migration to Finland has started later and the proportion of migrants in the population is lower. The migrant groups selected to this study represent different migration backgrounds: Russian participants are mainly voluntary migrants having migrated due to personal or work-related reasons, whereas Somali and Kurdish participants are forced migrants having migrated due to humanitarian reasons [23]. Migrants from Russia and other areas of the former Soviet Union form the largest migrant group in Finland [24]. In Russia, alcohol consumption per person is among the highest globally [25] and the prevalence of smoking is also higher than in other European countries, especially among men [1,26,27]. Somali migrants are the largest group of refugees and asylum seekers in Finland, and they represent a population where the majority are Muslim by religion. In contrast to Russians, abstaining from alcohol use is common [28], and the prevalence of daily smoking among men is similar to European men, but lower in women [1,26,27]. Kurdish is one of the largest foreign-language groups in Finland, and the majority of the recently accepted quota refugees in Finland were from Iraq and Iran when the survey was planned [23]. Specific information about the prevalence rates of alcohol use and smoking in the Kurdistan area or among Kurdish migrants is very limited, but the prevalence of lifetime abstainers from alcohol in Iran and Iraq has been reported to be high [28,29]. The prevalence of daily smoking is similar or higher in men in Eastern Mediterranean Region compared with European countries but much lower in women [1,26,27,[30][31][32]. However, there are also contradictory findings reporting lower rates of daily smoking in Iran and Iraq compared to the European general estimates [33,34]. The reported national annual prevalence rates of cannabis use are fairly similar in Finland, Russia, Iran and Somalia [35]. The Finnish alcohol use and smoking habits differ from the general European estimates to some extent: heavy episodic drinking is more common among both genders, while men smoke less compared to the general European estimates [33,34,36]. Previous literature and the public health importance of substance use highlight the need for population-based studies reporting substance use among adult migrant populations by ethnicity in the European context. --- Methods --- Aim In this study, we aim to report the prevalence of alcohol use, cigarette smoking and consumption of illicit drugs among three migrant groups in Finland in comparison with the Finnish general population. In addition, our aim is to study which socio-economic and migrationrelated background factors are associated with substance use among population groups studied. --- Study design and procedures The data used were from a comprehensive cross-sectional survey, the Finnish Migrant Health and Wellbeing Study [23,37], , submitted), carried out by the Finnish National Institute for Health and Welfare during 2010-2012, and targeted at migrants of Russian, Somali, and Kurdish origins. Migrant was defined as having been born outside Finland and having residence permit in Finland. The Maamu survey consisted of a health examination and a structured face-to-face interview on health and wellbeing, conducted by trained field staff in the participants' native language or in Finnish. The personnel were of Russian, Somali or Kurdish origin, and thus were proficient in both Finnish and the participants' native language. A short interview including the most essential items of the interview and health examination was generated to reach those who were unable to participate in the interview and health examination. --- --- Ethical approval Ethical approval was granted to both studies by the Coordinating Ethics Committee of the Hospital District of Helsinki and Uusimaa. Written informed consent was obtained from each participant in the health examination and interview. In the case of participating only in the short interview, oral consent was obtained from those participating face-toface or by phone, and for those answering by mail, returning the filled short questionnaire was interpreted as a consent among those answering by mail. --- Measurements --- Alcohol use As a screening question in the interview and short interview, the participants were asked if they had used alcohol during the previous 12 months . Alcohol use was measured with the three-item AUDIT-C questionnaire, a short version of the 10-item AUDITquestionnaire. AUDIT-C has been found to be valid for screening alcohol problems among adults [39,40], and also of different ethnicities [41]. The frequency of alcohol use was measured by the item "How often do you have a drink containing alcohol?", which was analysed as three-class variable: "Never" combining the negative response to the previous screening question; "Occasionally" including "Monthly or less" and "2 to 4 times a month"; "Frequently" including "2 to 3 times a week" and "4 or more times a week". The frequency of alcohol use was included in the short interview. The usual number of alcohol units consumed was analysed as a fourclass variable: "No alcohol use"; "1 to 2 units"; "3 to 6 units"; and "7 units or more". The frequency of binge drinking was analysed as four-class variable: "No alcohol use"; "No binge drinking"; "Occasional" including "Less than monthly" and "Monthly"; "Frequently" including "Weekly" and "Daily or almost daily". The usual amount of alcohol units consumed and frequency of binge drinking were not included in the short interview. Binge drinking was selected as the outcome variable for the logistic regression analyses and dichotomised . A total sum score for the AUDIT-C questionnaire with cut-off limits of 4 for women and 6 for men was applied to indicate risky drinking [42]. In the Health 2011 survey the AUDIT-C questionnaire was included in the self-administered questionnaire. --- Cigarette smoking The items of cigarette smoking were collected in the interview and short interview. Lifetime regular smoking was measured by a question "Have you ever smoked regularly ?" . Frequency of current smoking ?)". was included in the short interview and measured as a three-class variable: "No" including also the negative responses to the previous item; "Occasionally"; "Daily". The quantity of cigarette products consumed per day was analysed as a three-class variable: "No smoking" according to the previous responses; "1 to 10"; "11 or more". The variable measuring the frequency of current smoking was selected as an outcome variable for the logistic regression analyses, and dichotomised . No data on the general population were available in a fully comparable form for the item of quantity of cigarettes consumed. --- Illicit drug use Lifetime use of cannabis was asked in the interview, and it was followed by a follow-up item about cannabis use during the previous 12 months . Lifetime use of intravenous drugs was asked in the interview. Lifetime use of cannabis was selected as an outcome variable for the logistic regression analyses. No data on the general population were available for these items. --- Socio-demographic variables The socio-demographic background variables used were sex, age group , marital status , level of basic education , employment and a subjective evaluation of the economic situation . No general population data were available for the item of economic situation. The migration-related background variables selected were age at migration to Finland , duration of residence in Finland , grounds for the residence permit and self-reported language proficiency of the local languages . Duration of residence, age at migration and language proficiency were considered to describe acculturation. --- Data management and statistical analyses The statistical analyses were performed with Stata software version 13 IC, and the survey's sampling design was taken into account in all analyses. The effects of missing data were accounted for using inverse probability weighting. Weights were determined by the main predictive factors of nonresponse: migrant group, sex, age, research municipality and marital status. Finite population correction was applied because of the relatively small population sizes and inclusion of significant proportion of the total population. Adjustment for age was used in all the reported analyses because of the significantly differing age distribution between the studied groups, but all preliminary analyses were also conducted without the age-adjustment [43]. The age-adjusted prevalences for substance use were reported separately for men and women in each migrant group and for the general population. The comparisons were made between men and women in each migrant group, and between each migrant group and the general population by gender. The frequencies were determined using cross-tabulation and the age-adjusted proportions were produced using logistic regression analysis and predictive margins [44]. The associations between the background and substance use variables were analysed using univariate logistic regression analysis adjusting for age. The analyses were conducted separately for each migrant group by sex, each explanatory background variable and each outcome variable . The odds ratios and 95% confidence intervals are reported as age-adjusted figures, and a p < 0.05 was considered to be statistically significant. --- Results Altogether, 1846 individuals took part in at least some part of the survey, with a sufficient response rate to the items on substance use ; Somali 48% ; Kurdish 61% ). Because all of the items were not included in the short interview, the total number of participants varies between items in the tables. The characteristics of the study population are presented in Table 1. The majority of the Somali and Kurdish participants reported to be Muslim and nearly half of the Russian participants reported to be orthodox Christian by religion. Almost a third of the Russian and a fifth of the Kurdish participants reported not belonging to any religion. The majority of Somali and Kurdish participants had migrated as refugees or asylum seekers, whilst the majority of Russian migrants had other grounds for their residence permit including family ties and employment. --- The prevalence of alcohol use The age-adjusted prevalences of substance use by ethnic group and sex are presented in Table 2. The proportion of abstainers was large among Kurdish men and women . The frequency of alcohol use was significantly higher among men than women in Kurdish migrants and the general population, but not among Somali or Russian migrants . Almost all Somalis reported abstaining from alcohol use during the previous year, and thus further analyses could not be performed in the Somali group. Occasional or frequent binge drinking was reported by 65% of Russian men, 30% of Russian women, 27% of Kurdish men and 6% Kurdish women. These prevalences were significantly smaller than in the general population . In general, binge drinking was more frequent in men than in women. The prevalence of risky drinking was lower in all the studied migrant populations relative to the general population . Risky drinking was less common among Kurdish women than Kurdish men, but no difference between men and women became evident in Russian or general populations. --- The prevalence of cigarette smoking The prevalence of lifetime regular smoking was lower in migrant populations than in the general population. The prevalence of lifetime regular smoking and current daily smoking were higher in men than in women in all the studied populations . The prevalence of daily smoking was higher in Russian and in Kurdish but lower in Somali men than in the comparison group . In contrast, the prevalence of daily smoking was lower in Russian , Somali and Kurdish women compared with the general population . Smoking 11 or more cigarettes per day was reported by 32% of Russian and 22% of Kurdish men. Somali women were excluded from the further analyses because of the low prevalence of smoking. --- The prevalence of illicit drug use The prevalence of lifetime cannabis use was 21% in men and 14% in women of Russian origin, and lower among Kurdish origin participants . Somali participants did not report cannabis use. Somali migrants and Kurdish women were excluded from the further analyses. Cannabis use during the previous 12 months was reported by 9% of Russian men and 2% of Russian women. The lifetime intravenous drug use was very rarely reported in all the migrant groups . --- Factors associated with substance use Socio-demographic and migration-related factors associated with binge drinking, daily smoking and lifetime cannabis use are presented in Table 3 , Table 4 and Table 5 , with odds ratios and confidence intervals. Younger age increased the odds for both binge drinking and daily smoking in Kurdish men and for the lifetime use of cannabis among Russian participants. Not being married or cohabiting increased the odds for binge drinking among Kurdish women, and daily smoking for all migrant men and for women in the general population. Lower basic education decreased the odds for binge drinking for Kurdish women, but increased the odds for daily smoking for Russian and Somali men and both sexes of the general population. Among women, being economically inactive decreased the odds for both binge drinking in Kurdish and general population and daily smoking in Russian, Kurdish and general population. Unemployment decreased the odds for binge drinking in Kurdish men and women, but in contrast increased the odds for daily smoking in men of Kurdish origin and the general population. An unsatisfactory economic situation increased the odds for binge drinking in Russian men, and of daily smoking among Somali and Kurdish men. Underage migration to Finland in contrast to migration in adulthood increased the odds for lifetime cannabis use in Russian men. A duration of residence in Finland exceeding five years increased the odds for binge drinking, daily smoking and lifetime cannabis use in Russian women, and of daily smoking in Somali men. Poorer language proficiency compared with good language proficiency increased the odds for daily smoking in Russian and Kurdish men, whilst it decreased the odds for daily smoking and binge drinking among Kurdish women, and of binge drinking in Russian men. --- Discussion This study adds to the limited information on migrants' substance use in European context. Here, we describe the prevalence and associated background factors on substance use among Russian, Somali and Kurdish migrants in Finland in comparison to the general Finnish population. Our study is the first to examine the habits and background factors of substance use among adult Kurdish migrants. The prevalence of alcohol use was lower among migrants compared to the general population, but daily smoking was more prevalent among Russian and Kurdish migrant men than in the comparison group. --- Alcohol consumption Alcohol use and binge drinking were less frequent, and the proportion of abstainers was larger among all migrant groups compared with the general population. This is in line with previous European studies reporting lower level of alcohol use among migrant populations than among the general populations, especially concerning migrants from non-European countries, and also with previous preliminary findings from the Maamu survey [12-17, 20, 45-48]. The prevalence of alcohol use among Kurdish migrants in our study was lower compared to previous findings on Iranian migrants in Norway [17]. Our results are also in line with previous observations on differences in alcohol use between migrant groups [17,49]. Based on the data on alcohol use in Russia, the alcohol use of Russian migrants was expected to be higher or similar to the general population [25]. However, the alcohol use among Russian migrants was found to be lower than in the general population. Similar results have been recorded among migrants from the Former Soviet Union in the United States [50]. This could be explained by the higher level of education among Russian migrants compared with the general population, and the selective nature of mainly voluntary immigration from Russia. This may, in part, be explained by the "healthy immigrant effect", indicating lower levels of substance use among migrant population in comparison to the general population [51,52]. The prevalence of alcohol use among Kurdish migrants has not been previously reported. The prevalence of alcohol use in Somali migrants was low as expected based on cultural and religious factors. Taken into account the cross-sectional nature of the study, we found that low acculturation was associated with lower levels of binge drinking. Low socioeconomic situation seemed to be both risk and protective factor depending on the migrant group under study, gender and on the individual factors. Lower levels of binge drinking were associated with poor language proficiency among Russian men, shorter duration of residence in Finland among Russian women, unemployment among Kurdish men and a lower basic education and unemployment among Kurdish women. Higher levels of binge drinking were associated with an unsatisfactory economic situation among Russian men, a younger age in Russian women and Kurdish men and marital status being other than married or cohabiting among Kurdish women. Previous literature has reported similar findings between the higher level of acculturation and more frequent binge drinking among men of some migrant groups [53,54]. --- Cigarette smoking Our findings regarding the larger proportion of daily smokers among Russian and Kurdish migrant men compared to the general population are in concordance with previous European studies and also with preliminary findings from the Maamu survey, reporting a larger proportion of daily smokers among migrants in comparison to the general population [12,18,20,21,[55][56][57], and in concordance with previous studies reporting differences between migrant groups under study [11,12,18,46,47,49,55,56,[58][59][60]. In contrast, previous migrant studies from Spain, Denmark and United States, have reported less smoking or similar smoking behaviour than the general population [15,16,50,60]. It is important to notice, however, that the smoking rates of Finnish men are lower compared to European estimates [33,34], and the migrant populations under study and definitions of migration differed from one study to another. Compared with the general Finnish population, daily smoking was more common among Russian and Kurdish migrant men, but less prevalent among Kurdish and Somali migrant women, in concordance with previous literature on gender differences in smoking among migrants [56,60]. Here, the "healthy migrant effect" might apply to women, but not to Russian or Kurdish migrant men. Additionally, differences between lifetime regular smoking and current daily smoking in the general population and Russian migrants might indicate that previous regular smokers have been able to quit smoking, in contrast to Kurdish and Somali migrants among whom this type of pattern is not observed. Similar pattern of lesser smoking cessation among some migrant groups has been reported before, and it might be that there is a gap in educational and smoking cessation services for migrants compared with the services available for the general population [59]. In concordance with previous results, daily smoking among migrant men and men of the general population was associated with poorer socio-economic situation [59,61]. Daily smoking was associated with not being married or cohabiting in all migrant men. A lower basic education was associated with daily smoking in Russian, and general population men, and unemployment among Kurdish and general population men. Our results on the association of daily smoking and educational level were in line to previous findings from Iraq [32]. An unsatisfactory economic situation was associated with daily smoking among Somali and Kurdish men. Our results imply that a disadvantaged socio-economic position might predispose especially Kurdish and Somali migrant men to daily smoking, taking into account the limitations of cross-sectional studies. --- Lifetime use of cannabis The prevalence of lifetime cannabis use was low among migrants of Kurdish and Somali origin. The reported prevalence rates of lifetime cannabis use among Russian migrants were smaller compared with those of the general population reported elsewhere [62]. Our results are in line with previous research reporting lower prevalence of cannabis use among migrant adolescents compared with the general population [12,63]. Lifetime cannabis use was associated with younger age, underage migration to Finland, and a longer residence in Finland in Russian migrants; the implication here being that the adoption of the Finnish culture of substance use could increase the cannabis use among Russian migrants. This is in concordance with previous research reporting that higher level of acculturation might be associated with higher likelihood of cannabis use among migrant adolescents and young adults [19]. --- Gender differences Our results underline the gendered nature of binge drinking and daily smoking in all populations under study, as reported in previous literature [17,56,60]. The pronounced gender differences among migrants compared with the general population might reflect the cultural differences in social approval of substance use among women [60,64]. Economic inactivity was a protective factor for daily smoking and binge drinking among all women, possibly reflecting stay-at-home parenthood. Among Kurdish men, being unemployed seemed to be a protective factor for binge drinking and a risk factor for daily smoking. Russian men showed a trend towards economic inactivity being a risk factor for binge drinking although this finding did not reach statistical significance. Thus, clear gender differences were found in the associations between employment situation and substance use. Being married and having a lower basic education could be interpreted as producing a higher adherence to traditional cultural norms among migrant women. These factors together with poorer language proficiency as an indicator for lower acculturation were all protective factors against binge drinking among Kurdish women. The gender differences in the drinking patterns of Kurdish migrants were reminiscent of the gender differences in drinking patterns in the Finnish population some decades ago; nowadays, this has dissolved with the alarming increase in the alcohol consumption of Finnish women and the simultaneous significant gender role changes in Finnish society [65,66]. The prevalence of abstainers has decreased and the proportion of weekly drinkers has increased among Finnish women between 1968 and 2008 [67]. Similarly to previous research, we found lower acculturation to be a protective factor against daily smoking in Kurdish women, while it had a contrary effect among Kurdish men [22]. Higher education and adoption of more flexible gender roles as a result of acculturation have been hypothesised to change the patterns of substance use among migrant women, and our results show that this might apply to migrant women of Russian and Kurdish origin [22]. --- Strengths and limitations of the study Our study provides unique data from three different migrant populations to be compared with the general population in one European country , and it has a relatively high response rate. However, the data used does not allow for analysing the effects of living in rural vs. urban environments. Examining substance use by ethnicity or nationality, as conducted in our study, allows for taking cultural and other variation within the studied populations into account. Reporting substance use among all migrants as one population group or grouping migrants from the same continent as one migrant group may produce observation bias [68]. The comparison between migrant studies may be challenging because different methods used: varying study populations, definitions of migration and migrant groups, and differences in measuring substance use [8]. One of the limitations was that, information was available on frequency of smoking, and no timeframe for "current smoking" was defined. However, self-reported smoking behaviour has been shown to be a reliable measure to use in epidemiological studies [69,70]. Our study adds to the limited body of literature on habits of substance use among migrant populations, including background factors associated with it. In addition, it is among the first to report the prevalence of substance use among Kurdish migrants. Despite the relatively high response rate and the use of inverse probability weights, the effects of non-response on our results are difficult to estimate. Considerable social stigma is associated with substance use, especially among the Somali participants, and this might have resulted in the under-reporting of substance use among the participants. Various background factors analysed in this study are inter-related, and thus the association of individual factors is difficult to estimate. --- Conclusions Our results highlight the varying patterns of substance use and differences in associated sociodemographic and migration-related factors between the migrant groups and between the genders within migrant groups. Our results draw attention to the lesser binge drinking among migrants compared with the general population in all the three migrant groups in our study, and to more prevalent daily smoking among Russian and Kurdish migrant men compared to the general population. Therefore in the future, the possible treatment gap for harmful substance use among migrants should be evaluated, and the availability of culturally tailored interventions for smoking cessation [71] and for alcohol use reduction for migrants should be assessed, as proposed in previous literature [8]. Oulu, Finland. 4 Department of Psychiatry, University of Turku, FI-20014 Turku, Finland. 5 --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Substance use is a well-known public health problem, but population-based research on migrants' substance use in Europe is limited. Factors related to the cultural background and current life situation might influence substance use among migrants. Here, the prevalence of substance use in Russian, Somali and Kurdish migrants in Finland is reported in comparison with the general population, and the associations between substance use and socio-economic and migration-related background factors among migrants are analysed. Methods: Cross-sectional data from the Finnish Migrant Health and Wellbeing Study (Maamu) and comparison group data of the general Finnish population (n = 1165) from the Health 2011 Survey were used. The survey participants were of Russian (n = 702), Somali (n = 512), and Kurdish (n = 632) origin. Substance use included self-reported alcohol use within previous 12 months (AUDIT-C questionnaire), current and lifetime daily smoking and lifetime use of cannabis and intravenous drugs. Results: Binge drinking was less prevalent among all migrant groups than in the general Finnish population (Russian men 65%, p < 0.01; Russian women 30%, p < 0.01, Somali men 2%, p < 0.01, Kurdish men 27%, p < 0.01, Kurdish women 6%, p < 0.01, general population men 87% and women 72%). Current daily smoking was more prevalent among Russian (28%, p = 0.04) and Kurdish (29%, p < 0.01) migrant men compared with the reference group (20%). Younger age and employment were associated with binge drinking among migrants. Socio-economic disadvantage increased the odds for daily smoking in Russian, Somali and Kurdish migrant men. Several migration-related factors, such as age at migration and language proficiency, were associated with substance use. Conclusions: Binge drinking is less common among migrants than in the Finnish general population. However, current daily smoking was more prevalent among Russian and Kurdish migrant men compared with the general population. Younger age, level of education, employment, duration of residence in Finland and language proficiency were associated with binge drinking and daily smoking with varying patterns of association depending on the migrant group and gender. These findings draw attention to the variation in substance use habits among migrant populations.

Introduction --- T he COVID-19 pandemic has influenced numerous points of intersection between built, biotic, social, and health processes. In this discussion, we draw inspiration from conceptual developments in public health that focus on social and environmental 'relations' and the entanglement between biotic/ ecological, social, and health processes ; as well as integrative approaches to public health . By taking the relational approach one factor further to include the built environment; and by broadening disciplinary perspectives to architecture and design, philosophy, and systems biology, our interest is to explore how the built environment regulates multiprocess pathways. There has been thorough research about the junctures between built environments, ecological factors, social structures, and health . Our particular aim is to develop a conceptual framework that can embed built environment causal relevance because the representational adequacy of the built environment in scientific models and data practices has been limited, to say the least. For ease of navigation, section 'Diagnosing representational framework inadequacy: epistemological, ontological, and methodological problems' is directed toward researchers specializing in modeling and methodology in scientific practice; section 'Developing a relation-based representational precedent' is directed toward interdisciplinary and transdisciplinary researchers with skeptical inclinations about integrative public health concepts; and section 'Representing the causal relevance of the built environment', which we take to be the central conceptual contribution of this paper, is directed toward architects, designers, and public health researchers interested in analyzing and visualizing the dynamic roles of the built environment in public health. Our focus is twofold. First, it is part diagnostic. In section 'Diagnosing representational framework inadequacy: epistemological, ontological, and methodological problems', we draw attention to failures in representational infrastructure and practice, pertaining to complex processes. We present three intertwined problems with scientific representation in COVID-19 data-gathering practices. These problems pertain to missing data, idealized measurement outcomes, and shallow surrogate measures-all of which culminate in misrepresentations of COVID-19 complexities. The overarching general problem can be summarized as follows: to what extent do current measurement and data practices fail to robustly account for the complex intersections between built, biotic, social, and health processes? We argue that the first step in solving the presented problems requires the development of new conceptual foundations that set the ground for detailed causal interconnections. Our second, and larger, focal point is prescriptive. Section 'Developing a relation-based representational precedent' motivates the conceptual need to broaden our representational frameworks by looking at counterintuitive causal relationships in SARS-CoV-2 transmission. For instance, we describe that highdensity built environment contexts can maintain low transmission rates and high public health resiliency behaviors. Additionally, we describe how low-density environments can maintain high transmission rates, moderated by factors like built environment connectivity and even social belief systems. We suggest that further parameterization based on contextual factors should be explored in order to adequately represent the junctures between built environment, social, and biotic processes. In our discussion of the ontology of relationality, we argue why the representation of nuanced causal relations within dynamic contexts is paramount for accurate representation ; and we make a forthright point about difficulties and pitfalls of relation-based integrative public health concepts . In section 'Representing the causal relevance of the built environment', we synthesize concepts from different disciplines in order to address how the built environment, centrally, figures into the intersections between biotic, social, and health processes. In section 'Relationality and the built environment: built contexts and nudges' and 'Relationality and the built environment: contexts and affordances', we discuss at least three conceptual developments that can be applied to the built environment: 'nudge', 'affordance,' and 'interface'. The purpose of these concepts is to organize the multiple regulating functions of built environment contexts . We apply the concepts in order to show how BC's can be cohesively analyzed alongside social and biotic variables-in addition to showing how BC's can be visualized side-by-side with other contexts . Our main argument is that the built environment can be represented not only as a single discrete variable but also as an interface that reorganizes causal relationsconcurrently, deregulating factors, while leaving others unaffected. We conclude that to adequately represent the various moderating roles of the built environment, we must acknowledge who ought to be involved in the process of representation . Diagnosing representational framework inadequacy: epistemological, ontological, and methodological problems Throughout the pandemic there have been many failures in representational infrastructure and practice, pertaining to key social and built environment factors. We argue that expanding representational scope to embed process-intersections requires making deliberate conceptual choices about what to selectively represent. Before we can address adequate representation, we must address inadequate representation of complex processintersections. That is, to schematize useful representational suggestions it is necessary to first focus on broken representational methods and practices. We present three intertwined problems with scientific representation in COVID-19 data-gathering practices. These problems pertain to missing data, idealized measurement outcomes, and shallow surrogate measures-all of which result in the misrepresentation of COVID-19 complexities. Facing these problems requires reassessing the epistemological comprehensiveness of our representations, the methodological robustness of our measurement practices, and finally, how our scientific frameworks inform epistemological and methodological changes. We argue that the resolution of the presented problems is not "more" data but, rather, the development of new conceptual foundations that set the ground for a detailed causal topography. Throughout the pandemic there have been limited measurement practices to adequately account for disparities in COVID-19 morbidity and mortality based on structural racism; and there has been limited representational infrastructure to account for the causal roles of inequitable built environments as moderators for morbidity and mortality. We begin with the former issue. In Keyser and Howland , we describe the unmet need to adequately represent connections between various forms of systemic racism, like environmental racism, and pandemic health outcomes . One representational problem is the absence of data outcomes about race, ethnicity, and social determinants of health, which would require methodological modifications in data-gathering . Missing data outcomes result in overlooked disparities. For instance, not only has there been missing COVID-19 morbidity and mortality data about Indigenous populations, there is also methodologically correctable missing EHR data about symptoms, underlying health conditions, hospitalizations, ICU admissions, and mortality . It can be argued that synthesizing complex relations between systemic processes and health outcomes is reserved for the level of theory. However, initial methodological failure occurs at the stage of measurement by ignoring representational factors that could serve as the foundation for theory-building . For instance, meta-data work on COVID-19 health outcomes and population factors from 50 countries document population characteristics such as socioeconomic status . While socioeconomic status is causally relevant for the evaluation of health-related outcomes, it does not tell the full causal story. We suggest that data should be further partitioned beyond socioeconomic parameters. For instance, there are racial disparities in built environment pollutant exposure even when socioeconomic status is controlled for . Reparameterization could be a first step, although an insufficient one, for increasing causal resolution. In other words, the representational problems go deeper and are not resolved by a surface-level parametrization. The reason being that data partitioning merely based on race, ethnicity, and socioeconomic factors still risks using general surrogate measures and skewing causal complexity. There is an intertwined representational problem about the ontology used within our research programs: the absence of a representational framework that can embed causally relevant built environment processes that promote disproportionate health outcomes. For instance, currently there is limited infrastructure in electronic health records , used in machine learning models, to account for the causal role of inequitable built environments as moderators of morbidity and mortality . Built environment factors have seemingly important, but epistemologically opaque, regulating causal roles in our current research perspectives. Significant findings based on data from 3000 US counties predict that a 1 μg/m 3 increase in chronic PM 2.5 exposure is associated with a 9% increase in COVID-19 mortality . But such data fails to account for disproportionate pollution effects. For example, in a joint workshop report of ERS, ISEE, HEI and WHO, air pollution and its relation to COVID-19 severity is discussed but the only reference to social determinants of health is that they "…are clearly visible around the world" . Adequate representation of specific causal roles at the intersection between the built environment and social determinants of health is missing. Brandt et al. hypothesize disparate exposure to air pollution as a causal factor that contributes to the disproportionate impact of COVID-19 on Black and Latinx populations. The hypothesized moderating role of built environment pollution on health disparities is significant ; but there is still a need for conceptual and methodological infrastructures to account for the nuanced causal roles of the built environment. Brandt et al. present the methodological need to factor in new types of measures and models about 'accumulated lifetime exposure' to explain why the temporary decrease of PM 2.5 will not alleviate health outcome disparities. A set of ontological, epistemic, and methodological questions emerge: How do we represent the specific regulatory roles of the built environment; How do we operationalize those regulatory roles; and How do we create new methodologies to address those roles? An argument that data practices and theory-development will "eventually" lead to causal resolution is not sufficient because initial representational steps lack fine-grained details about causally relevant built environment factors. For example, current studies that link long-term exposure of PM 2.5 , PM 10 , NO 2 , and CO to decreased COVID-19 outcomes do not document race and ethnicity, let alone details like accumulated lifetime exposure or local variations in built environments and pollution. In other meta analyses that assess ambient air pollution and COVID-19 , multifactorial risks, race, ethnicity and social determinants of health are severely lacking or fully missing. Changes in measurement methodology and data-partitioning are necessary to understand the causal roles at the intersections between social and built environment factors. To clarify, this is a problem with selective representation in our data and modeling practices : certain variables are explored while others are ignored; and key process-intersections are fully unaccounted for. A humanities lens offers some perspective about selective representation in science. According to 'perspectivism' in the philosophy of science literature, measurement, experimentation, and modeling all produce limited representations of real systems; and those representations are accurate to aspects and degrees . For instance, a given representation can capture key properties of a system, while missing or idealizing other properties. Even if current COVID-19 research is pushing for the evaluation of new causal factors there is still a need to evaluate other ecological, built, and social factors. We take the view that a given real system is relationally complex. However, by creating robust measurement and modeling perspectives, we can aim to adequately represent important features and relations of a given system. The 'perspectivism' conceptual framework can be boiled down to a precise visual. Imagine a set of complex intersecting processes and multiple ways of selectively representing those processes . Our initial step in the diagnoses of scientific practice in COVID-19 is to understand how we are failing to expand perspectival/representational scope to intersections that involve built and social processes. Notice that although Fig. 1 is a general schematic of representational scope, there are a number of precise points of reference. First, causal factors can be represented as intertwined processes instead of as simple nodes. It is important to emphasize that a given built environment should not be reduced to a single factor; rather, it should be investigated as an evolving process Second, we draw attention to the importance of investigating process-intersections for the purpose of understanding compounded and emergent effects. For instance, built and social processes can concurrently determine social activity, relevant to public health outcomes. Chang et al. highlight the extent to which race, ethnicity, and socioeconomic status limit the set of mobility pathways within built environment contexts, creating incongruent risks in COVID-19 morbidity. Failures in widening representational scope over process-intersections may also indicate failures in interdisciplinary and transdisciplinary scientific collaboration. After all, the synthesis of built, biotic, and social factors within a given measurement practice requires multiple disciplines-an issue that we will return to in section 'Developing a relation-based representational precedent'. So far, we have discussed representational failures in datagathering and partitioning. But there are similar issues with selective representation, pertaining to predictive algorithms in COVID-19 contexts, where representational frameworks do not account for causally relevant variables. Selective representation of public health causal factors is sometimes applauded due to effective general prediction. In Howland et al. , we make the point that predictive success in machine learning does not translate to representational adequacy; and we suggest new ways of incorporating information about social determinants of health and inequitable built environments into machine learning practice. In electronic health record settings, it may be too much to expect someone to answer detailed questions about lifetime pollution exposure and other relevant built environment factors. But this need not deter us from incorporating built environment variables into EHR data practices. GIS data about built environment pollution is readily available and could be integrated into EHR. Reliable methods for GIS-EHR integration have been explored . But to apply those methods to relevant COVID-19 parameters, a representational precedent has to serve as an adequate guide. It is worth noting that other researchers have rigorously pursued 'small data' to investigate fine-grained causal relations in built environments-e.g., by investigating how built environment airflow processes moderate transmission . Kwon et al. have integrated viral genome data from local hospitals, GIS data, CCT footage, cross-referenced travel data, and recreated built environment conditions to trace how airflow pathways and seating arrangements can spread SARS-CoV-2 particles ~21 feet and infect individuals in around five minutes exposure. While there is much to improve, methodologicallyincluding stricter evaluation of genome matching and elimination of other social contact-such scientific work is fascinating because of its integration of methods and evaluation of multiple variables and scales-with the built environment serving a crucial moderating role. --- Social Process Fig. 1 Representational scope. This is a general schematic of representational scope, where a given system is depicted as intersecting processes ; and scientific perspectives can be used to selectively represent aspects and degrees of that system . Causal factors are represented as intertwined processes with a focus on process-intersections . For instance, a given built environment process and a given social process can concurrently determine social activity relevant to some biotic process. Beyond selective representation in epidemiological and EHR data, the same problem exists in computational models that attempt to predict the behavioral dynamics of social contact. Cutting-edge computational research presents a model of adaptive, optimal control of social contact rate . The model focuses on variables like economic and personal health costs, and crucially, information time delay. 'Optimal contact rate' is modeled as a utility function, parameterized by perceived risks and benefits of social contact. This adaptive social contact model usefully predicts the importance of early lockdown interventions, in addition to other interesting conclusions about oscillatory dynamics-e.g., dynamic regimes occur more often when there is a time delay . But there are a number of representational idealizations. Some are transparently stated. The authors admit that the model homogenizes individual responses to government policy. But there is a deeper representational issue. As we analyze in section 'Developing a relationbased representational precedent', socio-behavioral responses are associated with moral and social belief systems that are inextricably embedded within contexts. The authors state that, "Geographic and/or cultural variation in our parameter ['optimal contact rate'] are likely to affect how epidemic dynamics are affected by such tradeoffs" . We suggest that rather than taking the long view that geographic, cultural, and other variations will likely be "filled in" with later iterations of the model, it is crucial to understand how such parameters can be represented as differentially moderating the optimal contact rate in diverse contexts. For instance, applied empirical research differs with the Arthur et al. model implications, precisely for context-specific reasons-e.g., studies that focus on race, ethnicity, culture, and incongruent risks, moderated by mobility pathways, provide more nuanced early mobility restriction suggestions . Additionally, as we will see in section 'Representing the causal relevance of the built environment', there are counterintuitive causal relations between the earliness of lockdown, increased reactance, and reversal of the success of some types of 'nudge' interventions. The causal connections relevant for evaluating COVID-19 social contact rates are quite sensitive and necessitate a detailed analysis of social, built environment, and biotic factors-with a focus on how contexts shape relationships between factors. Adequately representing contextual variations requires scientific work that extends beyond merely filling in new variable values within the existing model. It requires a reconstruction of the model with a new set of variables that adequately reflect changing multivariable landscapes . Combining the aforementioned problems-failures of selective representation in epidemiological data and EHR, as well as failure to reflect relevant upstream causal connections-we present a general issue for scientific practices in COVID-19 contexts. The preceding problems stem from the reliance on surrogate measures within our measurement and data practices. For instance, when implementing predictive models that use binary questions about race, fully missing parameters on built environment pollution exposure, and comparatively inadequate biomarkers that impede deeper investigation into structural and social determinants of health , we cannot expect to improve causal resolution of multifactorial relationships. Generally, a surrogate measure/marker can serve as a substitute for a clinically meaningful endpoint , although in this discussion we characterize 'surrogate measures' more generally to describe substitute or inferential measures. A representational and methodological ideal is a surrogate measure that provides reliable information, regarding processes that cannot be investigated directly. For instance, pragmatically we cannot track each individual case of transmission to map out causally relevant factors, so surrogate measures are needed for ease of general prediction and explanation. Likewise, high-risk vs. lowrisk COVID-19 prediction can become causally cumbersome, so risk prediction models, like the one developed by Garibaldi et al. , provide a small set of surrogate parameters to make quick predictions-e.g., white blood cell count, absolute lymphocyte count, and respiratory rate-in addition to a set of questions about symptoms and risk factor parameters . The ideal scenario for a valid surrogate measure is that multifactorial processes are causally tied to some surrogate measure, which is strong in both specificity and sensitivity . 1 This ideal scenario is rarely met in biological surrogate measures. But we suggest that a combination of biological, social, and built factors makes surrogate measure use even more methodologically complicated. We propose that there are a number of parallel surrogate measure problems, similar to ones from biological surrogate measures , which are relevant for modifying COVID-19 scientific practices. Surrogate measures pose at least 3 abstract problems . Each problem pertains to the adequate representation of multifactorial upstream causes. We argue that surrogate measures alone do not provide the topographical causal story about relations; and hastily acquired surrogate measures can impede causal clarity and misrepresent relevant upstream causes -especially about social and built environment factors. Ignored upstream causes. As Keyser and Sarry outline, a precursor factor, context, or event can create branching parallel causal pathways, such that the surrogate measure does not tie into the causal pathway of the relevant health outcome . This would be akin to using a biomarker measure to differentiate high-risk vs. low-risk COVID-19 infection, while ignoring other biomarker pathways with specificity and sensitivity, relevant to the disease outcome. For instance, we have argued elsewhere that IL-8 has been missing as a key biomarker in recent machine learning models, even though it shows promising results for predicting mild vs. severe disease prognosis, especially in the context of multifactorial interleukin evaluation . One representational concern is that by focusing on limited biomarkers a different pathway, consisting of relevant upstream cascading factors , may be fully ignored . The same problem can be found in surrogate measures that lie only within one of many parallel causal pathways in multifactorial processes that are relevant to some health outcome (Fig. 2 --- . S2). We propose that the need to represent upstream complexity extends beyond a given biomarker's predictive usefulness. Adequately investigating upstream causes opens avenues to representing the roles of social determinants of health. For instance, analytical methods suggest that multiple molecular mechanisms are overexpressed due to social determinants of health . The overexpression of a number of genes in IL-8 signaling pathways-e.g., CXCR1, CXCR2, GNG10, and LIMK2-is robustly associated with a combination of social stressors . Important for our later discussion of ontology, Gaye et al. describe the significance of evaluating context-specific molecular effects-e.g., in inequitable "social environments", constituted by a number of "psycho-social" factors . Note here that a given set of social determinants can be represented as moderating the molecular mechanisms responsible for a given pro-inflammatory response. The relevant representational pathway can be characterized as upstream causal factors that begin with social and structural determinants , move to epigenetic processes , and culminate in pro-inflammatory IL-8 signaling and, subsequently, some relevant health outcome . Investigating only one biomarker can risk missing upstream complexity . So, while IL-8 can offer predictive utility as a quick EHR measure, it can also serve as an 'information lever' for the deeper epidemiological causal story about how social determinants of health are promoting health disparities . Notice that for the sake of simplicity, this example includes linear pathways only and does not address the difficulty of complex social and built process analysis, which we address in section 'Representing the causal relevance of the built environment'. The upshot to more nuanced causal representation is that the junctions between built environments and social determinants of health can be adequately tracked. Multifactorial relationships stemming from built environments. As Keyser and Sarry detail, a surrogate measure can be modified by some intervention, event, or context that has direct and/or indirect off-target effects on the health outcome of interest . This type of scenario is difficult to disentangle, causally, because the surrogate measure is modified by a given intervention at the same time as the intervention modifies other off-target effects. As we expand in section 'Representing the causal relevance of the built environment', even simple built environment changes/interventions can create branching causal pathways, which are difficult to represent. For instance, public park closure can prevent virus transmission within the population; but it can also inhibit resilience-promoting community movement behavior, while concurrently moderating molecular mechanisms relevant to lung inflammation. Although the configuration in Fig. 2. S3 provides a useful starting step for synthesizing built, social, and health factors, one representational problem persists: the oversimplification of built environment factors. It is important to note that a diagrammatic CE F 1 F 2 F 3 O S S1 CE F 1 F 2 F 3 CE' F 1 ' S F 3 ' CE" F 1 " F 2 " F 3 " O S2 CI F 1 ' F 2 ' F 3 ' F 4 ' S F 1 F 2 F 3 S O S3 CE F 1 F 2 F 3 F 1 ' F 2 ' F 3 ' CE" F 1 " F 2 " F 3 " S S0 O Fig. 2 Surrogate measures. The various forms of surrogate measures. CE : causal event; S: surrogate measure; F: causal factor; O: measurement outcome; CI: causal intervention. These representations are adapted from the account presented by Fleming and Powers and detailed in Keyser and Sarry . Beginning with S0, the ideal scenario for a valid surrogate measure is that multifactorial processes are causally tied to some surrogate measure, which is strong in both specificity and sensitivity. In S1, branching parallel causal pathways are represented such that the surrogate measure does not tie into the causal pathway of the relevant health outcome. In S2, surrogate measures lie only within one of many parallel causal pathways in multifactorial processes that are relevant to some health outcome. In S3, a surrogate measure can be modified by some intervention, event, or context that has direct and/or indirect off-target effects on the health outcome of interest. This scenario is difficult to causally disentangle because the surrogate measure is modified by the intervention, concurrently, as that intervention modifies other off target effects. structure like S3 fails to represent how built-to-social relations change within a given context. Filling in parallel built factors is just the initial step in visualizing how a given built environment regulates multiple factors that can, subsequently, produce amplified effects on a given health outcome; but the representational picture is still quite limited. In summary, surrogate measures alone do not provide a topographical causal story about relations between built, social, biotic and health processes. Additionally, hastily acquired surrogate measures can impede causal clarity. The purpose of discussing surrogate measure problems is to illustrate just how much representational work lies ahead, if we aim to go beyond shallow and limited measures in our data, EHR, and modeling practices. One might argue that finding valid surrogate measures and clarifying complex sociobiological mechanisms are separate scientific activities. But we think that these activities can inform each other. That is, by providing an adequate representational infrastructure and filling in causally relevant intersections, we can generate more robust surrogate measures, which can, in turn, dial up causal resolution in our models and measurement practices. Furthermore, in the next section we argue that solving these representational problems requires perspectival input from independent disciplines in order to piece together a representationally robust picture. We suggest that epistemic opacity goes hand-in-hand with a lack of infrastructure in interdisciplinary and transdisciplinary frameworks for the investigation of complex, intertwined processes. So far, we have presented a general problem: To what extent do current measurement and data practices fail to robustly account for the complex intersections between built, biotic, social, and health processes? Taking a fine-grained view, there are a number of interrelated problems: Measurement outcome problem. There is an absence of data outcomes about race, ethnicity, and social determinants of health, all of which would require a methodological modification in datagathering and/or analysis. While this may seem easily correctable, it points to two deeper problems. Representational framework problem. There is an absence of a representational framework that can embed causally relevant processes, responsible for disproportionate health outcomes. Related to the aforementioned problem, an absence of data outcomes can occur from a lack of representational infrastructure to account for the causal roles of inequitable built environments on health outcomes. Built environment factors can operate at many scales; and while some built environment moderating roles should be obvious, they are currently not being integrated into interdisciplinary and transdisciplinary research programs. As we will expand in sections 'Developing a relation-based representational precedent' and 'Representing the causal relevance of the built environment', built environment contexts, consisting of multiple variables, can moderate environmental and biotic factors, behavioral nudges, and norm changes in populations. Surrogate measure problem. Surrogate measures alone do not provide the topographical causal story about relations; and hastily acquired surrogate measures can impede causal clarity and misrepresent relevant upstream causes. The larger projects ahead of scientific communities include adequately representing multi-process intersections and incorporating new representational frameworks into scientific practices. But small steps can be taken to solve the Outcome, Representational Framework, and Surrogate Measure Problems. Our contribution to solving these problems is conceptual. We describe a representational precedent that focuses on 'relations' , and we clarify how to represent the unacknowledged moderating roles of the built environment . --- Developing a relation-based representational precedent Confronting the aforementioned representational problems requires building a conceptual precedent that can embed new ways of representing complex processes. In section 'Representing the causal relevance of the built environment', we discuss how the various roles of the built environment can centrally figure in the development of new modeling and methodological approaches. But first, it is important to expand the motivations for a conceptual precedent that focuses on an ontology of relationality: Why is the representation of nuanced causal relations important in COVID-19 contexts ; and what are some difficulties and pitfalls of relation-based integrative public health concepts '? These focal points are key for understanding the challenges of conceptual cartography ahead of us, if we hope to challenge the aforementioned representational problems. Why focus on relations in COVID-19 contexts? It might appear that representing nuanced causal interconnections may not constitute an effective scientific approach for quick prediction and explanation. Why not focus scientific efforts on pinpointing the most critical variables for SARS-CoV-2 transmission? For instance, population density is a key determinant of transmission and replication, so while interconnections between multiple factors might be relevant for piecing together a larger explanatory picture, such an approach is not necessary for proactive interventions. However, a deeper investigation into relationships reveals counterintuitive, critical variables at the intersection between the social, biotic, and built. Unusual relationships exist between high population density and low doubling times of virus spread. It turns out that virus transmission rates in dense populations are significantly associated with moral and social belief systems. First, high selfreported belief in freedom of assembly, association, and agency are correlated with increased COVID-19 transmission rates . In fact, internationallygathered data supports that individualistic belief cluster variables are more accurate predictors of transmission rates than GDP per capita, government effectiveness, preventative interventions, or quality of/access to healthcare . Additionally, even in low population density and high per capita GDP, civic and social participation, government effectiveness, preventative interventions, freedom of assembly and association, personal and family relationships, there is a significant increase in COVID-19 transmission rates . Furthermore, placing high value on doing something good for society and behaving properly do not decrease COVID-19 transmission; likewise, high transmission rates exist even in contexts where belief systems focus heavily on personal, family, and social network relationships . This is rather puzzling because, intuitively, it seems that consequentialist, deontological, virtue, social, care, and public good belief foundations should have a significant effect on transmission rate. Instead, societies that are "compliant" to both individuals and their governments showed the highest doubling times even in the context of increased population density . To frame this data, using comparative 'robustness analysis' , even when countries and environmental contexts vary, beliefs about doing something for the good of society, social group, and family, in addition to beliefs about proper behavior, are not sufficient for COVID-19 transmission decrease. Additionally, government conformity alone is not effective for transmission decrease. Moreover, comparatively lower scores on doing something for the good of society and behaving properly do not result in increased transmission rates. Interestingly, much of the Kapitány-Fövény and Sulyok data parallels independent pre-COVID data on the effectiveness, or lack thereof, of regulatory behavioral 'nudges' in different countries, which will be discussed in section 'Representing the causal relevance of the built environment'. This is just one smallbut-crucial focal point at the intersection between moral and social beliefs and biotic processes. If we scientifically represent the pandemic context as being regulated mostly by density and movement, we will miss key and counterintuitive causal relations, linking belief systems to transmission rates. This is also why a conceptual precedent has to be set for representing the varying 'contexts' of the pandemic. That is, it is not the case that contexts are homogenous with respect to causal variables-even if they appear to be similar. We turn to this point. Behaviors are shaped by environmental and built contexts. Kapitány-Fövény and Sulyok make a broad, speculative point about the connections between belief systems, social relationships, and socio-environmental contexts. But we argue that one can see direct examples of behavior within specific contexts by analyzing features of the built environment. Quite obviously, built environment contexts are causally relevant to density and movement. Large metropolitan areas have had higher infection and death dates compared to low-density communities . This is due to higher connections and social interactions . However, Hamidi et al. found that after controlling for metropolitan population, county density is unrelated to confirmed SARS-CoV-2 infection rates and is inversely related to confirmed virus death rates-putting into question city planning focused on sprawl for public health purposes, as well as the predictive and explanatory relevance of 'density' over more adequate built environment variables like 'connectivity'. In this case, attention is drawn to community movement and interaction, moderated by built environment variables beyond density. Noteworthy for exploring new variables in the built environment, higher COVID-19 morbidity and mortality rates result in higher outdoor activity restriction . Clearly, the built environment can constitute mediating and moderating structures for transmission. But it can also serve as a substrate for resilience-building behaviors. The representational focus on the latter has received minimal attention. The built environment can promote "community resilience" through healthy movement behaviors . Specifically, even though dwelling density is usually associated with decreased outdoor activity, access to parks within 1 km shows promising increased outdoor activity that promotes healthy movement behaviors in COVID-19 . Here, the representational implication is to focus on how relevant built environment contexts modify public health outcomes, transmission being just one relevant public health outcome. We make two points about representational infrastructure limitations. First, there has been a limited focus on the regulatory role of the built environment on social and biotic processesespecially the health-promoting roles of built environments. Second, RBEC evaluation is often subject to simplification and misrepresentation. For instance, take the already-limited set of RBEC's applicable to density and movement. Hamidi et al. discuss parameterization based on "compactness indices", which are the opposite of sprawl indices and consist of an amalgam of variables, measuring four dimensions of the built environmente.g., density, mixed use, activity centering, and street connectivity. Unaccountably, often only one of these variables is used in a given study . Such choices constitute epistemological and methodological limitations that prevents deeper analysis of built environment and health outcome relations. For instance, in the aforementioned RBEC studies on movement, it is assumed that a given RBEC, homogenously, affects population movement-whether county travel or healthy movement behavior. However, we have seen that built contexts can be modified by environmental racism, which can thwart resilience-promoting interventions. During the first wave of pandemic lockdown, it became apparent that public park access is not equitable for a number of reasons: lack of parks in close proximity to Black and Latinx populations; park entry fees; and various forms of racial discrimination . We suggest that built environment representations must focus on detailed causal relations within contexts. Well-rounded contextual representations could promote more nuanced interventions that are directed toward equitable park access-e.g., interventions proposed by the 2019 Urban Institute report, "Investing in Equitable Urban Park Systems" . It is important to avoid conceptual vagueness about 'context'. There have been thorough conceptual developments in areas of research such as, Urban Geography, Sociology, and Environmental Cognition. Gerson and Gerson's account broadens RBEC scope beyond limited variables like compactness indices. A given 'place', consisting of physical processes, modifies the interactions, conduct, and perspectives of its individual and collective occupants . Furthermore, there are feedback loops between the physical components of a place and the agents that occupy it, such that 'mutual adjustment' and, what we call, co-modification occurs: Places modify agents by shaping value systems, pace and type of interaction, as well as qualitative perspectives; and those agents, in turn, modify the built and natural features of a place . Additionally, we can add that different contexts-composed of physical, value-laden, temporal, and existential interactionsproduce emergent properties. Gerson and Gerson's 'place perspectives' can be utilized to explain non-homogenous causal relevance within a place. Even within the same location, one person may have access to public health-promoting pathways, while another may occupy a number of interaction-dependent contexts, where all roads are blocked . We explore a multitude of relational concepts in section 'Representing the causal relevance of the built environment'. For now, the benefit of such a conceptual precedent-one that focuses on causal feedback loops within places-is that it precisely organizes existing complex relationships instead of making homogenizing causal idealizations. To summarize, counterintuitive causal connections are informative even when critical COVID-19 variables like density appear as the obvious causal culprit. High-density contexts can maintain low transmission rates and high public health resiliency behaviors. We suggest that parameterization based on contextual factors should be explored, like the intersections between belief systems and biotic processes or the intersections between built environment contexts and biotic processes. Essentially, a conceptual infrastructure should be sought out to provide adequate characterizations of 'relation' and 'context'. Relation-based integrative concepts: warnings and developments. The general purpose of our discussion is to ground the importance of developing relational perspectives on complex entangled systems. But first, it is necessary to make a forthright point about using catching words-e.g., 'relational', 'integrative', and 'transdisciplinary'-in order to promote conceptual clarity. When specified, such concepts can be methodologically useful; otherwise, they may promote disciplinary silos. Additionally, our concern, in line with section 'Diagnosing representational framework inadequacy: epistemological, ontological, and methodological problems', is that by getting the ontology wrong the epistemology and methodology suffer. Definitions of 'transdisciplinarity' offer many conceptual pathways. For instance, there is a sharp distinction between transdisciplinary focus that is directed toward abstract epistemological holism, consisting of "unity of knowledge", compared to pragmatic participatory problem-solving approaches . The latter offers methodological heterogeneity-i.e., different viewpoints, experiences, and skills-in addition to mutual learning between scientists and participants . Others have developed similar methodological perspectives, where transdisciplinarity focuses on "bottom-up mutualization of methodologies and theories" . The upside of the participatory problem-solving approach is that two dimensions to 'transdisciplinarity' are emphasized: a focus on synthesizing multi-methods and an interest in coordinating scientists and other participants . The downside is that vague conceptions of 'transdisciplinarity' have led to methodological problems-such as, the fragmentation and reinvention of methods. As a result, new integrative concepts have emerged to synthesize disciplinary and stakeholder knowledge, and to develop new approaches for policy changes . Views like 'post-normal sustainability technologies' and 'integration and implementation sciences' attempt to fulfill a similar role to 'transdisciplinarity', e.g., by positing an iterative problem-solving process with integration and implementation at every stage. But new and multiplying concepts often miss the same key problems as their predecessors. One issue is about specifying methodology -i.e., how does bottom-up mutualization work relative to a particular problem? A second issue is that many transdisciplinary approaches refer to stakeholders and public parties beyond academia ; but larger integrative activities end up solely in scientific communities . That is, in these integrative approaches there are often vague suggestions for societal dimensions and culturally inclusive frameworks. If specified, both could promote conceptual and practical developments. But there is no specification for how applied activities work. Specific methodological suggestions with clear aims for public participation are particularly important in COVID-19 contexts, if our temporal goal is to intervene quickly and flexibly. However, vagueness in applied transdisciplinary activities can be a function of vagueness in conceptual foundations behind methodology. As we discussed in section 'Diagnosing representational framework inadequacy: epistemological, ontological, and methodological problems', specifying methodology requires fine-tuning conceptual foundations, for instance, by detailing how to parameterize intersecting systems in our measurement and modeling practices. Ideally, integrative concepts of health can transform epistemological and methodological realms to set the ground for future transdisciplinary practice . The benefit of such concepts is that a fully developed view of 'transdisciplinarity' is not necessary in order to achieve important epistemological and methodological approaches. However, Assmuth et al. point to a number of problems with integrative approaches in public health. Integrative health concepts and advances create disciplinary siloes-with expansion being dependent on funding privileges . Such concepts also implicitly and explicitly contain both anthropocentric and Eurocentric frameworks, resulting in dominance of concepts, approaches, narratives, and policies . Who shapes the integrative approach becomes paramount to the practice and method . Additionally, integrative health concepts lack the fine grain to analyze societal dimensions and natural systems . Finally, they have limited socioeconomic representation; and, for the most part, lack detail about contexts and processes . Particularly important for our discussion in section 'Representing the causal relevance of the built environment' is the lack of integrative concepts that focus on socio-ecological processes and contexts. However, getting a fully developed integrative concept that is applicable to changing pandemic contexts might be a multi-stage pursuit-such that, from the research-and-development vantage point, we have not even left the surveying stage. Even not-so-new interdisciplinary fields like ethnobiology-focusing on the relationships between humans, cultures, and the biophysical environment-are going back to the drawing board to figure out where the discipline should go, post-pandemic. Interestingly, the majority suggestion is to explore impacts on local communities and researcher-to-community relationships . Fret not, even at the surveying stage there is a clear conceptual goal, arising from all of the integrative concept discrepancies and shortcomings. We find the transparent academic call for reevaluation of relationships, namely the type and direction of relationships, particularly important as a starting step toward conceptual clarity. Assmuth et al. argue that many integrative concepts misrepresent relationships: concepts like 'One Health' , EcoHealth , 'ecosystem health' , and 'planetary health' have a notable limitation of anthropocentrism, as well as a limited focus on socio-ecological interconnections. For instance, in the literature on One Health and ecosystem/environmental health, the environment is often represented as an 'external influence' . Similarly, anthropocentrism often makes its way into sustainable process-views of the built environment. 'Green design' , 'ecodesign' , and 'cradle-to-cradle design' , offer relation-focused perspectives between built, ecological, and social environments-but with limited perspectives on complex ecological and social processes. In this case, the implicit representational problem constitutes treating ecological processes as 'external influences' and ignoring intricate socio-ecological relationships. We suggest that both biotic and socialsustainability conceptual infrastructures can benefit from the development of a more robust process ontology, focused on relations. Our motivation for this section is to be upfront about gaps in integrative concept developments. We are still in the early stages of incremental concept-and method-building. Our overall contribution to the construction process is to unpack 'relationality'-specifically, how it pertains to the built environment . However, it is also to propose a specific public-centered methodology for parameterizing the built environment . We are not proposing a new top-down integrative concept. Rather, we take a bottom-up approach of tracing fine-grained built environment regulating functions. We draw conceptual inspiration from Hinchliffe et al.'s discussion of 'healthy publics'. According to Hinchliffe et al. , "Healthy publics are collectives that take seriously the social and environmental relations that make health and well-being possible. These are dynamic collectives, composed of people, material processes, and ideas that can provide conditions for health and well-being" . Particularly important for our discussion, Hinchliffe et al. emphasize the role of relations in their account of healthy publics: "These [relations] may range, as we have suggested, from the social relations that characterize a community, to the ecological and material relations that make health possible to the systemic inequalities, institutional and other structural determinants that shape how opportunities for health and well-being are unevenly distributed " . The relationships between culture, social structure, and ecological settings produce emergent properties, which can shape access, opportunity, and outcomes in health . We believe that this makes the adequate representation of causal intersections all the more important for understanding the dynamics of health processes in complicated biotic and social environments. For instance, food insecurity, caused by inequitable social pathways, can be exacerbated by environmental contexts, resulting in reduced health outcomes, even in the presence of proper pharmaceutical interventions . 2 Upstream intervention within both social and environmental contexts becomes necessary before downstream pharmaceutical intervention can be effective. Understanding the multi-scale topography of biotic and social environments can guide effective health intervention planning. By taking the relational approach one factor further to include the built environment; and by broadening disciplinary perspectives to architecture and design, philosophy, and systems biology, our interest is to explore how the built environment regulates multi-process pathways. There has been thorough research about the connections between built environments, ecological factors, social structures, and health . Our particular interest is a conceptual framework that can embed built environment causal relevance. Specifically, representational adequacy of the built environment has been limited, to say the least. As Clifton and Perez note: The built environment has been represented in empirical analysis as discrete, disaggregate variables representing individual elements and as composite measures of the environment, such as indices or factors . The former offers insight for planners and designers seeking to create specific built environment policies, such as residential densities. But these may not adequately capture the ways that individuals perceive and react to their environments. Not only are there limitations in how we represent the various causal roles of the built environment, our representations often do not factor how individuals and social groups nonhomogenously perceive and respond to built environments. The aforementioned notion of 'place perspectives' is relevant for the conceptual cartography in section 'Representing the causal relevance of the built environment'. Individuals and social groups shape and are shaped by the built environment. There are various concepts worth exploring about how built environments regulate complex processes. For instance, built environment 'nudge'-based conceptual frameworks, relevant to public health, have been suggested . But even with the popularity of the nudge as an intervention tool during the pandemic, there are more adequate concepts to explore. Representing the causal relevance of the built environment How does the built environment, centrally, figure into the complex relations between biotic, social, and health processes? In this section, we discuss at least three conceptual developments that can be applied to the built environment: 'nudge', 'affordance', and 'interface'. The purpose of these concepts is to organize the multiple regulating functions of built environment contexts . We apply the concepts in order to show how BC's can be cohesively analyzed alongside social and biotic variables. This will set up our larger conceptual perspective-the built environment as an interface that regulates multi-process pathways. We conclude with the view that to adequately represent the various moderating roles of the built environment, we must acknowledge who ought to be involved in the process of representation. Relationality and the built environment: built contexts and nudges. Built environments can 'nudge' human health choices and outcomes , such that cues in the environment can alter specific behaviors in predictable ways without forbidding options . For instance, Birnbach et al. found that in-line-ofvision hand sanitizer dispenser placement resulted in 53.8% of physicians sanitizing their hands, compared to doorway-adjacent dispenser placement, which resulted in 11.5% of physicians sanitizing their hands . Hygiene-related behaviors may be more susceptible to built environment nudges-a possible explanation being that COVID-19-related hand-washing behavior directly depends on personal perception of risk . Additionally, perceived-risk environmental cues can easily be manipulated into value-negative nudges to influence behaviors like stockpiling . Because of the effectiveness and causal linearity of certain behavioral modifications, nudges are surprisingly effective in explicit forms. Hygienerelated behavioral changes can be nudged by creating explicit informational interventions about COVID-19 . We suggest that the 'nudge' provides an open conceptual infrastructure to represent public health behavioral modification; but we should be wary of the adequacy of the 'nudge' in accounting for the causal complexity of social and built environment contexts. The first issue with the nudging conceptual framework pertains to social contexts. Nudged behavioral changes can be both mediated and moderated by other contextual social variables. For instance, extended social distancing history can reverse the positive effects of the aforementioned informational intervention nudges . Even seemingly simple associations between risk perception and behavior are subject to complex relations: cultural context can modify certain types of risk perception, producing differences in risk-related behavior . Representational frameworks that focus on nudges are problematic because, behind the appearance of causal linearity, complex social norms and contexts drive nudge effectiveness . For example, how can we adequately represent non-homogenous contexts, where social movement is successfully nudged for some and fully inhibited for others? During the pandemic, as a result of inconsistencies in policy decisions to announce mask mandates and the atmosphere of xenophobia, a contradictory context emerged for Asian Americans: There was discrimination and violence as a function of both choice pathways-wearing a mask vs. not wearing a mask . That is, Asian Americans neither had the option of wearing a mask in public nor the option of not wearing a mask in public . In such a context of absent options, any mask-related nudging policy will fully discourage movement through built environments for some, while smoothly encouraging that behavior for others. A given nudge can be incorporated into an artifact-whether technological, built environment, or a simple object-to promote unconscious priming . This compliments Olszewska and Konecka's argument that there is no such thing as a neutral built environment nudge. This is quite interesting because the implication is that built contexts, intrinsically, possess nudge capacity. We agree with the general suggestion that the built environment should be evaluated for non-neutrality in unconscious priming; but we disagree that the 'nudge' is an adequate concept to explain the various forms of causal relevance at the intersections between built environments and human behavioral changes. Built environment nudges can be more complex than in-sight hand sanitizer and 6-foot-distancing markers. Additionally, we argue that built environment nudges are not intrinsic but, rather, relational. That is, context modifies the causality of a given built environment nudge. For instance, priming effects differ, drastically, depending on social context, as well as 'self-construal'-the implicit and explicit representation of relations between self and others . 'Self-construal' is fascinating because it depends on the representation of oneself relative to a given social norm; and such relative social location can regulate the effectiveness of nudge priming, delivered through environmental cues . So, because priming through environmental cues can be moderated by sense of self, self-construal, and social norms, we suggest that explicit representational attention should be paid to how contexts shape relations between built environment cues, social processes, and public health behaviors. Relevant to social contexts, a policy reliance on nudging has been used side-by-side with concepts like 'behavioral fatigue'almost to a comical degree-in order to both explain and intervene in a wide variety of COVID-related health behaviors . However, we warn that basing health policy decisions on the representational linearity of the 'nudge' glosses over the complexity of public health behavioral modifications. Namely, complex interactions produce emergent behavioral effects that might dilute or scatter behavioral interventions. For example, how do we explain and intervene in pandemic lockdown non-compliance? In some instances, policy relied on simple nudges to minimize 'behavioral fatigue'-with little success due to a misrepresentation of causes . When contextual variables were represented, a more adequate causal explanation emerged: reactance in combination with optimism bias can produce interaction effects that cause non-compliance . Social context becomes even more complicated and non-homogenous when we consider that belief systems, associated with 'reactant' attitudes, can be correlated with geographical regions, in addition to increases in COVID-19 transmission . As discussed in section 'Developing a relation-based representational precedent', relationships between belief systems and environments are relevant to transmission rates. But the further implication for nudging is that mitigating transmission rates depends on successful nudge planning that can address differential social contexts. We caution that misrepresenting public health changes as a function of implicit nudges may produce ineffective public health interventions that gloss over contextual complexities. Recall that we began this discussion with a general problem: To what extent do current scientific practices fail to robustly account for the complex intersections between built, biotic, social, and health processes? But it isn't just 'representation' that is the issue, it is also 'intervention'. To add onto the list of three representational problems from section 'Diagnosing representational framework inadequacy: epistemological, ontological, and methodological problems'-Measurement Outcome, Representational Framework, and Surrogate-there is a further problem, due to causal misrepresentation in nudge frameworks: Idealized intervention problem. The use of an inadequate causal representation that oversimplifies or homogenizes changing interactions between social, built, biotic, and health processes, in addition to the resulting oversimplification of emergent properties from such interactions, can result in inadequate public health interventions. To summarize, we make the specification that 'nudging' provides an incomplete representational framework for built environment and social contexts by idealizing causal connections. Nudges rely on homogenized behavioral modifications through heuristic, linear, desire-based choice interventions, often missing the contexts that serve as substrates to regulate the effectiveness of the nudge. We believe that the regulating functions of the built environment constitute deeper interactions than a simple opt-in vs. opt-out. The built environment organizes causal factors that can serve as mediators and moderators of larger processes. We address this in the next section. Relationality and the built environment: contexts and affordances. In section 'Diagnosing representational framework inadequacy: epistemological, ontological, and methodological problems', we presented a number of unacknowledged processintersections that require a new representational infrastructure. The purpose of an adequate representational infrastructure is to broaden ontology, epistemology, and methodology in COVID-19 investigation. 'Nudging' infrastructure-with a focus on linear causal relations between BC's, behavioral changes, and health outcomes-misses the ontological contextuality of representing how behavioral changes operate within a given built environment. There are at least three interrelated concepts, relevant to a more adequate representational infrastructure: 'context', 'affordance', and 'interface'. We detail the first two in this section, before visualizing the third in the next. In section 'Developing a relation-based representational precedent', we introduce that a given 'place' is ontologically complex, consisting of a series of contexts that organize physical and social interactions : contexts can determine social movement, rate of interactions, transmission of information, and value system changes . This is relevant for improving the conceptual backbones behind models. For instance, earlier we described the Arthur et al. model. It is worth exploring whether the parameters that constitute 'optimal contact rate'-i.e., the utility function, parameterized by perceived risks and benefits of social contacts-can take differential form, depending on the interactions that constitute a given context. Recall, for example, that cultural context can modify certain types of risk perception and risk-related behavior . To be explanatorily and predictively useful, the Arthur et al. model may require modification to account for how parameters change as a function of built environment and social contexts. The representational importance of 'place perspective' is that, generally, it highlights emergent causality; but it is important to be precise about the causal role constituted by contexts. If we take seriously Olszewska and Konecka's claim that built environment contexts cannot be causally neutral, then we can begin to explore more precise BC causal propensities. But to understand how such propensities can reside 'in' the built environment requires a sophisticated conceptual infrastructure: 'affordance'. Affordances are constituted by relational properties that emerge from the interactions between individuals/groups and objects, processes, or environments-such that the relationality produces 'capacities for action'-including, promotion and inhibition of behavioral pathways . A key benefit of the affordance framework is that it can be used to represent how public health behaviors are not homogenously promoted within built environment contexts. That is, environment-to-person interactions within a given place can produce different sets of affordances for both individuals and social groups. It is important to note that a given affordance is not merely a 'perception' or 'subjective experience' . Rather, we take affordances to constitute causal pathways that can operate differentially within a given place. This type of relationality is ignored in the 'nudge' conceptual framework and public health policy that uses the nudge framework. For instance, often unacknowledged at the policy level but faced by residents, environmental racism drastically alters the capacity for "healthy movement" nudges in contexts of inequitable public park regulation and polluted environments . In other words, the nudge for health-related movement becomes impossible within the context of environmental racism. Regional studies track multiple pollution zoning parameterse.g., PM, SOX, VOC-to document how permitted polluting sites disproportionately affect people of color, creating new forms of neighborhood redlining . Other studies yield deeper causal analysis pertaining to pollution-namely, that general car emission pollution decreases in COVID-19 have concealed industrial pollution increases that disproportionately affect long-term PM 2.5 exposure in Black communities . Novel concepts emerge to characterize pollution disparities as a function of structural racism-e.g., 'pollution burden' . Such concepts can be embedded within the affordance framework. Built environment burden is not just disproportionate exposure to pollutants within a given built environment context. Rather, built environment burden can constitute numerous BC affordances. For example, Louisiana's industrialized regions have been shown to hold significant associations between long-term PM 2.5 exposure, social stressors, and COVID-19 morbidity and mortality in Black residents . Here, the industrialized region BC can be represented as regulating social and public health burdens. To make the causal language precise, a given BC can moderate health processes at many scales-such as, BC moderation of environmental pollutants, as well as other social determinants of health that promote multiple pro-inflammatory mechanisms, e.g., via IL-8 Signaling, NF-κB Signaling, and Dendritic Cell Maturation . Notice here that we are using 'affordance' broader than the original actionbased framework. 'Affordances' can be conceptualized as public health capacities occurring at many scales within a given context. Some of those capacities are behavioral or action-based, and they can also be entangled with capacities at other scales. A given BC can also inhibit numerous beneficial public health capacities by preventing and discouraging access to beneficial resources. We often fail to acknowledge the sharp lines of disproportionate built environment obstructions in natural environments-such that, in addition to pollution pathways being amplified, biotic system pathways are muffled. For instance, in pandemic contexts our representational focus has been fine-tuned to transmission in BC's to such an extent that we often fail to explore the representation of resilience-promotion by the built environment. Quite obviously, it has been extensively researched that most residential and commercial HVAC systems are poorly equipped for virus particle filtration. 3 However, further representing built environment promotion and inhibition becomes critical in dense cities where a combination of factors may force residents to keep their windows shut, relying almost solely on central heating and cooling systems. Additionally, where affordances for social movement are limited or absent, understanding the regulatory roles of the built environment becomes essential. We suggest that in pandemic contexts, health-promoting airflow representation has taken on a new relevance. Perplexingly, indoor airflow processes are rarely considered in policy-planning relevant to the built environment and public health . Furthermore, various considerations in airflow are rarely amalgamated in public policy-related built environment representations and interventions. However, general intervention in healthpromoting mechanical airflow systems requires detailed representation of how the built environment regulates multifarious processes -e.g., energy, oxygen, and harmful and beneficial microbes. 4 For instance, central HVAC systems, even at ideal function, can create unsuitable health environments because they eliminate diversity of bacteria and create human disease-centered colonies . 5 Through the elimination of bacteria diversity, there is the concern that central HVAC systems create harmful microbe pumps, but we suggest that there is a deeper concern about inhibiting beneficial bacteria that may provide protective health against lung-related diseases-such as, bacteria from the Firmicutes genera . 6 HVAC system design that maximizes bacteria diversity, thereby promoting lung resiliency, has so far been fully unexplored in COVID-19 investigation; and thus, this is currently an unacknowledged causal nexus. This is why broader representational approaches are needed: Health-negating factors can be represented along with resilience-promoting factors -e.g., naturally occurring bacteria profiles that minimize lung inflammation. 7 Even in something as seemingly obvious as built environment airflow moderation, expanding the representational scope over processes beyond transmission could prove fruitful. Additionally, consistent with the discussion in section 'Developing a relation-based representational precedent', we suggest that amalgamating multi-process considerations is closely tied to developing new interdisciplinary and transdisciplinary practices-a point that we further illustrate in the next section. That is, the built environment regulates numerous processes; and only when we synthesize disciplinary lenses, do we begin to uncover more process-intersections. By taking wider microbiology, systems biology, architecture and design, logistics, and economic modeling approaches to airflow processes, we can reveal unaccounted-for relations that are relevant to population health. Interdisciplinary opportunities to take a new perspective on airflow are present but seldom integrated into causal analysis of the built environment. For instance, recent fields like 'pyroaerobiology' have emerged to investigate wildfire and airflow mediation of microbes . Our suggestion is that future investigation of built environment regulatory roles could benefit from disciplinary interactions with seemingly peripheral fields like pyroaerobiology-e.g., in order to study PM mediation and moderation of microbes between ecological and built environments. To summarize, the affordance framework can be used to represent multiple causal pathways that can operate differentially within a given place-e.g., in the form of dual representations of amplified 'built environment burden' and inhibition of beneficial biotic access. We propose that the 'affordance' conceptual framework offers the starting steps for the representation of built contexts by creating a structure that focuses on interactions. The next step is to summarize the regulating functions of the built environment in a cohesive visual representation. Visual models, heuristics, and metaphors can restructure ontology, while also presenting new epistemological and methodological avenues. Relationality and the built environment: visualizing interactions and interfaces. We return to the issues presented in section 'Diagnosing representational framework inadequacy: epistemological, ontological, and methodological problems' in order to make final suggestions about how representations can be constructed and who ought to have a role in shaping representations. In this section, by specifying visual metaphors and diagrammatic methods, we aim to create useful conceptual pathways for the exploration of ontological, epistemological, and methodological frameworks relevant to studying the built environment. So far, all of the visual representations in our discussion contain epistemological and ontological infrastructure limitations. For instance, Fig. 1 is limited because it idealizes points of intersection between processes. The multi-process interactions discussed in section 'Relationality and the built environment: contexts and affordances' are numerous, continuous, and operate on multiple scales. Also limited, Fig. 2 has the potential to embed multifactorial moderation and mediation; but the representational format risks simplifying built environment processes to single factors. Furthermore, there is no representational capacity to show how variable relations change as a function of contexts. The diagrammatic styles in Figs. 1 and 2 have other general limitations. They present little possibility for quantification, as well as representing multiple dimensions. So, perhaps there is a more adequate type of visualization that can capture important features of built, biotic, social, and health interactions. Our focus on 'interactions' makes the affordance conceptual framework amenable to a threedimensional visual representation that can embed relationships between multiple variables. We suggest a number of conceptual avenues to representing process-interactions in 3D. In the philosophy of biology, hypersurface representations have received considerable attention , traceable to Wright's and Waddington's iconic 'landscape'-style representations. For Waddington, the threedimensional landscape emerged out of an attempt to conceptualize an explanatory framework that could ontologically house numerous organizational structures and functions of biological development. Our purpose for discussing visual representation is similar to his: adequate visual representations can restructure ontology and facilitate new methodological routes. For Waddington , a biological system's developmental pathways and states are akin to a marble rolling through a threedimensional manifold. A marble moves from flat undifferentiated states to differentiated depressions, which represent alternative developmental states. The tension of the manifold slopes is determined by many interconnected gene products, or in a modern perspective-'gene regulatory networks'-that are attached to peg-like genes. We believe that Waddington's landscape metaphor is particularly fascinating because it can capture emergent properties-such as, 'restriction' and 'differentiation'-which are greater than the sum of their components. Similarly, because of the complexity of affordances, emergent properties relevant to differential built environment burden and inequitable health outcomes can begin to take three-dimensional representational form. That is, the built environment can be represented as more than just the value of a single variable. Recall that built environment representational choices often amount to just a single discrete variable or handful of variables-with limited focus on interaction and context . Here, ontology is limited by representational framework choice. The built environment can be represented as a set of causal factors, but we suggest that it can also be represented as an interface that reorganizes causal relations-concurrently, deregulating factors, while leaving others unaffected. A change in interface topography can modify multiple factors. Visual metaphors, like Waddington's landscape, can be scientifically useful for the development of more technical forms of representation. As a starting step for our more complex visualizations, we represent the built environment as an interface that regulates multiple variables , while being shaped by other variables . Imagine this interface as a catalyzing membrane whose causal topography can be co-modified based on its interactions . In the hand-crafted illustration in Fig. 3, we visualize the built environment behaving dynamically, like fascia within the architecture of physiology. Notably, in the interdisciplinary research on fascia, there is a shift away "…from a body made up of parts to the wholeness of the architecture holding them together" . Additionally, fascia has mediating and moderating roles, while also being shaped by other factors . As Dumit and O'Connor state, "Fascia can be described as a biomatrix that surrounds everything in our bodies, connects everything, and yet paradoxically cleaves and separates everything" . It is also structurally responsive to changing interactions . Finally, fascia maintains state flexibility and produces emergent properties that are greater than the sum of their parts . Similarly, we illustrate the built environment as a co-modifiable matrix with moderating functions . Why go through all of the trouble of creating a metaphor for the built environment? We think that such metaphors can direct scientific attention to the multifaceted roles of the built environment. The visualized interface-that reorganizes, connects, promotes, erodes, and inhibits pathways-spotlights key causal roles of built environments. As an exercise in representational versatility, there are options to explore the metaphor further-e.g., by spotlighting landscape connections. For instance, in Fig. 4 we take the reductionistic approach of Fig. 2 in order to idealize the built environment interface as individual built environment factors and their causal relevance to two multivariable landscapes. To reiterate, visualizations, like Waddington's landscape or the built environment interface , can be useful for the illustration of general causal roles, relationships, and properties, which can be further explored in more technical forms of representation. Choice in scientific representation requires emphasizing certain factors while idealizing others . Often, our scientific focal point can be shaped by practical considerations. The physiology of fascia had received little attention and had not been represented in textbook diagrams because disease in fascia had not been a key medical concern-simply put, "…you don't die from fascia" . As a parallel, we think that the built environment's causal import on other processes has received little attention, but not because "you don't die from the built environment". Rather, it is because the various moderating roles of the built environment on public health factors have been causally opaque or oversimplified for practical purposes. The importance of Waddington's version and new versions of 'landscape' diagrams, as organizational tools, is in the representation of relationships between multiple variables . For instance, as depicted in Fig. 5, a large number of dimensions can be represented in a smaller hypersurface, where a set of independent variables determines dimension n . Generally, this type of landscape is a mathematical function that associates values of a set of independent variables with the value of a dependent variable over a Euclidean space . Landscapes can also be translated into dynamical systems-although, with limitations as well as asymmetry in moving from dynamical systems to landscapes . For our use, the general representational purpose of a landscape is to depict the relations between multiple variables. A description of the time evolution can be supplementary. The built environment can be represented as an independent variable with other independent variables representing biotic or social factors, and the dependent variable representing a given health outcome . However, we also suggest that the built environment can be turned into its own landscape-for instance, by representing the variables involved in a particular type of built environment burden. We propose that this adaptable landscape structure allows for two useful diagrammatic methods in the representation of the built environment. First, this landscape structure promotes a flexible transition between multidimensional diagrams and surrogacy wiring diagrams to suit shifting explanatory demands. This provides versatility in representing complex relationships while being able to selectively depict certain features of the built environment for pragmatic purposes. Depending on the explanatory need, built environment burden can be represented as a multivariable landscape, series of landscapes, or it can be reduced to a single node within a wiring diagram. But importantly, we suggest beginning with complexity, rather than beginning with idealization, in order to avoid the surrogate measure problem described in section 'Diagnosing representational framework inadequacy: epistemological, ontological, and methodological problems'. After all, part of the practical challenge is how to represent multifarious processes-built, social, biotic, and health -cohesively, but also, flexibly. Second, with the use of this diagrammatic structure the built environment can be analyzed together with other multivariable landscapes in order to explore landscape connections. In robust data contexts, this can begin with the analysis of associations between the surfaces of multiple planes. In contexts with limited interdisciplinary data, this can prompt further empirical investigation of causal relationships between landscape variables. For instance, co-modification between social and built landscapes requires detailed analysis and empirical investigation. Social Fig. 3 Built environment as interface. Metaphors can restructure ontology, while also presenting new epistemological and methodological avenues. As a starting step for our more complex visualizations , we present a hand-drawn representation of the built environment, behaving as an interface that regulates the interconnections between two multivariable landscapes, while being shaped by each landscape. Here, each set of causally relevant variables is represented as its own landscape . The built environment interface is represented as a catalyzing membrane whose causal topography can be co-modified based on its interactions. The interface reorganizes, connects, promotes, erodes, and inhibits pathways. This general conceptual precedent opens avenues for the analysis of specific concepts relevant for understanding the moderating and mediating functions of the built environment-e.g., 'nudge', 'affordance', 'interface', and 'context'. landscapes can produce stark built environment disparities within a given place. Additionally, built and social landscapes can generate mutually reinforcing feedback loops. There are complicated relationships to explore-e.g., how social processes that create health disparities are further reinforced by inequitable built landscapes. Representing the accurate relations is paramount for understanding how built and social landscapes co-modify each other. Social landscapes can erode built landscapes over time, but they can also create quick, dramatic shifts in built landscapes-or both. For example, recent pseudo-environmental practices have resulted in pop-up toxic landfill sites, affecting Black and Latinx populations, while nearby White populations remain fully unburdened by permitted polluting sites . Importantly, such built environment sites are a continuation of a history of racist policies and built contexts. Something that is beyond the scope of this paper but could be an extension of our discussion is the application of multivariable landscape analysis for the purpose of clarifying the causal relationships in unfolding case studies. Within a general two-tier landscape diagram that shows the relationships between built and social landscapes, new iterations of systemic racist practices can be represented in detail by tracking how sudden multivariable spikes can deregulate a given health outcome. Newly formed toxic sites like 'Shingle Mountain', which is a six-story toxic site composed of particulate matter that took just six months to emerge, have created a compounded health hazard context . The topography of such contexts requires fine-grained multitiered causal analysis, otherwise we may risk an uninformative coarse-grained representation of average pollution levels within a homogenized built context-hiding patterns, similar to the aforementioned case study of industrial pollution spikes in a context of decreasing average pollution measures . It might seem obvious that multivariable interactions are important to analyze, just like it might seem obvious that built environment burdens, e.g., Shingle Mountain, receive adequate investigation into social and built process moderation. But this is false. Shingle Mountain emerged within six months, but it took years to be investigated and acknowledged by policymakers-only resulting as a function of unrelenting community efforts in data and advocacy . Illustrated by the Shingle Mountain example, community efforts can reveal unacknowledged interactions between built, social, and biotic processes. As we have discussed in section 'Diagnosing representational framework inadequacy: epistemological, ontological, and methodological problems', new research could benefit from a more robust conceptual framework that can shift ontological and methodological practices. We are advocating for greater causal resolution in how we represent complex relations between built and social landscapes. But there is a related issue about who represents complex relations. Relationality and the built environment: community participation in measurement practice. This brings us to an important point about accuracy and equitability in scientific practice: Indexical measurement questions -e.g., who measures, how they measure, and what they measure -are essential for the development of more adequate measurement/data practices. For instance, who determines what counts as a 'relevant' built environment context? We conclude with a suggestion for specific methodological needs, one of which is to develop new measurement methods and collaborations between scientists and communities. We suggest that one method to avoid the conceptual and methodological pitfalls mentioned in the first part of this discussion is to clarify transdisciplinary terminology sideby-side with direct collaborative goals. Rashid et al. Here, the built environment interface is reduced to built environment factors and their causal relevance to two multivariable landscapes. Note: We idealize 'opaque' vs. 'transparent' factors because our illustrative focus is on landscape connections. However, more detailed characterizations can specify continuums of opacity and transparency-detailing nuances in identification, uncertainty, and weak vs. intrinsic/essential limitations. continuation of transdisciplinary health research through crossagency collaboration and public partnerships in order to investigate the complexities of health disparities and to disseminate solutions. Unfortunately, even when epistemically effective and fruitful, such collaborative structures between researchers and the public have been cut . The relevant question here is, who has a say in diagnosing and shaping health environments? Community measurement methods can emerge from direct need. For instance, because of a relative lack of air quality monitoring infrastructure in polluted communities, Gabrys et al. proposed a new paradigm for iterative data-production and analysis. In this case, public participation in measurement practice is consistent with reflexive approaches to engaged community measurement practices for the pursuit of environmental justice . The Gabrys et al. paradigm of measurement can be used to re-envision the role of community engagement in scientific practice. The same inventive approach, arising out of a need, can be applied to COVID-19 data-generation. New research programs could synthesize epidemiological data on COVID-19 outcomes and social determinants of health with qualitative data, based on firsthand accounts about social-systemic and built environment factors. We think that it is imperative to mention that the need for and value of qualitative data is a developing topic in public health . If community members can autonomously report qualitative data to scientists and policymakers, this not only builds a more robust data picture, but it can also promote social autonomy. Community narratives can be used to inform public health modeling and policy-making. Such approaches have been attempted by using community qualitative data to generate affordance maps . We can structure and summarize the general method of public scientific measurement practice as follows. Data-gathering and modeling can extend into the sociotechnical collaborations between community members, scientists, and policymakers . Through access to technology, community members can generate data, which can subsequently be analyzed and visualized into various data models, referred to by Gabrys et al. as "storying processes". On our account, these storying processes would focus on finding new relevant variables for built environment and social landscapes. For instance, Shingle Mountain causal variables and their subsequent health impacts were wholly reported by community members. Before this, representations of the built environment within the region were homogenized, ignoring the various causal roles of environmental racism. We think that the storying process can contribute to scientific practices-such as, the co-development of detailed ontological frameworks that address emergent properties like built environment burden, or the creation of new data methods for the purpose of generating values for key variables. We conclude with a suggestion that the coordinated activity between multiple measurement perspectives can be expanded to include independent perspectives, generated by communities. These perspectives can be used for changes in intervention approaches , as well as theory-building practices . This type of iterative measurement process requires coordination between independent perspectives, including-scientists, communities, technological systems, and policymakers. The built environment can be represented as an independent variable with other independent variables representing biotic or social factors, and the dependent variable representing a given health factor . But the built environment can also be turned into its own landscape, represented with multiple independent variables, such as the variables involved in 'built environment burden'. As we discuss, landscape connections can be explored through associations between the surfaces of multiple planes, and they can also prompt further empirical investigation of causal relationships between landscape variables. Two things are important to note here. First, there is a technological challenge. Community engagement in the measurement process requires technological platforms for generation, organization, and dissemination of data. A given community can generate diverse data-through the creation of firsthand accounts, photographs, visuals, parameterized data sets, etc.; but equitable technological tools have to be provided as organizational platforms for the data. From a design standpoint, a possible start is to investigate the structure and equitability of technological platforms that provide real-time community-submitted data. The second thing worth noting about community participation in measurement practice is that data generation and storying methods indicate "responsibilities" that emerge from interacting with data . We make a more general point: Measurement practice choices are deeply entangled with normative considerations. That is, re-envisioning the coordination between scientists, communities, and technological measurement tools will require understanding new responsibilities in how we interact as part of the measurement process, as well as what kinds of measurements matter to what parties. --- Concluding remarks We have presented multiple problems in COVID-19 scientific practice that can be reduced to the following issue: to what extent do current measurement and data practices fail to robustly account for the complex intersections between built, biotic, social, and health processes '? To confront this general issue, we have presented a representational precedent that focuses on 'relations' . Additionally, we have analyzed at least three conceptual developments that can be applied to the built environment: 'nudge', 'affordance,' and 'interface' . The purpose of these concepts is to organize the multiple regulating functions of built environment contexts . We have applied these concepts in order to show how BC's can be cohesively analyzed alongside social, biotic, and health landscapes. Our main point is that the built environment can be represented not only as a single discrete variable but also as an interface that reorganizes causal relations-concurrently, deregulating factors, while leaving others unaffected. By clarifying the unacknowledged moderating roles of the built environment, we hope to inspire new conceptual and practical developments that broaden ontology, epistemology, and methodology in COVID-19 and beyond. Received: 2 August 2020; Accepted: 28 April 2023; Notes 1 High sensitivity conditions refer to minimizing false negatives, such that a given measure often achieves an expected value relative to a proximate clinical endpoint. High specificity refers to minimizing false positives, such that a given measure rarely achieves an expected value relative to absent clinical endpoint. 2 Whitmee et al. explicate the complexity and quality of health, social, ecological relations. They describe degrading health outcomes as a function of "inequitable, inefficient, and unstustainable patterns" in resources, environment, and technology . 3 Most residential and commercial HVAC systems require a minimum MERV of 8, which is measured to capture 70-85% of particles, with a range of 3.0-10.0 μm. But in aerosolized particle form, SARS-CoV-2 has been measured between 0.25 to 0.5 μm . 4 Broader representational frameworks can even be useful when applied to the most seemingly obvious airflow considerations. Transmission and energy are rarely synthesized within a single representational framework, directed toward public-policy interventions. Virus spread can be minimized by increased outdoor air fractions; however, this has the tradeoff of increasing energy supply and maintenance cost-the latter of which is often the focus in public and residential built contexts . But if we consider transmission and energy efficiency within a multi-process representation, an intervention dilemma emerges: whether to maximize air fraction and virus particle filtration at the expense of leaking energy and maintenance supplyi.e., transmission-minimizing interventions are pinned against efficiency-promoting interventions. There are numerous practical airflow dilemmas if we consider multiprocess pathways. For example, elevated humidity decreases virus spread; but the standard HVAC design prevents humidity moderation: Most HVAC systems do not have indoor humidification due to maintenance difficulty and risk of overhumidification, which causes mold growth. The dilemma is whether to include humidification in HVAC technological development, thus minimizing virus particle viability, but potentially creating maintenance breakdowns and mold hazards. Built environment dilemmas can be uncovered if we broaden representational scope in order to reveal process-intersections. 5 As National Academies of Sciences, Engineering, and Medicine thoroughly outlines, U.S. single-family residential HVAC systems "rarely incorporate outdoor air intake but instead recirculate interior air primarily for temperature control, typically with low-efficiency particle filtration". Often there is an increase in microbe distribution, e.g., through humidification systems and mechanical ventilation, rooftop HVAC components near standing water, and high-pressure differentials that influence interior microbial pathways via "stack effect" . 6 In a causal manipulation on mice, once mice were inoculated with these four bacteria, airway passage inflammation decreased . 7 Further research questions about unacknowledged process-intersections arise: How does pollution due to inequitable built environments interact with bacteria in airflow to produce non-resilience in overall physiological health? Research programs can investigate the intersections between built environment, microbiomes, pollution, and social processes. For example, Alderete et al. , show that there is a causal connection between air pollution exposure in adolescents and microbiome changes. Specifically, there is a connection between air pollution and increases in Bacteroidaceae and Corynebacteriaceae-both have been associated with intestinal inflammation, obesity, insulin resistance, and altered metabolism . --- Competing interests The authors declare no competing interests.

By broadening disciplinary perspectives to architecture and design, philosophy of science, and systems biology, this paper aims to explore the interconnections between built, social, biotic, and health processes with key attention to the moderating roles of the built environment. The focus is part diagnostic and part prescriptive. Initially, we specify failures in COVID-19 representational infrastructure and practice in accounting for built environment and social process impacts on public health factors. By presenting three intertwined problems with scientific representation in COVID-19 modeling and data-gathering, we examine to what extent current scientific practices fail to robustly account for the complex intersections between built, biotic, social, and health processes. We suggest that resolving the presented problems requires the development of new conceptual precedents for the analysis of causal relations in changing contexts. The second focal point is prescriptive. By discussing conceptual developments that spotlight relations-e.g., 'context', 'nudge', 'affordance', and 'interface'we organize the numerous moderating roles of built environment contexts, and we suggest practical applications to ongoing public health practices-such as, cautioning against nudge policies. Ultimately, we argue that the built environment can be represented not only as a single variable (or handful of discrete variables) but also as an interface that reorganizes multiple causal landscapes-concurrently, deregulating factors and leaving others unaffected. Because of the difficulty of representing emergent properties, relevant to differential built environment burden and inequitable health outcomes, we provide ways to visualize the built environment as interface in multidimensional form. We conclude that adequately representing the various moderating roles of the built environment goes a step beyond how to represent complexity, and it requires asking a deeper normative question: who ought to be involved in representing complexity.

Introduction Lesbian, gay and bisexual individuals are at an increased risk for mental health issues as compared to heterosexual individuals [1][2][3][4][5][6][7]. For example, mood, anxiety, and substance use disorders are at least 1.5 times more likely to occur in LGB individuals than in heterosexual individuals [2,3,6,8]. Among LGB individuals, gender differences in psychiatric morbidity have also been documented: compared to heterosexual participants, gay and bisexual men have been found to experience a higher risk for mood disorders [4,5], while lesbian and bisexual women appear to be more adversely affected by substance-related issues [2,3,5]. Overall, the risk for LGB individuals to be diagnosed with at least one disorder in the last 12 months appears to be twice as high as compared to heterosexual individuals [5]. --- In addition, among sexual minorities, bisexual individuals The members of WHO World Mental Health Survey collaborators are listed in acknowledgements. have been found to have especially high risks for experiencing adverse mental health outcomes [9]. Notably, these mental health disparities between LGB and heterosexual individuals have been well documented by national studies [4,5,7,8,[10][11][12], which have predominantly focused on Australian [10,11], European [4,5,7], and North-American [8,11] contexts. However, differences in socio-structural factors, such as the social acceptance of LGB individuals, have been found to be related to countrylevel effects on health [13,14] and happiness [15] among sexual minority individuals. Similarly, there is evidence of cross-national [16,17], and within-US state-level effects on psychiatric morbidity among sexual minority individuals [18,19]. However, because of cross-national differences in the prevalence of mental disorders generally, it is unclear if cross-national differences in mental health outcomes among LGB individuals are actually specific to LGB individuals or simply reflect more general differences in the prevalence of mental disorders across countries. Some cross-national evidence for disparities in mental health outcomes between LGB and heterosexual individuals also comes from meta-analyses [20,21]. However, studies assessing mental health disparities between LGB and heterosexual individuals across a broad range of mental disorders in a cross-national sample, including Western and non-Western countries, are lacking [21]. --- Minority stress and social support as a mediator In addition to societal stressors related to their stigmatized sexual orientation, there is consistent evidence linking the elevated rates of psychiatric morbidity among LGB individuals to the experience of individual-level stressors [22][23][24][25][26]. In addition to having a cumulative effect on psychiatric morbidity [23,27], it has been suggested that minority stress may result in elevated emotion dysregulation, interpersonal issues, and cognitive processes that ultimately increase LGB people's risk for psychopathology [27]. According to Hatzenbuehler's psychological mediation framework, sexual minority individuals may experience poorer social relationships as a consequence of social rejection and isolation due to their sexual minority status [28], which in return might reduce their resources to cope with general life stressors. Indeed, compared to heterosexual individuals, sexual minorities have been found to report lower perceived social support in general [11,29] and to report less social support from family as compared to friends [11]. In line with the psychological mediation framework, higher social support has been found to result in better coping and resilience among sexual minority individuals [29,30]. Specifically, support from family as compared to friends appears to be an important predictor of the mental health of LGB individuals [31]. Conversely, there is some evidence that lower social support may mediate the adverse effects of minority stress on health generally [32], and mental health specifically [33,34]. In addition, there are some findings that suggest that social support may mediate the relationship between sexual minority status and psychiatric morbidity among young men, but not women [28]. --- Current study The primary aim of the current study was to contribute to this literature by examining global mental health disparities between sexual minority and heterosexual individuals. The WHO World Mental Health Survey Initiative consists of a cross-national dataset that includes a comprehensive set of mental disorders. As such, it offers a unique opportunity to study disparities in mental health outcomes based on sexual orientation in the largest cross-national general population sample investigating the most comprehensive set of disorders to date. Based on previous national population studies, we expected LGB individuals to be more likely to report adverse mental health outcomes across multiple disorder groups as compared to heterosexual individuals. In addition, we explored the association between country-level social acceptance and the increased risk for mental disorders among LGB individuals compared to heterosexual individuals. The second aim of this study was to investigate the role of social support as a potential mediating factor for mental health issues among LGB individuals. --- Methods --- Sample Data came from the WHO World Mental Health Surveys [35]. Translated versions of the WHO Composite International Diagnostic Interview version 3.0 [36] were administered in 29 surveys across the world through stratified multistage clustered area probability household sampling between 2001 and 2012 based on Census area data, with the exact recruitment and data collection procedures varying somewhat by country [37]. Adults from the non-institutionalized population were selected to generate population-representative samples . A total of 13 surveys assessing sexual orientation were included in the current study. All surveys within the World Mental Health Survey Initiative are commonly presented by country income group; surveys were grouped into a low/middle-income country group , Mexico, Peru, and Romania) and a high-income country group , United States). Country income categories were based on the World Bank criteria at the time of each survey [38]. To reduce participant burden, the CIDI was administered in two parts. All participants completed Part I, assessing core mental disorders. Part II, assessing other disorders and correlates, was administered to all participants with any lifetime Part I diagnosis and a probability subsample of other Part I participants. Part II respondents were weighted by the inverse of their probability of selection. Trained lay interviewers administered surveys face-to-face at all survey sites. Translation, back-translation, harmonization, and quality control procedures were similarly standardized at all participating sites [39]. In accordance with the respective Ethics Review boards, verbal or written informed consent was obtained. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. --- Measures --- Socio-demographics The variables assessed were gender, age, marital status , --- Sexual orientation Sexual orientation was assessed using a single item in the part II sample, with the exception of Argentina and New Zealand, where it was assessed in part I. Participants who identified as something other than the answer options listed , who were not sure or did not know, as well as those who refused to answer the question or had a missing value were excluded from the sample; females who indicated 'homosexual' will be referred to as 'lesbian'. In all countries except Argentina, Australia, and New Zealand, participants who reported that they had never had sexual intercourse were not presented with the sexual orientation question, unless they indicated that they had biological children. --- Mental disorders Analyses were conducted with 12-month CIDI DSM-IV diagnoses. We included the following mental disorders: mood disorders , anxiety disorders , behavioral disorders , intermittent explosive disorder [IED]), substance use disorders , and eating disorders . --- Social support quality Social support quality was separately assessed for family and friends using a two-item measure. Participants reported both the frequency of contact as well as their general openness to talk to either family or friends about their worries . Participants who answered "don't know" or refused to answer any of the questions, as well as participants with missing values were removed from the sample for the moderation analysis. Social support quality was assessed in a subset of eleven countries , Japan, Mexico, Northern Ireland, Peru, Romania, Spain , United States). Notably, the openness item assessed openness with relatives or friends in general, and not specifically related to sexual minority status as the aim of this paper was to assess differences between heterosexual and LGB individuals. --- Social acceptance Data on social acceptance were taken from the LGBT Global Acceptance Index [40] which summarizes societal acceptance based on surveys across 141 different countries from 1981 until 2014. For each of the 13 countries included in this study, the value that most closely corresponded to the year that the survey was conducted was selected. --- Analyses We first examined the prevalence of sexual orientation subgroup to provide readers with an overview of the specific characteristics of the sample as it related to the current study. To this end, we also tested how sociodemographic variables were associated with sexual orientation using linear regression for continuous variables and multinomial logistic regression for categorical variables. --- First aim We used logistic regression to test the association between sexual orientation and 12-month mental disorders separately by gender. The decision to conduct gender-stratified analyses was determined a priori based on a review of the literature, which has suggested important gender differences in the associations between sexual orientation and mental disorders [2,3,5]. In addition, we used logistic regression to test the association between sexual orientation and 12-month mental disorders comparing LG participants and bisexual participants to heterosexual participants, as well as bisexual participants to LG participants, while controlling for gender. The size of sexual orientation sub-groups did not allow us to conduct this comparison as a gender-stratified analysis. Although our main analyses were completed in the full sample of countries, we additionally performed logistic regressions to test the association between sexual orientation and having at least one 12-month mental disorder separately for each country. --- Second aim To assess possible mediation of the association between sexual orientation and psychiatric morbidity by social support, we employed a hierarchical regression approach [28]. First, we repeated a logistic regression analysis with sexual orientation as a predictor and mental disorder as an outcome in the subsample with information on all variables involved in the possible mediation . Second, we performed four ordinal logistic regressions with sexual orientation as a predictor and each of the four indicators of social support as an outcome . Third, we performed a logistic regression analysis with sexual orientation and all four indicators of social support as predictors and mental disorder as an outcome . All models were performed separately by gender. From these three models, we calculated the indirect effect of each social support indicator and the total indirect effect by summing the four indirect effects. Estimates of the statistical significance of indirect effects were estimated in 10,000 bootstrap samples using Rao, Wu, and Yue's bootstrap weighting method for complex surveys as implemented in SAS [41]. Due to concerns about model convergence problems in at least some of the bootstrap samples for uncommon disorders or disorder groups, we examined possible mediation only for any mental disorder and for > 1 mental disorder. All analyses controlled for the country of origin of the participant. Because the data were clustered and weighted, standard errors were estimated using the design-based Taylor series method implemented in SAS [42]. Statistical significance for all analyses was evaluated at an adjusted α of 0.005, to account for the large number of tests employed in the analysis. All analyses were conducted in SAS 9.4. --- Results --- Sample descriptives Of the initial 48,250 participants , 1361 were omitted from the sample, due to reporting their sexual orientation as "something else" , because they were not sure about their sexual orientation or did not provide a response , or because they indicated never having had intercourse . The full sample demographics can be found in Table 1. Out of 46,889 participants , 0.8% reported being gay/bisexual and 1.0% reported being lesbian/bisexual . More women identified as bisexual , while more men identified as exclusively same-sex attracted . LGB participants were younger and were less frequently married compared to heterosexual participants. There were no differences in employment status among LGB and heterosexual participants; however, GB but not LB participants reported higher educational and income levels than heterosexual participants. --- Sexual orientation and mental disorders by gender --- Women Compared to heterosexual women, LB women were significantly more likely to report at least one 12-month disorder , as well as more than one 12-month disorder . Among LB women, we found increased prevalence rates for all disorder groups . LB women were also more likely to report each specific disorder , except for specific phobia and ADD . --- Men Compared to heterosexual men, GB men were more likely to report at least one 12-month disorder or more than one 12-month disorder . Specifically, compared to heterosexual men, GB men were more likely to report mood disorders and anxiety disorders , but they were not significantly more likely to have increased prevalence for any other disorder group . Among GB men, we also found increased prevalence rates for most specific mood and anxiety disorders , except for dysthymia, agoraphobia, and GAD . We did not find elevated prevalence rates for any of the specific disruptive behavior disorders or substance use disorders , with the exception of drug dependence . --- Prevalence of mental disorders by sexual orientation Compared to heterosexual participants, both LG and bisexual participants were significantly more likely to report at least one 12-month disorder as well as more than one 12-month disorder . Specifically, both LG and bisexual participants were more likely to report mood disorders, anxiety disorders, and eating disorders than heterosexual participants . Bisexual, but not LG, participants additionally were more likely to report substance use disorders . There were no differences for 12-month disruptive behavior disorders. While bisexual participants, as compared to LG participants, showed somewhat elevated risks for reporting each of the disorder groups, as well as for reporting at least one or more than one disorder, these differences were not significant . --- Country-level social acceptance and mental health Figure 1 shows the risk for reporting at least one 12-month disorder among GB or LB participants as compared to heterosexual participants by country. Countries are ranked by their increasing country-level social acceptance scores, illustrating that the relative risk for LB and GB participants to report at least one 12-month disorder did not appear to be associated with country-level LGB social acceptance. --- Mediation by social support The full results of the mediation analysis in women and men are provided in Supplemental Tables 4567. The results for Model 1 were very similar to those discussed before and hence are not discussed here. --- Women With regard to Model-set 2 , we found that LB women were significantly less likely to report high levels of contact frequency and high general openness with their family than heterosexual women . However, there were no differences between LB and heterosexual women in frequency of contact and general openness with friends . In Model 3, including sexual orientation and the social support indicators as predictors, lower openness with family was significantly associated with having at least one or more than one 12-month disorder . Additionally, a lower frequency of contact with friends was significantly associated with having more than one 12-month disorder . For at least one 12-month mental disorder, there was a significant indirect effect of openness with family and a significant total indirect effect of all social support indicators . Results were similar for more than one 12-month disorder: there was a significant indirect effect of openness with family and a significant total indirect effect of all social support --- Men In model 2, GB men reported significantly higher general openness with friends than heterosexual men . However, there were no significant differences between GB men and heterosexual men in general openness with family , or frequency of contact with either family or friends . In Model 3, including sexual orientation and the social support indicators as predictors, lower openness with family was significantly associated with having at least one or more than one 12-month disorder . Additionally, lower frequency of contact with family and higher frequency of contact with friends were also significantly associated with having at least one mental disorder. However, there were no significant indirect effects, indicating no evidence for mediation. Fig. 1 Association between likelidhood of reporting at least one 12-month disorder and country-level social acceptance, by sexual orientation. The social acceptance score for Northern Ireland was omitted, because the index only included an estimate for Great Britain as a whole and Northern Ireland was not included in this estimate. There were zero cases with a 12-month mental disorder among the sample for GB participants in Northern Ireland and both GB and LB participants in Romania --- Discussion The findings from this large general population-based study provide cross-national evidence for the disparities in psychiatric morbidity between LGB and heterosexual individuals. Our results are consistent with those reported in national population-based studies [4,5,7,11] and cross-national meta-analyses [20]. We extended these findings by illustrating that this increased risk is a cross-national issue that is present across a range of disorders , and further provided additional evidence partially supporting the psychological mediation framework [27]. --- Cross-national disparities in psychiatric morbidity In line with our first aim, our findings add to previous studies documenting the higher likelihood for both LGB women and men to report mood and anxiety disorders compared to heterosexual participants [5,7,10]. Furthermore, we found that LB women but not GB men were more likely to report a 12-month substance use disorder compared to heterosexual participants, consistent with previous national studies [2,3]. The exception to this finding was drug dependence, which was more likely among both sexual minority women and men compared to heterosexual participants. We also found that LB women, but not GB men, showed clearly increased risks for disruptive behavior disorders. Overall, we found a heightened risk for LGB individuals of any gender to report at least one disorder in the past year [5]. In contrast to previous, smaller national studies that found either no difference [5] or elevated risks for comorbidity only among GB men [4]; we found that both GB men and LB women were more likely to report more than one 12-month disorder compared to heterosexual participants. Contrasting previous studies [9], we did not find significant differences in the risk for psychiatric morbidity between LG and bisexual participants, despite the fact that bisexual participants showed elevated rates of psychiatric morbidity compared to LG participants. One possible explanation for this might have been the small size of sexual orientation sub-groups. While only exploratory and limited by the relatively small number of surveys included, our exploratory analysis suggests no clear relationship between country-level social acceptance and risk for psychiatric morbidity. Instead, an individual's sexual minority status appears to be a stable risk factor for psychiatric morbidity across countries. This finding is counter to both our expectations and to the literature suggesting a relationship between state-level LGB climate and improved mental health [19,28]. However, increasing social acceptance of sexual minorities might result in LGBspecific discrimination becoming more subtle rather than disappearing, as Sandfort and colleagues have suggested [5]. To address this, future cross-national research on the relationship between sexual minority stress and psychiatric morbidity should assess both country-level acceptance as well as person-level experiences of discrimination. --- Influence of social support quality Regarding our second aim, the exploratory mediation analysis found that, among LB women, lower levels of social support partially accounted for the relationship between sexual minority status and the heightened risk for reporting at least one or more than one 12-month disorder. This effect was primarily related to lower openness with family, and we found no evidence for a similar effect among men. These findings are partially consistent with the idea that sexual orientationbased stigmatization may lead to lower social support quality, making sexual minority individuals less resilient to life stressors [27,28,32,34,43]. Importantly, differences in the mediating role of social processes among LGB participants are in line with some studies that have shown stronger mediating effects among sexual minority women compared to men [44] but are inconsistent with those finding a mediating effect of social support only among sexual minority men [28]. It is noteworthy that, we found only small and inconsistent differences in the quality of social support between heterosexual and LGB individuals, contrasting previous research [11,29,34]. Future research should investigate potential differential mechanisms between sexual minority women and men in the mediating role of social support. --- Strengths and limitations The current study has multiple strengths. First, to our knowledge, this is the first cross-national study assessing mental health disparities between sexual minority and heterosexual individuals, highlighting that the adverse effect of minority stress experienced by LGB individuals is a global issue. In addition, using the CIDI, a validated and reliable interview, we were able to assess a wide spectrum of disorders. Moreover, participants for our study were sampled from the general population, thus preventing the biased estimates that can arise from sampling from organizations and groups within the LGB community [45]. This study also has several limitations. First, responses were collected between 2001 and 2012, meaning that recent social and legal changes are not reflected in the data. However, differences in prevalence rates of mental disorders between LGB and heterosexual participants reported in our study were similar to that of more recent studies [6,10]. Indeed, Sandfort and colleagues [5] have suggested that increasing social acceptance of sexual minorities might result in LGB-specific discrimination becoming more subtle rather than disappearing, which may explain the persistence of disparities in psychiatric morbidity between LGB and heterosexual individuals despite societal change. Furthermore, the overall small size of the LGB sample suggests underreporting of sexual minority participants, particularly in low/middle-income countries, where only about 0.5% of the population identified as LGB. This might be related to several methodological characteristics. First, participants might have been less likely to disclose potentially socially undesirable information in an interview compared to computer-assisted administration [46]. Second, in most countries, only participants who reported having had sexual intercourse were asked about their sexual orientation; sexual minority participants may be less likely to have had sexual intercourse or may not regard same-sex sexual activity as being sexual intercourse. Taken together, this means that self-identified LGB participants may not be fully representative of all LGB people, especially in low/ middle-income countries. Notably, it is possible that participants who did not disclose their sexual minority status might systematically differ in their risk of reporting psychiatric morbidity than those who did, leading to a potential underestimation of global mental health disparities. In addition, it is possible that those participants who were able to open up about their sexual minority status also experienced higher levels of general social support. Especially among GB men, this might have obscured a potential mediating effect of social support on mental health outcomes. We also grouped bisexual and lesbian/gay individuals together for the mediation analyses. This is an issue, as studies have shown that aggregating sexual minority groups can obscure differences in particular sub-groups [47]. Finally, the mediation analysis was based on cross-sectional data, so we have no information about the temporal ordering of events. Although it is reasonable to presume that sexual orientation preceded current social support and 12-month mental health problems, we cannot exclude the possibility that mental disorders actually caused a decrease in social support , rather than social support mediating the effect of sexual orientation on mental disorders [44]. --- Conclusion In conclusion, this study provides further evidence for an association between sexual minority status and a wide range of psychiatric disorders from a large cross-national sample. Our results are in line with findings from national studies and stress the increased risk for psychiatric disorders among LGB individuals. We found that the increased risk for a psychiatric disorder in the past year for LGB individuals was partially mediated by perceived openness with family among women, but not men. To fully understand the relationship between minority stress and psychiatric morbidity in sexual minority individuals, future studies should consider gender differences in the influence of social support. were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. --- Availability of data and materials Access to the cross-national World Mental Health data is governed by the organizations funding and responsible for survey data collection in each country. These organizations made data available to the WMH consortium through restricted data sharing agreements that do not allow us to release the data to third parties. The exception is that the U.S. data are available for secondary analysis via the Inter-University Consortium for Political and Social Research , http:// www. icpsr. umich. edu/ icpsr web/ ICPSR/ series/ 00527. --- --- --- Ethical standards The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author and the source, provide a link to the Creative Commons licence, and indicate if changes --- Authors and Affiliations

Purpose Lesbian,gay, and bisexual (LGB) individuals, and LB women specifically, have anincreased risk for psychiatric morbidity, theorized to result from stigma-baseddiscrimination. To date, no study has investigated the mental healthdisparities between LGB and heterosexual AQ1individuals in a largecross-national population-based comparison. The current study addresses thisgap by examining differences between LGB and heterosexual participants in 13cross-national surveys, and by exploring whether these disparities wereassociated with country-level LGBT acceptance. Since lower social support hasbeen suggested as a mediator of sexual orientation-based differences inpsychiatric morbidity, our secondary aim was to examine whether mental healthdisparities were partially explained by general social support from family andfriends. Methods Twelve-monthprevalence of DSM-IV anxiety, mood, eating, disruptive behavior, and substancedisorders was assessed with the WHO Composite International DiagnosticInterview in a general population sample across 13 countries as part of theWorld Mental Health Surveys. Participants were 46,889 adults (19,887 males; 807LGB-identified). Results Maleand female LGB participants were more likely to report any 12-month disorder (OR2.2, p < 0.001 and OR 2.7, p < 0.001, respectively) and most individualdisorders than heterosexual participants. We found no evidence for anassociation between country-level LGBT acceptance and rates of psychiatricmorbidity between LGB and heterosexualAQ2 participants. However, among LBwomen, the increased risk for mental disorders was partially explained by lowergeneral openness with family, although most of the increased risk remainedunexplained. Conclusion These results provide cross-national evidence for an association between sexual minority status and psychiatric morbidity, and highlight that for women, but not men, this association was partially mediated by perceived openness with family. Future research into individual-level and cross-national sexual minority stressors is needed.

Introduction It has been estimated that 7.5 million, or 10.5% of children younger than the age of 18 in the United States, live with a parent who had an alcohol use disorder in the past year . This is a major public health concern given the substantial evidence that offspring of parents with alcohol problems are at an elevated risk for a host of psychosocial problems . Here, we took a family systems perspective to examine parenting as a mechanism linking parental alcohol dependence symptoms to adolescents' risky drinking and conduct problems. Expanding upon previous research, we considered how fathers' and mothers' ADS might influence their own as well as the other parent's parenting behaviors , in relation to adolescent outcomes. We further examined whether and how these associations may vary across adolescent ethnicity and gender. --- Direct and Indirect Effect of Parental Alcohol Dependence Symptoms via Parenting Behaviors: A Family Systems Perspective Parental alcohol problems, both concurrent parental drinking problems and family/parental history of alcoholism, have been shown to be associated with a range of offspring maladaptive outcomes from childhood to adulthood . Among adolescents, parental alcohol dependence is associated with higher risk for early initiation of alcohol use , higher levels of alcohol consumption and heavy episodic drinking , more externalizing problems, such as aggression and conduct disorders , and more depressive and anxiety symptoms . In this study, we focus on adolescents' risky drinking, because adolescence is the developmental period when many individuals initiate alcohol use and risky drinking during adolescence is a precursor for alcohol problems later in life . We take a multivariate approach to also consider adolescents' conduct problems as they are prevalent and often co-occur with alcohol use and misuse among adolescents . Parents serve as a primary socialization agent for adolescents and play an important role in influencing adolescent outcomes such as substance use and deviant behaviors through parenting and parent-child interactions . Thus, in addition to direct associations , parental alcohol dependence may influence adolescent adjustment indirectly by disrupting positive parenting behaviors. Positive parenting includes parental behaviors that express and convey support, warmth, affection, and acceptance to their offspring . The extent to which parents exhibit positive parenting toward their children is associated with a variety of developmental outcomes, including problematic alcohol use and externalizing problems. For example, low levels of positive parenting such as low parental involvement and support , low parental warmth , poor parent-child communication , and low parent-child closeness , have been associated with more alcohol use and externalizing problems among adolescents. Parental alcohol problems consume parents' psychosocial and financial resources. Thus, parents with concurrent or past alcohol problems may be impaired in their ability to maintain good home environments and provide positive parenting behaviors, which in turn negatively influence adolescent outcomes. Indeed, parental alcohol dependence is associated with decreased parental monitoring , lower parental warmth and involvement , poorer parent-child communication , and lower parent-child closeness , which in turn are related to higher adolescent risky drinking and conduct problems. Many studies on the effects of parental alcohol dependence have focused on fathers' alcohol dependence or operationalized parental alcohol dependence broadly . Thus, there is a limited understanding of the unique roles of fathers' and mothers' alcohol dependence in influencing adolescent outcomes. This is important to note, given that the effects of parental alcohol dependence may differ between fathers and mothers. Ohannessian and colleagues , for example, found that fathers' alcohol dependence, but not mothers' alcohol dependence, was related to drinking behaviors among adolescents. It was also found that fathers' and mothers' problem drinking differentially affect substance use for boys and girls such that the association between parental problem drinking and adolescent substance use was stronger in same-sex parent-adolescent dyads . Furthermore, some studies suggest that fathers' and mothers' problem drinking may influence adolescents' substance use and externalizing problems via different mediating pathways . Moreover, even fewer studies have considered alcohol problems and parenting behaviors from both fathers and mothers as separate constructs and analyzed them simultaneously in the same model to examine whether and how one parent's alcohol problems may influence the other parent's parenting behaviors, which in turn can affect adolescent outcomes. This is a noteworthy gap, given that researchers have called for using a family systems perspective to better understand individual development . Family systems theory posits that individuals or subsystems in the family are interdependent and influence one another through processes such as spillover and/or crossover effects . Spillover refers to the transfer of affect or behavior within one person across subsystems , whereas crossover refers to the transfer of affect or behavior between people in the family . Here, we considered the unique contributions of fathers and mothers and the spillover and crossover effects between fathers' and mothers' ADS and parenting behaviors in relation to adolescent outcomes. --- Ethnic Differences Rates of adolescents' drinking and conduct problems vary across ethnicity. Compared to European American adolescents, African American adolescents are less likely to consume alcohol but have more conduct problems . In addition, the influence of risk and protective factors, including those from the family context, on adolescent outcomes may differ across ethnic groups . EA and AA families often differ in their neighborhood environments, socioeconomic status, and cultural norms, which may lead to differences in parenting behaviors and their effects on child development . For example, a high level of parental control may be benign or protective among AAs , because it is relatively normative in AA communities, particularly among those who live in at-risk neighborhoods . Indeed, an authoritarian parenting style characterized by low acceptance and high control was associated with lower increase in heavy episodic drinking over time for AA adolescents but not for EA adolescents . Deutsch and colleagues found that maternal support was more strongly associated with lower delinquent behaviors among EA adolescents than for AA adolescents . Other studies found no ethnic differences in the associations between parental problem drinking, parenting, and adolescent drinking and related outcomes . There is a clear need for more research to examine potential differences in pathways of risk/protection to alcohol use and related outcomes across ethnic groups. --- Gender Differences In addition to gender differences in adolescents' drinking patterns and prevalence of conduct problems , researchers have suggested that there may be gender differences in physiological, psychological, and social factors influencing alcohol use and related outcomes . There are mixed results from prior research on whether parental alcohol dependence/problems and parenting behaviors are associated with alcohol use and externalizing problems differentially in male and female adolescents . For example, some studies showed that parental closeness andparental support have a stronger effect on alcohol and substance use for adolescent girls than for boys. Other studies found parental monitoring to be more protective in reducing substance use and externalizing behaviors among adolescent boys than girls . Yet, others found no differences in the relations between parental alcohol problems, parenting, and adolescent outcomes . These mixed findings could be in part due to differences in sample characteristics and research methodology across studies. Furthermore, few studies have examined these offspring gender differences when considering the unique influences of fathers and mothers . --- The Current Study The primary goal of this study was to consider the potentially unique roles of fathers' and mothers' ADS in influencing adolescent risky drinking and conduct problems, both directly and indirectly via parenting behaviors. A secondary goal was to examine potential differences in these pathways of risk based on adolescents' ethnicity and gender. Grounded in family systems theory , we considered ADS and parenting behaviors from both fathers and mothers simultaneously to examine their unique roles, as well as the spillover and crossover effects between fathers' and mothers' ADS and parenting behaviors, in influencing adolescent outcomes. We hypothesized that, in addition to direct associations, fathers' and mothers' ADS would have indirect effects on adolescents' risky drinking and conduct problems via a pathway marked by disruptions to their own and the other parent's positive parenting behaviors. We did not have specific hypotheses regarding differences in these risk pathways across ethnicity and adolescent gender given the lack of consistent evidence from prior research. --- Materials and Methods --- Sample Participants came from the Collaborative Study on the Genetics of Alcoholism , a multi-site study that was designed to identify genes involved in alcohol dependence and related phenotypes. Probands were identified through alcohol treatment programs at six U.S. sites and were invited to participate if they had a sufficiently large family with two or more members in the COGA catchment areas. Population-based comparison families were also recruited. Data collection for COGA started in 1991 when adults in the target extended families were invited to complete the Semi-Structured Assessment for the Genetics of Alcoholism , a comprehensive interview that assesses alcohol use disorders and other psychiatric phenotypes such as major depressive disorder and conduct disorder . Approximately 5 years later, adults in the COGA extended families were invited again to complete a SSAGA interview ; thus some adults had one interview while others completed two SSAGA interviews. In 2004, COGA launched the Prospective Study that aims to examine how genetic risks unfold across development and in conjunction with the environment. Specifically, offspring of adults in the COGA extended families who were aged between 12 and 22 and had at least one parent who completed SSAGA in the original COGA adult interviews were recruited for the COGA Prospective Sample . These offspring participants complete SSAGA, or an adolescent version of SSAGA if they were younger than 18 years old, at enrollment and are re-interviewed at about 2-year intervals. For those COGA Prospective participants where only one parent had completed SSAGA, the other parent was invited for a SSAGA interview when his/her offspring enrolled in the COGA Prospective Study. For the purpose of the current study, we linked data from the COGA adult interviews and the Prospective Study. We only included adolescents who 1) completed their baseline assessment at enrollment, and their first follow-up assessment , 2) were aged between 12 and 17 at their first follow-up assessment, and 3) self-identified as European American or African American -the two largest ethnic groups in COGA. This strategy resulted in an analytic sample of 1,282 adolescents from 497 COGA extended families . This is a relatively early adolescent sample, as 47.4% of the sample was aged 12 at their baseline assessment . Fathers' age ranged from 28 to 70 and mothers' age ranged from 27 to 58 at the time of their adolescents' baseline assessment. --- Measures Parental alcohol dependence symptoms -Data on parental ADS were drawn from parents' SSAGA interviews. 425 fathers and 662 mothers completed SSAGA in the first wave of COGA; 648 fathers and 1011 mothers completed SSAGA in the 5-year follow up wave. Only a small subset of parents completed SSAGA when their adolescents enrolled in the COGA Prospective Sample, as many of them already completed SSAGA earlier and thus were not invited for the interview again. Some parents refused to participate and others were lost to contact . For parents who completed SSAGA twice, we used data from the interview in which they endorsed the greatest number of ADS to index their ADS. Parental ADS was operationalized as the number of alcohol dependence criteria endorsed according to the DSM-IV criteria . Our rationale for using ADS data from a single interview was that in high-risk sample such as COGA we wanted to measure parents' greatest expression of their predisposition/risk to alcohol dependence on which the sample was originally ascertained. Of the fathers who had data on ADS , 31.1% endorsed zero ADS and 40.9% endorsed 3 or more ADS, which met the clinical diagnosis for DSM-IV alcohol dependence. Of the mothers who provided data on ADS , 46.3% endorsed zero ADS, and 25.7% endorsed 3 or more ADS. We used ADS rather than alcohol dependence diagnosis in order to increase power for analysis. Of the families where ADS data are available both for father and mother , 25.2%, 8.2%, and 13.6% had only father, only mother, and both parents met diagnosis for alcohol dependence, respectively. Positive parenting-Data on parenting behaviors were drawn from the Home Environment section of adolescents' baseline CSSAGA assessment. Items accessing parenting behaviors were adapted from the Home Environment Interview for Children . We conducted an exploratory factor analysis with items that we considered broadly tap onto the positive parenting construct, which yielded three subscales that we labeled as parental involvement, parent-child communication, and parent-child closeness. In view of the previous findings that multiple dimensions of parenting better account for their influence on adolescent behaviors than a single dimension , we used these three subscales to index positive parenting behaviors as a latent construct. The parental involvement scale asked adolescents 5 questions about whether or not their father/mother figure helped them with schoolwork, chores, fun activities, shopping, and making plans. Responses were coded as 0 and 1 . Scores were summed to create variables for maternal involvement and paternal involvement. The parentchild communication scale contained 3 items asking adolescents whether or not they and their father/mother figure talked about news, their problems, or other things like movies, friends or anything else. Response options were 0 and 1 for each item. Sum scores were used to create variables for mother-child communication and father-child communication. The parent-child closeness scale included 2 questions that asked participants how well they got along with their father/mother figure most of the time and how close they felt to their father/mother figure. Response options ranged from 1 to 4 , and from 1 to 3 for the two questions, respectively. Scores were standardized and averaged across the two items to create variables representing father-child closeness and mother-child closeness. Correlations between the three scales ranged from .39 to .49 and from .50 to .57 for fathers and mothers, respectively. Some adolescents reported parenting behaviors for their nonbiological father or mother figures. Because we only had data on alcohol dependence symptoms from biological parents, adolescent-reported parenting behaviors of non-biological parents was coded as missing. Adolescent risky drinking-Data on adolescent risky drinking, as indexed by engagement in heavy episodic drinking in the past year, came from adolescents' first followup assessment. Adolescents responded to one question: "how often did you have 5 or more drinks in 24 hours during the last 12 months". Responses ranged from " 0 = never" to "12 = every day". Adolescents who reported never initiating alcohol use were assigned a score of zero. Because 85.8% of adolescents reported no heavy episodic drinking in the past year, we created a dichotomous variable to indicate engagement in risky drinking. Adolescent conduct problems-Data on adolescent conduct problems, as indexed by conduct disorder criterion counts , also came from adolescents' first follow-up assessment. CDSX was operationalized as the number of conduct disorder symptoms ever endorsed according to the Diagnostic and Statistical Manual of Mental Disorders and had a possible range of 0-15. In the current sample, CDSX ranged from 0 to 11. In total, 66.0% of the adolescents endorsed 0 CDSX, 18.3% endorsed 1 CDSX, 8.8% endorsed 2 CDSX, and 6.9% endorsed 3 or more CDSX. Because endorsing 3 or more CDSX meets the clinical diagnostic criteria for conduct disorder, we grouped adolescents into four categories to indicate their conduct problems . Covariates-We included adolescents' age, gender , and ethnicity as covariates given prior evidence that they are related to adolescent alcohol use and conduct problems . Because family structure may impact parental influences and family processes , we also considered family structure as a covariate in this study. Family structure was coded as 1 if adolescents indicated that they lived with both biological mother and biological father; all other family living arrangements were coded as 0. Given that family socioeconomic status has been associated with adolescent alcohol use and conduct problems and may confound the effect of ethnicity , we included parental education, indexing family socioeconomic status, as a covariate in this study. Parents reported their highest level of education as part of SSAGA, and scores were converted to the number of years typically required to complete that level of education. Parental education was calculated as the average of maternal and paternal education. In cases where information was only available for one parent, the available score was used to index parental education. Because alcohol use disorders often co-occur with internalizing and/or externalizing psychopathology , we also included paternal and maternal antisocial behavior and internalizing problems as covariates. Paternal/maternal antisocial behavior was operationalized as the number of DSM-IV Antisocial Personality Disorder criteria endorsed at the SSAGA assessment when fathers/mothers indorsed their maximum number of alcohol dependence symptoms. Paternal/maternal internalizing problems was measured using a 7-item Internalizing Symptom Scale recently developed by COGA researchers based on the SSAGA . This scale assessed symptoms related to agoraphobia, panic disorder, social phobia, obsessive compulsive disorder, post-traumatic stress disorder, major depression, and suicidality that took place in the context of depression; the scale has a possible range between 0 and 21. --- Analytic Strategy We began with preliminary analyses to examine the amount and patterns of missingness of our data, in order to determine the appropriate approach to handle missing data in subsequent analysis, followed by examination of descriptive statistics and correlations among study variables. We then examined the associations between parental ADS, parenting behaviors, and adolescent risky drinking and conduct problems by conducting structural equation models using Mplus version 7.31 . We conducted a multivariate SEM model to consider adolescents' risky drinking and conduct problems simultaneously given that they are related constructs . Fathers' positive parenting was modeled as a latent variable indicated by paternal involvement, father-adolescent communication, and father-adolescent closeness. Mothers' positive parenting was modeled as a latent variable indicated by maternal involvement, motheradolescent communication, and mother-adolescent closeness. To account for method effects, and to take into account the shared variances between fathers' and mothers' parenting , errors terms were correlated between fathers' and mothers' parenting subscales, as were the residual errors for mothers' and fathers' positive parenting latent factors. The residual errors for adolescent risky drinking and conduct problems were also allowed to be correlated. Standardized loadings for each indicator and correlated error terms for the latent constructs are presented in Table 3. Risky drinking and conduct problems were specified as categorical variables in the SEM model, and thus the robust weight least square estimator was used by default in Mplus. In the path model, fathers' and mothers' ADS were specified as predictors of their own and the other parent's parenting behaviors, as well as adolescents' risky drinking and conduct problems. Fathers' ADS and mothers' ADS were specified as correlated. Fathers' and mothers' parenting behaviors were specified as predictors of adolescent risky drinking and conduct problems. Adolescents' age, gender, and ethnicity, family structure, parental education, paternal/maternal antisocial behavior and internalizing problems were included as covariates to predict adolescent risky drinking and conduct problems. To evaluate indirect effects , we used the MODEL CONSTRAINT command in Mplus to specify new parameters that represent the multiplication of coefficients for the two specific paths , and father parenting → adolescent risky drinking ) that comprise the indirect pathway. These new parameters , which were tested along with other path coefficients in the model, provided a test of indirect effects. MODEL CONSTRAINT is the recommended method for estimating indirect effects in Mplus in the context of multiple imputation . The estimation of indirect effects via the MODEL CONSTRAINT command uses the delta method that takes into account the covariance between the a and b estimates and provides estimates of standard errors for the indirect effects. To test for potential differences in the associations between parental ADS, parenting behaviors, and adolescent outcomes across ethnicity and adolescent gender, we first conducted multigroup confirmatory factor analyses to consider measurement invariance for the paternal and maternal positive parenting latent factors across the comparison groups , followed by multigroup SEM analysis with Wald chi-square test of parameter equalities. MGCFA compared two alternative models, one with factor loadings for the latent constructs freely estimated across groups and another model where factor loadings were constrained to equality across groups. To the extent that the two alternative models do not differ significantly in model fit, as indicated by a non-significant chi-square difference test and/or difference in Comparative Fit Index smaller than .01 in absolute value , there is evidence of metric measurement invariance, which is sufficient for group comparisons in structural path coefficients . To test for potential differences in path coefficients between males and females, multigroup analysis was conducted by removing gender from the SEM model and then comparing a model with all remaining paths constrained to equality with one that had all paths freely estimated across male and female adolescents. Similarly, to test for potential differences in path coefficients across ethnic groups, multigroup analysis was conducted by removing ethnicity from the SEM model and then comparing a model with all remaining paths constrained to equality with one that had all paths freely estimated across EA and AA adolescents. We accounted for interdependence within family in all analyses using the CLUSTER command in Mplus. --- Results --- Missing Data and Imputation Of the whole sample, 477 adolescents were missing data for fathers' ADS, and 105 were missing data for mothers' ADS. Adolescents who had missing data on their fathers' and/or mothers' ADS did not differ in age, gender, risky drinking, conduct problems, and their self-reported parenting behaviors, compared to adolescents who were not missing any data for their fathers' and mothers' ADS. African American adolescents were more likely to have missing data on parental ADS than European American adolescents . 373 adolescents were missing data on fathers' parenting and 83 were missing data for mothers' parenting. AA adolescents were more likely to have missing data on fathers' parenting than EA adolescents . We used multiple imputation to account for missing responses, given that multiple imputation is more flexible to handle missing data for a mixture of categorical and continuous variables . Multiple imputation has also been shown to be robust and provide unbiased results even for high rates of missing data . Specifically, we used maximum likelihood estimation method to create ten imputed datasets in Mplus. The data imputation model included all of the observed study variables included in the SEM model. We conducted subsequent analyses with the imputed datasets and the final parameter estimates, standard errors, and goodness-of-fit statistics of the SEM model were obtained with the automatic aggregation procedure implemented in Mplus . --- Descriptive Statistics and Zero-order Correlations Table 1 presents sample sizes and descriptive statistics of the study variables by adolescent ethnicity and gender, along with results from independent samples T tests for mean differences across ethnic and gender groups. Parental ADS and positive parenting behaviors were similar between EA and AA adolescents. However, compared to EA adolescents, AA adolescents were less likely to be living with both biological parents and to be engaged in risky drinking, had lower parental education, higher paternal and maternal antisocial behaviors, and more conduct disorder symptoms. Family structure, parental education, and parental ADS were similar between male and female adolescents. However, compared to males, female adolescents reported higher levels of maternal involvement and mother-child communication and lower levels of mother-child closeness. Female adolescents also reported lower rates of heavy episodic drinking drinking and fewer conduct disorder symptoms. Table 2 shows bivariate correlations between variables considered in this study from analysis based on the imputed datasets. --- Predicting Adolescent Risky Drinking and Conduct Problems Results from the multivariate SEM model predicting adolescent risky drinking and conduct problems are presented in Figure 1. The model demonstrated mediocre fit to the data = 720.54; p < .001; CFI = .77; RMSEA = .07). Although the chi-square value was statistically significant and the CFI below typically accepted standards of "acceptable" fit , the RMSEA was indicative of acceptable model fit . We note that relatively low values of the CFI tend to occur in complex models where there are small correlations among variables. Given the small magnitude of several associations in the model , a somewhat low CFI is not unexpected. Kenny suggests that since the CFI is an incremental fit index, CFI may not be a useful indicator of model fit in cases where a null model produces RMSEA values < .158; ours was .102. As such, despite the relatively low CFI, the overall model fit statistics suggest acceptable fit. Standardized coefficients for the structural paths are presented in Figure 1. Consistent with the hypothesis, fathers' ADS were negatively associated with fathers' and mothers' positive parenting behaviors. Mothers' positive parenting, in turn, was negatively associated with adolescents' risky drinking whereas fathers' positive parenting was negatively associated with adolescents' conduct problems. Contrary to prediction, neither fathers' nor mothers' ADS were directly associated with adolescents' risky drinking and conduct problems. However, the indirect effect of fathers' ADS on adolescents' conduct problems via fathers' positive parenting was statistically significant . There was also an indirect effect of fathers' ADS on adolescents' risky drinking via mothers' positive parenting . Mothers' ADS was not associated with their own or fathers' positive parenting behaviors. Thus, there was no direct or indirect association between mothers' ADS and adolescents' risky drinking and conduct problems. --- Ethnic Differences in Pathways of Risk Results from MGCFA suggested the measurement model allowing factor loadings to be freely estimated across racial groups and the model constraining factor loadings to equality across racial groups were not significantly different in terms of model fit , suggesting that factor loadings were invariant across the EA and AA groups in the measurement model assessing fathers' and mothers' positive parenting. The Wald test for simultaneously constraining all of the key path coefficients in the model to equality across ethnic groups was not significant , suggesting that there were no ethnic differences in these path coefficients. --- Gender Differences Results from MGCFA indicated the measurement model allowing factor loadings to be freely estimated across gender groups and the model constraining factor loadings to equality across gender groups were not significantly different in terms of model fit , suggesting that factor loadings were also invariant across the gender groups in the measurement model assessing fathers' and mothers' positive parenting. Multi-group SEM analysis indicated no significant differences in path coefficients across gender . This suggested that the patterns of associations among the various constructs in the model were similar between adolescent males and females. --- Follow-up Analysis We conducted follow-up analyses to check the robustness of our results. As noted above, there was a relatively large amount of missing data for parental assessments of ADS in our sample. We conducted additional analysis with the subsample of adolescents having complete data on both parents' ADS, in order to examine whether the patterns of association are consistent or biased due to missing data. The overall pattern of associations was similar to results from analysis using the imputed datasets for the whole sample. Given that family processes may vary as a function of family structure/living arrangement, we also ran analysis with the subsample of families where adolescents lived with both biological parents to check whether the patterns of association were consistent. The overall pattern of associations was similar to those from the whole sample. Results from these follow-up analyses are not presented but available upon request. --- Discussion Grounded in family systems theory , the present study examined the unique roles and spillover and crossover effects of fathers' and mothers' ADS and parenting behaviors in relation to adolescent risky drinking and conduct problems, measured approximately 2 years after the parenting behaviors were assessed. We further tested for differences in these relationships by adolescents' ethnicity and gender. Findings suggested that fathers' ADS were associated with adolescent outcomes indirectly by disrupting their own and mothers' positive parenting behaviors, whereas mothers' ADS were not associated with adolescent outcomes either directly or indirectly via parenting behaviors. We found no ethnic or gender differences in these associations. Consistent with hypothesis and prior research , our findings suggest that parenting behaviors serve as a mediating pathway for the influence of fathers' ADS on adolescent risky drinking and conduct problems. Our findings revealed unique patterns of spillover and crossover effects for fathers and mothers. Consistent with the family systems perspective, fathers' ADS not only undermine their own positive parenting behaviors but also disrupt mothers' parenting behaviors. Perhaps fathers' ADS create high levels of strain and emotional tension in the home and/or impair the quality of relationship with the adolescents' mother , which in turn was associated with higher adolescent risky drinking and conduct problems. It is also likely the case that when the father exhibited ADS, the mother would need to compensate by increasing her responsibilities and duties , which may result in additional stress and time involvement that could potentially undermine mothers' positive parenting. In contrast, mothers' ADS did not negatively influence their own or fathers' parenting behaviors. Perhaps mothers are more likely to maintain positive parenting behaviors such as involvement and communication, even in the context of ADS, given that most mothers have primary caregiving roles in the family. Consistent with expectations, our bivariate correlations analyses indicated that both fathers' and mothers' ADS were correlated with higher likelihood of adolescent risky drinking and more conduct problems. However, results from our multivariate SEM model indicated that neither fathers' nor mothers' ADS were directly associated with these adolescent outcomes. These findings suggest that parental ADS may influence adolescent outcomes indirectly rather than directly, reiterating the importance of understanding the mediating pathways through which parental ADS exert an effect on adolescent adjustment. Findings from this study suggest parenting behaviors as an important mechanism through which fathers' ADS play a role in adolescents' risky drinking and conduct problems. Although our findings indicate no indirect effect of mothers' ADS on these adolescent outcomes via parenting behaviors, we note the possibility that mothers' ADS may influence adolescent outcomes via other mechanisms such as increased adolescents' environmental stress and negative affect and more family conflict . It is also likely that mothers' ADS influence other aspects of parenting such as monitoring, discipline, or negative parenting practices that were not considered in our model, which may in turn affect adolescent outcomes. For example, Chassin and colleagues showed that mothers' alcohol dependence was associated with lower maternal and paternal monitoring, which in turn was associated with higher adolescent substance use. Additional research is needed to further understand the pathways by which mothers' alcohol problems may impact adolescent outcomes. Interestingly, results from our multivariate SEM model indicated that fathers' but not mothers' positive parenting was associated with fewer adolescent conduct problems, whereas mothers' but not fathers' positive parenting was associated with adolescents' lower likelihood of engaging in risky drinking. This is unexpected, given that both paternal and maternal parenting behaviors have been shown to be associated with risky drinking and externalizing problems in adolescents . In our study, when paternal and maternal parenting behaviors were analyzed in separate models , both were indeed significantly associated with adolescent risky drinking and conduct problems, as also indicated by the bivariate correlations between these variables. Our model where fathers' and mothers' parenting analyzed simultaneously took into account the variances shared between fathers and mothers and provided an examination of the unique effects of fathers' and mothers' parenting. It is interesting that results from the fathers and mothers combined model and those from the analyses conducted separately for fathers and mothers were somewhat different. This suggests that while some of the effects of fathers' and mothers' parenting on adolescents' risky drinking and conduct problems exert through processes shared between fathers and mothers within the family, mothers' parenting seems to have an extra unique effect on adolescents' risky drinking while fathers' parenting has an extra unique effect on adolescents' conduct problems. Future research is needed to replicate these findings. Pathways of risk did not vary across European American and African American adolescents, suggesting the important role of fathers' ADS and its spillover and crossover effects between ADS and parenting behaviors across ethnic groups. This is consistent with some prior studies showing no ethnic differences in the associations between parental problem drinking, parenting, and adolescent outcomes , but contradicts other studies that found the associations between parenting behaviors and adolescent outcomes to vary across ethnicity . Overall, the path coefficients linking parental ADS and parenting behaviors to adolescent risky drinking and conduct problems also did not vary across adolescent gender. These findings are consistent with some prior studies that found no difference in the associations between fathers' and mothers' alcohol dependence or problem drinking and adolescent psychopathology for sons and daughters , but contradict others that suggested stronger influence of fathers' alcoholism on alcohol abuse for daughters than for sons , or stronger associations between same-sex parent-adolescent dyads . Future research is needed to further understand the role of ethnicity and gender in the association between parental alcohol problems and adolescent outcomes. --- Limitations Our findings need to be interpreted in light of several limitations. First, the assessment of parental ADS generally occurred years prior to the adolescent interview, and there is considerable variability in adolescents' age at the time of assessment of parental ADS . For the majority of the sample, we did not have data on parental ADS at the time when parenting and adolescent outcomes variables were measured, and thus did not know whether or not parental ADS were still present at the assessment of parenting behaviors and adolescent outcomes. Thus, it was not possible to determine whether or how long adolescents were exposed to parental ADS. We note, however, that both concurrent and history of parental alcohol problems have been shown to be associated with negative parenting behaviors and psychosocial outcomes in offspring . Prior research also suggests that parental alcohol dependence and related externalizing disorders have long-term negative influence on offspring alcohol problems, regardless of timing of parental exposure . Parental alcohol dependence , in part, represent parents' genetic predispositions toward risky and problematic behaviors and/or stable individual characteristics , which can play an important role in influencing their parenting behaviors and child outcomes. In addition, adult alcohol dependence tends to be relatively stable across time , and our post hoc analyses indicated that adding adolescents' age at the assessment of parental ADS as a covariate did not change the pattern of results. Second, we only focused on positive parenting because our measures of negative parenting were weak , only available for a subset of the sample, and lacked adequate reliability. However, parenting is a multidimensional construct, and positive and negative parenting may have differential effects on offspring outcomes . Thus, future studies should consider additional dimensions of parenting to further understand the effect of parental ADS on adolescent outcomes via parenting. In addition, although we used a latent variable approach to measure positive parenting that capitalized on multiple dimensions of positive parenting behaviors and helped account for measurement error, we acknowledge that some of the factor loadings were low . This limitation regarding the measurement of positive parenting may have attenuated the effects observed in our analysis. The large amount of missing data for fathers is another limitation. We note, however, that there was no significant difference in key variables between adolescents who had missing data for fathers and those without missing data for fathers. Our follow-up analysis indicated that the patterns of associations are largely similar between analysis with the subsample of adolescents having complete parental data and analysis with the whole sample. Thus, our findings were unlikely to be biased due to missing data. Despite the degree of missingness, which we accounted for using multiple imputation, the availability of data for both fathers and mothers allowed us to examine spillover and crossover effects within the family in the context of parental ADS, which also contributes to the relatively small literature on the role of fathers in adolescent development. Another limitation is that we only included EA and AA adolescents in this study. Future research needs to examine whether the pattern of associations observed here are generalizable to other ethnic groups. Because our sample is highly enriched both for alcohol dependence and larger families, replication of our findings in community or nationally representative samples will be important. Further, we only examined the role of parental ADS on adolescent risky drinking and conduct problems. Since parental alcohol problems affect a range of offspring outcomes, future research is warranted to extend the current investigation to other adolescent outcomes such as internalizing problems. Finally, we did not consider the role of prenatal exposure to alcohol because relevant data was only available to a small subset of adolescents in our sample. Given that prenatal exposure to alcohol has implications for later development , future research is needed to examine whether or not the pattern of association between parental ADS, parenting, and adolescent outcomes may differ for prenatally alcohol exposed and nonexposed adolescents. --- Conclusion Despite several limitations, this study has many strengths, including the clinical assessment of ADS from fathers and mothers, the family systems approach to consider the spillover and crossover effects of fathers' and mothers' ADS and parenting behaviors, prospectively measured parenting behaviors and adolescent outcomes, rigorous tests of ethnic and gender differences in pathways of risk, and careful consideration of a number of important covariates. Our results suggest that fathers' ADS put adolescents at risk for risky drinking and conduct problems, above and beyond the effects of parenting, parental education, parental antisocial personality disorder and internalizing problems, family structure, and adolescents' age, gender, and ethnicity. Disrupted fathers' and mothers' parenting behaviors such as lower involvement and poor parent-child communication may serve as indirect pathways of risk associated with fathers' ADS. Our findings emphasize the importance of considering the unique roles of fathers' and mothers' ADS in influencing family processes and adolescent outcomes. --- Supplementary Material Refer to Web version on PubMed Central for supplementary material. Gender was coded 1 = male, 0 = female; ethnicity was coded 1 = European American, 0 = African American; family structure was coded 1 = intact family, 0 = non-intact family. --- --- ---

Background-Parental alcohol problems are associated with adverse adolescent outcomes such as risky drinking and conduct problems. Important questions remain about the unique roles of fathers' and mothers' alcohol problems and differences and/or similarities in pathways of risk across ethnicity and gender. In this study, we used a family systems approach to consider spillover and crossover effects of fathers' and mothers' alcohol problems (number of alcohol dependence symptoms) and parenting behaviors in relation to adolescents' risky drinking and conduct problems. Methods-The sample included 1,282 adolescents (aged 12-17) and their parents from the Collaborative Study on the Genetics of Alcoholism. Parents completed the Semi-Structured Assessment for the Genetics of Alcoholism (SSAGA) and adolescents completed an adolescentversion of SSAGA. Data were analyzed using multivariate structural equation modeling.

Background Breast cancer is the leading cancer amongst women in India [1]. Although the incidence of BC in India is only one-fourth that of high-income countries , the mortality rate is similar, at 13.3 per 100,000 in India compared to 12.9 per 100,000 in HICs [1]. One of the major causes of this disproportionately high mortality is the advanced stage of cancer at the time of presentation [2]. Breast cancers in the Indian subcontinent also occur at younger ages compared to HICs [3]. The World Health Organization and Breast Health Global Initiative have recommended breast cancer awareness and clinical breast examination-based early detection programs for reducing the high mortality from breast cancer in Lowand Middle-Income countries such as India [4]. These programs aim to create breast cancer awareness at the community level and encourage people to attend clinical breast examinations . Most LMICs recruit frontline healthcare workers for these examinations at primary healthcare centres , or other centres near womens' places of residence [5]. The uptake of these screening and early detection programmes in India and other LMICs is low, as demonstrated by facility-and community-based studies [6,7]. In 2010, the Government of India, through its National Health Mission, first launched a comprehensive noncommunicable disease control program [8]. However, The National Family Health Survey, which is the largest household survey in India, documented that less than two percent of women have been screened for breast cancer in India [9]. Studies from LMICs attribute poor access to healthcare facilities, poor socio-economic conditions, as well as lack of awareness and education, and socio-cultural barriers as the main reasons behind limited uptake of screening services [10,11]. Furthermore, stigma is also considered a key social determinant of health, and increasingly recognized as a significant psycho-social health barrier [12]. There is an increasing focus on exploring and addressing cancer-related stigma in India [13,14]. Additionally, household factors such as family set-ups and autonomy to make health-related choices and decisions can also influence health-seeking behaviours, especially among women [15]. Universal Health Coverage provides financial protection to patients and their families from out-ofpocket expenditure where healthcare provision is concerned [16]. Indian populations do not benefit from UHC and this contributes to delays in seeking, reaching, and completing cancer care [2]. Women in India, like in other LMICs [17], grapple with issues related to poor access and affordability of healthcare. This study was conducted in an Indian cohort that is covered by UHC, which essentially eliminated the affordability and accessibility issues, bringing focus to the socio-cultural barriers. As India and many other countries progress toward implementing UHC on a wider scale, it is informative and insightful to have a more nuanced evaluation of social and cultural barriers to the uptake of an early detection programme in a UHC setting, where other barriers have been largely mitigated. This study therefore aimed to evaluate barriers to seeking early detection services for breast cancer in an urban population covered by a UHC scheme. --- Methods --- Study setting This study was conducted in an urban community cohort in Mumbai, India. This urban community of 100,000 people is covered under an employees' contributory health service scheme. Under this scheme, healthcare services are provided at a minimum flat contribution of one percent of the basic pay of the employee. Most families in this cohort were of middle socio-economic status and earned a fixed salary in the Department of Atomic Energy employee cohort. The health system is structured in the form of a two-tiered system within the eastern suburbs of Mumbai, India. It consists of 14 PHCs spread across the city. The PHCs refer patients to a central referral hospital when needed. The 14 PHCs served a population of 100,000, and 4 of these 14 centres were transformed into breast clinics for this study [18]. In June 2013, an awareness-based early detection program for breast cancer was implemented through a dedicated network of breast clinics across these PHCs, in collaboration with the International Agency for Research in Cancer . Approximately 22,500 women in the age group of 30-69 years were eligible for the early cancer detection program as per the Government of India Ministry of Health guidelines for screening in India [8]. Gadgil et al. have published further details about this healthcare system, cohort, and organisation, in a separate manuscript dedicated to methodology of establishing an early detection program [18]. The women in this cohort had not received any screening or early detection for breast cancer prior to the implementation of this program. They were given information on breast cancer awareness, signs and symptoms, and mail invites for early detection, and symptomatic as well as asymptomatic women were encouraged to report to clinics nearest to their homes. Some symptomatic women who participated were under the age of 30; however, as service providers, we could not deny clinical breast examinations to these patients, so they were also included in the study. None of the women were given any compensation or incentives for participation in the program. All participants resided in the catchment area around the PHCs and could access the screening activities without any cost [18]. --- Study design This prospective nested case-control study, conducted from June 2015 to June 2017, assessed the breast healthseeking behaviour of women who did not participate in the early cancer detection program, hereafter referred to as non-participants . These non-participant women were compared to women who availed the program, henceforth referred to as participants , or the control group. For the study, 80% power, 95% confidence intervals, and 12% participation rate in the program were attributed. Allowing for 10% attrition, n = 272 in each group was reached as the sample size. where, Z = Confidence level at 95% p = Estimated prevalence or proportions e = range of CI The study was approved by the Ethics Committee of Bhabha Atomic Research Centre Hospital before starting the recruitment process . All methods were carried out in accordance with relevant guidelines and regulations. --- Diagnosis and treatment protocol All women who were detected with abnormalities on clinical breast examinations were referred to the hospital situated in the same residential area for further investigation and treatment. Diagnostic mammography and cytology were carried out as a part of triple assessment, but not as a screening modality. CBE remained the sole screening modality for all women [18]. --- Recruitment and data collection We recruited women who visited PHCs for complaints other than breast-related symptoms and those who were accompanying their relatives to health centres as 'non-participants' . All women who had participated in the early detection program were identified for the control group . We generated a randomised list of unique identification numbers from the list of participants, and recruited these women for the study. Recruitment was conducted in keeping with a 1:1 allocation ratio between participants and non-participants. We contacted the recruited individuals over the phone and arranged a convenient time and date for in-person interviews. Informed consent was obtained from all women during their hospital visit prior to interview. When n = Z 2 p e 2 recruited women were illiterate, informed consent was obtained from their respective legal guardians. Investigators were medical professionals with graduate and above level qualifications, with an experience of greater than 10 years at the community healthcare level. They attended a two-day training programme at the Department of Surgery at the central referral hospital where the early detection program was being implemented. Trained investigators then conducted in-person, faceto-face interviews with all participants. If an individual in the P or NP groups refused to be interviewed or could not be contacted, the next individual in the initial randomised list was contacted, and so on. Recruitment and interviews were completed over two rounds. In the first round of six months, the target sample size number was not reached due to low response rates in both P and NP groups. More UIDs were generated and randomised again to recruit participants, and non-participants were recruited from PHCs, as explained above. This additional set allowed us to reach the target sample size in the second round. The investigators designed a questionnaire available in an additional file addressing the well-documented barriers to breast health seeking in the literature discussed earlier. The questionnaire comprised 19 questions and was adapted from existing literature and questionnaires on the known social, cultural, and perceived risk-related barriers to breast health-seeking [19,20]. Social desirability bias was controlled by appropriate framing of the questions included in the questionnaire. The questions were framed with both positive and negative wording to reduce response bias. The respondents were anonymized during analysis and confidentiality was maintained to further reduce social desirability bias. The questionnaire contained demographic variables and included age, education, partner status, employment status, and family type . Joint family set ups refer to two or more married couples of a single generation, or three or more married couples from multiple generations living in the same household. On the other hand, nuclear families refer to a single married couple living with or without their unmarried children [21]. Questions related to the utilisation of and trust towards the healthcare system in general, and the ongoing early detection program specifically, were also included. The investigators conducted interviews in English, and in the national language Hindi, where participants were not fluent in English. --- Statistical analysis SPSS Version 24.0 and Microsoft Office Excel 2013 for Windows were used to perform statistical analysis. Socio-demographic characteristics of women were represented across both groups. Responses were presented in absolute numbers as well as percentages. Chi-square test of significance was used, and p < 0.05 was set as the significance threshold. --- Results The total number of women recruited was 544: 272 participants and 272 non-participants. Figure 1 summarises the recruitment algorithm. The ages of women in this study ranged from 19 to 77 years. The mean age was 48.03 ± 10.37 years. Table 1 shows the comparison of characteristics of women between NP and P groups. Participants were significantly more educated than non-participants, with a higher percentage of women receiving education up to and above the level of a graduate degree . Women who belonged to a higher-risk group of developing breast cancer, with a family history of breast and related cancers, participated more in the programme compared to women with average risk . Additionally, women from nuclear families had significantly more participation in the early detection programme . Both groups were comparable in age, partner status, and occupation status. --- Awareness about breast cancer Women in both groups had high breast cancer awareness, but awareness was higher among participants . Awareness regarding breast lumps as a symptom of BC was comparable across groups. However, participant women were significantly more aware that cancer could also present as blood-stained nipple discharge , skin changes over the breast or a lump in the armpit . Furthermore, those who participated in the early detection program were significantly more aware of the need for breast examination in asymptomatic women . --- Social and cultural barriers Both groups of women agreed that their health was important and that their family and/or spouses would support their attendance at breast clinics for routine check-ups. Significantly more women in the NP group felt they did not have time to attend routine check-ups . A significant cultural barrier among participants was the social stigma of being labelled a "cancer patient" in the eyes of friends and neighbours, if seen attending breast clinics . Fatalistic attitudes and spiritual beliefs regarding BC were similar in participants and non-participants, with no significant differences noted. None of the women perceived the 'distance Though there was no perceived risk of overdiagnosis in either of the groups, participants were more worried about losing the entire breast if detected with a lump or cancer . There was a statistically significant difference in patients' trust in the doctors' ability to treat BC if found, between participants and non-participants . Furthermore, there was a significant difference in the reported rudeness and indifference of staff experienced by nonparticipants . Table 2 describes the socio-cultural barriers to breast health seeking in both groups. --- Discussion This study revealed that participants had higher levels of education, were more aware of the signs and symptoms of breast cancer, and the need for participation in early detection programmes even in the absence of symptoms. Non-participants belonged to joint families, and stated lack of time as a reason for not attending the programme. Despite higher education and awareness, participants feared social stigma and 'losing the entire breast if detected with cancer' but had significantly more faith in the health system, in that their doctors would be able to treat them if diagnosed with BC. Education and awareness have repeatedly been cited as facilitators of early health-seeking behaviour [22][23][24][25]. Similarly, in our comparative study, participants had a significantly higher level of education and demonstrated greater awareness . However, it is important to note here that although non-participants had lower knowledge compared to participants, their overall knowledge of breast cancer was still quite high -this is likely due to the fact that overall education levels were also quite high compared to other LMIC populations. Vieira et al. [22] examined the healthcare system related barriers in a systematic review in Brazil. They identified that people with less education had difficulty navigating and accessing the healthcare system for diagnosis and treatment. Low awareness of the presenting signs and symptoms of cancer is one of the major factors contributing to delayed presentation [23] and poor health-seeking behaviour among women [24]. Greater knowledge about the disease condition was noted to be one of the best predictors of adherence to recommended breast screening programmes in a study by Charkazi et al. [25]. In its early stages, cancer either has minimal symptoms or can be entirely asymptomatic. Thus, early detection is of utmost importance for better survival. However, without awareness of the disease, seeking breast examination becomes a low priority and is ignored, leading to very late-stage diagnoses. A recent systematic review on barriers to breast and cervical cancer screening [10] found that in addition to common accessibility and affordability barriers, awareness and anxiety about disease and diagnosis were the most frequent socio-cultural barriers to participation in screening. Our UHC-based study largely mitigated the geographical and financial barriers. In this context, awareness and higher education in the population can be important determinants of participation in early detection and screening programs. Both formal education and disease awareness are consistently associated with better participation [10,26], as is echoed in the findings of our study. Women are often expected to be primary caregivers for dependent children and the elderly due to set societal gender roles, especially in the Indian subcontinent. This could lead to self-neglect and not viewing their own health as a priority. Studies from Brazil and Chile have highlighted that self-neglect as a result of limited time, has led to a decrease in women performing breast self-examinations [27]. Furthermore, the time taken to attend clinics and undergo investigative procedures Were the staff rude or indifferent to your desire to get routine check-ups? 0 0.000 has also been cited as a barrier to regular breast checkups [28]. Families with dependent children also lead to decreased time for primary caregivers for healthseeking activities. These findings were similar to the results of our study, wherein we found that non-participants largely belonged to joint families and stated lack of time as a barrier to attending the early detection programme. Women from nuclear families had significantly greater participation, suggesting that joint family setups could compromise women's autonomy to take decisions about resource expenditures and healthrelated decisions. Mahalakshmi et al. [7] similarly documented that family responsibilities were a deterrent to women's participation in screening. Koirala [29] found that joint families were associated with fewer cancer diagnoses: the author elucidated that living in a joint family lowers the financial resources available to each member of the family to participate in health-seeking activities that would allow them to get diagnosed. In our cohort, where financial resources did not limit participation due to UHC, it is highly likely that lower autonomy in a joint family was the most influential factor. In a patriarchal society such as India, systemic gender inequities mean that women disproportionately take on household responsibilities, which in turn prevents them from having the autonomy to make healthrelated decisions. While there is intergenerational help and deep family solidarity in a joint family set-up, there is also a loss of privacy and autonomy. Traditional male authoritarian leadership shapes hierarchical relationships in a joint family in India: male sons must defer to their fathers, and women, especially daughters-in-law, are considered subordinate, lacking decision-making ability [30]. Recent work by the author team assessing the effect of women empowerment indices on screening uptake also showed similar findings: women who lack autonomy to make decisions in the household had poor screening uptake [15]. As a study from Ethiopia also suggests, improving women's autonomy can result in the improved use of health services [31]. Despite engaging in the early detection programme, and having higher levels of education and awareness, women who participated in the early detection program were noted to be more fearful of the social stigma of attending 'cancer clinics' . The fear of being diagnosed with cancer and being socially isolated after diagnosis has been documented in Indian literature on screening for breast cancers. A qualitative study from South India, quoted women's responses, "[I'm] scared, the family members may not mingle with us casually. If the society comes to know that is all, [I] fear [that], they will isolate us", or "If they say, we have the disease after the screening, our life is gone after that, the life is over, that kind of negative thoughts [prevent going]" [7]. This throws light on the pervasive fear of cancer diagnosis. Participants demonstrated more fear than non-participants, especially of losing their entire breast if cancer were to be detected. This could perhaps be attributed to the fact that participants attended awareness talks where they learned about mastectomies and surgical management of the disease. However, it is important to note here that participant women overcame that stigma and fear and presented themselves for examination. Information providers hence must take care to adopt a comprehensive approach to awareness building, emphasising early operative benefits, availability of breast conservation surgery and addressing concerns of patients when they present themselves for consultations. A comprehensive approach aided by empathic communication skills helps prevent fear and foster trust. Yet, further interviews of these participant women to understand mechanisms of coping with fear and stigma to present themselves for screening would be deeply insightful. The defining parameter of the participant group remained that they had trust in their healthcare providers' ability to treat them, if they were to be diagnosed with cancer. It is widely recognised that trust in physicians positively influences health-seeking behaviours [32]. Further, Flugelman et al. [33] describe that trust in physicians is the most important factor in reducing patient anxiety prior to an intervention. Another study by Gupta et al. concluded that trust in healthcare providers remained the most important determinant for cancer screening participation, after the already acknowledged barriers of access and affordability [34]. In our study too, despite describing feelings of fear and stigma initially, participants ultimately trusted that their doctors would manage their BC adequately, which helped mitigate their fears. In contrast, the non-participant group had expressed that the healthcare facility might not be able to adequately treat their BC if it were to be diagnosed and that healthcare workers were rude and indifferent. Given that this was a UHC setting and all patients are life-long members, it is possible that these perceptions among non-participants arose from their previous experiences at other outpatient departments or clinics prior to the study. Hence, they likely did not visit the breast clinics when they were invited to participate. The participants of the program on the other hand perhaps felt more cared for, when they responded to the invites by presenting themselves for breast examination. When they presented, receiving detailed information about cancer and careful handling of their questions and fears probably made them feel confident and reassured about the treating team. The clinics dedicated to breast examination also may have provided a less crowded environment, and the non-rushed, sensitive approach of the caregivers may have further contributed to their trust and confidence in the program. Haworth, Margalit [35] note that South Asian women value healthcare professionals who understand and respect values of personal modesty and shyness. In their study, patients discussed 'feelings of exposure' as barriers to attending cervical cancer screening. Trust building and delivering information in a culturally sensitive manner hence becomes a priority to encourage health-seeking behaviours from the patient's perspective. By improving soft skills of the screening and treating teams, to focus on patient-centric communication and careful redressal of their concerns, patients would be encouraged to trust the system: in order to ensure that the UHC system is utilised to its full potential, trust-building in the context of effective, sensitive communication is paramount. Future work could focus on the impact of communication skills training for physicians and other healthcare professionals in building perceived trust among their patients. This study has a unique strength in studying a population covered by UHC. As such, this paper was able to analyse socio-cultural barriers outside of other commonly described barriers such as affordability and accessibility. This is further emphasised by our study's findings that neither participants nor non-participants believed distance to or accessibility of the clinics to be prohibitive to their attendance, making a strong case for UHC. This is in stark contrast to most other publications on the subject. Pal et al. [36] noted financial constraints and infrastructure deficiencies along with low awareness, as significant barriers to seeking breast health services in India. This study also has certain limitations. It is crucial to note that this study was based in Mumbai, a metropolitan city in India, where access to transport is markedly better than in most rural settings. Hence the generalizability of our findings to rural populations may be limited. A lack of open-ended questions in the methodology also limited the scope of the discussions in this paper. In-depth interviews enabling qualitative analysis would reveal a far more nuanced understanding of socio-cultural barriers, and hence are needed going forward. Future work could also benefit from a more targeted approach and explore the barriers to the uptake of early detection programmes in first-degree relatives of patients with breast cancer, who have been identified as a high-risk cohort [37]. --- Conclusions Higher education, trust, and nuclear families were protective factors and thus facilitators to engaging in the early detection programme. The major socio-cultural barriers identified in the study were joint family setups, lower education and awareness, and a lack of trust in healthcare professionals. Identifying these socio-cultural barriers to seeking breast health services is essential in order to formulate context-specific solutions to mitigate them. Future work is needed to study these socio-cultural barriers more indepth. Our study has identified certain implications for changing practice: namely greater trust and awareness building. "Inter-sectorial co-ordination" has previously proven valuable in mobilising different stakeholders for organised cancer detection in low-resource settings [38]. In addition, the engagement of social organisations and communication skills training for healthcare workers will also be essential in building trust among the public. Such multi-pronged approaches could help tackle the vast pool of socio-cultural barriers that exist today, from individual awareness to systemic faults. --- --- Abbreviations --- --- Additional file 1. Additional file 2: Supplementary Table 1. Regression analysis of factors associated with breast cancer screening. --- --- Funding Open access funding provided by Karolinska Institute. This study was unfunded. --- --- --- Competing interests The authors 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: ---

Background Breast cancer is the commonest cancer among women in India, yet the uptake of early detection programs is poor. This leads to late presentation, advanced stage at the time of diagnosis, and high mortality. Poor accessibility and affordability are the most commonly cited barriers to screening: we analyse socio-cultural factors influencing the uptake of early detection programmes in a Universal Health Coverage (UHC) setting in India, where geographical and financial barriers were mitigated. Methods Two hundred seventy-two women engaging in an awareness-based early detection program were recruited by randomization as the participant (P) group. A further 272 women who did not participate in the early detection programme were recruited as non-participants (NP). None of the groups were previously screened for breast cancer. Interviews were conducted using a 19-point questionnaire, consisting of closed-ended questions regarding demographics and social, cultural, spiritual and trust-related barriers.The overall awareness about breast cancer was high among both groups. None of the groups reported accessibility-related barriers. Participants were more educated (58.09% vs 47.43%, p = 0.02) and belonged to nuclear families (83.59% vs 76.75%, p = 0.05). Although they reported more fear of isolation due to stigma (25% vs 14%, p = 0.001), they had greater knowledge about breast cancer and trust in the health system compared to non-participants.The major socio-cultural barriers identified were joint family setups, lower education and awareness, and lack of trust in healthcare professionals. As more countries progress towards UHC, recognising socio-cultural barriers to seeking breast health services is essential in order to formulate context-specific solutions to increase the uptake of early detection and screening services.

Introduction The COVID-19 pandemic has been a significant global public health challenge. The causative agent, SARS-CoV-2, which presents as severe viral pneumonia with a high case fatality rate, has caused a total of 605 million cases and 6.4 million deaths as of 11 September 2022, according to the World Health Organization [1]. As already mentioned in previous studies, the COVID-19 pandemic has been accompanied by an explosion of inaccurate information about the disease, making it difficult for the general public to make informed decisions [2]. The "infodemic" , which means too much information, including false or misleading information in digital and physical environments during a disease outbreak, has been the main obstacle in the fight against this pandemic [2][3][4][5][6]. The WHO termed the COVID-19 misinformation situation as an 'infodemic' swarming with conspiracy theories, propaganda, and unproven scientific claims regarding the diagnosis, treatment, and prevention of the disease [7,8], which has made reliable information more difficult to find and discern, and has allowed rumors to spread more quickly, putting public health at risk by hindering the implementation of effective preventive measures [8]. Health Literacy is essential to address this [9,10] since there are studies that identify that people with low HL have been found to show lower awareness, knowledge, and protective behavior, which might result in a greater risk of COVID-19 infection [11][12][13][14][15][16]]. An adequate level of HL facilitates the identification of reliable information on COVID-19, improves knowledge, and enables the practice of protective health behaviors [17]. Some studies identify HL as a "social vaccine" to face the COVID-19 infodemic [10,11,18]. The problem of the "infodemic" is especially critical for university students because of the frequent use of social networks and the internet [19][20][21], where information is rapidly disseminated and where a large amount of false or misinformation circulates [22][23][24][25][26]. They are also the most active online, interacting daily with an average number of 5 digital platforms . In order to better understand how they are engaging with technology during this global communication crisis, an international study was conducted covering approximately 23,500 respondents in 24 countries across five continents [27]. This study concluded that the challenge is in recruiting fake news and actively countering it rather than ignoring it. Similarly, university students are a relevant population group regarding health behaviors and risk perception. They are more likely to engage in risky behaviors due to their particular characteristics, such as being young people, their socialization in groups, and feeling invulnerable. Although university students are less likely to become ill, adherence to preventive behaviors is critical in this group as they can spread the disease to the population due to person-to-person transmission [28]. We cannot forget that health behaviors and the risk perception of a population in the face of an epidemic are essential when authorities design an action strategy. Indeed, the WHO is keenly aware of how important meaningful youth engagement is to the COVID-19 pandemic. They have an important role to play both in terms of helping to reduce transmission and in engaging in the response [29]. It has been emphasized that HL plays an important role in pandemic control, and the need to take it into consideration in public health messages to reach everybody in the fight against the virus [2,[30][31][32]. In fact, the COVID-19 pandemic has highlighted that poor HL is a globally underestimated public health problem [9]. In addition, studies to date suggest a positive association between HL, COVID-19-related knowledge, and preventive behaviors [14,17,[33][34][35][36][37][38]. People have difficulty evaluating the reliability of the information on COVID-19, and a population with limited HL is more likely to feel confused due to a large amount of information available in the media and the internet [2]. In the European context, HL has been defined as "the knowledge, motivation and competencies to access, understand, appraise, and apply information, to make judgments and decisions in terms of healthcare, disease prevention and healthy behaviors, to maintain and promote quality of life throughout the life course" [39]. This definition is based on a conceptual model that combines four dimensions referring to health information processing with three levels of domains that yield a matrix with 12 dimensions of health [39]. There is evidence that HL predicted health-protective behaviors in university students [12,34,[40][41][42]. In addition, HL is context-dependent. This means people with generally acceptable HL skills can still face HL challenges in some contexts, such as a pandemic. Therefore, regarding the infodemic, a comprehensive study into the HL levels of the target populations to determine the best strategies to improve it should be conducted [43]. On the other hand, according to the health belief model [44] , behaviors are the result of a set of beliefs and internal valuations that an individual uses in a particular situation. The desire to avoid illness and the belief that a specific healthy behavior can prevent illness would be the basis of the preventive behavior. Thereby, perceived susceptibility, perceived severity, perceived benefits, and perceived barriers are the dimensions of the health belief model [45]. The perceived susceptibility to a certain health problem fundamentally refers to the subjective perception that each human being has about the risk of falling ill. The perceived severity refers to beliefs about the seriousness of contracting a certain disease or leaving it untreated once contracted. Studies show that perceived susceptibility and perceived severity are significant determinants of health behaviors [46][47][48][49][50] and therefore affect the adoption of preventive behaviors. The perceived benefits are considered as the relative effectiveness that the different behaviors available in their repertoire may have when dealing with the disease; various studies show that perceived benefit can be a predictor variable of preventive behavior [51][52][53][54][55]. And the perceived barriers are those that oppose the execution of the behavior by the individual; different studies revealed that perceived barriers have a significant negative impact on COVID-19 preventive behavior [46,[55][56][57][58]. Therefore, understanding the risk perception of the university population and the sources of information that they trust is essential for allowing effective communication. Similarly, optimal HL in academic settings enables people to use reliable health information and empowers them to adopt preventive behaviors, helping to curb the transmission of SARS-CoV-2 by guiding idiosyncratic policies [29]. On the other hand, according to the health belief model [44] , behaviors are the result of a set of beliefs and internal valuations that an individual uses in a particular situation. The desire to avoid illness and the belief that a specific healthy behavior can prevent illness would be the basis of the preventive behavior. Thereby, perceived susceptibility, perceived severity, perceived benefits, and perceived barriers are the dimensions of the health belief model [45]. The perceived susceptibility to a certain health problem fundamentally refers to the subjective perception that each human being has about the risk of falling ill. The perceived severity refers to beliefs about the seriousness of contracting a certain disease or leaving it untreated once contracted. Studies show that perceived susceptibility and perceived severity are significant determinants of health behaviors [46][47][48][49][50] and therefore affect the adoption of preventive behaviors. The perceived benefits are considered as the relative effectiveness that the different behaviors available in their repertoire may have when dealing with the disease; various studies show that perceived benefit can be a predictor variable of preventive behavior [51][52][53][54][55]. And the perceived barriers are those that oppose the execution of the be- [39]. The frameworks associated with the three domains represent a progression from an individual towards a population level. However, despite the efforts made to assess COVID-19-related HL, there is little evidence of its determinants. There are population-based studies in different countries, such as the ones carried out in Spain [59], Germany [60], Turkey [61], and Canada [62]; some studies have analyzed the knowledge and attitudes of students from Indonesia [63], Denmark [18], Pakistan [40,64], Egypt [65], and Jordan [66]. However, to our knowledge, no study has examined the association between HL and COVID-19 preventive behaviors in Spanish university students. Therefore, this study aims to evaluate the level of HL related to COVID-19 , the risk perception, the misinformation, and the attitudes and behaviors adopted to prevent the spread of coronavirus in Spanish university students. --- Materials and Methods --- Desing and Sampling A cross-sectional descriptive study. The sample consisted of 499 Spanish university students through a nonprobabilistic convenience sample. The inclusion criteria were students who are currently enrolled in a university degree, master's degree, or doctorate in Spain and who voluntarily want to participate. The university population was invited to answer an online "Google Forms" questionnaire distributed via WhatsApp and social networks. --- Variables and Instruments The questionnaire used in the COSMO Spain study [61], the health literacy questionnaire related COVID-19 , was adapted and prepared for online distribution. It was based on the survey tool developed by the Regional Office for Europe of the WHO [59] and assessed the participants' self-perception. The initial questionnaire was sent to 5 people as a pilot before the link was sent to participants. It consists of 9 sections: Sociodemographic characteristics: The sociodemographic variables collected were age, gender, size of the population of residence , degree, and current academic year. Health Literacy Questionnaire related to COVID-19: HLCQ was adapted from the items included in the COSMO-WHO [67,68]. The questionnaire items, originally in English, were translated by professional translators, reviewed, validated, and slightly modified by the COSMO-Spain team [59,69]. The scale showed a Cronbach's alpha of 0.87, and the item total corrected correlation was 0.49-0.68 [69]. The questionnaire, which assessed citizens' perceived difficulty in accessing, understanding, evaluating, and applying health information follows the structure of the European health literacy survey questionnaire [70]. In the present study, a 9-item version was used. As can be seen in Supplementary Materials S1, the instrument begins with a general question followed by specific statements so that participants rate their perceived difficulty on a four-point scale: very difficult, difficult, easy, and very easy. In accordance with the original questionnaire, the HLCQ index was standardized from 0 to 50 [index = × ] using the mean of all items for each respondent. HLCQ scores are classified as "inadequate" , "problematic" , or "sufficient" [17,[71][72][73]. Information sources: The frequency and sources of information to obtain information about COVID-19 were assessed. The primary sources of information that were included were the news, radio and TV talk shows, press conferences, national press, social media, the internet, the Ministry of Health, the Consumer Affairs and Social Welfare website , and the WHO. The frequency of use of each of these was assessed on a scale from 1 to 5 . Confusion about the veracity of information: To assess the confusion about the veracity of the information, the participants were asked if they had received any information/news about COVID-19 that they had had difficulty distinguishing as to whether it was true or false. Knowledge about COVID-19: Knowledge was assessed by 13 items about the transmission and prevention of COVID-19. The first 6 questions include the possible transmission routes: droplets when coughing/talking , contact with contaminated surfaces , physical contact with someone infected , blood transfusion , insect bite , and contact with pets . They were also asked whether they believed that people without a fever can be contagious. The last 6 items focused on the beliefs of university students about the correct use of masks: washing hands before and after use, whether it should cover the nose and mouth, whether it should be touched only by the ear band, whether it should be removed for coughing or sneezing, and whether its function is to avoid infecting others and to protect oneself from being infected. Risk perception: Taking into account the health belief model to evaluate perceived risk related to COVID-19, 3 questions were asked: perceived severity: "How severe do you think the disease would be if you were infected with coronavirus?". Possible responses for this item are "very mild, mild", "intermediate", or "severe, very severe"; perceived susceptibility: "How likely do you think you are to get the coronavirus?" Responses range from 1 to 5 , and perceived invulnerability: "As things currently stand, managing to avoid getting infected with the coronavirus is...". The response is on a Likert scale from 1 to 5, with 1 being "very difficult" and 5 being "very easy". Preventive measures: Preventive measures included in the questionnaire were washing hands with soap and water frequently, using hand sanitizer or other disinfectants, not going to social/family gatherings, staying at home in case of symptoms, keeping a physical distance , using the mask as recommended, disinfecting surfaces, avoiding public transport, not going out of the house, and going out as little as possible. Participants had to select the measures taken in the last 7 days to prevent the transmission of COVID-19. Fear: The fear feeling produced by COVID-19 was evaluated through 3 questions. The feelings produced by the coronavirus were assessed on a scale of 1 to 5 , and the last two questions included concerns, asking how much you are worried about the coronavirus in general and how much you are worried about from 1 to 5 several specific aspects . Level of trust in information: The last question was related to students' trust in information from news programs, radio and TV talk shows, press conferences, national press, health professionals, conversations with family/friends, social networks, the internet, the MHCSW, the WHO, and COVID-19 hotlines. Each of these items was measured on a scale from 1 to 5 . --- Data Collection The data were collected through the online questionnaire "Google Forms" between 28 January and 28 February 2021, during the so-called "third wave" of the COVID-19 pandemic and with most university students in a non-face-to-face mode. During these dates, mobility restrictions and limitations on opening hours and seating capacity in commercial establishments in Spain were maintained. The use of face masks was mandatory for people aged 6 years or older in enclosed spaces and outdoors. The government opened all internal borders among autonomous communities as well as international travel restrictions between other European Union countries and the United Kingdom. Other restrictions related to mass gatherings and the closure of public spaces were handled by each autonomous community independently. --- Statistical Analysis A descriptive analysis was carried out to determine the sample distribution for each of the variables studied: sociodemographic variables, HL related to COVID-19, information sources, confusion over the veracity of information, knowledge about COVID-19, preventive measures, risk perception, fear, and level of trust in information. Prior to the analysis, the normality of the variables was studied through the Kolmogórov-Smirnov and/or Shapiro-Wilk test, depending on the sample size, and Levene's test was used to assess the homogeneity of variance. Quantitative variables are expressed in terms of summary and dispersion , and categorical variables in frequency and percentages. According to a study on the normality of the variables, the Mann-Whitney U test and the one-way analysis of variance were used to compare the mean difference between two groups or more. In the case of linear relationships, the Spearman's correlation coefficient was calculated. Statistical significance was determined for a value of p < 0.05. The statistical treatment of the data was carried out with the SPSS statistical package, version 23.0. --- Ethical Considerations This work was conducted in accordance with the Declaration of Helsinki [74]. The aim of the study and the anonymity of participants, as well as the voluntary character of their participation were all explained before the participants started answering the questionnaire and before their informed consent was obtained. The participants were informed that the data obtained would be used only for research purposes. --- Results --- Sociodemograpfic Profile The sample consisted of 499 university students, and the age of the students ranged between 18 and 36 years, with a mean age of 21.50 years . Most of the participants were students from the Health Sciences followed by Social and Legal Sciences . On the other hand, the percentage of students in the Sciences , Engineering and Architecture , and Arts and Humanities was low. Most students were enrolled in the second and fourth years of their university degrees . --- Confusion about the Veracity of Information A total of 78.4% of university students had difficulty distinguishing whether a news item or information about COVID-19 was true or false. If we divide the sample between students from the Health Sciences and those enrolled in other different degrees, the results show that, for the first group, the percentage was 81.5% , and for students of the rest of the degrees it was 75.2% . --- Risk Perception The majority of university students think that, if they were infected with coronavirus, the illness would be "very mild/mild" or "intermediate" . While only 3.6% think it would be "severe, very severe". These results are the same if we divide the sample between students in the Health Sciences and those who are enrolled in other different degrees. For the first group, 54.2% consider that, if they were infected with coronavirus, the illness would be "intermediate", and in the case of the nonhealth degrees, it was 54.4%. The mean score for perceived susceptibility to infection was 3.28 out of 5 points , and for perceived invulnerability was 2.62 . Likewise, among the students belonging to the Health Sciences, the perceived susceptibility was 3.31 , and in the rest of the students, it was 3.26 . As for the perceived invulnerability among students belonging to Health Sciences versus those belonging to other degrees, it was 2.54 and 2.70 , respectively. --- Preventive Measures Most of the students had carried out almost all of the preventive measures in the last seven days to avoid COVID-19 infection; the mean of the total number of preventive measures taken was 7.54 out of 9 points . The measures that were adopted by most of the students were the following. Using a mask as recommended , staying at home if they have symptoms , using hand sanitizer or other disinfectants to wash their hands , washing hands with soap and water , keeping a physical distance , avoiding public transport , avoiding going out, and going out as little as possible or teleworking . On the other hand, the measures least carried out by the students were the following. Avoiding going to social/family gatherings and disinfecting surfaces , with the latter being the measure that the lowest number of students carried out. Among the different groups according to the university degree being studied, some differences were found; among the students belonging to the Health Sciences, the measures most adopted were the use of a mask as recommended , the use of hand sanitizer , and staying at home if they had symptoms, and washing hands with soap and water ; in contrast, among the students from other university degrees, the order was the following. The use of a mask as recommended , staying at home if they had symptoms , the use of hand sanitizer , and washing hands with soap and water . --- Knowledge about COVID-19 The mean score was 10.40 out of 13 points . Statistical differences were observed between the groups according to the university degrees in which the students were enrolled . Most of the students had an optimal level of knowledge about COVID-19. However, many had misconceptions about some of the routes of transmission of the coronavirus. In particular, they incorrectly believed that transmission occurred via physical contact with someone infected , as well as with contaminated surfaces , and blood transfusion . The remaining items regarding the transmission routes were correct by more than 90%. Regarding mask use, the majority had adequate knowledge, with more than 93% of people answering every item correctly. The item that obtained a lower percentage for correct answers referred to masks being used to protect against infection. --- Health Literacy Related to COVID-19 The mean HLC index score of the total sample was 31.59 , which corresponds to a problematic HLC level. A total of 36.3% of the respondents had a sufficient HLC level, followed by those who had a problematic HLC level , and the remaining 30.7% had an inadequate level of HLC. In this case, the mean HLC index score of the Health Sciences students was significantly higher than that of the other degree programs . Participants found the following items challenging to understand: what authorities say about the coronavirus, understanding restrictions, and recommendations, assessing whether media information about the coronavirus is reliable, assessing when they need to go to the doctor for a problem unrelated to the coronavirus, and learning about restrictions related to the coronavirus. The mean scores for these statements ranged from 2.24 to 2.84. --- Correlation between HLC and Other Variables Concerning the sociodemographic characteristics, no significant correlation was found between HLC and age . No significant differences were observed for sex , size of the population in which they reside , course , confusion in the veracity of information , or with perceived severity . Although there were no significant differences between the HLC index and gender, the level of inadequate HLC was significantly more frequent in women . The results of the HLC index were significantly different according to university degree . The mean HLC index was higher for the Sciences . HL was not significantly correlated with perceived susceptibility, invulnerability, or consulting information on social media and talk shows . In contrast, there was a significant correlation with frequently consulting information from television news , press conferences , national press , the Internet , the website of the MHCSW and the WHO . Students who consulted information about COVID-19 on the Internet, the MSCBS website, and the WHO more frequently had a higher mean HLC index than those who consulted these sources less frequently . Regarding the different preventive measures, there were significant differences in the HLC index according to the following measures : not going to social or family gatherings , keeping a physical distance , using the mask following the recommendations , not going out of the house, and going out as little as possible or teleworking . Students who carried out these measures had a higher HLC index. --- Correlation between the Preventive Measures Taken and the Different Variables The total preventive measures were significantly associated with university degrees. Health Science and Science students were the ones who practiced the most preventive measures ( x = 7.83 vs. x = 7.45), and Engineering and Architecture students were the ones who adopted the least preventive measures . Adopting preventive measures was significantly correlated with fear of COVID-19 . The more fearful they were of the disease, the more preventive measures they took. The mean number of preventive measures for the highest level of fear was 8.22; for the lowest, it was 6.15 on average. Even though during the third wave of the pandemic, the preventive measure of disinfecting surfaces was dismissed, we observed a statistically significant relationship between fear of COVID-19 and the use of these measures . Similarly, a statistically significant relationship was found between feelings of concern and the implementation of this preventive measure . As for the relationship between preventive measures and perceived risks, it is significantly related to the perceived severity of contagion , and a significant correlation is observed with perceived susceptibility , but no significant correlation is found with perceived invulnerability . The higher the severity and perceived susceptibility, the more preventive measures students will take. Specifically, university students who consider the illness mild or very mild took an average of 7.32 out of 9 preventive measures, compared to an average of 7.75 for those who consider the illness severe or very severe. The average number of preventive measures taken by students who think they are not at all likely to catch the disease was 7, while the average number of preventive measures taken by those who thought it was very likely to catch the disease was 7.8. --- Correlation between Knowledge and the Different Variables The level of knowledge related to COVID-19 was significantly associated with washing hands with soap and water and not going to social gatherings . Students who reported washing their hands with soap and water achieved a higher level of knowledge than those who did not wash their hands ; additionally, those who did not go to social gatherings had a higher score than those who did . Finally, no significant correlation could be established between the level of knowledge and the total number of preventive measures adopted or the HLC index. However, we did find a correlation between the HLC index and the total number of preventive measures taken . The higher the HLC index, the more preventive measures were practiced. University students who took only three preventive measures had a mean HLC of 21.76, while those who took all measures had a mean HLC of 33.13. --- Discussion To our knowledge, few studies have investigated COVID-19-related HL and the adoption of preventive measures in university students. This study aimed to evaluate the level of HL related to COVID-19 , the risk perception, the misinformation, and the attitudes and behaviors adopted to prevent the spread of the coronavirus among Spanish university students. The results showed that a high number of preventive measures were taken , and the level of knowledge was moderate among the participants. On the other hand, the HL regarding COVID-19 among university students was not adequate, as the mean HLC index did not reach the level of HLC "sufficient". This result is similar to the one obtained in the COSMO-Spain project with a representative sample of the Spanish population [59], where the mean index score was 33.89, and more than 60% did not have adequate levels of HL, akin to the 63.8% in the present study. These findings also seem to agree with the results of the study by Yuan et al. [73], who reported that 45.9% of university students did not have adequate levels of HL. Furthermore, Okan et al. [17] found that, in the general population in Germany, just over half of the sample had inadequate COVID-19-related HL. In all three of the studies mentioned, the level of HL is higher than the score in our study. However, in the study by Shaukat & Naveed [40], the proportion of university students with inadequate HLC levels was even higher than in the present study, as they found that 81.55% of university students had low to moderate levels of HL related to COVID-19. These findings were quite surprising and worrying because if university students belonging to a population group with higher education do not have sufficient HL, we should ask ourselves whether individuals with less education, the illiterate, and the rural population might have even lower levels of HLC. On the other hand, we are also concerned about the consequences for a population that will be more prone to risk-taking. Inadequate HL among university students may be due to inappropriate sources of information. The most commonly used sources among the participants are the internet and social networks, with the attendant risk of finding unreliable information [22][23][24][25][26]. Moreover, more official and, therefore, more reliable sources, such as the MSCBS website and the WHO, were the least consulted. Students who frequently consulted the MSCBS website and the WHO for information about COVID-19 had a higher HLC score. Similarly, in a study by Rosário et al. [74], it was found that the frequent use of public agency websites increased the likelihood of sufficient digital health literacy. However, it is also true that they had little trust in social media and the internet and relied more on information from health workers, the MSCBS, and the WHO. Nonetheless, 78.4% of the university students had difficulty distinguishing whether a news item or information piece about COVID-19 was true or false, presenting higher results than those obtained in other studies [18,75,76]. Okan et al. [17] found that 56% of the general German population was confused about COVID-19 information, showing more confusion among university students than in the present study. This may be due to university students' frequent use of unreliable sources and their low trust in them. The difficulty for a high percentage of the sample to identify misinformation about COVID-19 indicates the need to support students by providing high-quality information that allows them to assess the reliability of the source and counter misinformation about COVID-19 on social media. The mean HLC scores show that the biggest challenges for the students were understanding what the authorities said about coronavirus, understanding the restrictions and recommendations, assessing whether the information given by the media about coronavirus was reliable, assessing when they needed to go to a doctor for a problem unrelated to coronavirus, and learning about the restrictions related to COVID-19. These difficulties coincided with those obtained in round 4 of the COSMO-Spain project [59]. The majority of students found it difficult or very difficult to assess whether the information about COVID-19 provided by the media was reliable, obtaining higher results than the ones in the studies by Okan [17] and Dadaczynski et al. [75], where 47.8% and 42.3% of respondents, respectively, reported that it was difficult or complicated to as-sess the reliability of the information provided by the media. When students access misinformation or false information and have difficulty making judgments about the veracity of the information, they are unlikely to identify that information as "fake news", and this may prevent engagement in effective health behaviors. These findings suggest the need to implement health education measures to strengthen students' HL skills [77]. Age and the HLC index did not correlate significantly in our study, as the sample has a specific age range with a minimal margin. The association with the rest of the sociodemographic data was not significant, except for with university degree. The HLC index was higher in the Health Sciences, Sciences, and Engineering and Architecture degrees. In the Social and Legal Sciences and Arts and Humanities, more emphasis should be placed on health education. Although there was no significant association between the HLC index and gender, the "inadequate" HL level was significantly more frequent in women . These results can be explained by the attendance of mostly women in the sample. Students with higher levels of HLC took more preventive measures. The same correlation was found in other studies [13,16,73,[78][79][80], where the higher the level of health literacy, the more preventive behaviors were carried out. In general, students reported adequate prevention measures. Those who did not attend social or family gatherings, kept their distance, wore masks as recommended, and left the house as little as possible had a higher index of HLC. The results of the study by Falcón et al. [69] compared to the results of the present study showed a lower frequency in the adoption of preventive measures in participants from the general Spanish population. This difference stands out for measures such as staying at home if presenting symptoms , avoiding public transport , and not attending social gatherings . In other words, it seems that university students are more aware of adopting preventive measures than the general Spanish population. This difference can also be explained by the fact that the data collection was carried out when the said population was more relaxed about taking precautions against COVID-19 due to the arrival of summer. The high number of preventive measures taken among the participants could also be explained by the fact that most of them study the Health Sciences; students from this area have more health knowledge, as well as the possible experience of the consequences of the pandemic during clinical practices. According to Shaukat et al. [34,40], the present study showed that, as fear and concern of COVID-19 increases, the number of preventive measures practiced increases . Regarding risk perception, the higher the susceptibility and perceived severity of the university students , the more preventive measures they follow; similar findings have been found in numerous studies [46][47][48][49][50]. Less than 4% of the students considered COVID-19 to be serious or very serious. This was very low compared to the percentage obtained in the fourth round of the COSMO-Spain project [59]. In contrast, the average score for susceptibility and perceived invulnerability was approximately 3 points out of 5 in both cases. In the case of perceived susceptibility, the results tended more towards being likely to become infected, while invulnerability tended towards the consideration that infection was easy to avoid. The latter could be explained by the aforementioned feeling of invulnerability of young people, the low mortality in this age range, and their knowledge about preventive measures. Paradoxically, there is no significant correlation between the HLC index and the level of knowledge. This may be because they are all university students, which a highly educated population group, who have been provided with much information about COVID-19 by public administrations and educational centers. There was also no correlation between the level of knowledge and the number of preventive measures taken. However, students who reported washing their hands with soap and water and those who did not attend social gatherings achieved a significantly higher level of knowledge. Finally, the mean knowledge score was similar to that in Yuan's study [73]: 14.68 out of 21, or 9.9 out of 13, compared to 10.40 out of 13 in the present study. Despite being a sample consisting of nursing students, they showed less knowledge. This may be because the questionnaire date was almost a year earlier . Regarding mask knowledge, the majority of the participants had adequate knowledge about the usefulness and the correct use of the mask. However, 16% did not know the mask was used to protect them from infection. This was lower than in round 1 of the COSMO-Spain project [59], where 30% were unaware of the usefulness of the mask. Therefore, we can see that the population is becoming more aware of the use of masks over time. --- Limitations and Strengths This study is not without its limitations. In addition to the limitations inherent to a cross-sectional study, the sample was chosen by nonprobabilistic convenience sampling. A convenience sample can lead to the over-representation of particular groups within the sample. Considering that HLC is context-dependent, that is to say, it may vary in different settings, generalizing the results should be carried out with caution. In addition, the data collection was conducted during the "third wave" of COVID-19, when limitations and more restrictive measures were maintained due to the spike in cases resulting from the Christmas holidays. After enduring restrictions for long periods and the emergence of contradictory information on COVID-19, the adherence to prevention measures and COVID-19-related HL may have been compromised. Despite the limitations encountered, our study has strengths to be considered. Thus, it is one of the first studies to measure HLC in university students, providing relevant data on the levels of HLC, risk perception, misinformation, and the attitudes and behaviors adopted to prevent the spread of coronavirus in Spanish university students. Given the difficulty of data collection due to restrictions, a sizeable sample was drawn to overcome the shortcomings of the nonprobabilistic sampling method of convenience. --- Conclusions The level of HL in the surveyed university students is primarily inadequate and problematic. Despite this, they generally follow preventive measures, having more difficulty keeping social distance, avoiding social/family gatherings, going out as little as possible, and avoiding public transport. On the other hand, most of the surveyed university students use new technologies to gather information despite their distrust. The level of knowledge of most of the participants was moderate. They incorrectly believed that physical contact with someone infected, contaminated surfaces, and blood transfusion are all transmission routes. Efforts should be made to reach university students through their usual information channels, increasing the quality of health information available on the internet and social media, and implementing fact-checking strategies on social media, checking that all information is based on scientific evidence. Between our participants, the profile of university students who find it challenging to follow preventive measures is that of those who are least afraid of COVID-19 and consider that the disease is not severe and that they are not likely to contract it. The HLC index increases significantly when more preventive measures are taken in total. As the scientific literature has demonstrated, infodemia requires strengthening the HL skills of the general population to encourage their adherence to COVID-19 prevention behaviors, particularly with university students. This should include public policy strategies to address the toxic spread of misinformation and misinformation about COVID-19 [9,17]. In addition, university awareness is critical to assist with emergency responses and increase students' capacity to respond to the HL related to COVID-19. To this end, universities can offer courses on HLC and provide health information to their students and disseminate reliable news about COVID-19 through their websites and social media. --- Data Availability Statement: Not applicable. --- Acknowledgments: The authors thank all the university students who participated in the study. --- Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/ijerph192215370/s1. Questionnaire Health Literacy Questionnaire related to COVID-19 . Institutional Review Board Statement: This work was conducted in accordance with the Declaration of Helsinki. The aim of the study and the anonymity of participants, as well as the voluntary character of participation, were all explained before the participants started answering the questionnaire and their informed consent was obtained. The participants were also informed that the data obtained would be used for research purposes only. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. ---

The "infodemic" is one of the main obstacles in the fight against the COVID-19 pandemic. In order to face it, health literacy (HL) is essential since it allows for knowledge about COVID-19 and the practice of preventive measures to be fostered. This is especially relevant in university students due to their idiosyncrasy. This study aims to evaluate the level of HL related to COVID-19 (HLC), risk perception, misinformation, and the attitudes and behaviors adopted to prevent the spread of coronavirus in Spanish university students. An online questionnaire was administered to 499 Spanish university students. The HLC index presented a mean of 33.89 out of 50; a total of 63.8% had an inadequate level of HLC. They practiced a mean of 7.54 out of 9 preventive behaviors, and the mean knowledge score was 10.40 out of 13. The HLC showed significantly different scores for the degree, the practice of preventive measures, and some sources of information. The level of HL correlates with the adoption of preventive measures. The higher the severity and perceived susceptibility, the more preventive measures are taken by the students. Therefore, there is a need to strengthen the HL skills of university students and address the dissemination of misinformation. Although caution should be taken when generalizing these results due to the limitations inherent within a cross-sectional study and the convenience sampling, our results can guide the establishment of health education strategies and policies for the management of the infodemic in pandemic situations, according to this target population.

Introduction Globally, 3.3 million children under 15 years of age were living with human immunodeficiency virus infection at the end of 2012 . India has the third largest number of people living with HIV worldwide, and children under the age of 15 years comprise 7% of the estimated 2.1 million PLH in India . Previously, access to antiretroviral therapy was limited in India, and most HIV-positive children failed to survive into adulthood. Recent emphasis on early infant diagnosis and scaling up of the ART programme has led to drastic reductions in childhood and adolescent mortality due to HIV. However, improvements in paediatric HIV diagnosis, treatment, and longevity pose further challenges for health systems in low-and middle-income countries such as India, including improving and maintaining adherence to ART, reducing secondary transmission, and comprehensive improvement of the psychological and physical health of children living with HIV . Informing a CLH about his or her diagnosis is considered critical from the perspective of disease treatment and overall psychosocial development . The decision about disclosing HIV status to a CLH is one of the principal dilemmas for caregivers . As with other life-threatening diseases, disclosure of HIV status to a child or adolescent can be either partial or full. Partial disclosure may omit some of the disconcerting or stigmatizing aspects about HIV , whereas full disclosure involves discussing the specifics of HIV . However, recent guidelines suggest that disclosure to children should be a continuing process rather than all at once, so that complete disclosure occurs over a period of time and increases the amount of information to be shared with children as their mental faculties mature . Findings from cognitive research conducted in developed countries endorse implementation of a processoriented disclosure framework that promotes developmentally appropriate information sharing and takes into account emotional, social, and developmental needs of children and caregivers . Although there is still no consensus on the appropriate age of disclosure, guidelines from the World Health Organization and the American Academy of Pediatrics recommend that school-going children be made aware of their HIV status. However, these guidelines also suggest that efforts should be made to ensure a child's right to privacy and reduce possible discrimination that might arise from disclosure. Compared with Western countries, a much smaller proportion of CLH in low-and middleincome countries receive disclosure about their HIV status . A number of barriers to HIV disclosure in low-and middle-income countries have been recognized, including apprehension about discrimination , concerns about the impact on the child's psychological health , young age and/or perceived immaturity of the child , and caregivers being unprepared to face questions from the child . Prior research in low-and middle-income countries identified certain benefits for disclosure, including better adherence to treatment , higher self-esteem , and ability to cope with stigmatizing behaviour . However, some have noted a few negative impacts of disclosure, including rejection and blaming of the caregivers , difficulties in social interactions , and stigmatization . Our literature search revealed only two studies in India that specifically addressed HIV status disclosure to children . However, neither qualitatively explored the knowledge, attitudes, and practices of caregivers or various barriers/facilitators related to the process of disclosure. The aim of this study was to explore the perceptions of caregivers in West Bengal about HIV status disclosure to children and, in turn, to inform India's national HIV programme on formation of a framework for childhood disclosure. --- Methods --- Study setting Participants in this study were primary caregivers of CLH aged 8 to 15 years who were residing in the districts of Purba Medinipur and Paschim Medinipur in West Bengal. Access to the study population was gained with the help of a community-based organization , the Society for Positive Atmosphere and Related Support to HIV/AIDS , an organization composed of and managed by PLH and their friends, which has been working with children and adults living with HIV in rural and urban settings of West Bengal since 2000 . --- Participant recruitment The SPARSHA prepared a roster of its service recipients residing in the two study districts, and participants in this study were a convenience sample from that roster. Primary caregivers of CLH on this roster were contacted by outreach workers from SPARSHA, who used a recruitment script to briefly inform them about the study objectives and procedures and determine eligibility. Eligible caregivers who expressed willingness to participate were invited to come to the nearby SPARSHA field office for an interview. At the field office, a trained interviewer explained the study procedures in detail and obtained informed consent. If a participant wished to be interviewed in his or her home, an interviewer, accompanied by a SPARSHA outreach worker, visited their home by appointment for consent and interviews. Participants were assured that their decision about participation would not affect their relationship with SPARSHA or any other aspects of their lives. --- Ethical approval The study was approved by the Institutional Review Board of the University of California, Los Angeles, and the Institutional Ethics Committee of the National Institute of Cholera and Enteric Diseases , the collaborating research institute located in Kolkata, West Bengal. --- --- Interviews Between July and September 2014, caregivers responded to a 1-h in-depth interview with a study staff member. Interviews were conducted in a private room at the SPARSHA field office or in participants' homes. Permission to take written notes and digitally record the conversations was specifically obtained prior to initiating the interviews. Written notes were used to emphasize important points in the interview and record details about participants' characteristics, emotional expressions, body language, and so on, as well as for recording interviewers' comments. All the interviews were conducted in Bengali, the native language of the area. A semi-structured interview guide with broad questions was used and aided by open-ended probes to draw out information from participants. Questions/issues included in the guide were formulated from the review of relevant literature and probable themes that were anticipated to emerge on the basis of prior interactions with CLH and their caregivers. The sample size was determined according to the Green and Thorogood method, with a target of 40 interviews , or until thematic saturation was reached, whichever was lower. In total, 20 biological parents and 14 non-parent caregivers were interviewed. --- Data management and analysis Audio recordings of the interviews were transcribed verbatim into MS Word in Bengali. The notes taken during the interviews were used to facilitate transcription and to enhance the transcripts by insertion of relevant comments . These were also utilized later for content analysis. Electronic transcription for each interview was completed within 24 h of the interview. The transcripts were then reviewed for preliminary coding by the investigators. This process continued until the study team felt that theme saturation had been achieved . The interview transcripts were then imported into the ATLAS.TI 7.5 software package for coding and data analysis. We analysed the interview transcripts for emergent themes in accordance with the grounded theory principles . At the time of initial readings, an 'open coding' technique was used, followed by 'axial/thematic coding' during re-readings. The Bengali transcripts were used for the coding process, and only relevant quotations were later translated into English. Any disagreements regarding coding were resolved by discussion between study team members. Quotations and codes were grouped according to perceived themes for further comparison. --- Results --- Participant characteristics Major characteristics of the participants are presented in Table 1. Forty-eight eligible primary caregivers identified from the SPARSHA roster were approached for participation, and 34 agreed to be interviewed. The main reason for refusal was lack of time on the part of the caregivers. Among the 34 caregivers interviewed, 20 were biological parents , and the other 14 were grandfathers, grandmothers, uncles, or aunts. There was a single case where a male child orphaned by both parents was being cared for by a neighbour who was also HIV infected. Among participating caregivers, 17 were HIV positive, and among the CLH, only four were infected through blood transfusions for thalassemia care, and the rest reportedly were infected by vertical transmission. --- Emerging disclosure themes The broader themes that emerged from content analysis were perceived difficulties with disclosure, rationale for disclosure, reluctance of caregivers to disclose, preferred individuals to perform disclosure, and best times and places to disclose. We could not identify any important differences between perspectives of biological parents and nonparents. Table 2 provides participant quotes according to themes. --- Perceived difficulties with disclosure For many, disclosure of HIV status to CLH was likened to revealing a dark secret. Caregivers were afraid to face questions on the topic from the child and were often unsure about how to answer. Nevertheless, some caregivers expressed their perceptions and experiences of how external assistance could ease discussing HIV. Parents who were HIV positive mentioned various awareness programmes and interventions that helped them talk to their children. --- Rationale for disclosure Rationale for disclosure of HIV status varied among study participants. We grouped them as follows: perceived health benefits of disclosure for CLH and others; anticipation of caregiver's impending death; and fear of negative repercussions owing to late disclosure. --- Perceived health benefits of disclosure for CLH and others A commonly perceived benefit of disclosure was that knowledge about the disease might motivate CLH to maintain their own health, and many caregivers felt that awareness about HIV would improve their health-seeking behaviours and motivate them to take their medications regularly. Participants were also concerned about protecting others from becoming infected. Most said that it was the responsibility of PLH to avoid transmitting HIV to others, and caregivers of CLH should teach them how to avoid transmitting. They felt that such knowledge would prepare CLH for the future so that they could avoid further threats to their health and/or avoid transmission of HIV to others. --- Anticipation of caregiver's death Some of the caregivers who were HIV positive said that they often thought about their own deaths and worried about the care of their child after they died. They felt that timely disclosure might prepare the child to live a better and healthier life. Even HIV-uninfected but aged caregivers were concerned about the future of their CLH after their deaths. They were worried that their child would be at a disadvantage if their caregiver did not tell them about their disease and its implications before they died. --- Fear of negative consequences owing to late disclosure A few caregivers, mostly biological parents, were worried that if they did not start priming their children about their HIV status at an early age, there would be negative repercussions when the CLH would come to know about it later. Responses revealed that late disclosure might be uncomfortable for both the caregiver and the CLH. The caregiver might feel awkward about answering questions and the child might be traumatized. --- Reluctance of caregivers to disclose Among the 34 participating caregivers, 16 said that they did not want their CLH to know about their HIV status in the immediate future. These caregivers were apprehensive about potential negative consequences following disclosure. Contextual exploration revealed three main concerns, namely, distress caused by disclosure, apprehension of blame, and fear of discrimination. --- Distress caused by disclosure Among the reasons that caregivers are reluctant to disclose, the most common reasons given were the distress of the CLH and feelings of sorrow, hopelessness, and shame. These caregivers did not want their child to be burdened by stress due to disclosure. --- Apprehension of blame HIV-positive caregivers who were the biological parent of a CLH were concerned that the child might blame them, unless the child had been infected through a contaminated blood transfusion. --- Fear of discrimination Caregivers feared discrimination about being HIV positive, and that following disclosure, the CLH, owing to lack of cognitive maturity, might not completely understand the social implications of being HIV positive. Thus, they might fail to keep it confidential or inadvertently reveal HIV status, resulting in stigma and social isolation. --- Preferred individuals for performing disclosure Participants were asked who they considered to be most suitable for disclosing HIV status to the children under their care, which revealed two major sub-themes: identifying suitable individuals for disclosure and letting CLH discover their HIV status by themselves. --- Identifying suitable individuals to make the disclosure As indicated in Table 3, the majority of caregivers, either the biological parents or nonparents, felt that healthcare providers were best suited to disclose to CLH. The principal reason was that HCPs were much more well informed about the disease, and they would be able to explain it better. Some caregivers thought that hearing it from the HCPs would hurt the child less. Some felt that children would pay more attention to the advice of a doctor . Most of the caregivers who did not prefer an HCP felt that responsibility of disclosure should be on the caregivers themselves or other family members or friends. Although they recognized the difficulty associated with it, they reasoned that they understood their children better than anyone else and were better equipped to handle the disclosure process. Some of these caregivers also appeared to believe that it was solely the obligation of parents living with HIV to share the information with their children because it was no one else's business. There were some caregivers who said that they did not have any particular preference regarding the choice of disclosing persons. One non-parent caregiver suggested that even if there was a preference, that could change based on individual scenarios. While describing her ideal scenario of disclosure, a mother of a thalassemic son suggested that children would be more at ease with the revelation about their disease status if that information was shared by CLH who had a similar experience. She thought that such peerto-peer information sharing could also help in raising awareness. --- Letting CLH discover their HIV status by themselves A few caregivers said they would prefer to leave the onus of disclosure to the child under their care. They felt that their school-going children were getting a better education than them and were more exposed to the outside world. Their children, they thought, were capable of discovering their HIV status by themselves when they matured. Caregivers also suggested that the child learning by himself or herself could probably lead to a better understanding of the disease and its implications. --- Best times and places to disclose Another broad theme that emerged was caregivers' perceptions about the ideal scenario or circumstances for the disclosure process. The 'how' and 'when' of approaching disclosure had two sub-themes: suitable environment and way to initiate disclosure and appropriate age of the child. --- Suitable environment and ways to initiate disclosure Caregivers expressed their thoughts about how to initiate the disclosure process and the best environments for it. Most participants said that to make the process easier, they would talk to the child like a friend. Widowed mothers who were infected by their husbands and mothers whose children were infected from contaminated blood transfusions felt that blaming the father or contaminated blood might make it easier to talk about it. Opinions varied about the right environment for disclosure. An aunt of a CLH commented that it was better to undertake disclosure in the evening, as it was the time of the day when kids are more likely to be calmer. A mother said that her disclosure would be aided by television commercials on HIV awareness. --- Appropriate age of the child for disclosure Overall, the age of the child emerged as the most common and probably the most important factor affecting caregivers' decisions about disclosure. There was consensus among the caregivers who had yet to tell their child about HIV that disclosure should occur only after the child had reached a certain age, or when suitably mature. Many participants had an age range in mind rather than a specific age. Some felt that disclosure should be initiated when the child could understand about HIV/AIDS, without referring to any particular age . Some caregivers wanted to delay disclosure until the child reached marriageable age or might start seeking a partner. Caregivers were aware of the risk of inadvertent sexual transmission if a CLH was not aware of being HIV positive, and they wanted to disclose HIV status as a preventive measure when the children reached reproductive age. --- Discussion Disclosure of HIV status to CLH is an emotional process, for both the caregiver and the child. This is especially true for low-and middle-income nations such as India, where anticipation of stigma often prompts HIV-infected individuals to conceal their diagnosis . It has been reported that non-disclosure can undermine efforts to reduce discrimination, make the child secretive, reduce health-seeking behaviours, and even increase the possibility of accidental sexual transmission . Thus, the current consensus is, if done in an age-appropriate way and as a continuing process rather than a single interaction, disclosure is likely to be beneficial for both CLH and their caregivers . The findings of the current study reveal the complexity of disclosure of HIV status to children. In terms of rationale pertaining to disclosure, a number of perceived benefits were cited by caregivers, and they intended to obtain those benefits for themselves and their children through disclosure. Improved treatment adherence was an important perceived benefit that motivated caregivers for disclosure. This is similar to findings from other lowand middle-income countries . However, the association between children's awareness about their own disease status and medication adherence has been reported to be inconsistent . As reported in studies from Asian countries , other positive goals of disclosure included willingness to protect healthy individuals from getting infected and avoiding damage to caregiver-child relationships owing to late disclosure. Similar to a study in China , caregivers who anticipated dying in the near future felt that it was their responsibility to disclose to their child to better prepare him or her for the future. As has been suggested by others , caregivers probably assume that in the event of their own death, they may miss the chance to communicate with their children, which might deny the child's potential sources of future support. Another dimension of our findings was related to emotional and cognitive barriers that might prevent caregivers from supporting disclosure, such as perceived immaturity of the child that makes him or her unable to understand the information shared about HIV. However, this indicates that caregivers probably perceived disclosure as a single event rather than an ongoing and evolving process. As has been suggested in the literature, the results of onetime disclosures are often unsatisfactory, as a child needs to adapt to the new information and adjust his or her understanding all at once, often leading to confusion, isolation, and depression . Thus, any programmatic intervention on systematic disclosure to children should be recommended as a continuous process rather than a single event. There was no consensus among participants regarding the appropriate age of disclosure, similar to what was suggested by a recent review . Overall, in the current study, biological parents were found to favour earlier disclosure than non-parents. Although a prior study in India reported that mid-teens was the preferred age for disclosure, we did not find any age group that was clearly favoured by the caregivers in our study. Interestingly, the age of the child was perceived as both a facilitator and a barrier for disclosure. Some caregivers, mostly biological parents, reasoned that children should be primed about their disease status from an early age, as late disclosure might lead them to blame their parents. On the other hand, as mentioned earlier, many considered immaturity to be a barrier for HIV status disclosure. Parent-child communication in India tends to be less open and direct than in Western countries , and it is unlikely to be different for a sensitive issue such as HIV. Thus, it is essential to develop context-specific interventions that endorse communication about the disease, while taking into account the child's age or developmental maturity, domestic environment, and cultural factors, as has been suggested by Wiener et al. . To facilitate the communication process, studies from other lowand middle-income countries have proposed that any such planned interventions need to reflect the nuances of the communication process within families . As the age of the child has been reported to be an essential element, formulation of age-and culture-sensitive guidelines on the nature and content of communication is an absolute necessity . Caregivers expressed numerous personal and societal fears as hindrances for confiding to the child. On a personal level, they feared that disclosure might cause psychological stress to the child, but this has not been substantiated in the published literature. It has been suggested that although children might suffer initially after being informed about their HIV status, possibilities of any long-term psychological adverse effects, such as depression and behavioural problems, are not greater for CLH who have undergone disclosure than those who were yet to be informed . On the other hand, some HIV-positive parents reported their concern that they might be subjected to emotional backlash from the child, who may end up blaming his or her parents for his or her condition. Many parents thus considered themselves guilty and expressed their inability to discuss HIV with their children out of this fear of repercussions. As has been suggested by a study of caregivers of CLH in Singapore , self-blame might be more pronounced in Asia, where parents are expected to assume a central and protective role in the family. We found that some widowed mothers wanted to blame the child's father, possibly to avoid self-blame and/or being blamed by the child. Thus, it is important to implement interventions that can help caregivers come to terms with their own HIV status and their relationships with their children, in order to improve the likelihood of disclosure to children. There was also widespread concern that the caregiver and/or child might be subjected to discrimination if the child, owing to his or her lack of understanding of the implications, revealed his or her HIV status to others. This finding was in agreement with reports from various studies conducted in different settings . It has been suggested in other studies conducted in Asia that caregivers often feared that they might lose their reputation or social status if their child inadvertently told others about their diagnosis . We maintain, as has been recommended by others, that national HIV programmes need to address apprehensions about discriminatory behaviour and psychosocial impact to improve the acceptability of childhood disclosure . Interview excerpts revealed that HCPs were suggested by most caregivers to be their preferred choice as the disclosing person, which is in contrast to previous findings from India and China that reported parents/caregivers as primary choices. The principal reason for this choice was that HCPs were perceived to have greater knowledge about HIV that better equipped them to perform disclosure. It can be reasoned that the low educational level of our study participants, with about 90% receiving only primary schooling or less, led them to this 'disease-centric' choice for the disclosing person. Informal caregivers were favoured as disclosing persons by those who felt that they 'understood their child best', and HIV-positive caregivers who considered sharing disease status and related information as their duty. Some caregivers wanted to avoid disclosure and leave their child to learn about their HIV status by themselves. The policy gap on the issue of HCPs' assistance to informal caregivers about disclosure has been identified in many studies, according to a recent review on childhood disclosure . We suggest that this be addressed through future operational research. Being prompted by the varied context-specific findings revealed in this research, we propose that the ongoing HIV prevention and care programme in India should provide multi-component strategies for disclosure of HIV status to children. Several factors might limit interpretation of the findings in this study. Although we interviewed 34 primary caregivers , which may be considered adequate, based on the broader categorization , some demographic categories could be considered under-represented. Among the parent caregivers, only one of the 20 parents was the father , making mothers' perspectives dominant in our findings. However, this was not unexpected, as local cultural beliefs dictate that it is the mother's duty to look after her children. In addition, non-parent caregivers participating in this study were a heterogeneous population, with representation from a group composed of grandparents, uncles, aunts, and even neighbours. As the number of interviews from each group was small, it could not be determined whether opinions varied among different groups of non-parent caregivers. Moreover, in terms of broader representativeness, the participants were recruited through a single CBO that provides various preventative and awareness services to adults and CLH. In the Indian setting, recipients of such services mostly belong to the lower socioeconomic strata of society, as indicated by the low educational levels of the participants. Thus, we cannot assume that our study participants were representative of the caregivers of CLH from all strata of society. Capturing the perceptions of both parent and non-parent caregivers was a major strength of this study, as prior studies in Asia were often restricted to parents only . The caregiver's own health-related conditions would affect his or her perceptions about the child's quality of life. Interviewing both biological parents and non-parents helped us explore differing concerns that these caregivers had about the CLH under their care. Not all caregivers participating in this study cared for children who were perinatally infected, as some had been infected through contaminated blood transfusions. Thus, it was possible to record the disclosure-related perceptions of a varied group of caregivers. Additionally, CLH and their caregivers were recruited from the community through a local CBO, and interviews were conducted in an informal setting, in either homes or the field office of the CBO. As these events occurred away from hospital or treatment settings, we believe that the interviewees were more at ease because they did not fear any negative consequences to treatment or discriminatory behaviour based on their participation. Despite these limitations, the current research, to the best of our knowledge, is the first study in India to qualitatively explore caregivers' perceptions about diagnosis disclosure to CLH and may thus inform policymaking. Several organizations such as WHO and Medecins Sans Frontieres have suggested frameworks on disclosure of HIV status to CLH . There are also guidelines from different countries with culturally appropriate recommendations regarding the process . However, as in other low-and middle-income countries , there are few context-specific paradigms in India, and a standardized, age-appropriate procedure for disclosure is rarely followed. This study is in agreement with the existing literature that any intervention for childhood disclosure of HIV demands a multifaceted approach and should consider not only the disease but also the entire family/social structure. --- Key Messages --- • Childhood disclosure of HIV status is a complex issue that depends on multiple factors. --- • Caregivers are often apprehensive about negative impacts from disclosure, such as distress, stigma, and damaged parent-child relationships. --- • Young age/immaturity of children is perceived by some caregivers as an obstacle for disclosure, while other caregivers preferred initiating the disclosure process at an early age. --- • Healthcare providers were preferred by caregivers to perform disclosure. --- • A culturally appropriate guideline on disclosure of HIV status to Indian children is highly recommended. HIV-positive mother of an 11-year-old boy -'In order to make the kids take the drugs [ART], one has to tell them. Once you get started on the drugs, you'll have to know it.' HIV-negative mother of a 14-year-old thalassemic boy -'I think I should inform my child about HIV….and explain to him about the preventive measures… so that he becomes conscious…and could protect himself and others.' HIV-positive neighbour of a 14-year-old both parent orphan boy -'I told him bluntly…. to get her medicines, tell her.' HIV-negative uncle of an 11-year-old girl -'It is better if the doctors tell the children about their HIV. The counsellors or nurses may also tell them. But the children will take the doctors seriously.' HIV-positive mother of an 8-year-old girl -'No, it's better that I say it. If anyone else tells her [her daughter] that you have the disease, however good he explains it… if anyone else tells her, I may not like it…she will not like it too. You know it's always better to hear the bitter truth from your own mother, rather than an outsider.' HIV-positive widow and mother of a 9-year-old girl -'I think when children grow up, it is the responsibility of the parents to tell, "see, we have such a condition." We should explain facts about living a healthy life. Take care that they do not transmit HIV to anyone else. That they should take care of themselves. It is the parents' responsibility.' HIV-positive neighbour caring for a 14-year-old boy orphaned by both parents -'I think parents should say it, whenever the time is right….depends on situation, overall mood etc……‥ Parents have an advantage as they are always around. Whenever they think that it's a suitable time…… that is if the child is told at that time it may cause less hurt, then they should [say it]……But in other cases, if parents are finding it difficult, they may have to resort to other options……………… Say counselling, or maybe tell their children through friends… depends on what you have at your disposal.' HIV-negative mother of an 8-year-old thalassemic boy -'My suggestion is that a [HIV-positive] kid may tell others [other HIV-positive children] that "I have got this disease, you should take such and such precautions, so that it doesn't get any worse." A kid should be able to share that with other kids…' Letting CLH discover their disease by HIV-positive mother of an 11-year-old boy -'When such kids grow up, they will know by themselves…‥ they --- Das et al. Page 20 Theme Sub-theme Representative quotes themselves will become aware. Nowadays kids study a lot, they will definitely become aware. All the kids who are on ART have become aware, my kids will know about it, too.' HIV-negative mother of an 8-year-old thalassemic boy -'When he asks, "Mother, what has happened to me? What disease do I have? How come you don't have it but I have?" He asks me again and again. I just tell him when you will grow old you will understand yourself.' --- Best times and places to disclose Suitable environment and way to initiate disclosure HIV-positive widow and mother of a 9-year-old girl -'While talking about it [disclosure related discussion] mother has to become a friend to her daughter. It doesn't matter whether it's a son or daughter, she has to interact like a friend while saying it. Do it when both of you are healthy, and maintain your cool during the process.' HIV-positive, separated from husband, mother of an 8-year-old boy -'One has to treat them like a friend and discuss about the disease. Then the child will easily understand and the mother would feel relaxed, too.' HIV-positive mother of an 11-year-old boy -'I'll tell him…. I have to take him into confidence first by saying that "I did not do it to you, it has come from your dad. You can't undo whatever has happened, but you need to undergo treatment." That's how it should be explained.' HIV-negative mother of an 8-year-old thalassemic boy -'When I'll tell him [about the disease] for the first time I'll say, "see you have got the disease from the blood that you received [for thalassemia]." He might ask, "how come I have the disease but you don't?" I'll reply that it came through blood, and nobody can help it.' HIV-negative aunt of a 14-year-old girl -'One should sweetly ask them to sit in the evening, and say that "you have this disease, try to take care of yourself." He will understand better if told in the evening. Your mind can be diverted by other things during the day, while you are more likely to be at peace in the evening.' HIV-positive mother of an 8-year-old girl -'The commercial that comes on TV… ‥ that's what I mean… [I told her] "you have the same disease, take care of yourself, eat properly." That's how I could make her understand easily.' --- Appropriate age of child HIV-positive widow and mother of a 9-year-old girl -'It is our [parent's] responsibility to disclose about the disease when they turn 12 years old, because at twelve, I think, she would be mature enough to understand.' HIV-positive widow and mother of a 15-year-old girl -'I will let her know when she is of marriageable age.' HIV-negative grandfather of an 8-year-old boy -'This disease…… when he is of age of 16-17-18, then there will be chance that he might transfer it to others. Then I'll tell him, "you can't do it, you should know about prevention." ' --- --- Author Manuscript Author Manuscript Das et al. --- ---

Introduction-Informing the children living with HIV (CLH) about their disease (disclosure) is important from the perspective of disease treatment and overall psychosocial development. There are no published studies that qualitatively explored HIV disclosure-related issues among CLH in India. Our aim was to provide insights into the perceptions of informal caregivers of CLH regarding childhood disclosure. Methods-Children were defined as those aged <16 years. In-depth interviews were conducted with 34 primary caregivers of CLH aged 8 to 15 years old who were residing in West Bengal, India. The participants were recruited with the help of a community-based organization that provides need-based services to people living with HIV. Results-We obtained caregivers' perspectives on the motivators and barriers of childhood disclosure. Health benefits such as medication adherence emerged as an important motivator, while distress caused by disclosure and potential for stigma were identified as barriers. Health care providers were the preferred disclosers for most caregivers, followed by the caregivers themselves. Some caregivers wanted their child to learn about his/her HIV status by him/herself. There was no consensus among the caregivers about the ideal age for disclosure. Many preferred to wait until the child attained maturity or was of marriageable age. Discussion-Disclosure of HIV status to children is an emotional issue, both for the caregiver and the child. Like most low-or middle-income countries, no standardized, age-appropriate disclosure guidelines exist in India. Our findings advocate adoption of a multi-faceted approach, including increased availability of social and familial support, for childhood HIV disclosure.

Background Childhood overweight and obesity is increasing rapidly in Sri Lanka. In 2016, the global prevalence of childhood obesity was 7.8% in boys and 5.6% in girls [1,2]. A recent survey among 5-18 year olds in urban Sri Lanka showed an obesity prevalence of 10.3% and overweight prevalence of 11.3% [3]. Obesity has a negative impact on all aspects of child and adolescent health, including their physical, psychological and social well-being. Lifestyle factors and metabolic complications associated with childhood obesity have been studied previously in Sri Lankan children [4][5][6], but there is a paucity of research on impact on their quality of life including physical and psychosocial well-being. Quality of life describes the subjective wellbeing of an individual centered on his/her perception of life, based-on expectations, goals and standards [7]. QoL is a multidimensional variable including physical, psychological and social aspects. Both generic and diseasespecific tools are available to assess health-related QoL [8]. Generic tools are especially useful for making comparisons between healthy and unhealthy individuals [8]. Assessment of QoL differs between children and adults, due to differing influence from family, school, and friends, and reduced ability of children to make significant changes to their context by themselves [9]. Thus special generic tools have been developed to measure QoL of children such as the Pediatric Quality of Life Inventory and Child Health Questionnaire , which assess children's quality of life from their own as well as their parents' perspective [8,10]. Previous studies from high income countries have shown that QoL of children with overweight and obesity are lower than children of normal weight, and that QoL deteriorates with increasing body mass index [11,12]. Indeed, one study revealed that QoL of severely obese children was comparable to children suffering from cancer [13]. However, despite childhood obesity being a rising problem in South Asia, there is a lack of studies on QoL from the region. Further, there are considerable socio-cultural differences such as high parental expectations and different societal perceptions on ideal body weight and lifestyle standards [14]. Thus QoL data from other regions cannot be directly extrapolated to South Asian children. Furthermore, previous researchers have highlighted the need of exploring potentially modifiable factors affecting QoL of children with overweight and obesity, to identify and implement strategies to improve QoL [11,15]. Therefore, we conducted this study to assess physical and psychosocial QoL in Sri Lankan children with overweight and obesity aged 8-12 years using a generic international QoL questionnaire and explore social factors including potentially modifiable factors associated with lower QoL. We hypothesized that physical and psychosocial QoL would decline with increasing BMI, and be lower in obese compared to overweight children. --- Methods --- Study design and setting This cross-sectional, observational study was conducted between July to August 2016 at the Professorial Unit specialized obesity and nutrition clinic of Lady Ridgeway Hospital, Colombo, the largest and main tertiary care pediatric referral center in Sri Lanka. Ethical clearance was obtained from Ethics Review Committee of the Faculty of Medicine, University of Colombo and Lady Ridgeway Hospital. Information sheets explaining purpose, risks and benefits of the study were provided, and informed written consent obtained from all participating families. --- Study population The study sample consisted of 110 children aged between 8 and 12-years with a body mass index >+1SD, presenting to the obesity/nutrition clinic during the study period. Exclusion criteria included major disability, chronic medical complications and receiving obesity-related interventions > 1 year. --- Data collection and measurements --- Quality of life Child Self-Report and Parent Proxy-Report forms of the PedsQL™ 4.0 Generic Core Scales [10], a validated international questionnaire [16,17] was used to assess QoL. Original English questionnaires were cross-culturally translated into Sinhalese as per standard linguistic validation guidelines. Forward translation was done by two independent translators and a single reconciled Sinhalese version was produced which was backtranslated into English and compared with the source instrument to make amendments to the forward translated version. Pilot testing was done using the respondent debriefing method, and confirmation of appropriate linguistic validation was obtained for the final translated questionnaire. The PedsQL questionnaire for 8-12 year old children is a self-administered questionnaire which contains 23items classified under four subscales . For each item, 5 options were provided: never a problem; almost never a problem; sometimes a problem; often a problem; or almost always a problem, and 0-4 marks allocated accordingly. Appropriate items were reverse scored and linearly transformed to a 0-100 scale. Higher scores indicate better QoL. Physical QoL was derived from physical functioning subscale, psychosocial QoL was derived from emotional, social and school functioning subscales and overall QoL was derived from all four subscales [10]. --- Socio-demographic data and factors associated with QoL An additional questionnaire was used to collect sociodemographic data and possible/perceived modifiable social factors influencing QoL of children with overweight/obesity with similar body habitus, being on diet control, obtaining adequate exercise, facing difficulties in finding suitable clothes that fit). Social factors were based on information gathered during prior group-counseling sessions conducted on random clinic days for children with overweight/ obesity at the clinic, by the last author in the year proceeding this study. The translated, self-administered PedsQL™ Child Self-Report questionnaire was provided to each child and the self-administered Parent Proxy-Report to the accompanying parent. Parents and children were seated separately, to facilitate children's independency. An investigator was present to clarify any doubts. The additional questionnaire was filled in collectively by parent and child. --- Anthropometric measurements Height of children was measured using a stadiometer and weight using a weighing scale complying with standard techniques. BMI was calculated using the standard equation; dividing weight in kilograms by height in meters squared . World health organization BMI z-score charts for age for girls and boys were used to categorize children as overweight and obese [18]. --- Data analysis As QoL data showed non-normal distribution, central tendency was described using median and inter quartile range , and analyzed by appropriate nonparametric tests, using SPSS statistical software. Childreported and Parent-reported QoL scores were compared using Wilcoxon test, while associations between the two were assessed by Spearman correlation. Childreported QoL was used for analysis of associations with BMI and social factors. Comparison between overweight and obese children , and influence of social factors were analyzed by the Mann-Whitney U test while associations with BMI were assessed by Spearman correlation. Significance level was set at 0.05 for all tests. Effect size of correlations was evaluated by Cohen's standards [19]. --- Results --- Socio-demographic characteristics One hundred and ten children participated in the study. Average BMI was 24.15± 2.85 kg/m 2 . Majority of the participating children were obese 77 with 8 being severely obese, while 33 were overweight. Table 1 describes the basic sociodemographic characteristics of participating children. There was a male preponderance . Participants from all main ethnicities and religions in Sri Lanka were included. 2. Children's Self Report QoL scores were higher than parental perception scores. However, there was medium to large positive correlation between children's self reported and parents' perception of QoL. --- Association between PedsQL child self-report scores and BMI As hypothesized, BMI showed small yet significant inverse correlation with physical QoL, as well as tendency for significant inverse correlation with psychosocial and overall QoL . Considering psychosocial subscales, BMI showed significant negative correlation with social functioning, but not with emotional or school functioning . When comparing BMI groups, overall QoL was lower in obese than overweight children (78. --- Discussion --- Summary of findings Sri Lankan children with overweight and obesity in this study had lower physical QoL than psychosocial QoL . As hypothesized, obese children had greater impairment of overall and physical QoL compared to overweight children, and increasing BMI had a negative effect on physical QoL. On the contrary, psychosocial QoL did not show significant correlation with increasing BMI except in the social functioning subscale, and showed stronger associations with modifiable social factors, such as bullying, having difficulty finding appropriate clothes, and lack of regular exercise Parental perception of QoL showed good correlation with, but was lower than, children's self-reported physical and psychosocial QoL. --- Comparison with previous research Comparison with QoL of children with overweight/obesity from other regions Sri Lankan children with overweight/obesity appeared to have similar physical QoL compared to children with Correlation between PedsQL Parent-Proxy-Report Scores and PedsQL Child Self-Report Scores. Effect size was evaluated using Cohen's standards: with correlation coefficients of 0.10-0.29, 0.30-0.49 and ≥0.50 representing small, medium and large associations respectively [19]. *Statistically significant at p<0.05 level **Statistically significant at p<0.01 level QoL, Quality of Life; IQR, Inter Quartile Range overweight/obesity elsewhere in the world, with other studies reporting mean/median PedsQL™ physical QoL ranging from 67.2 to 88.4 [13,[20][21][22][23][24][25]. Our study population however, appeared to have higher psychosocial QoL compared to mean/median psychosocial QoL from other studies which generally ranged from 62.5 to 79.9. Further, most other studies reported lower psychosocial QoL, compared to physical QoL, which differed in our study [13, 20-22, 24, 25].. Therefore, Sri Lankan children with overweight and obesity appear to have equally compromised physical QoL compared to similar children from other regions of the world, but they appear to show better preservation of their psychosocial QoL. Cultural attitudes towards childhood obesity in the Sri Lankan population could have played a role in the comparatively higher psychosocial QoL in Sri Lankan children with overweight and obesity. In Sri Lanka, being overweight and obese is not considered as an illness, and many parents prefer their children to be plump rather than thin [26]. Further, increased weight/BMI is sometimes considered as an indicator of strength, attractiveness and even prosperity in children and adolescents, whereas thin children, are considered to be weak [26]. Thus, children with overweight and obesity might be less psychologically and socially disadvantaged in contrast to the cultural context in western countries. With regard to parental perception of QoL, many studies mirrored our findings, demonstrating that parents perceived lower QoL compared to that self-reported by children with overweight and obesity, albeit with good positive correlation between the two [11,12,23]. One explanation suggested is that parents could be more conscious of long-term negative consequences of obesity in their children, while children may have a more short-term view, and thus a more positive outlook [27]. --- Comparison with QoL of children with other conditions When considering PedsQL physical QoL described by Varni et al., Sri Lankan children with overweight and obesity appeared to have lower physical QoL than healthy children , more comparable to QoL of acutely ill children of similar age in United States . However, psychosocial QoL score appeared more comparable to healthy children , than acutely and chronically ill children [16]. Closer to home, a Sri Lankan study which assessed QoL using PedsQL™ in children with asthma and healthy controls aged 12-14 years, reported mean physical QoL of [28], which appeared higher than the scores we obtained for Sri Lankan children with overweight and obesity, although children with asthma had lower psychosocial QoL [28]. Thus Sri Lankan children with overweight and obesity appeared to have lower physical QoL than similar aged healthy children, which was lower /similar to that of children with asthma and acute illnesses, while their psychosocial QoL was more comparable to healthy children and better than children with acute/chronic illness based on average PedsQL scores. While childhood obesity is sometimes considered favorably socio-culturally in Sri Lanka and other South Asian countries, asthma is socio-culturally considered a more serious and debilitating chronic illness by South Asians [29]. This could perhaps explain why Sri Lankan children with asthma have lower psychosocial QoL, while psychosocial QoL of Sri Lankan children with overweight and obesity remain more comparable to healthy children. Physical QoL in Sri-Lankan children with overweight and obesity, where socio-cultural influences will not confer a similar protection, however, is lower than in healthy children, and more comparable to that of acutely ill children reflecting the extent of difficulties in physical activities and physical discomfort faced by these children due to obesity. --- Factors associated with QoL When considering the relationship between QoL and BMI, a decrease in QoL with increasing degree of obesity, as seen in this study, has been described by previous studies [22,23]. Further, as seen in our study, several previous studies have reported a strong to moderate negative relationship between degree of obesity and QoL in physical and social functioning, with emotional and school functioning remaining relatively less affected [12,22,23,30]. These results suggest the relative resilience of children with overweight and obesity, to preserve their school functioning and emotional wellbeing in the context of increasing severity of obesity, irrespective of geographical or cultural background. Similarly, decrease in physical QoL noted with increasing BMI also appears to be a universal problem, irrespective of socio-cultural millieu. Further, while psychosocial QoL in Sri Lankan children with overweight/obesity appears to be higher compared children from western countries, both groups experience a decline in their social functioning with increasing severity of obesity. Studies looking at QoL in relation to gender have shown controversial results, with some studies reporting a higher QoL in boys than girls [11,12] and others reporting no difference [13,25]. Our study findings collude with the latter. The complex and multidimensional nature of interaction between gender and obesity related QoL have been reported, and could explain these discrepancies [31]. There is limited published data on the association of social/modifiable factors with QoL of children with overweight/obesity. Previous studies have shown peervictimization/ bullying to be associated with lower QoL in children overweight/obesity, which was mirrored by our study [11,32]. Further we found a positive association between getting regular exercise and better psychosocial QoL, which corresponds with a previous study reporting lower psychosocial QoL in overweight children who are less physically-active [33]. Our study findings thus contribute valuable information on modifiable social factors associated with lower QoL in children with overweight/obesity to the global literature. --- Implications for clinical practice Research on social factors associated with impaired QoL in children with overweight/obesity is important to gain a deeper insight into the issue, and to identify new approaches to improve QoL [11,15]. In this study, we studied associations of QoL in children with overweight/ obesity with some novel social factors such as: difficulty in finding/ fitting in to age-appropriate clothes; being on diet control; and having siblings or friends with overweight/obesity; which have not been reported before. Interestingly, we found both physical and psychosocial QoL to be lower in children with overweight and obesity who had difficulties in finding suitable clothes. This finding is further supported by a previous study showing that buying clothes was an activity that obese children disliked [34]. This is a less well recognized factor that, perhaps, could be addressed by highlighting this issue to stakeholders in the clothing industry in the locality, to help ally a common problem faced by children with overweight/obesity in everyday life, which is impairing their QoL. The strong association of bullying with lower psychosocial QoL of children with overweight and obesity demonstrated in this study, reinforces findings of several previous studies, and highlights the urgent need for better measures to reduce obesity-related bullying to improve the wellbeing of these children. Better classroom support, encouraging peers to give better support, and increasing coping skills of children with overweight and obesity could be helpful strategies [34][35][36]. Importance of engaging in regular physical activity/exercise, and its positive associations with psychosocial QoL of children with overweight and obesity has been demonstrated in this study, and as well as previously [33]. Therefore, promoting exercise among children with overweight/obesity appears important, not only for weight management and metabolic health, but also for improving psychosocial wellbeing. Further, given the findings of this study, on relative preservation of psychosocial QoL especially in emotional and school functioning subscales in Sri Lankan children with overweight and obesity, future research and programs for management of childhood obesity, could try to understand and enhance the resilience of these children in emotional and school functioning, in addition to focusing on improving physical and social wellbeing in these children [14,23]. --- Strengths and limitations Strengths of this study include using a pre-validated, international generic QoL questionnaire, adhering strictly to standardized translation procedures as outlined, which increased validity and enabled comparison with similar QoL data from children from other regions and with other conditions, to obtain a more in-depth appreciation of the impact of overweight/ obesity on QoL. Limitations of this study are the relatively small study sample, lack of a control group, and the study being conducted in a clinical rather than a community setting. Inclusion of a control group of healthy children would have provided useful information. Availability of a crossculturally adapted PedsQL questionnaire will be helpful for the conduct of future studies in this area. However, while the questionnaire was translated according to standard protocol with cross cultural adaptation, and certified as linguistically validated, psychometric validation was not conducted. Reliability of obtained results would increase if the questionnaire was psychometrically validated. With regard to the study setting, a few previous studies have compared QoL in children with overweight/obesity between community and clinical settings, where some studied have shown no difference in QoL scores in hospital-based versus community settings [23], while others have shown lower QoL in clinical settings [11]. Conversely, the study setting also helped to increase the generalisability of the results. As this study was conducted at a tertiary care referral centre, 40% of participants were referred from districts other than Colombo district. Our study population thus included families living throughout the country as well as representing all main ethnic and religious groups in Sri Lanka. Thus, we believe that the study findings are generally applicable to children with overweight/obesity in Sri Lanka. --- Conclusions To the best of our knowledge, this is the first study to assess QoL in Sri Lankan children with overweight and obesity. Our findings emphasize the importance of addressing impairment of QoL faced by Sri Lankan children with overweight and obesity, especially in physical functioning where most distress was documented. Further, parental perceptions were lower, but showed good correlation with child self-reported QoL, which signifies that parents are quite attuned to their children's issues, and possibly more concerned about impaired wellbeing of their child, than the children themselves. This emphasizes the importance and value of parental involvement in management of childhood obesity. This study documents for the first time, novel modifiable social factors associated with impaired QoL of children with overweight/obesity in Sri Lanka, such as difficulties in finding age-appropriate clothes that fit, as well as previously documented factors such as bullying. It also demonstrates for the first time, the protective effects of regular exercise on psychosocial QoL of Sri-Lankan children with overweight/obesity and the relative resilience of these children in emotional and school functioning, even in the context of increasing BMI. Prevention of peer victimization of children with overweight/obesity, promoting physical activity and enlightening the clothing industry on the need of a range of children's clothes to fit larger children, are potential strategies which could improve QoL of children with overweight/obesity in Sri Lanka, as well as elsewhere. --- --- Abbreviations QoL: Quality of Life; PedsQL: Pediatric Quality of Life inventory; BMI: Body Mass Index; WHO: World Health Organization; IQR: Inter Quartile Range. Authors' contributions SG drafted the manuscript, performed the statistical analysis and was involved in acquisition of data. CBG was involved in acquisition of data and contributed to drafting the manuscript. MSH was involved in acquisition of data and contributed to drafting the manuscript. SNS conceptualized the study, supervised and provided expert advice, critically revised the manuscript to produce the final manuscript. All authors have read and approved the final manuscript. --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: While childhood obesity is rising rapidly in South Asia, there is limited research on quality of life (QoL) of children with overweight and obesity from the region. This study assessed physical and psychosocial QoL in Sri Lankan children attending a specialized obesity clinic, from both children's and parents' perspective, and modifiable social factors affecting QoL. Methods: We performed cross-cultural translation of the Pediatric Quality of Life Inventory (PedsQL™) 4.0 (Child-Self Report and Parent-Proxy forms), and assessed self-reported and parental-perception of physical and psychosocial QoL in 8-12 year-olds with overweight and obesity (n=110), referred for obesity management at a tertiary-care children's hospital in Sri-Lanka. Body mass index (BMI) and pre-selected social factors affecting QoL were also assessed. Data were analyzed by non-parametric tests (Mann-Whitney U test, Wilcoxon test and Spearman correlation). Results: The median physical QoL was lower than psychosocial QoL (78.1vs81.7, p=0.032) and physical QoL was inversely correlated with BMI. Parental-perception of children's physical and psychosocial QoL correlated with childreported QoL, but was lower. Being bullied (p=0.001) and not getting regular exercise (p=0.031) were associated with lower psychosocial QoL. Both physical and psychosocial QoL were lower in children having difficulties in finding suitable clothes (p< 0.001). Conclusions: Children with overweight and obesity from Sri Lanka appeared to have greater impairment of physical QoL than psychosocial QoL. Higher BMI, bullying, lack of regular exercise and lack of suitable clothing, negatively affected QoL. Potential strategies to improve QoL include promoting regular exercise, addressing bullying and promoting availability of children's clothes in larger sizes to fit children with overweight and obesity.

Introduction Managing migration flows and the integration of migrants is a major challenge for our societies. Recent crises and conflicts have led to large flows of refugees. For example, in 2015 alone, more than one million refugees arrived in Germany, largely from Syria. In addition to the challenges of short-term crises, there are also long-term demographic changes that contribute to the migration debate. Longer lives and lower fertility levels mean that population aging is an inevitable consequence. Immigration is often seen as a stopgap measure to address population aging, which would otherwise strain the economy and public finances. In this context, immigration is expected to become a major driver of population dynamics. Understanding the processes of assimilation and integration of migrants has become a priority for countries all over the world. Currently, United Nations member states are developing a global compact for safe, regular and orderly migration. 5 This is an intergovernmentally negotiated agreement to cover all dimensions of migration. Policy action is informed by a number of indicators of integration that international organizations produce [8]. Traditional indicators include measures of education, language acquisition, poverty, intermarriage, and other aspects [11]. These indicators evaluate how opportunities and outcomes for migrants and their children differ from the ones of the host population. Cultural assimilation is a more elusive quantity to measure with traditional methods. However, perceived cultural differences often fuel negative sentiments towards immigrants. At the same time, immigrants contribute to the cultural diversity of the host society. It is thus important to evaluate processes related to cultural assimilation. In this paper, we use anonymized, aggregate data from Facebook users, available via Facebook's advertising platform, to evaluate cultural assimilation of Arabic-speaking migrants in Germany. Using this data, we compute an assimilation score and compare this score for different migrant populations in Germany, including migrants from Austria, Spain, France and Turkey. To the best of our knowledge, this paper is the first to make use of online advertising data from Facebook to show the feasibility of measuring migrant integration at scale. --- Related Work Understanding migrant integration and the effectiveness of policy measures to favor assimilation is a longstanding challenge. A wide range of aspects such as 'civic integration policies' [4] or multiculturalism [12] have been analyzed. These studies developed evaluation metrics based on concepts such as political trust [6] as well as lack of electoral participation or composite measures of civic integration [12]. For a review of empirical and theoretical challenges see [10] . New information, like Web and social media data, are a main source of innovation in the context of migration studies. Research in this area has focused on using online data to improve estimates of migration flows and stocks. After Zagheni and Weber used geo-located Yahoo! e-mail data to estimate international migration flows [14], several platforms have been used to understand the network structure of migration, including Facebook [5] and Google+ [7]. Geo-located Twitter data has proved useful for studying the relationship between internal and international migration [13], as well as short-term mobility versus long-term migration [3,1]. LinkedIn data has provided insights into global patterns of migration for professionals [9]. More recently, Facebook data for advertisers have been used to create estimates of stocks of international migrants [15]. These data are a promising source for generating timely migration statistics at different levels of spatial granularity. In this paper, we expand the use of these data to study cultural assimilation. --- Data Facebook's advertising platform allows advertisers to programmatically target their ads to a specific population, e.g. based on age, gender, country of residence, or spoken language. 6 Using the Marketing API, advertisers can obtain estimates of the number of people who belong to a certain demographic group and show certain interests based on 'likes', pages that they visit and other signals. 7 As an illustrative example, consider the number of Facebook users speaking Arabic, living in Germany, being in the age group 18-65 and interested in football. As of May 3, 2018, Facebook Ads manager reports that there are 530k users matching these criteria. The complete list of available targeting options is available on the Facebook Ad API page. 8We used an open source Python implementation to access the Facebook Marketing API and collect such audience estimates. 9We base our analysis on Facebook's interests. For each population, we obtain the audience size for each interest and we compare a measure of prevalence across populations. As an oversimplified example, we can compare the fraction of Arabic-speaking migrants in Germany who are interested in typically German interests such as Bundesliga -the German soccer league -to the same fraction of Germans in Germany interested in this topic. Arguably, if a migrant population has similar levels of interest as the host population, this is an indicator of assimilation in terms of cultural taste. Facebook Marketing API has hundreds of thousands of interests to target ads. In this paper, we collected data for 2,907 interests10 with the biggest global Facebook audiences. Our main criteria in selecting this subset is that their audiences should be sufficiently large so that the estimates are reliable. Then, we obtained the audience sizes for those interests for adults from different populations . 11 In this paper, we focus mainly on the assimilation of Arabic-speaking migrants in Germany. To achieve this, we obtained data for Arabic-speaking migrants living in Germany, and nonexpats in Arab League countries. 12 Finally, for comparative purposes, we also collected audience estimates for the same set of interests and demographic groups for other migrant populations in Germany, from Austria, France, Spain and Turkey. --- Methods: Quantifying assimilation Our goal is to obtain an assimilation score that could serve as a proxy for the assimilation of a group of migrants to a local population in terms of interests expressed by both groups. To reach this objective, for a given list of interests, we need the audience sizes of three groups: the destination country , the target group in the destination country , and the target group's home country . We will denote those groups as Dest, Target and Home respectively. We will focus on the case where Dest is German non-expats. We then proceed in two steps: we identify 'distinctly German' interests -interests that are more popular in Germany compared to in Home; we use the audience sizes for those interests to compute the assimilation score. The selection of 'distinctly German' interests is a necessary preprocessing step to compute a meaningful score. If we were showing that migrants had the same level of interests as local people in generic interests that are prevalent across countries, like "Technology" or "Music", this would not necessarily be a sign of assimilation. First we describe how, for each of the 2,907 initial interests, we evaluated if they are 'distinctly German'. Estimating how popular an interest is among a population could be simply done using the percentage of people in that population with the interest. However, this approach would be biased as a result of differential online activity level, since more active Facebook users also have more interests. Though the exact methods used by Facebook to calculate these interests is not disclosed, it is known to make use of a user's activity. 13To correct for this activity level bias, we can instead use the normalized audience percentage within each population given by IR p = A p n i=1 A p , where A p and IR p denote, respectively, the audience size and what we call interest ratio for interest i in population p. Tables 1 and 1 show an example of the audiences for three populations and the corresponding interest ratios. IR can now be compared between populations to obtain typical interests. We can identify 'distinctly German' interests by comparing the IRs for Germany and Home. 'Distinctly German' interests are defined as those that have a larger IR for Germany than for Home ). For added numerical stability, we extract among those 'distinctly German' interests the ones that are the 'most German' ). This is done by dividing the IRs for German interests by the ones for Home, for each of the 'distinctly German' interests, and keeping the top k% interests with highest value of the relative IRs. After selecting the 'most German' interests, we compute the IRs for the target population only for these interests. Then, we define an assimilation score per interest i by dividing the IRs for Target by the ones of Dest. AS T arget = IR T arget IR Dest where AS T arget is the assimilation score of Target for interest i. For instance, for the interest 'Brewery' in the example in Table 1, the assimilation score for the target group is 0.011/0.081=0.14. A specific interest i is considered to be fully assimilated by the migrant population if AS T arget ≥ 1.0. Finally, to have a single score for the target population, we aggregate the per-interest scores by taking the median across all the 'most German' interests. --- Results We start by validating some of the assumptions in our data collection. Although Facebook identifies expats of some countries of origin, such as Spain, most Arab League countries cannot be targeted individually this way, though there is a catch-all "Expats ". Thus we use a proxy for this group by instead obtaining estimates for the Arabicspeaking residents in this "Expats " group. To test if it is indeed a good approximation for expats from Arab League countries, we compared the number of migrants in the 16 German states from a recent report by the Brookings Institute 14 with the estimated number of migrants per state using the data from Facebook ads manager. We find that there is a near perfect correlation between the two sets of values . Then, we evaluate the selection of 'distinctly German' interests. Table 2 shows the top and bottom of the list of interests sorted in descending order according to IR Germany /IR Arab League countries . Note that the top of the list showing some 'distinctly German' interests is obtained when using the set of the 2,907 interests that are most popular worldwide. This is why even more typically German interests such as Oktoberfest do not show up. Similar lists were computed to validate this process when using a different Home but are not shown here due to space constraints. In the remainder of this section, we evaluate our assimilation score. We set the 'most German' parameter k to 50% and we recall that this list of interest is always computed using the Home corresponding to the Target being analyzed. 15 Our first line of analysis compares Targets coming from different countries -Austria, France, Spain and Turkey. Since there is a sizable minority of Turkish-speaking non-expats in Germany16 , we Fig. 1: The number of migrants per square kilometers in each German state according to two sources; in blue, from a study of the number of refugees by the Brookings institution; in orange, from the estimated audiences by the Facebook ads manager using the Arabic-speaking expats as target population. divide the Turkish population into two: All Turkish speakers in Germany, and Turkish-speaking non-expats in Germany containing residents of Germany who speak Turkish).The assimilation scores for different sub-populations from these countries is shown in the first part of Table 3 in comparison to Arabicspeaking migrants in Germany. The results show that European migrants have a higher assimilation score than Arabic-speaking migrants and Turkish speakers. Next, we compared the assimilation scores for different sub-groups among the Arabicspeaking migrants. More precisely, we divided this population according to gender, age and education level. The results for those sub-groups are shown in the rest of Table 3. We explored all attribute combinations to see if there were any additive effects. Due to space constraints, assimilation scores for only a small selection of attribute combinations are included. Women appear to be less assimilated in terms of their Facebook interests than men. We also observe that the assimilation score for university graduates is slightly higher than that for non-graduates. Note that for both educational levels, the scores are lower than the one for the whole Arabic-speaking migrant group. This statistical phenomenon, akin to Simpson's Paradox, is due to the aggregation process of the per-interest scores. Finally, in our analysis, young people between 18-24 are the most assimilated. --- Discussion The work that we presented in this article has important limitations that we would like to acknowledge. At the same time, we also want to emphasize the potential for further research that computational social scientists can perform in this area. Some limitations are related to our methods. For example, the results may be sensitive to the total number of interests that we consider, currently 2,907. To check the .553 Table 3: Assimilation score for different choice of Target in Germany using the top 50% distinctly German interests of 2,907 interests. Lines 1-5 of the first column correspond to non-Arab populations. The remaining cells, all with the "A:" prefix, correspond to Arabic-speaking migrants and sub-groups of this population. robustness of our approach, we computed the assimilation scores using random subsets of sizes varying from 100 to 2,900, in steps of 100 within our chosen subset. Figure 2 shows the variation in assimilation score. The figure shows that for any subset of size 500 or bigger, the assimilation score becomes stable. The maximum relative change from the average score across all considered populations was 10.4%, and the average relative change was 4.7%. This indicates that our results are relatively stable with respect to the number of interests being used. Some limitations are related to the type of data. Our work relies on audience estimates produced by Facebook. The procedures used to infer users' interests are not well documented. For example, there could be differences in how well content in the Arabic or German language is processed, leading to artificial differences in interest profiles. Additionally, Facebook data do not provide information about the number of years that people have spent in the country, a key variable for the study of assimilation processes. Fig. 2: Variation of the assimilation score for different Arab and non-Arab populations in Germany using the top 50% distinctly German interests when changing the size of the set of interests. 'E' denotes expats and 'NE' non-expats. In our analysis, we grouped together arabic speaking migrants from all 22 countries of the Arab league into a single group, which might introduce biases. Though the arabic speaking countries typically share a lot of cultural similarities, different confounding factors can create differences. Our choice to group these users was because of a limitation by the Facebook Marketing API which does not allow us to target individuals from specific arabic speaking countries . Also, since a majority of Arab migrants in Germany are from Iraq and Syria [2], we think this bias would highly influence our results. Despite these limitations, this article opens important lines of research that computational social scientists are well-positioned to address with Facebook data for advertisers. First, migration affects the host society and these data can be used to evaluate the extent to which the host society absorbs and embraces exposure to diversity. Second, assimilation processes can be studied at different levels of geographic granularity and in relation to contextual variables like political orientation of various sub-regions. Third, the idea that we presented for the specific case of Germany can be scaled to many countries of the world and used to study macro-regional processes like integration in the European Union. --- Conclusion We presented a methodology that uses anonymized, aggregate data from Facebook's advertising platform to compare the interest profiles of different migrant groups to that of the German host population. Based on the interest similarities, we derive an assimiliation score and observe that this score is lower for Arabic-speaking migrants compared to several European reference groups. We also show that the score varies among sub groups with younger and more educated men scoring highest.

Migrant assimilation is a major challenge for European societies, in part because of the sudden surge of refugees in recent years and in part because of long-term demographic trends. In this paper, we use Facebook data for advertisers to study the levels of assimilation of Arabic-speaking migrants in Germany, as seen through the interests they express online. Our results indicate a gradient of assimilation along demographic lines, language spoken and country of origin. Given the difficulty to collect timely migration data, in particular for traits related to cultural assimilation, the methods that we develop and the results that we provide open new lines of research that computational social scientists are well-positioned to address.

Introduction Adult children are the most common source of assistance for older adults to achieve ageing-in-place [1]. Drawing on the data from the Health and Retirement Survey during waves of 1995 to 2010, it was estimated that approximately 17% of adult children in the United States will become caregivers at some point in their lives and will provide an average of 77 h of care per month [2]. According to the report from the National Alliance for Caregiving [3], 50% of surveyed caregivers were providing care for their parents or parents-in-law. In China, it is estimated that 90% of older people rely on their families for eldercare [4], with majority of this care being provided by their adult children. Both anticipated and actual caregiving for parents can be sources of stress for adultchild caregivers. As the potential primary caregivers for their parents, adult children tend to be concerned about how much help they can provide in the future and whether they can manage the burden of providing care [5,6]. Studies also discovered that providing care to older parents can generate multidimensional stress and burden for adult-child caregivers, including financial, physical, and mental burden [2,3]. Since adult-child caregivers are more likely to be working [3], the conflict between their work and family roles is likely to contribute to their caregiving stress. Studies investigated the care preparation activities that older adults engage in as a proactive coping strategy, and they have identified their effectiveness for mitigating future long-term care challenges [7][8][9]. Applying the theories of decision-making, problemsolving, and planning to everyday situations, Sörensen and Pinquart [10] distinguished four types of preparation activities and developed the Preparation for Future Care Needs Model for studying care preparation among older adults. The model encompasses the following care preparation domains: awareness/avoidance of future care needs, information gathering, decision-making, and concrete planning [11,12]. Individuals first anticipate or avoid the need for future care; thereafter, they begin to collect information, which involves actively seeking out information from other people or the media and making decisions about future care arrangements. The final step in their planning process is concrete planning, which refers to the activities that will help them implement their plans. Scholars have also suggested that adult children should prepare themselves for their future caregiving role and participate in care preparation activities for effectively reducing care anxiety and preventing their future caregiving burden. While studies on adult children's caregiving preparation have been conducted for decades, there is currently no systematic review on this topic. This study seeks to fill this gap by conducting a comprehensive overview of the literature in this field, synthesising the available evidence, and identifying knowledge gaps and opportunities for future investigations. The following research questions will be addressed in the systematic review: --- Methods --- Search Strategy The EBSCOhost, PsycINFO, ProQuest, Web of Science, PubMed, and CNKI databases were systematically searched. A lexicon of keywords was generated after a pilot search of the selected databases. With the exception of CNKI, the following combinations of keywords were used to identify and retrieve studies on anticipatory care preparation from the databases: 'adult child*' AND AND . For the search of the CNKI database, the following combinations of keywords were used: '子女' AND '养老照顾' AND OR '计划' ). Relevant studies dated from the inception of the databases until November 2022 were identified and retrieved. This broad period was applied because few systematic reviews have explored this research topic since its emergence. --- Inclusion and Exclusion Criteria Studies were included in the present review if they: were reports, reviews, book chapters, conference papers, dissertations, or journal articles related to the caregiving preparation of adult children for their parents, were published in English or Chinese, and were published between the inception of the databases and November 2022. Studies were excluded if they were commentaries, unpublished reports, not empirical studies, or published in languages other than English or Chinese. Studies were also identified from the reference lists of the articles already included in the present review. The abstracts or summaries of all included studies were reviewed, and the eligibility of each study was determined by two researchers who evaluated the relevance of each study to the study topic. --- Data Collection Figure 1 presents the Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram describing study selection. The initial literature search yielded 924 studies, of which 145 duplicates were removed. Subsequently, the abstracts of the remaining 779 studies were screened, and 726 irrelevant studies were excluded; thus, 53 articles were included in the full-text assessment. Next, a number of studies were excluded for various reasons; specifically, five studies did not have English or Chinese full text, six studies were not empirical studies, eighteen studies were related to actual caregiving experiences instead of care preparation, three studies pertained to the care preparation from the perspective of older adults, nine studies were related to end-of-life care planning, two studies were related to care transition, and three studies were about perceptions of care provision. Subsequently, 11 studies were retrieved from the reference lists and added to the present review. Consequently, a final sample of 18 articles was included in the present review. The screening and selection processes were conducted by two researchers, independently, and their results were cross-checked. text, six studies were not empirical studies, eighteen studies were related to actual caregiving experiences instead of care preparation, three studies pertained to the care preparation from the perspective of older adults, nine studies were related to end-of-life care planning, two studies were related to care transition, and three studies were about perceptions of care provision. Subsequently, 11 studies were retrieved from the reference lists and added to the present review. Consequently, a final sample of 18 articles was included in the present review. The screening and selection processes were conducted by two researchers, independently, and their results were cross-checked. --- Data Analysis First, two researchers independently evaluated the quality of the extracted studies by using the Mixed-Methods Appraisal Tool, which has been widely used in systematic reviews to evaluate the quality of qualitative, quantitative, and mixed-methods studies. Inter-rater reliability was determined by calculating intraclass correlation coefficients, and the results indicated a high inter-rater reliability . All the studies yielded an overall score of 3 or higher, indicating an acceptable research quality. Next, thematic analysis was conducted to synthesise the findings of the reviewed studies. All studies were read by two researchers. Data on the sources of the literature, themes, research methods, participants, and key findings of the reviewed studies were coded and summarised. Data synthesis was conducted by comparing the findings of the included studies and identifying differences and similarities. --- Results Table 1 lists the characteristics of the reviewed studies. Among the 18 reviewed studies, 83.3% were journal articles, and 16.7% were dissertations. Half of the studies adopted quantitative methods, including questionnaire surveys and experiments , four studies employed qualitative methods, four applied mixed methods, and one study adopted a case study design. The systematic review findings were analysed based on three themes: conceptualisation of caregiving preparation, prevalence of caregiving preparation, and the influencing factors for and consequences of caregiving preparation. The details on the participants, research methods, and the main findings of the reviewed studies are listed in Table 2. Table 3 provides a summary of the reviewed studies' conceptualisations or measurements of the caregiving preparation conducted by adult children for their parents. The daughters were more prone to anticipate providing care when a long-term crisis occurred than when a short-term crisis occurred. More than 90% of the daughters anticipated providing advice or information if a crisis occurred, whereas financial assistance or hiring others to help their mothers was anticipated more frequently in response to a long-term crisis. Fowler and Afifi [17] 173 adult children aged >40 years --- Experiment Filial anxiety was correlated with the likelihood of engaging in discussions with parents regarding their preferences for future care and coping efficacy. Fowler and Fisher [18] 128 adult children aged >40 years; 113 nonrelated parents aged >60 years Face-to-face questionnaire survey Attitudes toward shared autonomy and ageing anxiety, concerns about the negative effects of caregiving, and perceptions of a limited future were associated with caregiving preparation. Gui and Koropeckyj-Cox [19] 20 Chinese young adults aged 23-31 years --- Qualitative interviews Some migrant adult children planned to settle down in Canada and bring their parents over, whereas others planned to return to China. The adult children were expected to take care of their parents in the future but also to consider their dilemmas. Adult children and ageing parents rated several domains of parent care readiness, including the medical, legal-insurance-financial, family-social, and spiritual-emotional domains. The adult children's emphasis on the domain of care preparation could differ from that of their ageing parents, but both generations still emphasised reconciliation within their families. Paulson and Bassett [24] 165 women prospective caregivers aged 45-65 years Online questionnaire survey Filial obligation fully mediated the association between close attachment style and caregiving preparation. Radina [25] 10 Mexican American sibling dyads --- Qualitative interviews The preparation process for parent care among Mexican Americans comprised the three conceptual components of caregiver selection/designation, anticipation, and planning. Sörensen [26] 33 multigenerational families comprising grandmothers, mothers, and daughters Qualitative interviews and face-to-face questionnaire survey Few of the participants made concrete caregiving plans. The planners were more satisfied with the amount of discussion and planning that they engaged in with their family than the nonplanners were. The daughters' discussion of future caregiving with their family members was predicted by their internal locus of control and the mother's age. Sörensen, Webster, and Roggman [9] 141 adult children Face-to-face questionnaire survey Attachment style was associated with caregiving preparation, which in turn was associated with feelings of preparedness. Sörensen and Zarit [27] 33 multigenerational families comprising grandmothers, mothers, and daughters Qualitative interviews and a face-to-face questionnaire survey The daughters' anticipation of household helping was predicted by their relationship with their own daughters, and affective solidarity moderated the association between filial responsibility and the daughters' anticipation of household helping. --- Qualitative interviews The adult children believed that later-life care planning conversations alleviate the anxiety of older adults toward future care arrangements, thereby helping these older adults to establish a sense of control and access family support. Planning helped a family to learn about appropriate care arrangements. This awareness led to increased calmness among the families, helping them to alleviate their doubts and eliminate their stress and potential conflicts. The adult children required information to make informed choices but had limited access to information. Initiating conversations early could reduce barriers to effective decision-making. Yoo and Kim [29] 124 adult children of immigrants aged 22-57 years Qualitative interviews and a face-to-face questionnaire survey The adult children felt a strong sense of responsibility toward their ageing parents, and they were prepared to support their parents' financial, healthcare, and long-term care needs. Although both daughters and sons expressed the desire to care, daughters were more likely than sons to engage in in-depth discussions with their parents to clarify their concerns and worries. --- Author Conceptualisations/Measurements Bromley and Blieszner [13] The decision-making process comprises four activities, namely considering, discussing, preliminary planning, and making a final decision. Considering refers to whether an adult child has ever considered their parents' future needs with respect to conditions that may limit the parents' ability to live independently. Discussing refers to whether an adult child has discussed their concerns with their parents. Preliminary planning refers to activities involving obtaining information, asking for advice, or making arrangements for services if one's parents cannot live independently. Making a final decision refers to whether a plan is in place for the future care of one's parents when necessary. Brown [14] Caregiving preparation comprises four main categories and several subcategories: Child feelings refer to the feelings of an adult child regarding various decisions, emotional experiences, and expectations, and their feelings in regard to recognising change and loss. Expectations refer to the expectations of an adult child's parents and the adult child's own expectations, roles, actions involving thinking ahead, feelings, and expectations. Decisions refer to an adult child's feelings regarding their decisions and their awareness of their parents' feelings, emotional well-being, and physical well-being. Recognition of change and loss refers to positive focus, family involvement, and comparisons. Conway-Turner and Karasik [15]; Conway-Turner and Karasik [16] In these studies, the planning and adjustment inventory was used. This inventory comprises information on when and how frequently the daughters investigated in these studies considered caring for their mothers . Fowler and Afifi [17] This study is based on the theory of motivated information management, which pertains to the decisions of individuals to seek out or avoid information regarding particular issues. The measurements used in this study include the importance of an issue , the uncertainty discrepancy related to parental eldercare preferences , outcome expectations , and information-seeking behaviour ). Fowler and Fisher [18] The four items from the 'becoming aware' subscale and four items from the 'gathering information' subscale of Sorensen and Pinquart's measure of preparation for future care needs. Items of the original versions for older parents were reworded and applied to adult children . 'Discussion of future care needs' were measured by asking adult children to report on how much they had discussed with their parent a range of issues pertaining to future care needs. Gui and Koropeckyj-Cox [19] The expected care arrangements made by adult children for their parents. Hansson et al. [20] This study discussed the following concepts: The likelihood of intervention index . Elderly parent consciousness ). The family involvement index . --- Author Conceptualisations/Measurements Lieberman [21] This study discussed: the perception of parent change , the degree of parent concern , and the orientation of the participants toward behaviours related to parental ageing . Miller [22] This study explored questions pertaining to the awareness of factors relating to long-term care planning and to risk aversion. It also examined how these concepts affected the way in which a participant advised their parents on the concept of long-term care planning . --- Myers et al. [23] A 50-item parent care readiness assessment instrument was used, covering the medical, legal-insurance-financial, family-social, and spiritual-emotional domains. Paulson and Bassett [24] Sörensen and Pinquart's Preparation for Future Care Short Form was used . --- Radina [25] The model employed in this study comprises two primary components: decision-making and the caregiver socialisation/preparation process. Sörensen [26] Guided by Sörensen's [30] conceptual framework for caregiving preparation, this study examined the anticipation of caregiving, exploring whether a participant had anticipated: their mother's need for help with household tasks and their mother's need for help with personal care. It also examined a family anticipation of providing care to the mother in the family. Sörensen, Webster, and Roggman [9] In this study, preparation for future caregiving is defined as the mental and physical actions involved in anticipating, making decisions, making concrete plans, and engaging in role socialisation relating to providing care for an older person in the future . Care planning was measured using the 14-item scale of the short form of Sörensen and Pinquart's [11] Preparation for Future Care Needs Model. Sörensen and Zarit [27] Guided by Sörensen's [30] conceptual framework of preparation for caregiving, the questions asked in this study were: ( [28] In this study, three overarching themes were derived from interviews, namely role considerations, practical considerations, and emotional considerations. Under each of these themes, older adults, family members, and healthcare providers contributed different ideas. Yoo and Kim [29] In this study, the following key themes were derived through the application of grounded theory: adult children were likely to talk about caring for their ageing parents in the future as a form of repayment for the care and support that they received from their parents while they were growing up, adult children were concerned about the language and financial barriers faced by their ageing parents, and adult daughters often felt more pressure to consider the details of their parents' future caregiving needs. --- Conceptualisations of Care Preparation Studies proposed various conceptualisations of the caregiving preparation of adult children. Of the 18 reviewed studies, 10 conceptualised it as a multiphase process. Bromley and Blieszner [13] identified four decision-making activities as the sequential steps of caregiving preparation. Hans-son et al. [20] reported that adult children's awareness regarding their parents' needs progresses gradually over time; initially, they consider these needs, and then they learn about ageing and monitor their parents regarding specific areas of concern. Brown [14] described that caregiving preparation involves the dimensions of awareness, decision-making, and planning, whereas Fowler and Fisher [18] described that it comprises awareness, information collection, and discussion. Stolee et al. [28] conceptualised caregiving preparation as a process in which adult children converse with their parents regarding care planning, and this process also involves communication and decision-making. Two studies specifically developed models for studying the caregiving preparation of adult children [25,30]. Sörensen [30] developed the Preparation for Caregiving framework to explain care preparation of multigenerational families by applying the life course perspective, role theory, and several theories of planning and decision-making derived from cognitive psychology. Caregiving preparation is defined as the mental and physical actions that are related to the four dimensions of anticipation, decision-making, concrete planning, and role socialisation, and they are performed by an individual before they assume the role of caregiver for an older individual [26]. On the basis of Sörensen's [30] framework, the theory of proactive coping, and the literature on the traditional values of Mexican Americans and Hispanic/Latino cultures, Radina [25] investigated how Mexican Americans prepared for parent care and developed the Caregiver Preparation Model. Compared with Sörensen's [30] Preparation for Caregiving Model, which focuses on both the caregiver and the care recipient, the Caregiver Preparation Model is centred on the care preparation of caregivers and comprises the two main components of decision-making and caregiver socialisation/preparation. Caregiver socialisation/preparation further consists of three secondary components, namely caregiver selection/designation, anticipation, and planning. As indicated by this model, the primary component of decision-making influences each of the three secondary components. The rest of the studies mainly examined caregiving preparation as a single-step event , or as-expected care arrangements related to multiple care domains [19,22,23,29]. The measurements of caregiving preparation were mostly self-developed questions, and few studies have validated the scales used. Several studies used scales developed for measuring older adults' care preparation [9,18,31] to assess adult children's caregiving preparation. --- Prevalence of Care Preparation Based on the self-report of 103 adult daughters aged between 30 and 55 years, Conway-Turner and Karasik [15] discovered that almost all of them reported that they had considered the possibility of providing care to their ageing mothers. Approximately 68.3% of them reported that they had considered this issue frequently , and this high level of consideration was unaffected by their age. Moreover, the mean age at which they started to consider this issue was 28 years, indicating the early occurrence of the anticipation process, which could be due to the high media coverage of caregiving responsibilities and the increasing visibility of the mother cohort caring for their own mothers [15]. In another study involving the same sample, Conway-Turner and Karasik [16] further reported that over 90% of the adult daughters anticipated providing advice or information if a crisis occurred. Sörensen and Zarit [27] conducted a study that examined a sample of 33 multigenerational families, and they reported that substantial numbers of both mothers and daughters in the sample anticipated the need for care for their grandmothers. Bromley and Blieszner [13] surveyed 169 adult children aged between 20 and 59 years and discovered that over 80% of them had thought about the future dependency needs of their still-healthy parents and 39% of them had discussed the topic of future care with their parents. In a survey of an older sample, Hansson et al. [28] reported that 82% of 242 adult children had started to seriously consider issues and concerns related to caring for their older parents. However, most adult children only engaged in the early stages of caregiving preparation , with few acquiring concrete planning experience related to making future caregiving arrangements [26,27]. Bromley and Blieszner [13] reported that ageing families rarely conducted preliminary planning and final decision-making related to caregiving preparation. --- Influencing Factors and Consequences of Caregiving Preparation The age and gender of adult children are associated with their caregiving anticipation [13,21]. Lieberman [21] investigated a sample of 807 adults who were aged between 20 and 70 years and had at least one living parent or parent-in-law. They reported that age, gender, and race were associated with the awareness of future caregiving; specifically, adult children who were older, female, or Black were more likely to acknowledge the dependency needs of their parents. They also discovered that those from workingclass families were more likely than those from middle-class families to feel that discussing dependency issues of parents is appropriate. For adult daughters, the order of birth, the characteristics of the sibling network, the geographical proximity between mothers and daughters, and the health status of the daughters have been reported to be significantly related to their caregiving anticipation [15,32]. Filial anxiety has been reported to be correlated with the likelihood of adult children discussing their parents' preferences for future care and their coping efficacy [17]. Attitudes towards shared autonomy and ageing anxiety, concerns about the negative effects of caregiving, and perceptions regarding a limited future have been reported to be associated with caregiving preparation [18]. However, Bradley et al. [33] did not identify a significant correlation between adult children's filial anxiety and the extent of their engagement in caregiving preparation. In a small sample of 33 three-generation families, Sörensen [30] discovered that adult daughters' anticipation of household help for their mother was influenced by adult daughters' relationships with their own daughters, with affective solidarity moderating the association between filial responsibility and the anticipation of household help. In a separate investigation of the same sample, Sörensen [26] reported that whether adult daughters discussed future caregiving with other family members was influenced by their mother's age, but not by health; additionally, the internal locus of control was revealed to be associated with increased preparation for caregiving. Furthermore, in a study of 141 adult children [9], securely attached individuals were marginally more likely than insecurely attached individuals to make plans. A comparison of individuals who are already providing care with those not currently providing care indicated that secure attachment has a greater influence on the preparation activities of individuals not currently providing care. In a sample of 165 women who were prospective caregivers, Paulson and Bassett [24] discovered that filial obligation fully mediated the close attachment style and preparation for caregiving. Broomley and Blieszner [13] reported that the age, personal authority, and family stressors of adult children were positively correlated with whether they considered their parents' future care needs. Specifically, daughters were more likely to consider care arrangements and engage in discussions with their parents, and family stressors and personal authority negatively and positively, respectively, influenced the likelihood of discussion. Miller [22] and Stolee et al. [28] asserted that information deficiency is the key barrier preventing adult children from conducting long-term care planning for their older parents. Hansson et al. [20] reported that the awareness of ageing issues and caregiving anticipation among adult children were the most often triggered by the health crises of their parents and by their perceptions of vulnerability, associated with their parents' psychological adjustment, personality, and support resources. The age of an adult child's mother is also significantly correlated with their caregiving anticipation [15,32]. Bradley et al. [33] reported that the age of parents significantly influenced whether respondents engaged in minimal, partial, or substantial planning behaviours. Few studies investigated the consequences of caregiving preparation. In a qualitative study, Stolee et al. [28] revealed that long-term care planning discussions can alleviate the anxiety of older adults regarding future care arrangements, helping them to establish a sense of control and access family support. Planning can also help family members gain knowledge on appropriate care arrangements, which can lead to increased calmness, help alleviate doubts, and eliminate stress and potential conflicts. Moreover, initiating conversations early on can reduce the barriers to effective decision-making. Sörensen and Zarit [27] reported that few respondents indicated that they engaged in care preparation activities; however, the respondents who were planners were more satisfied than nonplanners with the amount of discussion and planning with their families. --- Discussion The present study is the first to systematically review the literature on the caregiving preparation of adult children for their older parents, and synthesised research findings regarding the conceptualisation, prevalence, influencing factors, and potential consequences of caregiving preparation. The findings of the present study can provide future directions for this research topic and have implications for service providers and policymakers to design measures for supporting adult-child family caregivers. The findings of the systematic review revealed that most studies conceptualised caregiving preparation as a multistage process. These studies' multidimensional conceptualisations generally corresponded to the key components in the Preparation for Future Care Needs Model [11,12]. They mostly covered one or several components, including awareness of parents' ageing or future care needs, gathering information related to eldercare, discussion or communication with older parents regarding future care, decision-making about parents' future care arrangements, preliminary planning, and role socialisation. It is noteworthy that majority of studies examining caregiving preparation explicitly or implicitly address the concepts of anticipatory role socialisation and proactive coping. These are two useful strategies for families to undergo transitions [34] and cope with potential stressors [31]. Derived from the role theory, role socialisation refers to the process of gradually learning the norms, scripts, attitudes, values, and subtle rules that an individual must acquire to effectively function in society [35]. Applied to the family context, anticipatory socialisation reflects the learning that adult children engage in before assuming the caregiver role and the application of the knowledge they acquired through learning. It also involves the acquisition of new abilities and skills; in some cases, it may involve changes to an individual's reference or social group [34]. When a person can anticipate a situation and learn from their experience , they are more likely to efficiently undergo transitions when they encounter similar situations in the future [34]. Moreover, according to proactive coping theory [31], people can anticipate or recognise potential stressors and act in advance to modify their effects [7]. Proactive coping involves five interrelated stages to detect and respond to potential stressors, namely, resource accumulation, attention to/recognition of potential stressors, initial appraisal, preliminary coping efforts, and elicitation and use of feedback [31,36]. The recognition of stressors related to future eldercare, such as parents' ageing and anticipation of future care needs, may initiate the caregiving preparation. With the assumption that future caregiving tasks are stressful events, the proactive coping theory has been more frequently used to study the care preparation among older adults [7,12] than adult children. Future studies may consider systematically examining adult children's caregiving preparation experiences from the perspective of anticipatory socialisation and proactive coping. A more comprehensive and theory-guided conceptualisation of caregiving preparation activities can help future research better investigate adult children's essential involvement in the different stages of the caregiving preparation process and examine the relationships among different components of caregiving preparation and their impacts. Understanding the intrinsic connections and transitions between the different components can help provide targeted interventions to promote the development and maturation of the whole process. Additionally, attention should also be paid to the interactions and communications between adult children and older parents at multiple stages of the care preparation process. Although some studies highlighted the communication and discussion between adult children and older parents, few adopted an intergenerational perspective to understand their different roles, mutual influences, and practical initiatives to care preparation. Examining adult children's caregiving preparation experiences from an intergenerational perspective can provide a more comprehensive overview of care preparation that considers the familial context and how different generations influence each other. This would help to develop customised intervention programmes to assist these two groups in overcoming difficulties they encountered in the care preparation process. The results of the systematic review also identified a lack of comprehensive and validated measurements for the entire care preparation process. Existing studies basically only assessed certain phases, especially the preliminary stage, such as the awareness of caregiving responsibility and how adult children consider providing care to their parents. The lacking evidence and assessment of subsequent stages, such as the final decision-making stage, leads to a lack of evaluation of the whole process. In view of the limitations in the validities of the measurements, future studies should consider developing and validating measures of comprehensive caregiving preparation activities of adult children. Although the reported prevalence of caregiving preparation cannot be compared between studies considering the variations in the types of activities examined, most studies reported a high prevalence of caregiving preparation. Studies also highlighted that many adult children were anticipating future caregiving needs and responsibilities, but few adult children were making definite plans [27,33], as evidenced by the low frequency of preliminary planning and final decision-making [13]. This phenomenon could be due to the absence of apparent health problems and urgent care needs among the parents of these adult children. Further research is required to investigate the caregiving preparation experiences of adult children, to further explore the barriers and challenges to their caregiving preparations, which can serve as a basis for the development of key strategies for addressing contingencies. The influencing factors for caregiving preparation have rarely been explored from a theoretical perspective, with only a few studies applying attachment theory or role theory to guide their investigations. Furthermore, considering that all the reviewed studies had a cross-sectional design, limited knowledge is available regarding how the caregiving preparation of adult children progresses over time and what are the causal effects of the aforementioned influencing factors. Future studies should adopt an integrative framework, preferably combined with a longitudinal research design, to investigate the predictors of multistep care preparation activities among adult children. Notably, few studies examined the potential effects of the caregiving preparation of adult children. Although Western scholars have frequently discussed the potential benefits of care preparation in reducing the potential stress due to caregiving, there is limited empirical evidence supporting the role of care preparation in reducing the stress caused by anticipating or actually taking on the caregiving role among adult children. Thus, scholars should empirically investigate the potential effects of caregiving preparation, especially the reduction of caregiver stress, concerns about providing eldercare, and the enhancement of caregiving preparedness and other aspects of the well-being of adult children and their older parents. Overall, the literature on the caregiving preparation of adult children is still limited and primarily focused on Western countries, and few studies have examined the care preparation activities and processes of Asian adult children. Considering that Asian societies emphasise filial piety and filial care, adult children play a crucial role in meeting the caregiving needs of older adults, and they are more likely than Western adult children to become primary caregivers of older parents. Cultural sensitivity is crucial to care preparation activities in Asia. Therefore, scholars should further investigate the unique characteristics of care preparation of Asian adult children. --- Conclusions The present study systematically reviewed the literature on the caregiving preparation of adult children for future eldercare needs. The identified studies have several limitations, including the lack of consistent conceptualisation of caregiving preparation, comprehensive examination of the process of caregiving preparation, intergenerational perspectives, validated measures, longitudinal research design, representative sampling, guiding frameworks for examining influencing factors and consequences, and the underrepresentation of Asian populations. Future studies should aim to address these research gaps to enhance the understanding of caregiving preparation and facilitate the development of interventions to promote care preparedness and well-being in ageing families. --- Data Availability Statement: The data that support the findings of this study are available from the corresponding author upon reasonable request. --- Author Contributions: C.L. contributed to data collection, data analysis, and manuscript writing. J.H. contributed to manuscript writing and reviewing. X.B. performed critical revisions of the manuscript. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. --- Institutional Review Board Statement: This study is part of a larger research project with the ethical code of HSEARS20200820002, approved by the PolyU Institutional Review Board. Informed Consent Statement: Not applicable. ---

With the increasing life expectancy and ageing population, long-term care has become an urgent policy issue worldwide. The informal care provided by family members, particularly adult children, is a key aspect of long-term care. However, socioeconomic transformations have resulted in changing family and demographic structures and increased geographic mobility, reducing the capacity of families to provide informal care and meet the caregiving needs of older adults. For ageing families, care preparation can be an effective method for coping with eldercare challenges, and care preparation is attracting increasing attention from researchers. This study seeks to conduct a systematic review for studies on caregiving preparation by adult children that were retrieved from six databases, to synthesise the available evidence, and to identify knowledge gaps and opportunities for future investigations. The characteristics and main themes of eighteen reviewed articles were identified and analysed. This study discussed various conceptualisations of caregiving preparation by adult children, the prevalence of caregiving preparation, the factors related to caregiving preparation, and the related consequences. On the basis of the systematic review findings, several limitations of the literature and directions for future research were proposed to promote care preparedness and the well-being of ageing families.

Introduction --- Background on Non-Binary Identities in the Early 21st Century The early 21st century marks a significant era in the recognition and understanding of non-binary identities. These identities, transcending the traditional binary concept of gender, have become increasingly acknowledged and celebrated in contemporary discourse. The linguistic evolution toward inclusivity reflects broader societal changes, challenging established norms and fostering a more comprehensive understanding of gender diversity. This cultural shift is not confined to semantics; it mirrors a profound reconsideration of societal perspectives on gender. Non-binary individuals, whose experiences defy the conventional male-female categorization, have become integral to discussions about gender identity. Against this backdrop, literature emerges as a powerful tool for articulating and exploring the multifaceted experiences of non-binary individuals, providing nuanced representations that contribute to a more inclusive cultural narrative. --- Intersectionality of Feminism and Non-Binary Representation The intersectionality of feminism and non-binary representation encapsulates the dynamic interplay between gender and various other facets of identity, such as race, class, and sexuality. At its core, feminism is a movement dedicated to dismantling patriarchal structures and advocating for the rights and dignity of all genders. Within this framework, the inclusion of non-binary voices is not just a natural progression but a vital expansion of the feminist dialogue. Non-binary representation in feminist literature becomes a conduit for challenging binary gender paradigms and advocating for a more expansive, fluid conceptualization of gender. This intersectionality recognizes and amplifies the diverse narratives of non-binary individuals navigating intersecting systems of oppression. It underscores that the pursuit of gender equality encompasses the liberation of all gender identities. As we commence our exploration in this literary critique, our aim is to unravel the specific ways in which non-binary identities are portrayed and understand the consequential impact on the broader narrative. Additionally, we will scrutinize the intricate dynamics between non-binary representation and other social categories, contributing to a richer comprehension of gender representation in feminist literature during the early 21st century. --- Literature Review --- Historical Context of Non-Binary Representation in Literature To appreciate the contemporary landscape of non-binary representation in literature, an exploration of its historical roots is imperative. The 20th century laid the groundwork for challenging gender norms and expanding the literary portrayal of diverse gender identities. Notable works, such as Virginia Woolf 's Orlando, hinted at the fluidity of gender, anticipating the nuanced approach that would later define non-binary representation. The late 20th century witnessed an increased focus on gender diversity, with authors like Leslie Feinberg contributing groundbreaking works like Stone Butch Blues. This novel not only highlighted the struggles of a non-binary protagonist but also initiated a broader conversation about gender non-conformity. Understanding this historical context is essential for contextualizing the emergence of non-binary representation in the early 21st century. --- Feminist Literature Trends in the Early 21st Century Examining the trends within feminist literature during the early 21st century reveals a notable evolution in the exploration of gender identities. A surge in intersectional feminist perspectives, championed by writers like Chimamanda Ngozi Adichie and Roxane Gay, has broadened the scope of feminist discourse. These authors bring attention to the interconnectedness of gender with race, sexuality, and other dimensions of identity. The rise of digital platforms and social media has empowered marginalized voices, including non-binary individuals, to share their narratives. Online spaces have become hubs for discussions on gender diversity, influencing the thematic choices of contemporary feminist literature. The digital era has facilitated a democratization of storytelling, allowing non-binary voices to be heard and recognized in mainstream literary conversations. --- Gaps in the Literature Despite the progress made in recent years, gaps persist in the literature concerning the representation of non-binary identities. The historical trajectory of non-binary representation in literature is marked by sporadic moments rather than a consistent, sustained exploration. There remains a need for more expansive and nuanced portrayals that move beyond stereotypes and tokenism. While the digital age has amplified diverse voices, it has also introduced challenges. The rapid dissemination of information online may result in the oversimplification or commodification of non-binary identities. This raises questions about authenticity and the potential for misrepresentation within the literature. As we navigate the early 21st-century literary landscape, these gaps provide a roadmap for understanding where further exploration and nuanced representation are required. This literature review sets the stage for our subsequent analysis, offering insights into the historical foundations, contemporary trends, and existing gaps in the representation of non-binary identities within feminist literature. As we delve into specific works, we aim to contribute to the ongoing dialogue and address these gaps through a critical examination of the narratives that have shaped and reflected the experiences of non-binary individuals. --- Methodology --- Criteria for Selecting Literary Works The selection of literary works for analysis is guided by a set of criteria designed to ensure a diverse and representative exploration of non-binary representation within feminist literature. The following criteria shape the inclusion of works in this study: Publication Period: Literary works considered for analysis fall within the early 21st century, aligning with the temporal focus of this study. This timeframe captures the contemporary evolution of feminist literature alongside the increasing visibility of non-binary identities. Genre Diversity: To encompass a broad spectrum of perspectives, the chosen works span various genres, including novels, poetry, essays, and digital media. This approach acknowledges the diverse literary landscape and the multitude of ways non-binary identities are expressed within different genres. Authorial Diversity: Recognizing the importance of authentic representation, priority is given to works authored by individuals who identify as non-binary or possess an in-depth understanding of non-binary experiences. Additionally, efforts are made to include works from a variety of cultural backgrounds, ensuring a global and intersectional perspective. Critical Acclaim and Relevance: Works selected exhibit a degree of critical acclaim and relevance within feminist and literary circles. This criterion ensures that the chosen works have had a notable impact on shaping discussions around gender diversity and have contributed to the broader literary discourse. --- Methods for Data Collection and Analysis The methodology employed for this study involves a multifaceted approach to data collection and analysis, aiming for both depth and breadth in understanding non-binary representation in feminist literature. Close Reading: A foundational aspect of the analysis involves close reading of selected literary works. This method allows for an in-depth exploration of language, themes, and characterizations related to non-binary identities. Close reading enables the identification of nuanced portrayals and the extraction of subtle meanings embedded in the texts. Thematic Analysis: Thematic analysis is employed to identify recurring themes and patterns within the selected works. This method facilitates the recognition of overarching narratives, common motifs, and the evolution of non-binary representation throughout the early 21st century. Thematic analysis provides a structured framework for understanding how literary portrayals contribute to the broader discourse on gender diversity. Intersectional Lens: An intersectional lens is applied throughout the analysis to recognize the interconnectedness of non-binary identities with other social categories, such as race, class, and sexuality. This approach ensures a holistic understanding of the complex ways in which identity is constructed within the selected literary works. Through the combination of these methodologies, this study aims to offer a comprehensive and nuanced examination of non-binary representation in feminist literature, contributing valuable insights to the ongoing dialogue surrounding gender diversity in literary spaces. --- Non-Binary Representation in Selected Works in Early 21st Century --- Analysis of Specific Examples Akwaeke Emezi's Freshwater immerses readers in a narrative that transcends traditional notions of identity. The protagonist, Ada, grapples with a fractured sense of self, inhabited by multiple spirits. While the novel doesn't explicitly explore non-binary identity, Ada's fluidity challenges established norms, creating a canvas for understanding the intricate relationship between mental health, spirituality, and non-binary themes. Emezi's narrative choices, such as the use of multiple perspectives and mythic symbolism, contribute to a nuanced portrayal of a gender-fluid existence. Rivers Solomon's An Unkindness of Ghosts propels readers into a speculative future where societal hierarchies echo historical inequalities. The character of Aster Grey disrupts gender norms within the confines of a spaceship society. Solomon employs the sci-fi genre to create a nuanced exploration of gender identity, challenging real-world societal structures. Through Aster's journey, Solomon navigates the impact of gender fluidity on power dynamics, contributing to the broader discourse on non-binary representation. Jeffrey Eugenides' Middlesex offers a sprawling family saga that delves into the life of Calliope Stephanides, later known as Cal, who is born intersex. While not explicitly non-binary, Cal's journey explores the complexities of gender identity within a broader historical and cultural context. Eugenides weaves a narrative that challenges binary perspectives on gender, examining the intersections of identity, heritage, and societal expectations. --- Portrayal and Impact on the Overall Narrative In Freshwater, Emezi's portrayal of Ada's fractured identity serves as a metaphor for the fluidity of gender, inviting readers to question normative constructs. The impact on the narrative is profound, as Ada's various selves interact and influence the overarching story. Emezi's stylistic choices and narrative techniques create a literary landscape that challenges and expands the understanding of non-binary representation within the early 21st century feminist literature. Similarly, in An Unkindness of Ghosts, Solomon's portrayal of Aster Grey challenges gender norms within a futuristic society. The impact extends beyond individual identity, shaping the broader societal critique woven into the narrative. Solomon's exploration of how gender fluidity intersects with power structures offers readers a compelling lens through which to examine the complexities of non-binary representation in the early 21st century. In Middlesex, Eugenides' portrayal of Cal's intersex identity disrupts conventional gender narratives. The impact on the narrative is seen in the multigenerational exploration of identity, challenging societal expectations and illustrating the fluidity of gender across time. Eugenides' narrative choices contribute to a nuanced understanding of non-binary themes within a historical context. --- Intersectionality with Other Social Categories Exploring the intersectionality of non-binary identities with other social categories enriches our understanding of representation. In Freshwater, the intersectionality is nuanced, blending mental health and spirituality with gender identity. Ada's struggles are not solely about gender but also encompass cultural and spiritual dimensions. Emezi's narrative unfolds within a complex web of intersectional experiences, contributing to a holistic representation of non-binary themes. An Unkindness of Ghosts intersects non-binary identity with issues of race and class within the spaceship society. Solomon's narrative choices highlight the interconnectedness of gender with broader societal structures, offering readers a thought-provoking exploration of how non-binary representation intersects with and challenges existing power dynamics. In Middlesex, the intersectionality extends to questions of heritage and cultural expectations. Eugenides explores how intersex identity intersects with familial and societal expectations, adding layers to the understanding of non-binary themes. The narrative unfolds against the backdrop of historical events, further emphasizing the intersectional dimensions of gender identity. This section illuminates the multifaceted nature of non-binary representation in early 21st-century feminist literature. Through the critical analysis of three selected works, the discussion aims to deepen our comprehension of the complexities inherent in portraying non-binary identities and their intersections with other social categories. --- Themes and Patterns --- Common Themes in the Representation of Non-Binary Identities The exploration of non-binary identities in early 21st-century feminist literature reveals several recurring themes that contribute to a nuanced understanding of gender diversity. Across the selected works -Freshwater by Akwaeke Emezi, An Unkindness of Ghosts by Rivers Solomon, and Middlesex by Jeffrey Eugenides-distinctive patterns emerge: --- Fluidity and Multiplicity Representation in Freshwater: Akwaeke Emezi portrays fluidity through Ada's fractured identity, where multiple spirits coexist. Representation in An Unkindness of Ghosts: Rivers Solomon explores fluid gender norms within a futuristic society, challenging binary distinctions. Representation in Middlesex: Jeffrey Eugenides illustrates the multiplicity of identity through Cal's intersex experience, challenging rigid gender categorizations. --- Intersectionality of Identity Representation in Freshwater: Emezi intertwines mental health and spirituality with gender identity, emphasizing the interconnectedness of experiences. Representation in An Unkindness of Ghosts: Solomon addresses issues of race and class in conjunction with non-binary identity, offering a layered exploration. Representation in Middlesex: Eugenides delves into the intersection of intersex identity with familial and cultural expectations, enriching the narrative with diverse layers. --- Challenges to Societal Norms Representation in All Works: Each narrative challenges societal norms related to gender, presenting non-binary identities as valid and worthy of exploration. --- Evolution of Themes Throughout the Early 21st Century The evolution of themes surrounding non-binary identities reflects the changing landscape of societal attitudes towards gender diversity. In the early 2000s, literature began to move beyond binary frameworks, allowing for more expansive and authentic portrayals of gender. The themes evolved from mere questioning and exploration to a robust challenge against normative structures, advocating for the acknowledgment and celebration of diverse gender expressions. Early 2000s: Works such as Middlesex laid the groundwork for questioning binary norms, introducing mainstream audiences to intersex experiences. Mid-2000s to 2010s: The narratives in Freshwater and An Unkindness of Ghosts represent a shift towards fluid and intersectional explorations, emphasizing the multiplicity of identity and challenging societal norms. --- Implications of Themes in Feminist Literature The emergence and evolution of these themes hold profound implications for feminist literature: Inclusive Narratives: The representation of non-binary identities expands the scope of feminist literature to be more inclusive, embracing a spectrum of gender experiences beyond the binary. Intersectional Feminism: By intertwining non-binary identities with other social categories, these narratives contribute to an intersectional feminist discourse, acknowledging the interconnectedness of gender with race, class, and cultural factors. Deconstruction of Gender Norms: The challenge posed to societal norms around gender fosters a feminist literature that actively participates in deconstructing and reimagining traditional gender categories. In conclusion, the identification of common themes and the evolution of these themes across early 21st-century feminist literature provide a comprehensive understanding of how non-binary identities have been portrayed and the transformative impact on the broader feminist discourse. These themes not only reflect societal shifts but also propel feminist literature into a more inclusive and dynamic realm, redefining the boundaries of gender representation. --- Critique of Representation The effectiveness of non-binary representation in early 21st-century feminist literature is evident in the progress made towards a more inclusive and diverse portrayal of gender identities. The selected works -Freshwater by Akwaeke Emezi, An Unkindness of Ghosts by Rivers Solomon, and Middlesex by Jeffrey Eugenides-contribute to the ongoing dialogue about gender diversity. Empowerment and Visibility: The representation of non-binary identities empowers individuals who find resonance in these narratives, offering visibility to experiences often marginalized in mainstream discourse. Challenging Binary Constructs: By challenging binary constructs, these narratives contribute to a broader societal understanding of gender, fostering inclusivity and acceptance. Diverse Lenses: Each work provides a unique lens through which to explore non-binary experiences, acknowledging the multifaceted nature of gender identity. Intersectionality: The intersectional exploration of identity in these narratives broadens the scope of feminist literature, fostering a more comprehensive understanding of the complexities within gender diversity. While the selected works contribute significantly to positive representations, it is crucial to acknowledge areas where misrepresentation or problematic portrayals may arise: Stereotypical Tropes: Some works may unintentionally reinforce stereotypes associated with non-binary individuals, limiting the depth and authenticity of representation. Tokenism: Instances where non-binary characters are included merely for the sake of diversity without substantial development can perpetuate tokenistic portrayals. Overlooking Complexity: Certain works may oversimplify the intersectionality of non-binary identities, neglecting the nuanced interplay between gender, race, class, and other social categories. The role of authors and publishers is pivotal in shaping non-binary representations and influencing their reception: Authenticity and Research: Authors bear the responsibility of conducting thorough research to portray non-binary identities authentically, avoiding reliance on stereotypes. Intersectional Understanding: A nuanced understanding of intersectionality allows authors to create characters that reflect the diversity of real-world experiences. Editorial Guidance: Publishers play a crucial role in providing editorial guidance that ensures accurate and respectful portrayals, fostering a collaborative effort between authors and editors. Market Trends: Commercial considerations may influence publishers, potentially leading to the prioritization of market trends over authentic representation. This can impact the diversity of voices presented to readers. Impact on Perception: The way representations are received by readers shapes societal perceptions. Positive portrayals can challenge biases, while problematic ones may reinforce stereotypes. In conclusion, the effectiveness of non-binary representation in early 21st-century feminist literature is evident in the positive impact on visibility and the expansion of perspectives. However, it is crucial to critically examine instances of misrepresentation or problematic portrayals, recognizing the role of authors, publishers, and readers in shaping and interpreting these representations. A continuous dialogue and commitment to authenticity are essential for ensuring that non-binary representation in literature remains a catalyst for positive change and understanding. --- Intersectionality and Inclusivity --- How Intersectionality Contributes to Nuanced Representation The concept of intersectionality, introduced by Kimberlé Crenshaw, emphasizes the interconnected nature of social categories such as race, gender, class, and sexuality. In the context of non-binary representation in early 21st-century feminist literature, intersectionality contributes to a nuanced portrayal in the following ways: Blending of Identities: Intersectionality acknowledges that individuals embody multiple identities simultaneously. This is evident in the works analyzed -Freshwater, An Unkindness of Ghosts, and Middlesexwhere characters navigate not only their non-binary gender but also intersecting aspects of race, class, and culture. Societal and Historical Context: Intersectionality allows for a more profound exploration of how societal and historical contexts shape non-binary identities. The characters' experiences are not isolated but influenced by a complex interplay of social factors. Counteracting Stereotypes: An intersectional approach helps counteract the risk of stereotyping non-binary individuals. By considering the multifaceted nature of identity, authors can create characters that defy simplistic categorizations. --- Importance of Inclusivity in Feminist Literature Inclusivity is foundational to feminist literature as it strives to represent the diverse experiences of individuals, acknowledging their unique intersections of identity. In the realm of non-binary representation, inclusivity holds significant importance for the following reasons: Validation and Recognition: Inclusive literature validates the experiences of individuals whose identities may be marginalized or overlooked. Non-binary readers, particularly those with intersecting identities, find recognition and resonance in narratives that authentically reflect their lives. Expanding Readers' Understanding: Inclusive literature broadens the perspectives of all readers, challenging preconceived notions and fostering empathy. Readers are exposed to a rich tapestry of diverse experiences, contributing to a more comprehensive understanding of the human experience. Amplifying Voices: Inclusivity ensures that marginalized voices are not only heard but also amplified. By providing a platform for non-binary individuals with intersecting identities, feminist literature becomes a catalyst for social change and empowerment. Intersectionality as Core Tenet: Inclusivity aligns with the principles of intersectional feminism, recognizing that feminism must address the interconnected nature of oppression and privilege. Non-binary representation becomes a vehicle for dismantling systemic inequalities. Educating and Informing: Inclusive literature serves as an educational tool, challenging societal biases and fostering a more informed readership. It promotes dialogue around the complexities of identity, contributing to a more inclusive and equitable society. In conclusion, intersectionality enriches the representation of non-binary identities in feminist literature by acknowledging the complex interplay of various social categories. Inclusivity is not just a moral imperative but a transformative force within feminist literature, shaping narratives that resonate with a diverse readership and challenging ingrained societal norms. The ongoing commitment to intersectionality and inclusivity ensures that feminist literature continues to evolve as a powerful tool for social understanding and advocacy. --- Impact on Readership and Society --- Investigating the Influence on Readers' Perceptions The representation of non-binary identities in early 21st-century feminist literature has a profound impact on readers' perceptions, shaping their understanding of gender diversity and fostering empathy. The selected works -Freshwater, An Unkindness of Ghosts, and Middlesexcontribute to this influence in significant ways: Personal Connection: Readers often form a personal connection with characters navigating non-binary identities. This emotional engagement fosters empathy, allowing readers to step into the shoes of individuals with diverse gender experiences. Challenging Preconceptions: The nuanced portrayal of non-binary characters challenges readers' preconceptions about gender. By presenting characters with intersecting identities, literature encourages a more expansive and nuanced understanding of the complexities within the non-binary experience. Validation for Non-Binary Readers: Non-binary readers, in particular, find validation and affirmation in literature that authentically represents their experiences. Seeing diverse and authentic portrayals contributes to a sense of visibility and acknowledgment. --- Examining Societal Attitudes Toward Non-Binary Individuals The societal impact of non-binary representation extends beyond individual readers to influence broader attitudes and perceptions: Catalyzing Cultural Change: As literature reflects and shapes culture, the authentic portrayal of non-binary identities contributes to cultural shifts. It challenges traditional norms and encourages society to embrace a more inclusive and accepting perspective on gender. Destigmatizing Non-Binary Identities: Positive representation helps challenge stigmas and misconceptions surrounding non-binary individuals. It counters harmful narratives and fosters an environment where diverse gender expressions are recognized and respected. Influence on Advocacy Efforts: Literature can serve as a catalyst for advocacy efforts, influencing policy changes and legal recognition of non-binary individuals. By shaping public discourse, literature contributes to the dismantling of discriminatory practices. Integration into Education: Non-binary representation in literature has educational implications, influencing curricula to be more inclusive. By exposing students to diverse narratives, educational institutions contribute to the normalization of non-binary identities. Impact on Media Representations: The influence of literature extends to other forms of media and popular culture. Positive representations in literature may inspire similar portrayals in film, television, and other media, further shaping societal attitudes. In conclusion, the impact of non-binary representation in early 21st-century feminist literature is twofold: it influences individual readers by fostering empathy and expanding understanding, and it contributes to broader societal changes by challenging norms, destigmatizing non-binary identities, and catalyzing cultural shifts. As literature continues to play a pivotal role in shaping perceptions, the positive representation of non-binary identities emerges as a potent force in advancing inclusivity and acceptance within contemporary society. --- Conclusion The exploration of non-binary representation in early 21st-century feminist literature reveals a dynamic landscape where authors engage with the complexities of gender diversity in nuanced ways. Through the analysis of selected works -Freshwater by Akwaeke Emezi, An Unkindness of Ghosts by Rivers Solomon, and Middlesex by Jeffrey Eugenides-this paper has sought to unravel the multifaceted dimensions of non-binary identities and their portrayal within the feminist literary discourse. The selected works demonstrate a commitment to nuanced portrayals of non-binary identities, acknowledging the intersecting factors that shape gender experiences. Characters in these narratives defy stereotypes, contributing to a more authentic representation. Intersectionality emerges as a crucial lens through which non-binary identities are explored. The works analyze how gender intersects with race, class, and culture, emphasizing the interconnected nature of individual experiences. Themes within the selected works evolve over the early 21st century, reflecting a progression from questioning binary norms to challenging societal structures. The narratives contribute to a broader conversation about the fluidity and complexity of gender. Inclusivity is identified as foundational to feminist literature, with positive representation contributing to empowerment and validation for non-binary readers. The impact extends beyond individuals to influence societal attitudes, challenging stigmas and fostering cultural change. The evolving landscape of non-binary representation calls for a continued dialogue within feminist literature. Authors, publishers, and readers must engage in conversations that push the boundaries of understanding and challenge persistent norms. The intersectional exploration of non-binary identities is essential for providing a comprehensive understanding of the diverse experiences within this community. Future works should strive for a more nuanced portrayal that considers the interplay of various social categories. The positive impact of non-binary representation in literature has the potential to catalyze advocacy efforts and contribute to broader social change. Continued efforts in media, education, and policy advocacy can further dismantle discriminatory practices. Comparative studies analyzing non-binary representation across different cultural and geographical contexts could provide insights into the universality or specificity of certain themes and challenges. In-depth studies on reader reception, including diverse perspectives and reactions, can offer a more comprehensive understanding of the varied ways in which non-binary representation resonates with audiences. Exploring the direct impact of literature on legal frameworks and policies related to non-binary recognition and rights can provide a clearer understanding of the influence of cultural representation on institutional change. In conclusion, the exploration of non-binary representation in early 21st-century feminist literature is not only a scholarly endeavor but a reflection of societal shifts and aspirations for a more inclusive and accepting world. As literature continues to evolve, it remains a powerful tool for challenging norms, fostering empathy, and contributing to the ongoing journey toward a more equitable and diverse society. --- Copyrights Copyright for this article is retained by the author, with first publication rights granted to the journal. This is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license .

This academic paper examines the representation of non-binary identities in early 21st-century feminist literature, focusing on the works Freshwater, An Unkindness of Ghosts, and Middlesex. The analysis explores the nuanced portrayals of non-binary characters, considering the intersections with race, class, and culture. The paper identifies evolving themes throughout the early 21st century, emphasizing the fluidity and complexity of gender in these narratives. Inclusivity is highlighted as foundational to feminist literature, contributing to empowerment and validation for non-binary readers. The impact extends beyond individuals, influencing societal attitudes, challenging stigmas, and catalyzing cultural change. The conclusion emphasizes the need for continued dialogue, intersectional exploration, and advocacy efforts to promote inclusivity in literature and contribute to broader social change.

Introduction Ageing, dependency and the distribution of resources for long-term care are issues of vital importance in today's modern society. Europe, in particular, has undergone a demographic shift, with birth rates declining while longevity increases. In fact, it is estimated that, by 2060, over 30% of the European population will be over 65, with 12% of the total population being over 80 [1]. This has led to an increased interest from governing institutions in understanding how and where exactly this change is happening so that policy, both at a national and European level, can be better tailored to meet the needs of this evolving situation [2]. Of particular interest is the long-term care needs of older individuals, particularly of those who have disabilities that prevent them from performing certain tasks that relate to daily life. Individuals who require such long-term care are said to be dependent. Dependency is highly complex and intersectional with many factors. These can be physical ; psychological ; cognitive ; economic ; social ; etc. [3]. It is important to view dependency from a holistic approach, understanding all factors that impact on an individual's level of dependency. Being considered dependent by the state is often the only way an individual can gain access to a variety of public services related to long-term care. However, how an individual is designated "dependent" or not is remarkably heterogeneous across Europe [4,5]. Long-term care services are essential for older people with dependency and include a range of services, including domestic help, personal care, nursing care, general health care, and services such as "meals-on-wheels" [6]. In addition, there is a large reliance on informal care, in particular in Southern European countries, which is affected by national policy [7,8]. Another commonly accepted definition of dependency is that of the Recommendation No. R 9 of Council of Europe that states dependency is "such state in which people, whom for reason connected to the lack or loss of physical, mental or intellectual autonomy, require assistance and/or extensive help in order to carry out common everyday actions" . Many other definitions exist in other countries or institutions, such as that of the World Health Organization , which states that dependency relates to any restriction or lack of ability to perform an activity in the manner or within the range considered normal for a human being [9]. While definitions of who is dependent differ from country to country and institution to institution, it is without doubt that disability increases with age. Given the demographic changes discussed above, it is essential that European countries prepare for the ageing population and their increased reliance of public long-term care services. As this change is European wide, understanding who and where the most dependent populations are can inform government institution and shape how and where any policy work should be directed. In this paper, we use the Survey of Health, Ageing and Retirement in Europe as a useful tool to track this phenomenon across Europe. SHARE is a multidisciplinary and cross-national panel database of micro data on health, socio-economic status and social and family networks of about 140,000 individuals aged 50 or older . It covers 27 European countries and Israel. It was founded in 2002 and is coordinated centrally at the Munich Center for the Economics of Aging , the Max Planck Institute for Social Law and Social Policy . SHARE is harmonized with its role models and sister studies the US Health and Retirement Study and the English Longitudinal Study of Ageing . This model has sparked and informed exciting new survey research on ageing all over the world, e.g., Japan , China , Brazil , South Korea , and India , which puts SHARE into a truly global perspective. The SHARE interview is ex-ante harmonized and all aspects of the data generation process, from sampling to translation, from fieldwork to data processing, have been conducted according to strict quality standards. As a result, SHARE has the advantage of encompassing cross-national variation in public policy, culture and history across a variety of European countries. SHARE has therefore become a major pillar of the European Research Area with over 9000 researchers registered as SHARE users. More than 1350 journal papers have been published using this data source, and over 2300 publications have cited SHARE data in their papers. Further details can be found on their website . The main contribution of this paper is to propose a procedure to visualize profiles of individuals, that is, groups of individuals with similar characteristics, using mixed-type survey data . In particular, we design four health and socio-economic wellbeing composite indices that are used to obtain the profiles and essential to better understand and interpret them. Next, all socio-economic and demographic variables about the respondents are combined with these indices via the k-prototypes algorithm [10] and clustered to create several profiles. The methodology proposed is wide enough to be extended to other surveys or disciplines. This paper aims to provide a useful guide to the social and economic researchers in the visualization of profiles of individuals, that is, groups of individuals with similar characteristics, from very large mixed datasets using existing tools. --- Literature Review Today, there is still an open debate about the concept of wellbeing and how to measure it properly in individuals. This concept of wellbeing can incorporate several variables depending on the field of study . The most common characterization considers no mutually exclusive approaches, such as subjective, objective, material and relational wellbeing. See references [11][12][13][14][15] for different approaches on categorizing wellbeing. There are plenty of studies measuring wellbeing, either focusing on general population or on particular groups such as children, the elderly, women, minorities and others. See references [16,17] among others. Some studies of note include references [18,19], which discuss how feeling useful in life halves the risk of developing disabilities related to ADLs. Walker [20] highlights the importance of objective health and wellbeing in the quality of life of older people. It can be seen in references [21,22] that health status shows a strong relationship with quality of life, mainly during the second half of life. According to the worldwide study by Steptoe et al. [23], wellbeing and health are closely related to age. Thus, wellbeing seems to play a protective role in being healthy, although other factors may be considered when studying longevity. In fact, mobility restrictions and other functional disorders related to ADLs and IADLs can be considered relevant issues in wellbeing and should be considered in any study on the subject [24,25] Cristea et al. [26] considers a major challenge for the European Union: public health expenditure is mainly focused on older, dependent people. Sustainable economic development relies on a well-balanced workforce of young and old. If an imbalance in favor of older people occurs, productivity will suffer, and older people demand more health services. Social protection systems , according to Eurofound [27], are intended to protect people against the risks associated with old age, illness and disability, unemployment, costs of children and housing. In particular, article 2 of the EU Treaty identifies the promotion of a high level of social protection as a key task of the EU. Social protection systems include benefits both in cash and in kind. As such, the European Pillar of Social Rights highlights access to long-term care, healthcare, old age pensions, childcare, and social housing alongside access to unemployment benefits and minimum wage policies. In Esping-Andersen [28], different social welfare models are distinguished . Specifically, this author speaks of Anglo-Saxon, Corporative, Scandinavian, and Southern-Europe models. Today, regarding long-term care, Central and Northern European countries are adhered to the "north-continental" long-term care system, which can be seen as inheritor of Corporative and Scandinavian social welfare models. This long-term care model is financed via tax and public funds with a medium-high development of private insurance sector. On the other hand, Southern European countries tend to have the "Mediterranean" long-term care system, which is financed via tax and co-payments with a very low development of the private insurance sector. Additionally, this second long-term care model has a stronger dependency on informal care by households and families [7,8]. The social characteristics of care, such as the availability of formal care, and the proportions of people receiving various types of care differ greatly in European countries [6]. The increasing of life expectancy and declining fertility rates in Europe makes care for older people one of the most important issues at a policy level [29]. A big increase in the need for care and a smaller number of potential informal carers can be expected due to the rapid ageing population. Barczyk and Kredler [7] make a comparison of the regulations of eleven public national or regional long-term care programs in Austria, Belgium, the Czech Republic, France, Germany, Italy, and Spain. The comparison is made in terms of their degree of inclusiveness with respect to vulnerable older individual's health status. These authors provide detailed information on assessment and eligibility frameworks for long-term care programs in Europe and compare them by using the SHARE data. According to Hlebec et al [29], research must focus on developing instruments and methods that produce reliable measurements on health that are comparable across countries. These tools help to better understand the health problems of older adults globally, and to facilitate appropriate health and social policy responses. Additionally, developing these tools will contribute to the collection of comparable health and ageing data across countries that in turn facilitate international comparisons and learning from the experience of others [30]. Other interesting papers about long term care, health, ageing, and wellbeing include references [31][32][33][34]. This paper aims to fill a literature gap by searching for a comprehensive, cross-national overview that takes into account key issues related to objective and subjective health and long-term care and wellbeing. In particular, this paper contributes to the study of wellbeing and health combining objective and subjective variables of mixed type from the SHARE dataset in order to visualize differences between groups of Europeans. Having an overview of the situation in Europe is relevant for the development of socioeconomic and health policies. Most countries aim to enable people to live "longer and better." Increasing life expectancy is not, in itself, a guarantee of a better quality of life [35]. --- Methodology Mixed-type data comprise both numeric and categorical features, and mixed datasets frequently occur in socio-demographic surveys as well as in other disciplines, such as economics, health, finance, marketing, etc. One of the challenging aspects of dealing with mixed-type data is to find structures and to group similar individuals for further analysis. Additionally, the availability of large and very large datasets adds complexity to this matter. The method that we propose is based on two stages: The first stage consists of the design of several composite indices, each of which focusses on a particular issue of interest. In the second stage, these thematic indices are combined with other descriptive variables through a clustering procedure able to cope with large datasets of mixed data. In this work, we use the k-prototypes clustering algorithm [10], and we illustrate the methodology through the SHARE database. --- First Stage In this stage, the survey information is split into two groups: descriptive variables and thematic blocks. Next, indicators are designed within each thematic block. The procedure consists of the following steps: Select the group of descriptive variables. This group does not contribute to the indices and it is left unchanged. The kind of variables to be included in this group is usually related to socio-demographic information such as gender, age, education, employment, etc. Thematic blocks: Group the rest of information in particular issues of interest yielding thematic blocks. For example, group the variables in four or five aspects to be studied. For each thematic block, construct a composite index. To do so, Variables within thematic blocks must be encoded. Here it can be of help to review some related literature with a view on how to treat each of the pertinent variables. Transform all the included variables into binary variables, always maintaining the same polarity. For example, assign 1 to the worst-case scenario and 0 to the best-case scenario. i. Always maintain consistency. For categorical variables, when a scale is identical for two or more of them, the same cut-off must be designated for these variables. For example, this means that any question where the response options are "Poor", "Fair", "Good", "Very good", or "Excellent", these variables must be transformed in the same manner using the same cut-off point. ii. For quantitative variables, thresholds such as the median or even more restrictive percentiles can be used. Leave binary variables unchanged. Finally, construct the thematic index by adding all binarized variables and rescale to 0-10. This index design has several advantages: First, the researcher can use both quantitative and qualitative variables. Second, qualitative variables can be measured in different scales. Third, the transformation into binary variables serves to unify scales and helps to identify patterns of interest. Fourth, the indices are additively constructed; thus, relevant new variables can be added if necessary. --- Second Stage Once thematic composite indices are constructed, in the next stage we combine them with the collection of descriptive variables, clustering the individuals in groups with similar characteristics and, thus, obtaining the profiles. Standard hierarchical clustering methods can handle data with numeric and categorical values by using Gower's similarity coefficient [36]. For instance, this measure was used in Albarrán et al. [4], and profiles were obtained via multidimensional scaling. However, this approach may be unfeasible for large datasets. Hierarchical clustering methods such as this have a computational complexity of O, where n is the number of observations, and with n > 10,000, this reinforces the unfeasibility of applying this methodology to large and very large datasets. In our application, n = 60,020. A popular algorithm in unsupervised learning and clustering is the k-means clustering method [37], which has proven to be efficient for processing large data sets, although it only works on numeric data, i.e., the variables measured on a ratio scale. In Huang [10], a combination of the k-means with the k-modes clustering algorithm by Huang [38] is proposed, yielding to the k-prototypes clustering algorithm, with a reduction of the computational cost to Okn), where n is the number of observations, k the number of clusters and T the number of iterations. As << n, we can see this is much less computationally intensive than, for instance, calculating Gower's similarity coefficient and applying multidimensional scaling. See references [39,40] for comprehensive reviews of clustering methods for mixed data. d 2 = p j=1 x ij -x lj 2 + γ m j=p+1 δ x ij , x lj where δ = 0 for p = q and δ = 1 for p q. Note that, when γ = 0 clustering only depends on numeric variables. The influence of this parameter is discussed in Albarrán et al. [15], where it is recommended to select its value in accordance with the average standard deviations of numeric variables, σ; that is, the author suggests to take γ between 1/3σ and 2/3σ. See also Huang [41]. As with any non-hierarchical clustering approach, the number of clusters, k, must be determined before running the algorithm. A variety of techniques are available to assist in the decision of the number of clusters to be selected. However, determining the optimal number of clusters is an inherently subjective measure that depends on the goal of the analysis. In this work, we use the "elbow" method. This method consists in running the algorithm with a varying k and calculating the cost function for each run. Next, cost function values are plotted in a line-graph. At the point where there is a turning point, or "elbow", the improvement in the cost function levels off. That is to say, the improvement by adding an additional grouping in the data is marginal, and the added complexity of understanding the profiles outweighs any benefits given. Finally, the description of the profiles can be done by looking at the "average" individual of each profile, regarding wellbeing indicators and descriptive variables. The modal class can be used for categorical variables, and for quantitative ones, either the mean or the median may be useful. --- Data and Methods Wave 6 of the Survey of Health, Ageing and Retirement in Europe is a rich panel database of individuals aged 50 or over in 18 European countries and Israel. This survey took place in 2015 and asked questions ranging from the respondents' financial situation to their self-perception of their health levels [42]. This survey is the only of its type that collects homogeneous information of this nature. This massive dataset aims to be representative; the 60,020 cases selected for this analysis include a weighting variable that scales to represent a target population of 124,313,623 aged 55 years or over individuals in Europe [43]. The data also contained many descriptive information about the respondent, such as their level of education, their marital status and others. A presentation of the descriptive variables can be found below and a description of how they were treated is in Section 4.1. To create the profiles, the survey responses were recoded into binary variables, and four indices were created. These were then used in the final clustering algorithm, along with the categorical data, to create the profiles. The process of creating the indices is described in Section 4.2, below. Descriptive socio-economic variables included in the analysis and their possible values or categories: Household receives benefits or has payments: Payments and no benefits, No benefits and no payments, Benefits and payments, Payments and no benefits. --- Descriptive Variables Age was split into brackets of five years, starting at 55 years. Marital status and employment status have been classed into binary responses following Albertini and Arpino [44]: either the individual has a spouse or they do not and similarly either they are working or they are not . Education has been classified into four classifications following Becchetti et al. [45]: no education, primary education, secondary education, and university education. Finally, as the focus of this paper is less on the financial aspects of ageing, the financial metrics were summarized into a combination of whether the respondent received payments from the government and whether they had out-of-pocket medical expenses. Binary variables, such as whether the respondent had children or not, were left as is. An interesting variable is financial distress, which is a binary variable on whether the household has trouble making ends meet from month to month. --- Index Creation In order to cluster the data efficiently, four indices were created from 28 variables listed in Table 1. The data was grouped by variables that dealt with the respondents' self-perception of their health, their physical health and nutrition, their mental agility, and their level of dependency. To encode these variables, papers using the SHARE dataset were reviewed with a view on how to treat each of the pertinent variables. All the included variables were transformed into binary variables. For each variable, the worst-case scenario was given the value 1, while the better scenario was given the value 0. To maintain consistency, when a scale was identical for two or more variables, the same cut-off was designated for these variables. For example, this meant that any question where the response options were "Poor", "Fair", "Good", "Very good", or "Excellent", these variables were transformed in the same manner using the same cut-off point. As with the descriptive variables, binary variables were left unchanged. Once the variables had been recoded to binary values, they were summed and rescaled from 0 to 10. We start by describing which variables were used in the creation of which index. A description of the process and reasoning follows. • Index 1: Self-perception. Life satisfaction; Life happiness; Self-perceived health; of health EURO depression scale; Satisfied doing no activities last year. --- Index 1: Self-Perception of Health This index included all the variables that related to what an individual thought about their health rather than the physical reality; this is quite a subjective measure and captures some aspects of subjective wellbeing. Questions in this section were more around how satisfied the respondent was with their life, whether they were feeling depression, etc. Havari and Peracchi [46] analyzed the SHARE dataset and recoded many variables which we have also investigated in this paper. Their focus on self-perceived health was important for the construction of this particular index. Perhaps most importantly, they coded a 0 if the respondent answered "Good", "Very good", or "Excellent" in the self-perceived health question and a 1 for those who answered "Fair" or "Poor." This is supported by Côté-Sergent et al. [47] who follow the same levels for recoding. This convention was followed in all variables that used the same scale. Both Havari and Peracchi [46] and Hank [48] classified a respondent with a score of 4 or less on the EURO depression scale as a 0, else a 1. Similarly, for the life happiness variable, if the respondent was "often" happy, they were given a value of 0, else a 1. Both life satisfaction and the "Satisfied doing no activities last year" questions were on a scale of 0 to 10, the cut off was determined to be 6 or less given a 1, else a 0. --- Index 2: Physical Health and Nutrition This index was created to capture the risk of suffering or developing serious health problems. The SHARE dataset also contains many questions about the physical health of respondents, such as BMI, nutrition, chronic conditions and others. If an individual had 2 or more chronic condition, they were given a value of 1. This convention is followed in Havari and Peracchi [46], Côté-Sergent et al. [47] and Nie and Sousa-Poza [49]. For the diet related metrics, [49] designated a 0 for individuals who responded "Less than once a week" to any of the categories. This has been applied to the "How often do you eat vegetables?" or the "How often do you eat meat, fish, and poultry?" questions, which were included in the database. Abeliansky and Strulik [50] had a unique approach to the BMI and grip strength variables. They combined these variables to create a "fragility" variable that is dependent on this combination of variables as well as the gender of the respondent. Those who were deemed "fragile" were assigned 1. Finally, if an individual had spent at least one night in hospital in the previous year, they were encoded with a 1 [51]. --- Index 3: Mental Agility This index was created to capture the mental acuteness of the respondents and is related to cognitive functions. The SHARE survey contained certain tests around numeracy, orientation and linguistic fluency; these variables were included in this index. Both the numeracy test and orientation tests were scaled on a 0-5 scale. This was encoded to a 1 if the respondent scored a 1, 2, or 3 [46]. For the fluency and word list tests, Hank [48] combined the scores of these three metrics, then used the median as the cut-off to convert this to a binary variable. --- Index 4: Dependency As stated in Section 1, dependency is measured and defined in surprisingly heterogeneous ways across Europe. Thus, this index has combined a variety of these methods to standardize and ensure comparability across individuals. For all the non-binary variables included in this index, an individual was given a value of 1 if they reported suffering from at least one of the corresponding conditions. Many papers follow this convention for ADL and IADL, and this was extended to number of mobility limitations [5,52]. --- Profile Construction Once the health and wellbeing indices are constructed, the next stage of the procedure is to combine them with a collection of descriptive variables, which contain the socio-economic and demographic information, clustering the individuals in groups with similar characteristics and, thus, obtaining the profiles. To do so, we used the k-prototypes clustering algorithm. In particular, we used the k-prototypes clustering algorithm included in the k-modes Phyton package https://github.com/nicodv/kmodes/blob/master/kmodes/kprototypes.py. To select the number of clusters we run the algorithm for k = 2, 3, . . . , 10. The cost function showed and "elbow" around k = 4 or k = 5 . Having both 4 or 5 profiles was investigated, and k = 5 was chosen, as k = 4 led to too broad results. --- Results --- Description of Profiles and Findings With the clusters created, a profile of the average member of each cluster can be made. To do so, the profiles have been ranked by taking the mean of the mean values for each index. The profiles with higher mean values in the indices can be said to be more disadvantaged than those with lower index scores. Figure 1 in the Appendix A show the distribution of each profile for the categorical variables used in the k-prototypes clustering algorithm with k = 5. A broad summary of each profile follows. Using Figures 1 and2 and the summary of statistics of each profile included in the Appendix A , the following broad profiles were made. From Figure 1, it can be seen that there are three female profiles and two male profiles. This is the modal class for gender for these profiles; it is not to say that every member of these profiles is female or male. From Figure 2, it can be observed that there is quite a difference between Profile 1 and the other profiles across index 1, index 3, and index 4. Index 2 is more balanced, with the mean value of this index differing only slightly across the profiles. The female profiles rank much worse in index 1, relating to mental health issues. While women are not more likely to have more mental health issues, they are more likely to be affected by depression and anxiety, in particular as they age [53]. Index 4, which relates to dependency, shows the highest variance both within each profile and between the profiles. In fact, all profiles have individuals scoring 0 and others scoring 10 for this index. When investigating deeper into the base dataset, it can be observed that on average members of Profile 1 suffer 3 times more limitations in ADL and IADL than the sample mean, while members of Profile 5 suffer only 0.1 times the sample mean limitations. See the Appendix A for all profile results. Int. J. Environ. Res. Public Health 2020, 17, x From Figure 1, it can be seen that there are three female profiles and two male profiles. This is the modal class for gender for these profiles; it is not to say that every member of these profiles is female or male. From Figure 2, it can be observed that there is quite a difference between Profile 1 and the other profiles across index 1, index 3, and index 4. Index 2 is more balanced, with the mean value of this index differing only slightly across the profiles. The female profiles rank much worse in index 1, relating to mental health issues. While women are not more likely to have more mental health issues, they are more likely to be affected by depression and anxiety, in particular as they age [53]. Index 4, which relates to dependency, shows the highest variance both within each profile and between the profiles. In fact, all profiles have individuals scoring 0 and others scoring 10 for this index. When investigating deeper into the base dataset, it can be observed that on average members of Profile 1 suffer 3 times more limitations in ADL and IADL than the sample mean, while members of Profile 5 suffer only 0.1 times the sample mean limitations. See the Appendix A for all profile results. Profile 1 and Profile 4 are the worst off for each gender , although Profile 4 is still better off than Profile 1. This is supported by the literature [54][55][56] which acknowledges the well-researched fact that women both live longer and with a poorer quality of life. These profiles also have a higher share of individuals aged 76 years or more old. For the constructed indices, they both perform badly in index 3, which relates to mental agility. Profile 5 is the least disadvantaged profile and is heavily skewed towards younger individuals; over 50% of the individuals in this profile are aged between 55-65, while those over 76 make up less than 10% of this profile. They are the most male heavy group and also the most likely to still be working. Continuing to work in later life has been correlated to positive health outcome [2]. Across all the health indices, they score low, indicating good health. These individuals are in the least need for current social assistance. Education is often used as a proxy for socio-economic status and its impact on later-life health outcomes is well researched [45,[57][58][59]. This education disparity can be seen in the profiles. Over 75% of respondents in Profile 1 either have zero education or only primary education; for Profile 4, this number sits around 50%. Conversely, Profiles 2, 3, and 5 have barely any individuals with no education and similar levels of university level educated individuals . --- Profiles across the EU The country of residence of the respondent was not included in the clustering algorithm and had no impact on the formation of the clusters. It is therefore interesting to see how the profiles are distributed across the continent and gain insight into how areas of need. This is shown in Figure 3. Profile 1 and Profile 4 are the worst off for each gender , although Profile 4 is still better off than Profile 1. This is supported by the literature [54][55][56] which acknowledges the well-researched fact that women both live longer and with a poorer quality of life. These profiles also have a higher share of individuals aged 76 years or more old. For the constructed indices, they both perform badly in index 3, which relates to mental agility. Profile 5 is the least disadvantaged profile and is heavily skewed towards younger individuals; over 50% of the individuals in this profile are aged between 55-65, while those over 76 make up less than 10% of this profile. They are the most male heavy group and also the most likely to still be working. Continuing to work in later life has been correlated to positive health outcome [2]. Across all the health indices, they score low, indicating good health. These individuals are in the least need for current social assistance. Education is often used as a proxy for socio-economic status and its impact on later-life health outcomes is well researched [45,[57][58][59]. This education disparity can be seen in the profiles. Over 75% of respondents in Profile 1 either have zero education or only primary education; for Profile 4, this number sits around 50%. Conversely, Profiles 2, 3, and 5 have barely any individuals with no education and similar levels of university level educated individuals . --- Profiles across the EU The country of residence of the respondent was not included in the clustering algorithm and had no impact on the formation of the clusters. It is therefore interesting to see how the profiles are distributed across the continent and gain insight into how areas of need. This is shown in Figure 3. One thing that is immediately of note is the distribution of the most disadvantaged profiles: Profile 1 and Profile 4. There is an over-representation of these profiles in the Southern European countries . As mentioned above, there is a strong relationship between education and health outcomes. This is also influenced strongly by geography [47,57]. Namely, having lower education in the Southern and Western European countries has a more negative impact on your health and long-term care needs than similarly educated individuals in the Northern European countries. This is reinforced by the findings here. Profile 2 and Profile 3 are quite similar, with a few small differences in the distribution of their categorical variables. Where they differ more is in their geographical distribution. We can see that Profile 2 is more frequently found in the Northern and Eastern European countries , while Profile 3 is more prevalent around the Central European countries . As mentioned above, there is a strong relationship between education and health outcomes. This is also influenced strongly by geography [47,57]. Namely, having lower education in the Southern and Western European countries has a more negative impact on your health and long-term care needs than similarly educated individuals in the Northern European countries. This is reinforced by the findings here. Profile 2 and Profile 3 are quite similar, with a few small differences in the distribution of their categorical variables. Where they differ more is in their geographical distribution. We can see that Profile 2 is more frequently found in the Northern and Eastern European countries , while Profile 3 is more prevalent around the Central European countries . They also differ in constructed indices, with Profile 2 being slightly worse off in the majority of the four indices. In particular, Profile 2 ranks much worse in index 1, which contains variables related to depression and mental health. --- Discussion --- Key Messages and Implications In the modern European context, it is more and more important to understand the demographic shift that has been happening for the past decades. Not only will this impact the makeup of European society, it will also present challenges for policy at a European and national level. Of particular importance are older, dependent individuals. In an attempt to homogenize the classification of these --- Discussion --- Key Messages and Implications In the modern European context, it is more and more important to understand the demographic shift that has been happening for the past decades. Not only will this impact the makeup of European society, it will also present challenges for policy at a European and national level. Of particular importance are older, dependent individuals. In an attempt to homogenize the classification of these individuals, five profiles were created using the respondents to the Survey of Health, Ageing, and Retirement in Europe survey data. To create these profiles, a clustering algorithm useful for mixed datasets, known as the k-prototypes algorithm, was implemented on the large dataset available. These profiles represent broad groupings of older Europeans and the similar struggles they face in day-to-day living. Two of the profiles, Profiles 1 and 4, can be said to be the most disadvantaged females and males across Europe. These individuals are older, less educated, much more likely to be dependent or have issues with ADL or IADL, and are an important subsection of society when considering policy regarding aged individuals. Interestingly, they have a similar geographic distribution, with an over-representation in Southern Europe. In fact, the well-documented Northern to Southern Europe health gradient has been observed in the distribution of the profiles. The better off profiles, such as Profiles 3 and 5, are more likely to be found in the Northern and Central European countries. Analogous patterns were found by Pappadà [60] when studying the social welfare systems in the EU from ESSPROS Eurostat dataset. A plausible explanation may be found in the diversity of social protection systems for long-term care operating in the EU. Similar findings were pointed out by several authors. In Muir [61], it is concluded that the number of people who need help in the course of their daily lives is similar in most countries, but whether and how that help is provided varies. Although the probability of someone needing long-term care increases with age, the cross-country variation in long-term care is not in general driven by demographics. For example, the Spanish population is much older than that of the Netherlands, yet far fewer people in Spain receive long-term care. Nor is it driven by disability rates, which are thought to be similar in most OECD countries [62]. According to the Social Protection Investment in Long-Term Care project, social investments mean policies designed to strengthen people's abilities and capacities and support them to participate fully in employment and social life. The key policy areas included are education, quality childcare, medical care , training, job search assistance, and rehabilitation [63]. In Glanz and Fernández [64], the authors consider that the social investment approach has considerable potential to allow decision makers to strengthen long-term care systems across the EU and, therefore, help address the challenge of population ageing. Thus, after more than thirty years of belonging to the EU wellbeing inequalities still persist in the Southern European Countries. Indeed, going further, it seems that SDG-3 of United Nations 2030 Agenda is hardly fulfilled in these developed countries. Undoubtedly, long-term care is expensive, and its cost varies widely between countries, being always high in relation to typical income, which means that long-term care is often unaffordable in the absence of social protection. Thus, countries should strengthen social protection systems to ensure that long-term care is not only available to the relatively wealthy [61]. A reasonable starting point maybe the proposal by Colombo et al. [65], who propose to promote healthy ageing and prevention, given that one of the most obvious ways to reduce costs in long-term care systems is to reduce potential dependence in later life by promoting health throughout life. Nevertheless, much more should be done to guarantee long-term care services across the EU. --- Strengths, Limitations, and Future Research In this paper, we provide a useful guide to the social and economic researchers in the visualization of profiles of individuals, that is, groups of individuals with similar characteristics, from large and very large mixed datasets by using existing tools. Our method is based in two stages: the first one consists in the design of several composite indices, each of which focusses on a particular issue of interest, yielding to thematic indicators. This index design has several advantages, namely the researcher can use both quantitative and qualitative variables, qualitative variables can be measured in different scales, binarization is used to track events of interest and finally their additivity allows to add relevant new variables if necessary. In the second stage, thematic indicators are combined with other descriptive variables, like socio-demographic information, through a clustering procedure able to cope with large datasets of mixed data. As a result, individuals are clustered in groups with similar characteristics and, thus, profiles are obtained. Finally, profiles can be described by looking at the "average" individual within profiles, regarding wellbeing indicators and descriptive variables. The methodology we propose is wide enough to be extended to other surveys or disciplines. Necessarily, the statistical analysis presented in this paper is limited to the variables available in SHARE wave 6 survey data. For instance, excessive alcohol drinking and/or drug use may have an important impact on the inequality of health in the elderly. However, this information was not available in wave 6. --- --- Appendix A Summary statistics of the variables used to create the profiles, per profile. For categorical variables, the mode is supplied, as well as the proportion of that class in the cluster. For numeric variables , both the mean and median are supplied.

The main objective of this paper is to visualize profiles of older Europeans to better understand differing levels of dependency across Europe. Data comes from wave 6 of the Survey of Health, Ageing and Retirement in Europe (SHARE), carried out in 18 countries and representing over 124 million aged individuals in Europe. Using the information of around 30 mixed-type variables, we design four composite indices of wellbeing for each respondent: self-perception of health, physical health and nutrition, mental agility, and level of dependency. Next, by implementing the k-prototypes clustering algorithm, profiles are created by combining those indices with a collection of socio-economic and demographic variables about the respondents. Five profiles are established that segment the dataset into the least to the most individuals at risk of health and socio-economic wellbeing. The methodology we propose is wide enough to be extended to other surveys or disciplines.

The purpose of this study is to identify multiple dimensions of oppression experienced by impoverished Black women who use drugs in New York by examining several settings in which participants experience oppression. Informed by the literature described above, oppression is defined as a system of power imbalance and exploitation maintained by unrealistic social norms and expectations within contemporary American society that make it impossible for certain groups of people to fit in and excel. Dimensions constitute ways in which oppression manifests itself in different areas or settings. --- Method Three case studies of drug-using, impoverished Black women were randomly selected from two large-scale, consecutive ethnographic studies conducted in New York City-Violence in Crack User/Seller Households: An Ethnography and Transient Domesticity & Violence in Distressed Households . In-depth interviews and observations were collected from 1998 to 2005. Code names will be used to ensure confidentiality. Grounded theory was used to analyze the data in the following steps: We read baseline interviews for each participant, extracted sections that seemed to illustrate oppressive experiences, and categorized participants' experiences of oppression into settings where the experiences took place; we examined each setting and extracted dimensions of oppression experienced by participants; and we consolidated dimensions across settings to ensure that each dimension included the full spectrum of oppression it described. The following cases were included. The first two women, Brenda and Olivia, are sisters. Brenda is 36 years old and Olivia is 38. Both of them identify themselves as Black, and they were born and raised in New York. Their mother graduated from high school and went to college for about 1 year. She was a waitress until she had her first child. From that point on, she never worked again and the family was supported by welfare. Olivia and Brenda's mother and their oldest brother died of AIDS. Olivia is now the eldest living child, and Brenda is the fourth child of the original seven: four sisters and three brothers. Brenda and Olivia have different fathers, and neither woman knows her father. All of Brenda and Olivia's brothers and sisters receive welfare. Olivia and Brenda live together at Brenda's home with one brother and Brenda's children. Brenda served 5 years in prison for attempted murder. She got her general equivalency diploma and attended college while in prison, though she never graduated. She dropped out of school because she was getting into fights and cutting class. Brenda had eight children; her first child was born when she was 14 years old. Her eight children have three different fathers. Brenda never had a regular job in her life. She worked for her welfare check as a custodian, but she had to stop because she did not have babysitters for her children. Brenda started using alcohol regularly at age 12 when her aunt offered her a drink. She quickly started experimenting with other drugs, especially marijuana. She stopped using drugs, but she still drinks twice per week. Olivia stopped going to school in sixth grade and never returned. According to her, she simply stopped going, and the school never notified her mother about it. She had two sons by two different men: One son is 20 years old and lives with Olivia's grandmother; the other is 3 years old and lives in foster care. The father of her older son is deceased, and the father of the 3-year-old does not contribute financially or emotionally. She is trying to complete drug rehabilitation and get a one-bedroom apartment so that she can get her son back. Olivia started using and selling drugs at 16. Like Brenda, she used to take prescription drugs from her mother and drink alcohol. As a teenager, she was selling marijuana but not smoking it. At 18, Olivia started experimenting with multiple drugs. She quickly became an alcoholic and addicted to angel dust and crack. She has been clean for 4 years now and is supported by disability payments from Supplemental Security Income because she is HIV positive. Carmen is a 23-year-old Black female born and raised in New York. She has one sister who lives in a shelter in Manhattan. She might also have a paternal brother, but she is not sure. Her sister uses crack and marijuana. Her mother graduated from high school and attended at least 1 year of college. Carmen's mother used to sell incense and oils when she was young. Carmen does not know what her father did for a living or how much schooling he had. Carmen grew up in the projects and moved to tenements at around 16 years of age. Her mother was evicted, and the family had a hard time acquiring stable housing. Carmen dropped out of school during 10th grade but later obtained her GED. She dropped out because other kids would tease her for not having good clothing. Carmen has a 2-year-old son, of whom she shares custody with his father. Carmen has had two legitimate jobs and one informal job, all temporary positions that did not last more than 4 or 5 months. She earned minimum wage on the legitimate jobs and less than minimum wage on the informal job. In order to make ends meet, Carmen also hustled, selling crack, stripping, and prostituting. Her main source of income is welfare. Carmen first started drinking with her friends at age 14. Unlike Olivia and Brenda, Carmen denied ever using hard drugs; however, she admitted smoking marijuana until recently. At the time of her interview, Carmen reported only using alcohol every other day. --- Findings Findings are divided into two distinct sections. First, settings in which oppression occurred are discussed and literature that explains how settings contribute to oppression is reviewed. Then oppression dimensions are presented, including examples retrieved from the data. Analysis revealed eight settings in which oppression occurs and five dimensions of oppression. --- Settings in Which Oppression Occurs School system-The school system was a setting of oppression for women in the study as U.S. educational policies and expectations are usually based on Eurocentric and patriarchic assumptions . In order to meet school expectations, students had to be able to purchase appropriate clothing, belong to a family in which parents were involved and available to participate in their children's education, and have their basic needs of food and shelter met by their parents. Reality was quite different for this study's participants,1 all of whom dropped out of school as a result of oppression. Correctional system-The ostensible purpose of the correctional system is to serve the community and maintain public safety. However, participants viewed the correctional system suspiciously and purposefully avoided interactions with law enforcement. Participants' suspicions are justified when considering the history between African Americans and the police, reaching back to the era when law enforcement was responsible for capturing and brutally punishing runaway slaves . Since then, numerous accounts of police brutality against African Americans have been registered. Considerable research has documented racism, sexism, and classism within the correctional system, demonstrating that the police often work to protect society from impoverished African Americans rather than viewing impoverished African Americans as members of the society the correctional system must serve . All study participants reported multiple encounters with the correctional system, yet the police were seldom contacted to address violence in the community because participants did not trust law enforcement. In order to survive and maintain their freedom, participants consistently avoided and deceived authority figures. Welfare system-The welfare system includes Temporary Assistance for Needy Families, disability income , Child Protective Services, food stamps, and public housing. Considerable research has discussed how the welfare system has oppressed applicants and recipients due to assumptions based on a Eurocentric, patriarchic society. That is, based on assumptions of personal responsibility and plentiful opportunities, Americans who are poor and in minority groups are blamed for their own misfortunes . In this study, the welfare system had high expectations that were often impossible for participants to overcome. Women were expected to work, take care of their children, and develop work skills in order to become financially independent. Yet welfare did not provide the means to fulfill these expectations, forcing women to go into debt, lose their electricity, engage in illegal activities, and/or play the system in order to survive. All participants reported receiving public assistance throughout their lives. They struggled financially and explained that lack of child care was the main obstacle for them to find and maintain employment. Housing and neighborhood-Housing and neighborhood were key settings of oppression because they exposed participants to violence, unsanitary living conditions, and drug use at quite young ages. Americans are expected to obtain their own housing, but the high cost of living in New York makes it inaccessible to poor New Yorkers. Their only alternative is to turn to the city's overcrowded low-income housing program. However, lowincome housing is often located in violent and drug-infested neighborhoods that offer impoverished schools to residents. Housing and neighborhood have also been connected to racial segregation, which leads to worse living conditions among African Americans, especially those living in poverty . All participants experienced homelessness at one point of their lives. They grew up in impoverished and crowded tenements and projects, which they described as violent, infested with mice, and overcrowded. Landlords neglected the buildings and drugs were present everywhere. The same issues were present in participants' descriptions of their current housing experiences. Relationships with men-This setting included participants' experiences with their male relatives, friends, and partners. Its oppression is rooted in the lack of value placed on Black women's welfare as evidenced by sexual exploitation, physical abuse, and inadequate legal protection. For instance, research shows that significantly fewer rapists of Black women are convicted than rapists of White women . In this study, all participants reported being victims of repeated sexual abuse and rape while growing up as well as engaging in at least one violent romantic relationship. Similar patterns were reported among participants' parents and friends. Accounts were quite graphic, illustrating the violence and lack of protection these women encountered while growing up. Perpetrators were usually family members, neighbors, or friends who were often much older than participants. Many of the incidents included multiple rapists. Abuses were rarely reported to the police, and those that were reported often did not result in conviction, due to lack of evidence. Participants reported not telling anyone about rape and abuse because they felt responsible and were scared and ashamed. Participants often saw their attackers regularly on the streets and dealt with these encounters by trying to avoid the perpetrators. Family-This setting was oppressive to impoverished Black women because it incorporated assumptions that moral and appropriate families follow a Eurocentric, patriarchic, heterosexual model. In this study, mothers were the heads of household and they had children with multiple men. The roots of these patterns have been linked to slavery. It was nearly impossible for a slave family to be formed and to remain together because children, husbands, brothers, and sisters were often sold separately. Moreover, the brutality to which slaves were exposed left deep scars and practices that survive to this day, such as whipping to discipline children . All participants reported experiencing conflict, violence, and neglect by their parents due to poverty and drug use. All three of the women whose stories are analyzed here reported being raised by a loving mother who was addicted to crack/cocaine. When the mother was unable to care for the family, the eldest female child often became responsible for taking charge of the household. Experiences with drug use-This setting includes participants' use or witnessing the use of drugs and/or alcohol. Moreover, it included the impact of their family's drug use on their lives. All participants reported either having used drugs in the past or currently using drugs. Two participants reported experiencing more violence in their homes after their mothers became addicted to crack. This setting's oppression is rooted in the expectation mat individuals struggling with drug use should be able to perform as well as those who do not use drugs. Such an expectation is particularly unrealistic in light of recent emphasis on substance abuse and addiction as a chronic disease of the brain that impairs the individual's ability to function in society . It is not reasonable to expect that someone who is struggling with substance abuse and/or addiction will be able to care for a family, obtain and maintain employment, and support himself or herself without help. Moreover, research has documented the role of stigma in exacerbating negative consequences of drug use, such as inhibiting users' ability to recover and develop positive self-esteem . Employmen-Employment included participants' experiences with formal and informal work. This setting was oppressive because of the assumptions that there are plentiful employment opportunities available in the United States and that employment provides enough financial support for employees to avoid poverty. In fact, research has demonstrated that employment is not plentiful for everyone and that employment does not guarantee one's ability to support a family . In this study none of the participants managed to maintain legitimate employment for a significant period of time. However, all participants worked on minimum wage-paying, informal jobs and engaged in various hustles to supplement their income. Hustles included doing laundry for others, prostituting, and selling drugs. Informal jobs available to participants included stacking fish boxes, breaking wooden crates, and working as a cashier, teaching assistant, nursing aid, and messenger. All participants reported working as custodians for the Work Experience Program . Respondents reported leaving these jobs due to lack of child care, domestic violence, relocating, or being laid off. --- Dimensions of Oppression Classism-Classism consists of expectations that in the United States people have equal opportunity and consequently will be able to acquire material resources to support themselves. Participants were expected to feed and clothe themselves properly based on the assumption that in the United States food and clothes are plentiful. Classism further assumes that a high school education is available to all and that it fosters social and economic ascension. These assumptions completely disregard the harsh reality of impoverished Black families. Participants' poverty significantly contributed to their inability to graduate from school and obtain the skills necessary to secure employment. In this sense, poverty was a vehicle of oppression in participants' lives. Participants reported growing up in violent, crime-infested neighborhoods because they could not afford to live anywhere else. Some participants also experienced homelessness. Two of the participants reported experiencing hunger while growing up, and none of the participants had proper clothing. Olivia believes that her lack of proper doming might have contributed to her dropping out of school: I thought I was dressing old-fashioned. So they [her classmates'] shit compared to mine made me feel like I was looking stupid. They had on the new stuff … like the marshmallow shoes that came out. And I was wearing some ole' Payless or whatever you call it. [Laughter] And I didn't like it. Maybe that's part of what kept me out of school too. Because the girl would look at me like if I had, you know, had it like that so why would I be wearing Payless shoes. The teasing of students can be seen as an expression of internalized repressive norms and assumptions of classism, and evidence that such norms and assumptions are reproduced by the school system. For example, Carmen explains how school policy contributed to her inability to feed herself as a young teenager: I used to be starving. And then come home and still be starving for all that. I just starve [at] home and hopefully somebody come bring me food. And then part of the way through high school, I couldn't even get the school lunch, because just to find my mother to copy the paper, you know. You got to do the paper with your financial thing to get school lunches. So I even was messed up with that. Governmental agencies theoretically designed to help people excel placed further obstacles for participants by ignoring their special needs. Olivia discussed her family's experience with Child Protective Services. According to her, the agency disregarded her mother's previous good behavior and her struggle with drug addiction and poverty. They removed her children: Welfare, she was like she would get cut off and then um my mother was a good woman she kept all her appointments. … You know, my mother's house wasn't right cause underneath the sink was damaged … so they call um ACS on her and they took her two kids … and my son. … And my grandmother was there within a half hour to get them. … There was a hole under the sink the house was unmanaging [unmanageable] … because they said that she was unfitted some body called up on her and told some junk on her … and um that was that … they I mean they just came and took her. … None of the participants were ever able to secure employment that paid more than minimum wage. Olivia was willing to work hard for little money, but better paying jobs were not available to her because she did not have her high school degree. In order to make ends meet, she had to engage in hustles and apply to welfare and SSI: Uh … 1 worked doing outside work on my own like stacking fish boxes and breaking down wooden crates. It's a big yard and they bring in all kind of crates … baskets that collard greens and stuff come in. They empty them and we knock them down, tie them up, and put them in bundles. You have to open them [the wooden boxes] up, bend them down with your hands. I did that for about two years or a year. That was off the books because it was hard for me to get welfare. … Then when I got on welfare they sent me to WEP. I worked in the welfare center doing filing for about four months. They got rid of me very quick! They didn't need no more help and I was just getting ready to get into the computer. You know? I was like, Oh man! No they didn't send me home! You know? I haven't worked WEP since then. I'm off welfare now. I'm on SSI. Olivia's experience suggests that the welfare system reproduces oppressive classist assumptions by presenting work "opportunities" that remain undeveloped or do not offer meaningful routes to self-sufficiency. Sexism-Sexism consisted of unrealistic expectations of how women must behave and perform sexually. Sexism is a complex and multifaceted dimension. In addition to meeting unrealistic gender roles in the family household, participants were expected to sexually satisfy their partners, enforce condom use to prevent sexually transmitted diseases and unwanted pregnancy, and not trade sex for money. Participants reported that men usually refuse to wear condoms. Brenda described how she takes responsibility for ensuring their compliance by describing her mother's struggle with HIV: If we get in a little too far into it and I ain't heard that little crackle, crackle, of the paper; and when they really is persistent about it, then I tell them about my mother. And after hearing that, they always have the look like, why did she tell me that? I don't want to know. But I don't feel bad about it; I love my mother. I'm not saying it to down her; but listen, you gonna use the condom, because I already got to worry about my mother, because she's got HIV. And I'm not trying to be like her. So after I say that, they really don't have too much arguing to do. In order to make ends meet, Carmen has hustled selling crack, working in a club as a stripper, and prostituting. She described her trauma working in the club as a prostitute due to the stigma attached to prostitution: Well, I tried to make it something weekly, like two or three times out of a week, but that is traumatizing. Because the whole environment will kill you. It's just so much negativity. Like, you can't just get into that kind of job, if you like sentimental or something, because it's like everybody, nobody cares. It's just about the money. The extremely low worth attached to impoverished Black females in American society was glaring in their experiences of being raped. All participants were raped multiple times throughout their lives. All of them were abused as children and therefore did not have the choice of when and how to initiate their sexual lives. Their rapists were never prosecuted, and participants were forced to live in the same neighborhoods as their aggressors. Carmen described being repeatedly raped by a man in her neighborhood. In describing the first attack, she seemed to feel the need to justify it as rape: I had met him and, um, this is like right after the problems happened with my mother. And this is something that I blamed on my mother, too. I had met him one day, and he said, well, come see me. I came down there one day to visit him, and I don't know how we started messing with each other. He, okay, I don't know if you can call it rape, but I call it rape. When it's like, first of all, there was a gun involved; and I was scared. Second of all, he let his friends have sex with me. And yeah, I didn't exactly just fight with him and get punched. I was dumb scared, so I just laid there, you know what I'm saying. And I blocked that out. And these are guys that I still seen, to this day, some of them, you know what I'm saying. But after he finished, he tried to make me have oral sex with him. And back then, that was really a sin, to have oral sex, you know what I'm saying, because, you know. And I had just started having sex, so I wasn't really having sex like that. But anyway, I told him, if you stick that-after going through all that, I just basically told him, if you stick your penis in my mouth-then neither one of us are gonna leave. That's that. Participants often felt responsible for being raped. Carmen was raped several times by the same man while no one ever attempted to stop it. While telling her story, she indicates that she believes she was responsible for some of the attacks: Anyway, I figured I should have just not been-I figured it was my fault, too. Because another time I went to a block party, and he seen me there, and he dragged me to a building. I never had sex outside. I wasn't that type of person. But really, he would threaten me, pull my hair, hit me in my mouth. And another time I was on a street in the neighborhood, he literally dragged me ten blocks. My cousin could have helped me, but he was in jail. And other than that, like I said, I was scared. What am I supposed to do? I really didn't know. And I mean that was really forced. But like a year later, he raped this other girl, and got her pregnant. And she had to give her baby up for adoption. And he's been in jail since then. Male/female relationships were another setting where impoverished Black women endured oppression. There was a complex exchange of power between the men and the women where they both met some of their needs and exploited their partners. The men were often transient and unable to contribute financially to the family budget, though sometimes they would provide needed child care. Domestic violence seemed to be commonplace within these relationships. The women were usually the heads of the households, and they were able to kick the men out by claiming their home. Monogamy was not necessarily expected, and men enjoyed bragging about their sexual relationships to their friends. Familism-Familism consisted of expectations mat a family is composed of a heterosexual couple and their children, where parents are married and responsible for raising and providing for their children. Participants' understanding of family was quite different than mainstream definitions. None of the participants or their parents were married or able to provide financially and sometimes emotionally for their families. In fact, the eldest children would often take on adult roles and be expected to run the household themselves while their parents were out chasing drugs. The males in the family were transient, often violent, and usually unable to provide financially. For example, Olivia described her mother's expectations of her while growing up: I didn't play a lot. When we really started getting up to the fourth and fifth grade, it was like do the laundry, fold the socks, fold the clothes, get ready for dinner, and eat. After you eat, take a bath and sit your ass down and watch TV. [Laughter from both] For real! My mother was on some shit! And don't leave no clothes in the laundry when you go! Bring everything back that you took! Drugs played a significant role in overloading the children with adult responsibilities. Carmen explained how her mother changed after she became addicted to crack and heroin: And you go from my mother cooking every night to I got to stand outside in front of the building and wait for you to get home; because I don't got no key to get in the house. Talk about dinner, and I'll tell you to go to a friend's house if I wasn't home? Yeah, but I ain't know you meant if you spent the night out, because you ain't never did that before. So we literally jumped from me, from sugar to shit, literally. But you know, it probably was an ongoing process with her; but I was a child. I didn't know that I didn't know it was that bad. Participants were ultimately responsible for providing for their families. Their male partners sometimes provided child care; however, their contribution was not reliable due to their transience. While parents are usually expected to protect and care for their children, this was unrealistic for many of participants' parents because of their drug addiction, poverty, and lack of voice in society. As Olivia and Brenda explain, their family struggled with domestic violence after their mother moved in with their stepfather: Well when she got with this last man she was bug head I mean my mother wasn't really mean but she when she started chasing that coke cocaine behind this man … she thought that she could whip. … You know what I mean she thought that she could really work earn and all the time she was getting beat up herself by these men. Though all participants reported witnessing their parents engage in violence, drug use, and crime at young ages, they loved their parents and were willing to withstand great hardships to maintain their families intact. When Carmen's mother became dependent on drugs, her relatives suggested she move south to live with them. She would have had her basic needs met if she had accepted. However, at age 12, she chose to stay with her mother: Yeah, but you asking a person that was, like I said-you went from neglect to nothing; but when it was going through that process, you asking me to give away, to leave my mother. And this was before, how am I supposed to know that I'm gonna be starving every day? I-this problem just now occurred; and soon as the problem occur, you want me to jump up and leave my mother, while everybody, it seemed to me, is whispering bad things about my mother. I don't want to hear that. And I don't want to be with you, if you gonna sit there and talk about my mother. Carmen's account suggests she is aware that as a child she internalized the familistic assumption that her mother would care for her and resented relatives' claims that she would not. Racism-Racism consisted of expectations that governmental agencies will not serve and protect all people independently of their racial background. This dimension was mostly salient in participants' interaction with law enforcement. The mainstream expectation that the police would serve and protect their well-being was not shared by the Black community. Many participants did not believe that the police would protect them, and they did not trust police officers. In fact, the only time Brenda reported being raped to the police, nothing happened. Brenda blamed herself for the outcome and sought justice herself: We got up there and he flipped on me and tried to make his friend have sex with me. He said, go ahead; take your pants off. And I was scared, because he was saying, I'm gonna throw you off the roof, you know. So I was scared. So I pull them down and his friend had sex. Then he told the girl. And then he said, okay, it's your turn. And then I jumped up. But she never got to do it, because when he said that, I said, no. He said, well, we gonna throw you off the roof. And I just jumped up some kinda way, pulled on my pants and jerked away and ran down to the next flight and banged on people's doors. They wouldn't let me in, but this lady called the cops for me. When me cops came, I just told them his name. But I didn't really know where he lived at, because he was living here and there and everything. So they just took me to die train and I went home. But I got die girl. My friend, I was with a friend, and I tore her up. I kicked her butt. Whereas Brenda sought help from the police, unsuccessfully, after being sexually assaulted, Carmen avoided contacting law enforcement, even after the third time she had been raped by the same man: And after that, I just avoided mat block. I was scared. I didn't know to call the cops. Black people-and back then, my mother didn't sit there to let me know if somebody did somedthing, call the cops. And as for young people, cops was a nono; because they don't care. I'm-and then when the cops come, where's your mother? So that would have been in the subject. Even though they might have arrested him, okay, let's take you home. Okay, when you go home, or go to the precinct, your mother don't come for five days. Oh, you're going to a foster home. I wasn't trying to call no police. Theoretically, there is the social expectation that the police will be called when there is violence. Participants witnessed and experienced a lot of violence but seldom contacted the police. There was an unspoken understanding that participants must defend themselves. Police officers were seen as the enemy and were seldom involved in these violent disputes. History of police racism is oppressive in that it leaves Black people vulnerable to victimization because they see no reason to expect protection. Another expression of racism in this context is targeting of Black men for arrest and prosecution; the women in the study give voice to a less obvious form of racism as a dimension of oppression. Drugism-Drugism consisted of expectations that those addicted to drugs should be able to function and take responsibility for themselves and their families as healthy people. Participants seemed to have strong negative perceptions of cocaine use. They seemed to struggle between blaming their parents for what their addiction made them do and defending their behavior. Generally, participants felt that those who use crack were weak and responsible for their addiction. Brenda discusses her views about her mother's addiction: Okay, my mother, well, I always drought she was just weak. She was so nice and stuff. I mean, at first she was strong; she was going to school, vibrant, happy. And then when things started, like when her husband left her, she was using drugs then, but I didn't know. She was smoking crack; she wasn't a crackhead back then, you know, because there is a difference. And it was alright then, but after she started abusing the drugs, and you know, things went down. My sister to me, she was always like a troubled child. And then she was smart, but then she went and started smoking crack. So how smart? I mean, I think I was actually the strongest one, because they was weakened so easily, well, my sister. Because how could you grow up knowing, you know, what's going on, and then go and do the same thing? It's like, don't be ridiculous. Even though I might be falling into that cycle, but I mean, she was smoking crack. But I'm looking at it differently. I never did that, so, in my opinion, you know. But she was a very smart person. It's just that I think she just got hurt so much in life that-it's like she just lives on a high or something. This expectation of functional behavior on the part of drug addicts was also evident in their interactions with a methadone clinic. Clients were expected to show up at the methadone clinic within a small time frame to take their methadone dose. If they missed that window, they did not get their dose. This was not realistic for this population, considering the chaotic environments in which they lived. Yet they had to cope with these expectations, as illustrated by Carmen's mother's experience: Cooking, getting up in-only thing I ever heard her complain about was going to a methadone clinic. You know, if she was missing it. That's the only thing I really heard her complain. Oh, I don't want to miss my clinic. Or if she missed it, it was like she was annoyed; but other than that, I really never heard her complain. The quotation above illustrates limitations of current drug use polices based on a punitive perspective associated with beliefs that drug use is an issue of deficient moral behavior. While the government recognizes methadone as an appropriate method to treat drug addiction, it imposes multiple controls to change patients' behavior and to punish them when they are unable to meet such expectations . --- Discussion The purpose of this study was to identify multiple dimensions of oppression experienced by impoverished Black women who use drugs in New York by examining several settings in which participants experience oppression. Analysis revealed five dimensions of oppression occurring within eight distinct settings. Dimensions of oppression included classism, sexism, familism, racism, and drugism. Settings included the school system, correctional system, welfare system, housing and neighborhood, relationship with men, family, experiences with drug use, and employment. Dimensions of oppression are embedded in the rules governing the achievement of success in every setting, taking the form of unrealistic assumptions and expectations. Findings demonstrate the unprecedented disadvantages that this population faces in their daily lives. From childhood to adulthood these disadvantages accumulate, having a snowball effect that makes it impossible for participants to realize economically and emotionally stable lives. They were born into impoverished families where the head of the household was addicted to crack and other drugs. Their homes and neighborhoods exposed them to drugs, violence, and sexual behavior at extremely young ages. Their childhood was stripped from them by the need to step into their parents' role and provide for themselves and their families. These adverse conditions made it impossible for them to attend school and graduate, reducing their ability to secure employment in mainstream society. Thus, in order to survive they engaged in street hustles selling drugs and/or their bodies or engaging in odd jobs whenever possible. These hustles exposed them even further to state controls such as welfare and law enforcement that often resulted in an arrest and/or the removal of their children from their homes. Their financial struggle and the only role models available in their lives motivated participants to apply for welfare, only to once again encounter further state controls and meager financial aid. Mainstream society aggravates and perpetuates participants' struggles by imposing on them Eurocentric and patriarchic assumptions that blame participants for their own plights and justify society's lack of support for those in need. This becomes clearer when considering the impact of U.S. slavery, emancipation, and oppression history in the lives of Black women . Specifically, the trauma of slavery and its impact may be passed down as the family legacy even to children born after the trauma. Consequently, many Black women end up fulfilling the roles assigned to them by society . Other research has demonstrated how White privilege has contributed to contemporary racial disparities in the form of accumulated wealth, punitive policies that are based on stereotypical images of different racial groups, and preferential treatment due to covert prejudice . For instance, Shapiro demonstrated how, historically, privilege allowed White immigrants to accumulate wealth while making social ascension difficult for African Americans. Whites owned the land and profited from African slave work that was never compensated. After emancipation, policies such as the GI Bill allowed White Americans to purchase their own homes while indirectly excluding African Americans. The current study builds on this research by demonstrating how dimensions of oppression have contributed to participants' inability to compete for success in contemporary American society.

Oppression against Black women continues to be a significant problem in the United States. The purpose of this study is to use grounded theory to identify multiple dimensions of oppression experienced by impoverished Black women who use drugs by examining several settings in which participants experience oppression. Three case studies of drug using, impoverished Black women were randomly selected from two large scale consecutive ethnographic studies conducted in New York City from 1998 to 2005. Analysis revealed five dimensions of oppression occurring within eight distinct settings. While dimensions constitute different manifestations of oppression, settings represented areas within participants' lives or institutions with which participants interact. Dimensions of oppression included classism, sexism, familism, racism, and drugism. Settings included the school system, correction system, welfare system, housing and neighborhood, relationship with men, family, experiences with drug use, and employment. Findings have important implications for social justice, welfare, drug, and justice system policy.

INTRODUCTION --- Background Individuals who encounter cultural barriers to accessing care may be more likely to seek health information online. 1 Data from the Health Information National Trends Survey have found that presumably cisgender sexual minority people are more likely to seek and be exposed to incidental health information online, more likely to watch online health-related videos on YouTube, and less likely to first seek health information from a physician compared with their heterosexual peers. [2][3][4] Less is known about online health information seeking among transgender and gender diverse people who are estimated to number at least 1.4 million in the United States. 5 TGD people have gender identities or gender expressions that may not align with those commonly associated with their sex assigned at birth and may identify as transgender men, transgender women, trans men, trans women, men, women, or other gender identities. 6,7 Cisgender people have gender identities or gender expressions that align with those commonly associated with their sex assigned at birth. From an informatics perspective, the availability of datasets that capture detailed information about gender identity to accurately represent TGD people have been limited. 8 Accordingly, the collection of national data on TGD populations was declared a priority objective for U.S. public health infrastructure in Healthy People 2030. 9 TGD people may have unique health information needs relating to supporting gender affirmation such as gender-affirming hormone therapy, which may motivate them to seek health information online. 10 In addition, TGD people utilize the Internet for community building and information sharing. 11 However, there is a paucity of research that explores how online health information seeking among TGD people may be associated with personal health decision-making, like whether or not to receive a vaccine. Moreover, eHealth literacy-the ability to use electronic health information to make health decisions-has been explored in presumably cisgender sexual minority people but not among TGD people. 12 Human papillomavirus , the most common sexuallytransmitted infection in the United States, is known to cause 90% of cervical and anal cancers; 70% of oropharyngeal, vaginal, and vulvar cancers; 60% of penile cancers; and is largely preventable by vaccination. [13][14][15] Few studies have focused on HPV vaccination among TGD people even though these communities are at increased risk for HPV infection compared with the general population. 16,17 --- Objective The purpose of this study is to describe online health information seeking among a sample of TGD people compared with cisgender sexual minority people to explore associations with HPV vaccination, and whether general health literacy and eHealth literacy moderate this relationship. --- MATERIALS AND METHODS --- Theoretical framework We adapted the Integrative Model of eHealth Use to guide our study. 18 IMeHU posits that online health information seeking is associated with health behavior outcomes, and this relationship is influenced by individual factors, such as general health literacy, eHealth literacy, Internet use, health knowledge, situation factors , and demographics . --- Study design and sample We employed a cross-sectional design to explore the association of online health information seeking and HPV vaccination among TGD and cisgender sexual minority people. Between February and May 2020, we launched an online survey to an existing researchready cohort. E-mail and text message invitations to participate were sent to 17 036 participants in The Population Research in Identity and Disparities for Equality Study , a longitudinal, U.S.-based, national health study of sexual and gender minority people. 19 The PRIDE Study launched in 2017 and recruits adults aged 18-years and older, who are English speaking, and reside in the United States or its territories, who self-identify as a sexual and/or gender minority person. Details of The PRIDE Study longitudinal cohort, its digital health research platform, and dataset are described elsewhere. 20,21 Participants who completed the survey were entered in a drawing to win 1 of 10 $50 gift cards. This study was approved by the Institutional Review Boards at Columbia Irving University Medical Center and Stanford University Medical School . --- Survey administration Data were collected using Qualtrics , and the survey was hosted on The PRIDE Study Web-based participant portal. 20 We used a modified Dillman 22 method for survey reminders that were issued by The PRIDE Study web portal via email and opt-in text message. --- Measures --- Online health information seeking and Internet use We used 23-items from the Health Information National Trends Survey 23 to assess preferences for online health information seeking and Internet use . HINTS is an instrument administered by the National Cancer Institute to understand how adults obtain health information. By using HINTS items, we sought to better understand preferences for information seeking, especially among TGD people, as these communities were not recruited in previous adminstrations of HINTS. We assessed the primary independent variables of interest-online health information seeking related to vaccines and general online health information seeking-using 2 HINTS items: HINTS_B3) In the past 12 months, have you used the Internet to look for information about vaccines for yourself?; and HINTS_B5a) In the past 12 months, have you used a computer, smartphone, or other electronic means to do any of the following? Looked for health or medical information for yourself. Several HINTS items were modified from their originally validated form in order to include language related to vaccines instead of the original language that referred to cancer. --- General health literacy We assessed general health literacy using 3 discrete subjective items proposed in the health literacy literature. 24 These items were validated as a brief alternative to longer format instruments that address reading and understanding of written health information. The 3 Likert response items addressed confidence in filling out forms, difficulty understanding written communication, and needing help to read written material from a doctor or pharmacy. --- eHealth literacy We assessed eHealth literacy using the Electronic Health Literacy Scale , 25 an 8-item scale that uses Likert responses to yield a total score of 8 to 40 ; corresponding with selfperceived eHealth literacy in 6 domains; traditional literacy, health literacy, information literacy, scientific literacy, media literacy, and computer literacy. eHEALS has been validated in numerous settings and populations including presumably cisgender sexual minority people, such as men who have sex with men. 12,25 HPV knowledge, HPV vaccination Access to Care, Preventative Care, Barriers to Care --- Demographic characteristics We assessed HPV knowledge, HPV vaccination, access to care, preventative care, barriers to care, and demographic characteristics using items taken from The PRIDE Study Annual Questionnaire 2018. Although previously administered, the items in our survey were posed again to ensure contemporaneous accuracy with other items in the cross-sectional survey. The PRIDE Study Annual Questionnaire 2018 is publicly available for review and use at pridestudy.org/ collaborate. Our final survey included 74 items and could be completed in 15 to 20 minutes . Participants could skip any item and pause and resume the survey during the study period by logging into their existing web portal accounts with The PRIDE Study. --- Classification of TGD and cisgender participants Self-reported gender identity and sex assigned at birth distinguished TGD participants from cisgender participants; Boolean logic accounted for multiple responses to gender identity. We categorized participants as TGD if they indicated their gender identity was woman or transgender woman, and their sex assigned at birth was male; if they indicated their gender identity was man or transgender man, and their sex assigned at birth was female; or if they indicated their gender identity was genderqueer, another gender identity, transgender man, or transgender woman. Participants were categorized as cisgender if they indicated their sex assigned at birth was male and their gender identity was man; or if they indicated their sex assigned at birth was female and their gender identity was woman. We excluded individuals who did not report their sex assigned at birth. --- Statistical analysis Analyses were performed in SAS 9.4 . Descriptive statistics for demographics were calculated, including means with standard deviations. Pearson's Chi-squared test, Fisher's Exact test, and paired t-tests were employed to examine differences in categorical and continuous variables . Alpha was set at .001 for bivariate comparisons to control for multiple comparisons. The distribution of continuous variables and potential outlier values for age and eHEALS scores were identified using boxplots. We used variance inflation factor statistics to assess for multicollinearity. A VIF value less than 5 suggested no multicollinearity. 26 We used multivariable logistic regression to model the association of health information seeking on HPV vaccination. TGD were compared with cisgender participants as the reference group. We performed post hoc testing using the Bonferroni-Holm sequential procedure for adjusted alphas. 27 To test for any moderating effects, we ran a separate model for each interaction term; including each of the 3 categorical general health literacy variables and the eHEALS score variable. We evaluated the effect of each of the interaction terms on the logistic regression models using likelihood ratio tests. --- RESULTS There were 3339 completed responses . We excluded 81 of the responses due to missing sex assigned at birth, gender identity, or age. Of the remaining 3258 participants, 1172 were classified as TGD and 2086 as cisgender . The median age of participants was 31 years . TGD participants were slightly younger than cisgender participants . A greater proportion of TGD participants were female sex assigned at birth compared with cisgender participants . Just over a third of TGD participants indicated their lived gender day-to-day was sometimes man, sometimes woman, or third gender other than man or woman. The proportion of white participants was slightly less for TGD participants compared with cisgender participants . The proportion of TGD participants of Black, African American or African race was less than half of cisgender participants . Similarly, the proportion of TGD participants of Hispanic, Latino, or Spanish ethnicity was less than half of cisgender participants . The proportion of TGD participants who had more than a high school education , was slightly less than cisgender participants . All participants who identified as intersex answered the sex assigned at birth and gender identity questions and were thus classified as either TGD or cisgender. A greater proportion of TGD participants identified as intersex than cisgender participants . --- Online health information seeking and internet use The groups had similar online health information seeking behavior and Internet use . Nearly all participants used a computer, smartphone, or other electronic device to look for health or medical information in the past 12 months. Just under a third of TGD participants and 33.7% of cisgender participants used the Internet to look for information about vaccines in the past 12 months . Nearly all participants visited a social networking site, like Facebook, in the past 12 months. --- General health literacy and eHealth literacy In terms of the 3 discrete general health literacy items, 96.5% of TGD participants and 92.1% of cisgender participants reported they were "quite a bit" or "extremely confident" filling out medical forms by themselves . Additionally, 88.9% of TGD participants and 94.1% of cisgender participants said they "never" or "occasionally" had difficulty understanding written communication . Moreover, 95.1% of TGD participants and 98.1% of cisgender participants said they "never" or "occasionally" needed help to read written material from the doctor or pharmacy . TGD participants' self-perceived eHealth literacy measured by their eHEALS score was lower than cisgender participants' eHEALS score . --- HPV knowledge and HPV vaccination Nearly all participants had heard of HPV. A greater proportion of TGD participants reported HPV vaccination compared with cisgender participants . A smaller proportion of TGD participants than cisgender participants reported that a doctor refused to give them the HPV vaccine when they requested it . Access to care, preventative care, barriers to care In terms of access to care, the majority of all participants reported having a primary care provider . Nearly all participants had health insurance, but a smaller proportion of TGD participants had Medicaid insurance than cisgender participants . With regards to preventative care, a a smaller proportion of TGD participants reported receiving 3 or more vaccines since 18 years of age than cisgender participants . Considering barriers to care, over 43% of TGD participants reported delaying necessary medical care in the past year compared with 24.5% of cisgender participants . --- Modeling online health information seeking and HPV vaccination To explore the relationship between health information seeking and HPV vaccination, we performed multivariable logistic regression. In our sample of 3258 participants, 1528 reported HPV vaccination. We performed a bivariate analysis of 30 predictors on the outcome HPV vaccination . Predictors that did not meet our entry criterion of P < .25 were removed and were not proposed in the preliminary main effects model. Predictors were selected based on the IMeHU categories . We summarize the reduced model in Table 4. Some variables were kept in the model even though they did not meet the entry criterion because they were important variables of interest . We performed a post hoc Bonferroni-Holm correction for multiple comparisons on the final reduced model. --- Predictors: Online health information seeking and internet use After controlling for covariates including age, race/ethnicity, and education, we found that, compared with cisgender participants, TGD participants had decreased odds of reporting HPV vaccination when they looked for information on vaccines in the past 12 months but over twice the odds of reporting HPV vaccination if they visited a social networking site like Facebook in the past 12 months . TGD participants had decreased odds of reporting HPV vaccination if they used a computer, smartphone, or other electronic means to track healthcare charges and costs or to make appointments with a healthcare provider . --- Predictor: HPV knowledge Compared with cisgender participants, TGD participants had over twice the odds of reporting HPV vaccination if they had heard of HPV . Predictors: Access to care, preventative care, barriers to care Having Medicaid insurance was associated with decreased odds of reporting HPV vaccination . Con-versely, TGD participants who reported receipt of 3 or more vaccines since 18 years of age had 3.5 times the odds of reportng HPV vaccination compared with cisgender participants . Having had an HIV test in the past 12 months was as- sociated with increased odds of reporting HPV vaccination . However, having had an anorectal cancer screening was associated with decreased odds of vaccination . TGD participants who had delayed medical care in the past 12 months had 1.5 times the odds of reporting HPV vaccination compared with cisgender participants . Having less than 50% of healthcare providers be aware of participants' sexual orientation or gender identity was significant in preliminary logistic regression modeling, but was not associated with reporting HPV vaccination in the reduced model. --- Predictors: Demographics Overall, TGD participants had 1.5 times the odds of reporting HPV vaccination compared with cisgender participants . TGD participants younger than 27-years-old had decreased odds of reporting HPV vaccination than cisgender participants in the same age group. Race/ ethnicity and education were not significant in the reduced model. --- Interaction of general health literacy and eHealth literacy To test for moderation effects, we assessed for the interaction of general health literacy and eHealth literacy and online health information seeking. We added interaction terms 1 at a time to the logistic regression model for each of the 3 general health literacy variables and eHEALS score. We found no interaction of general health literacy or eHealth literacy with online health information seeking and HPV vaccination. --- DISCUSSION Our study found that, in a sample of 3258 TGD and cisgender sexual minority people participating in The PRIDE Study, TGD people overall reported increased HPV vaccination compared with cisgender sexual minority people, but decreased vaccination after they used the Internet to search for information about vaccines. Decreased HPV vaccination after searching for vaccine information online may reflect concerns about the safety, efficacy, or necessity of vaccines manifested as vaccine hesitancy; 28 the delaying or refusal of vaccination that may be context-specific and related to factors like complacency, convenience, and confidence. 29 The quality of information that individuals encounter when searching for vaccinespecific information may also influence personal health decisionmaking surrounding vaccines, especially if the quality of the content is poor, or the content increases anxiety from misinformation. Antivaccine web content, even some specific to the HPV vaccine, has proliferated in recent years, and pro-HPV vaccine YouTube videos were 4 times more likely to report accurate information than anti-vaccine videos. 30 Moreover, online content that is not transgender-inclusive and affirming may pose an additional barrier to HPV vaccination among TGD people if they feel they vaccine is not appropriate for them. 31 Conversely, our finding that visiting social networking sites, like Facebook, increased HPV vaccination among TGD participants is notable, given the ubiquity and widespread use of social media today. Moreover, the prevalence of health information seeking on social media is increasing with peer interactions and the need for social and emotional support contributing to social media use. 32 In the era of COVID-19, social media can negatively and positively affect health information related to vaccines. On the one hand, social media facilitates the spread of misinformation that further contributes to vaccine hesitancy; on the other hand, social media is used to promote information accuracy campaigns to counteract vaccine misinformation. 33 Peer norms are a facilitator for preventative vaccination among presumably cisgender sexual minority people, and this could have implications for vaccination information sharing through social media platforms that have growing use among TGD and sexual minority communities. 34,35 Research that explored information sharing using HINTS data found that use of social media for sharing health information declined over time, whereas use of social media to exchange medical information with a health professional increased. 36 These findings were based on a general population sample; the motivations for health information sharing among peers and/or healthcare professionals may be different for TGD people, especially those who have experienced discrimination and stigma in healthcare. 37 A limited number of studies have investigated the association of online health information seeking and vaccination among presumably cisgender sexual minority people. A study of presumably cisgender men who have sex with men showed an increase in HPV vaccination among those who searched online for sexual health information. 38 Related research found that MSM had higher perceived benefits of HPV vaccination when they exhibited higher levels of health information orientation ; however, the sources of health information were not identified. 39 These findings are inconsistent with our primary finding that TGD paricipants were less likely to report HPV vaccination than cisgender sexual minority participants if they looked for vaccine information online. This difference could possibly be attributed to additional factors that increased perceived advantages of HPV vaccination among MSM, such as the presence of a perceived threat. 39 In addition, our study found that TGD participants reported increased HPV vaccination after using social media, which may have been an information source for vaccination. Our findings suggest that TGD people may be engaging with online health information differently than cisgender sexual minority people. Having heard of HPV increased the likelihood of reporting HPV vaccination among TGD participants, which is consistent with the literature that has examined knowledge of HPV and vaccination among sexual and gender minority communities. 31,40,41 Our findings that users who tracked healthcare costs and healthcare appointments using a computer, smartphone, or other electronic means was associated with decreased reporting of HPV vaccination warrants further investigation. Out-of-pocket costs for healthcare may be a perceived barrier to individuals who track costs closely. The cost of other vaccines is a barrier in presumably cisgender sexual minority people. 42 This would especially be true with lower income individuals who have Medicaid, which is consistent with our findings of decreased reporting of HPV vaccination among TGD participants with Medicaid. Decreased reporting of HPV vaccination among TGD participants who had made medical appointments online is possibly related to a perceived barrier to obtaining care if preventative care appointments are not readily available. In contrast, having had an HIV test in the past 12 months was associated with increased reporting of HPV vaccination, which corroborates studies that have shown an increase in preventative vaccination in presumably cisgender sexual minority people when HIV and STI testing were bundled with vaccination. [43][44][45] The greatest effect sizes for reporting HPV vaccination were observed among TGD participants who had received 3-or-more vaccines since 18 years of age. This is consistent with the literature that has demonstrated that when other vaccines such as hepatitis A/B are bundled together it can increase vaccine uptake among sexual and gender minority communities. 46 Although we found no moderating effects of general health literacy or eHealth literacy, this is likely related to the highly healthliterate sample who had generally high eHEALS scores and few challenges understanding health information. The lack of variability in general health literacy and eHealth literacy further limited any moderating effects. --- Strengths/limitations This study has several strengths. To our knowledge, this is the first study to investigate the relationship between online health information seeking and HPV vaccination using a large sample of TGD people. From an informatics perspective, use of The PRIDE Study and its digital health research platform enabled our study team to leverage a novel national dataset that empowers TGD people to describe diverse gender identities and gender expressions. The Integrative Model of eHealth Use is a theoretical framework that has never been adapted to examine a specific health behavior outcome among TGD communities. In addition, we took a novel approach to operationalize the theoretical model and incorporate general health literacy and eHealth literacy as moderators. The study is not, however, without its limitations. Although our cross-sectional survey was composed of items from previously validated instruments, our survey as a whole may not be considered a validated instrument because of the modifications made to items and mixture of items from multiple sources. The PRIDE Study is a convenience sample; since the majority of participants were white and had greater than a high school education, our sample was not representative of TGD and cisgender sexual minority people residing in the United States. TGD people were compared with cisgender sexual minority people in aggregate and comparison groups were categorized using sex assigned at birth and gender identity. However, comparison groups were not further stratified by specific gender identities and factors that are associated vaccine among different gender groups warrants further investigation. HPV vaccination by self-report may be subject to recall bias which may worsen over time. Lastly, the cross-sectional nature of the study limits our ability to derive any causal relationships. --- CONCLUSION In summary, our study of online health information seeking and HPV vaccination found that compared with cisgender sexual minority people, TGD people reported increased HPV vaccination overall, but were less likely to report vaccination after they searched for vaccine information on the Internet. Factors most associated with HPV vaccination were having visited a social networking site like Facebook, having received 3-or-more vaccines since 18 years of age, and having heard of HPV. We found no moderating effects from general health literacy or eHealth literacy. Future studies should investigate potential deterrents to HPV vaccination in online health information to enhance its effectiveness, and further explore which aspects of social media might increase vaccine uptake among TGD and cisgender sexual minority people. --- FUNDING ATP received funding support from the Robert Wood Johnson Foundation Future of Nursing Scholars Program. AF's work on this project was partially supported by K23DA039800 from the National Institute on Drug Abuse. The content is solely the responsibility of the authors and does not necessarily present the official view of the National Institute on Alcohol Abuse and Alcoholism, National Institute on Drug Abuse, or the National Institutes of Health. Research reported in this article was partially funded through a Patient-Centered Outcomes Research Institute . The statements in this article are solely the responsibility of the authors and do not necessarily represent the views of Patient-Centered Outcomes Research Institute, its Board of Governors or Methodology Committee, or the National Institutes of Health. --- --- SUPPLEMENTARY MATERIAL Supplementary material is available at Journal of the American Medical Informatics Association online --- ACKNOWLEDGMENTS The PRIDE Study is a community-engaged research project that serves and is made possible by LGBTQ community involvement at multiple points in the research process, including the dissemination of findings. We acknowledge the courage and dedication of The PRIDE Study participants for sharing their stories; the careful attention of PRIDEnet Participant Advisory Committee members for reviewing and improving every study application; and the enthusiastic engagement of PRIDEnet Ambassadors and Community Partners for bringing thoughtful perspectives as well as promoting enrollment and disseminating findings. For more information, please visit pridestudy.org/pridenet. --- DATA AVAILABILITY STATEMENT Members of the lesbian, gay, bisexual, transgender, and queer communities have experienced significant stigma and discrimination from society including the medical and investigational communities. As such, we are ethically bound to upholding the principle of nonmaleficence; we promise our participants to not let any data fall into the hands of people who may use it to publish stigmatizing results about the LGBTQþ communities. For example, someone could look at the gender identities of racial/ethnic minorities and make claims that a specific racial/ethnic minority should be targeted to 'cure' people with a specific gender identity. As such, The PRIDE Study has developed an Ancillary Study process in which investigators interested in using PRIDE Study data submit an application which is reviewed by both a Research Advisory Committee and Participant Advisory Committee to affirm appropriate data use and work in collaboration with The PRIDE Study for planning, analysis, and dissemination. Details about the Ancillary Study process are available at www.pridestudy.org/collaborate or by contacting us at [email protected] or 855-421-9991 . --- CONFLICT OF INTEREST STATEMENT JO-M has served as a consultant for Sage Therapeutics , Ibis Reproductive Health , Folx , and Hims . MRL has served as a consultant for Hims and Folx . AF has served as a consultant for Hopelab, a not-for-profit research group. None of these roles present a conflict of interest with this work as described here. All other other authors have no conflicts of interest to report.

Objective: The purpose of this study is to describe online health information seeking among a sample of transgender and gender diverse (TGD) people compared with cisgender sexual minority people to explore associations with human papillomavirus (HPV) vaccination, and whether general health literacy and eHealth literacy moderate this relationship. Materials and Methods: We performed a cross-sectional online survey of TGD and cisgender sexual minority participants from The PRIDE Study, a longitudinal, U.S.-based, national health study of sexual and gender minority people. We employed multivariable logistic regression to model the association of online health information seeking and HPV vaccination. Results: The online survey yielded 3258 responses. Compared with cisgender sexual minority participants, TGD had increased odds of reporting HPV vaccination (aOR, 1.5; 95% CI, 1.1-2.2) but decreased odds when they had looked for information about vaccines online (aOR, 0.7; 95% CI, 0.5-0.9). TGD participants had over twice the odds of reporting HPV vaccination if they visited a social networking site like Facebook (aOR, 2.4; 95% CI, 1.1-5.6). No moderating effects from general or eHealth literacy were observed. Discussion: Decreased reporting of HPV vaccination among TGD people after searching for vaccine information online suggests vaccine hesitancy, which may potentially be related to the quality of online content. Increased reporting of vaccination after using social media may be related to peer validation. Conclusions: Future studies should investigate potential deterrents to HPV vaccination in online health information to enhance its effectiveness and further explore which aspects of social media might increase vaccine uptake among TGD people.

Background International migration1 is increasing globally. There were 281 million international migrants worldwide in 2020, nearly half of whom were women [1]. Migrant women are often young and of childbearing age [1]. Pregnancy is a period of increased vulnerability for migrant women since the transition to motherhood and migration to a new country both involve more profound changes in their psychological, social, and physical domains than any other developmental stage of the family life cycle [2,3]. In Switzerland, foreigners account for 25.3% of the permanent resident population, of which more than half have lived for more than ten years or were born in Switzerland [4]. On average, the fertility rate of migrant women in Switzerland is higher than that of Swiss women [4]. However, migrant women from non-European countries are more likely to report having a negative childbirth experience and being less satisfied with their maternity care than Swiss women [5,6]. They were reported to have a higher risk of perinatal complications , mental and physical illnesses before or after birth, and higher infant and maternal mortality rates than native women [7]. The reasons leading to maternal health disparities have been identified as personal issues and structural problems . Maternal health among the migrant population is a public health issue that can result in significant mental distress, disrupt maternal bonding, and have long-term impacts on the well-being of the woman, the baby, her family, and wider society [11,12]. Therefore, a better understanding of migrant women's reproductive health is of great public health concern in the Swiss healthcare system [6]. Chinese migrants constitute a relatively small minority in Switzerland. The number of Chinese women giving birth is relatively small but steadily increasing, with 3902 Chinese mothers having given birth in the past ten years -the triple when compared with two decades ago [13]. Pregnancy, childbirth, and the postpartum period are a psychological and physical continuum [14]. According to our first study on the pregnancy experiences of Chinese mothers in Switzerland, they felt dissatisfied with the follow-up pregnancy complications and lacked information on medical procedures and prenatal courses [15]. Thus, we emphasize the importance of follow-up research on their childbirth experiences. Furthermore, Chinese migrants constitute the third largest Asian migrant group in Switzerland [15]. Women from Asia tend to experience greater cultural and language gaps than those from European countries. We believe that some of the findings of this study may contribute to a better understanding of the childbirth experiences of mothers from Asia in Switzerland. The objectives of this study were to examine the childbirth experiences of Chinese migrant mothers and their family members and to identify their intrapartum care needs. Our study responds to the call of the Swiss Federal Office of Public Health for specific studies of migrant populations [7]. The findings of this study may help address health inequities and promote equal maternity care services for the migrant population from non-European countries in Switzerland. This study also enriches the body of knowledge about migrant studies on reproductive health globally. --- Methods --- Research design and participants This is the second study in a series about the pregnancy, labor and birth as well as postpartum experiences of Chinese migrants in Switzerland [15]. A qualitative study employing in-depth, semi-structured interviews was conducted with Chinese mothers and their family members on their childbirth experiences in Switzerland. To be eligible for the study, Chinese mothers were required to give birth in Swiss hospitals. The newborns were less than one year old at the time of data collection. Women's family members, including their partners, parents, or in-laws, did not have to be Chinese nationals. Participants were recruited from eight cantons of French-and German-speaking Switzerland using a Chinese social media platform, named WeChat, by the first author . Research flyers containing information about the study objectives, interview procedures, and eligibility criteria were posted in several WeChat groups of Chinese migrants in Switzerland. Potential participants were contacted by the first author to address their inquiries regarding interview procedures and data confidentiality. Once they agreed to participate in the study, interviews were scheduled either at home or online. Each participant provided their written consent to use audio recordings and transcriptions in data analysis, paper writing, and publication. The study was approved by the cantonal research ethics committee . --- Interview procedures In-depth interviews took place between September 2019 and April 2020 with 27 participants, including 14 Chinese women and 13 family members. Individual interviews with 14 Chinese women and 10 Chinese family members were conducted in their native language and three foreign partners in English. Twenty-two participants were interviewed face-to-face at the participants' homes whereas interviews with the other five participants were performed online. Interviews with women and their family members were conducted using two separate interview guides with both broad open-ended questions and specific follow-up questions. Topics covered included choice of labor hospitals, admission to the hospital for labor, birth modes, labor pain, obstetric interventions, birth companions, breastfeeding initiation and practices, infant and maternal care, medical treatments, etc. The interview lasted an average of 41 minutes. All interviews were audio-recorded, transcribed, and translated into English for data analysis . --- Data analysis Thematic analysis was employed to generate a detailed description. Six steps of the analytic process from Braun and Clarke were followed [16], using qualitative analysis software MAXQDA [17]. The participants in the research team come from a number of disciplines. The doctoral candidate in Nursing Sciences is specialized in maternity care. A postgraduate specializing in psychology assisted with data analysis. Three senior researchers in Nursing Sciences , Social Sciences , and Clinical Psychology have expertise in maternity care, perinatal mental health, qualitative research, and sinology. The data were analyzed by the two researchers separately and then jointly discussed with AH. Regular discussions between the two researchers were organized to refine the codes and reconcile discrepancies. Both researchers worked during the analytic stage under the supervision of AH. --- Results The 14 Chinese mothers, aged 34 on average, had been in Switzerland for one to twenty-one years. 13 out of 14 were married. All Chinese mothers had legal status and were covered by mandatory health insurance. Interviews were conducted within seven postpartum months on average. More information about the sociodemographic characteristics of the participants can be found in [15]. Birth-related information is summarized in Table 1. Three main themes and seven sub-themes were extracted from the transcripts: Sense of security, Intrapartum care, and Needs beyond childbirth. The outline of the main themes and sub-themes is shown in Table 2. --- Main theme one: Sense of Security Chinese mothers in our study prioritized giving birth in a physically and psychologically safe environment. Their choice of the hospital and the timing of their admission to the hospital during labor were determined by their need to feel secure. Four mothers preferred to give birth in smaller state or private hospitals. They generally appreciated the friendly inpatient environment, and the personalized care from a dedicated team. They reported that they had abundant opportunities to interact with healthcare workers during the labor process when they were in doubt or seek additional support for comfort. In addition, one mother who gave birth in a private hospital highly valued the ability to actively participate in the decision to administer epidural analgesia without the need to follow the rigid guideline based on the extent of cervix dilation. "We chose a small hospital on purpose. [ • • • ] When we needed something, the nurses could quickly respond, which made us feel at ease. [ • • • ] I had a struggling 13-hour long birth and my midwife kept patiently explaining and encouraging me. She helped me to overcome anxiety and stress toward the end of the birth. " " --- I stayed at hospital X and they handed us a form to be filled during prenatal checkups to give consent for epidural administration. They explained that the consent would grant me prompt adoption of pain-relief should I need it. There was no need to wait until certain cervix dilation indications. I felt very comfortable and a sense of selfcontrol thanks to the ability to apply pain control according to my personal comfort. " Three mothers gave birth in private hospitals because their obstetricians during their pregnancies were able to be present when they were giving birth there. The continuity of care and the trusted relationships made them feel confident and secure. --- "My private obstetrician worked there. My first baby was born there. I found him to be very reliable and trustworthy. " --- Timing of labor admission Except for one mother who had an elective caesarean section with a scheduled admission date, all Chinese mothers who planned a vaginal birth were asked to follow the admission criteria provided by their birthing hospitals and to make their own judgments about when to request labor admission. However, they reported difficulties in determining whether they met the admission criteria and were in active labor because nine of them were first-time mothers without prior childbirth experiences. Consequently, they had to return to hospitals multiple times to confirm with midwives, which caused them great anxiety and insecurity while waiting to be admitted. --- "I felt contractions and was not sure if I was about to give birth. Therefore, I called the hospital and told them I was on the way. The midwife who met me did some examinations and informed me that I would not be giving birth immediately. She advised me to go home and wait for labor to start before coming to the hospital. She stated that it was normal for firsttime mothers to make several trips to the hospital. " One mother explained that her symptoms were atypical and did not meet the list of indicators of birth signs given by the hospital, leaving her to feel confused and stressed. She finally gave birth on the way to the hospital due to a series of misjudgments. --- "My symptoms were very different from the other mothers' . When I had my second child, all I had was back pain and a sore, tight back. I did not know that was a sign of labor. [ • • • ] I called the hospital at 3 a.m., and they told me that I did not need to come to the hospital. [ • • • ] One hour later, my water broke, and I started to have very strong contractions. My husband and I decided to go to the hospital immediately [ • • • ] My baby was born in my pants while we were waiting for the taxi downstairs. " --- Main theme two: Intrapartum Care Chinese mothers in our study expressed a desire to achieve a physiological birth with minimal obstetric intervention and to avoid an instrumental birth or an emergency caesarean section. They generally appreciated pain control and continuity of practical and emotional support from a birth companion. Three sub-themes emerged during their labor and birth process including pain management, birth mode, and birth companion. --- Pain management In general, Chinese mothers reported they had access to comprehensive pain management and utilized various pain relief methods or techniques throughout their labor process. For instance, non-pharmacological analgesic methods, such as moving around, changing positions, deep breathing, and birthing balls, were introduced to them in the early stages of labor. Pharmacological analgesia, including intravenous anesthesia, inhaling nitrous oxide, and morphine injections, were also given when deemed appropriate or necessary. They stated that they could also proactively request epidural analgesia if the cervix had been dilated to two to three centimeters or if those mild pain relief methods were inadequate. In our study, all 12 mothers who tried a vaginal birth used different methods of analgesia at different stages of labor, except for one mother who did not have time to benefit from analgesia due to an emergency vaginal birth. --- "At first, they gave me a small amount of intravenous analgesia, and then I inhaled nitrous oxide, but neither worked to relieve my pain. [ • • • ] Finally, they administered an epidural. " When asked about the impact of the epidural on their labor, Chinese mothers generally gave positive feedback. Eight mothers expressed that it not only allowed them to rest and regain strength between contractions but also made it easier for them to cooperate with birth attendants. --- "I felt like I was resurrected all of a sudden . " However, four mothers thought that their late epidural, excessive doses, and technical problems as risk factors that led to undesirable childbirth experiences. --- "When I had the epidural, my legs were anesthetized, not my belly. I kept feeling pain for 20 minutes. [ • • • ] Then they decided to give me a higher dose of anesthesia, but they added too much. I could no longer feel the pain, but I also could not feel the contractions. I could not feel anything. [ • • • ] When the con- tractions came, my midwife told me to push. When she told me to push, I pushed, and I completely listened to her throughout the labor. " To avoid this situation, Chinese mothers claimed that family members, such as in-laws, pressured them not to use epidural analgesia. "My mother-in-law, who was in China, talked with my husband on the phone. She insisted a lot that I could not use epidural anesthesia. However, we decided to use it after my husband repeatedly confirmed its safety with the doctor. " --- Birth mode Except for one mother who had an elective caesarean section due to advanced maternal age , the other mothers stated that their birth attendants and obstetricians had always encouraged them to have a vaginal birth. They experienced various obstetric interventions, such as artificial rupture of membranes, numerous vaginal examinations, or prolonged use of oxytocin, during the vaginal birth attempts. --- "I was examined vaginally by interns more than 20 times during the first night. Then I had a high fever, and my baby's heart rate became erratic. They performed an emergency caesarean section, which made me very unhappy. " Among those mothers who attempted vaginal births, three underwent instrumental births, and three had emergency caesarean sections following a prolonged and unsuccessful vaginal birth. Four of them had a traumatic birth experience and were suffering from post-traumatic stress reactions. In addition, they developed long-term health issues, such as postpartum anemia, prolonged body pain, urine leakage, and constipation, following childbirth. --- "I was given intravenous oxytocin for three days at the hospital, and I bled vaginally on the third night. They rushed me to the delivery room and resumed the oxytocin injections. Then I suffered 20 hours of painful contractions. I requested an epidural when my cervix dilated to two fingers. Soon after, I began having nosebleeds. The doctor rushed me to an emergency caesarean. [ • • • ] I suffered too much pain from vaginal birth to an emergency caesarean section. [ • • • ] It had long-term consequences for my physical and mental health. " Nine of the fourteen mothers considered their childbirth experiences as poor. The main reasons for their negative experiences were birth complications associated with instrumental births and emergency interventions, which did not match their expectations based on their smooth pregnancy process or their ideology about which birth mode had better childbirth outcomes. Mothers who had instrumental births felt that they should have gone straight to caesarean sections to avoid the prolonged struggles and traumatic experiences during the vaginal birth process, as well as postpartum suffering. Mothers who had emergency caesarean sections expressed dissatisfaction with the sudden change in the birth mode because they believed that such an emergency intervention was only necessary for those mothers with complications during the pregnancy, which they did not have. Some mothers and their family members felt there was a lack of transparency in information exchange and the decision-making process when it came to finding the optimum birth mode appropriate for the mothers according to their personal circumstances and precipitations. --- "What I very much regret is that they were not transparent enough about what was happening. They performed an episiotomy on my wife, and I had no idea until I saw what was going on. [ • • • ] They did not explain to me anything from start to end. I did not think it was a good practice, anyway. " "The nurses told me that the doctor would come the next day to discuss with us the possibility of a caesarean section. However, the next morning, the doctor informed me I would be having a caesarean section right away. I thought, 'What? There was no discussion. It was an already-made decision. ' [ • • • ] I was very disappointed, but as minor- ity migrants, we did not speak the language fluent enough to argue with them. " Some mothers were hesitant to give birth vaginally due to various concerns, such as a previous caesarean section, a large fetus, or severe vaginal bleeding. They considered that a better birth outcome might be achieved if they were allowed to have an elective caesarean section . --- "Compared to my previous childbirth experience with an elective caesarean section, I suffered more and had a lengthy recovery this time due to an instrumental birth. I think the caesarean section would have been better for me. " --- Birth companion In our study, all mothers, regardless of vaginal birth or caesarean section, had their intimate partners as birth companions at various stages of the labor process. Ten Chinese mothers who gave birth vaginally stated that practical support, such as interpreter assistance, nonpharmacological pain relief, and emotional support, were crucial to making them feel secure and relaxed. "My husband was extremely helpful, especially when I was in pain. I took prenatal classes to learn how to breathe to relieve pain, but I forgot how to do it when I was in actual labor. My husband kept encouraging me by massaging, holding my hands, and cheering me. The presence of my loved one put me at ease. " Four Chinese mothers who had caesarean sections reported that their partners were permitted to accompany them during certain stages of the caesarean section, such as after anesthesia was administered, the operating table was prepared, or the caesarean section was performed. They wished for flexibility regarding when their partners could be present during the caesarean section. Granting earlier companionship from their partners was considered beneficial. --- "It was too late for my husband to come in! . [ • • • ] He was not allowed in until after my belly had been opened. If he had been there earlier, I would not have had to keep an eye on the medical personnel around me, and I could have relaxed. " Generally, fathers of Chinese or non-Chinese nationals all confirmed that being present during the birth process had a positive physical and psychological impact on mothers. Although some Chinese fathers described the labor and birth process as "brutal" and "unpleasant", they believed that witnessing the birth was a highly meaningful and emotional event. --- "I found being a birth companion a special and meaningful experience for me. The labor process was relatively long and exhausting because it was our first child. [ • • • ] At first, I thought I would not be too excited to see the baby, but after I was with my wife for more than 10 hours, I cried when I saw the baby come out. I cannot imagine how anxious I would have been if I had just waited outside the delivery room. But after I went through the entire process, all I felt was moved and heartwarming. I felt that as a father, it was very precious to witness the first moments of my child coming into this world. " --- Main theme three: Postpartum needs Chinese mothers and their families in our study desired professional support and culturally sensitive care. They expressed concerns about support for infant care and breastfeeding initiation, in particular those with caesarean sections. In addition, they desired to retain some traditional cultural practices that were still meaningful to them following childbirth. --- Professional support Although standardized neonatal care was practiced following childbirth, several mothers reported shortfalls in the guidance and support they received in establishing early mother-infant bonding. Mothers with cesarean section or assisted virginal birth reported issues ranging from inadequate skin-to-skin contact, and lack of mother-infant rooming-in, to weakness in practicing early bonding after excessive blood loss. They considered instructions given by maternity care workers insufficient and they often did not know what to expect after the newborn were out. --- "My son was delivered by cesarean section. They didn't explain much about the process. I had to ask my husband whether the baby was out. When the baby was out, they brought the baby to me but only rubbed him against my face a few times before taking him away with my husband. I didn't even have time to look at my baby. " Early sucking and breastfeeding initiation were perceived to be very challenging for most mothers immediately following childbirth and during their hospital stay. Various difficulties were raised by delayed milk production, inverted nipples, poor infant-mother attachment, or inappropriate breastfeeding positions. Complications for breastfeeding developed in one case due to earlier formula feeding administered by maternity care works with bottle nipple, leading to baby's nipple confusion. --- "I had no milk at all in the first three days after giving birth, even with breast pumping. My baby's weight dropped significantly. He didn't get anything, even though he kept sucking. The midwives suggested I turn to formula milk. They gave him formula milk in a bottle. He drank very fast. He didn't want to suck for breast milk anymore after that. Because it was much easier for him to suck out milk with bottle nipples. " Support was deemed even more imperative by mothers following caesarean-section, whose mobilities were greatly restricted due to post-operative pains. --- "My baby started breastfeeding on the third day, but he could not suck on my nipples at all because I have inverted nipples. In addition, I had a large caesarean section wound and no idea how to breastfeed. [ • • • ] The midwives told me, 'I will only teach you once, and then you should do it on your own, ' so I was hesitant to seek their help again. " "They told me to go upstairs to the inpatient unit for some medical examinations after the caesarean section. This was extremely difficult for me because I was extremely weak. The wound was extremely painful, every step I took hurt terribly. --- [ • • • ] I thought their service was poor, and I was particularly dissatisfied. " Unlike in China, where family members are encouraged to stay throughout the day in the hospital to assist with the care of newborns and mothers, Chinese mothers realized that Swiss hospitals do not allow family members to stay overnight. Consequently, some mothers reported that their postpartum hospital stays were very challenging, particularly for those who underwent a caesarean section. --- "The nights were the most difficult for me because I did not have anyone to help with baby care. He cried all the time while he was in the stroller. I could not even move toward the bed after the caesarean section, let alone hold and comfort him. " A few Chinese mothers and their family members believed that their healthcare providers were occasionally critical rather than helpful. --- "In general, the midwives would not help us with infant care. [ • • • ] When the baby kept crying, the midwife came and asked, 'Why have you not calmed him down?' " --- Culturally sensitive care Chinese mothers and their family members reported that their needs for culturally sensitive care were primarily connected with their traditional postpartum practices. They claimed that it was not always possible to strictly follow their cultural customs regarding hygiene and cleanliness restrictions, as well as dietary and drinking habits. Participants explained that their healthcare providers were not aware of their specific cultural customs and thus could not provide culturally sensitive care. In addition, the lack of effective communication between mothers and their healthcare providers sometimes resulted in misunderstandings or confusion for both parties. Nevertheless, with proper guidance, Chinese mothers managed to adapt to the norms of the host culture while maintaining some cultural practices that were still meaningful to them. "My midwife advised me to take a shower the day after the caesarean section. I told her that in our culture, mothers are usually not allowed to shower for the first month after giving birth. The midwife warned me that if I did not shower, I would stink and bother the other mothers in the room. Eventually, I took a shower upon the midwife's request. " "Chinese women are supposed to keep warm and eat hot food after giving birth. However, the hospital here was unconcerned. They served cold drinks and food to my daughter . We brought hot water and food from home to the hospital every day. " --- Discussion Like most women worldwide [14,18], Chinese mothers who gave birth in Switzerland desired a physically and psychologically safe environment. Compared to Swiss mothers, who prioritized professional and obstetric skills, an intimate atmosphere, and continuity of care from a private obstetrician when choosing a birthing hospital [19][20][21], the safety of both mothers and babies during the labor process was weighed heavily among Chinese migrant mothers. University hospitals were their first choices as they associated their safety with the availability of interdisciplinary collaborations in general hospitals to provide emergency interventions for both mothers and babies in case unexpected events occurred during labor and birth. A few mothers, especially from densely populated major Chinese cities, placed a higher value on woman-centered care [22,23] from the entire maternity care team during the labor process in smaller-scale state hospitals or private hospitals. They appreciated the maternity care that prioritizes pregnant women's caring needs, personalized aspirations for comfort, and the ability to collaborate with maternity care professionals for decision-making. All unlikely to be accessible for the general population in their home country due to "assemblyline birth" [24] in Chinese hospitals. Evidence-based recommendations for late labor admissions can avoid early obstetric interventions, emergency caesarean sections, and instrumental births [25,26]. However, unlike other migrant mothers who delayed hospital admission for fear of caesarean sections [27], Chinese mothers in our study perceived early admission as a safe guarantee for handling labor uncertainties during the early stage, compromising the increased probability of early and frequent medical interventions. It is tied to the strong beliefs in Chinese culture that childbirth is a very dangerous event and that a woman who gives birth to a child is passing through a ghostly gate [28]. It is important to ensure the safety of women and babies so that the event takes place without danger. Furthermore, our study also showed that the criteria for labor admission based on active birth symptoms as instructed by healthcare personnel were considered ambiguous and that one mother even experienced atypical labor symptoms. All of these posed challenges for Chinese mothers to correctly interpret their early labor symptoms and left them stressed about the proper timing to request admission. In view of supporting mothers and improving the efficacy of maternity care services, it is advisable to encourage healthcare providers to provide more practical and clear guidelines and admission procedures to migrant women. It is imperative to educate them to truly understand that the early labor process can be lengthy. Moreover, it would be beneficial to provide migrant mothers with timely phone support to facilitate the recognition of early labor reactions at home. In contrast to previous studies of Chinese mothers giving birth in some foreign countries [29,30], which found that Chinese mothers perceived labor pain as inevitable and tolerated it silently, almost all Chinese mothers in our study highly valued labor analgesia for pain control. The reason for such a disparity in Chinese migrant mothers' practices could be partly attributed that epidural analgesia is widely regarded as an effective and safe for labor pain relief in developed countries [31][32][33][34]. In the meanwhile, with the evolution of childbirth ideology among Chinese people over the past decades as China underwent rapid social and economic growth and reformation, China has been advocating the implementation of epidural labor since 2015 to reduce the caesarean section rate, as fear of labor pain is the primary reason for Chinese mothers to opt for caesarean sections [35,36]. This national campaign has also dispelled the widespread beliefs among Chinese women and their families that epidural analgesia is harmful to the mother and the baby. Consequently, Chinese mothers, as shown in our study, became more self-aware of their physical and psychological comforts rather than silently tolerating and suppressing their pain relief needs. It should also be noted that a few Chinese mothers in our study had less desirable pain control outcomes, and they attributed this to inappropriate timing of pain management at certain stages, excessively administered pain-relief medicine doses, or technical problems. This necessitates further research to understand any underlying deficiencies in the assessment of proper obstetric interventions or ineffective communications between mothers and healthcare providers. Our research found that Chinese migrant mothers were much more likely than Swiss women to have instrumental vaginal births [37]. Nine out of ten mothers who gave birth vaginally in our study were subjected to various obstetric interventions during their early labor stage that are not recommended by WHO intrapartum care guidelines [18]. Three Chinese mothers switched to a caesarean section after a long and unsuccessful trial of vaginal birth. These interventions probably played a role in an abnormal fetal heart rate, maternal fatigue, or fever, necessitating an instrumental birth or an emergency caesarean section. To avoid the traumatic childbirth experiences reported by Chinese mothers in our study, in particular for difficult instrumental births, it may have been preferable to proceed directly to a caesarean section [38]. It should also be noted that traumatic and instrumental births had been reported as a common issue in the Swiss maternity care system, not only experienced by Chinese migrant women but also Swiss maternal women in general [39,40]. However, migrant women are more likely to confront such an issue due to communication difficulties and poor maternity care support [7]. The previous research on Chinese migrant mothers in the United Kingdom and Canada, conducted nearly 20 years ago, showed that mothers hesitated to have their partners present during childbirth to avoid embarrassment out of fear of their loss of control and emotional outbursts over labor pain [29,41]. However, Chinese mothers in our study highly appreciated the birth companionship, which provided them with a great sense of security and confidence. They placed a high value on psychological safety and received practical and emotional support from their birth companions. This is particularly important for migrant mothers in an unfamiliar healthcare environment. This should be viewed in the context of the unique setting for Chinese mothers giving birth in Switzerland. Unlike in English-speaking countries, where Chinese mothers could initiate more in-depth communication with health professionals in English, a widely taught second language in China, language barriers faced by those Chinese mothers in Switzerland impede their communication with their healthcare providers [15]. Partners of those Chinese mothers who either speak the official Swiss language or have relatively adequate language skills could greatly aid communication between birth attendants and mothers. Another factor contributing to the positive experiences from partner engagement in the childbirth process is Chinese mothers' ability to maintain self-control while cooperating with their birth companions and birth attendants in our study, thanks to epidural analgesia. It is noteworthy that our study indicated that overall childbirth experiences of Chinese mothers in Switzerland were perceived as less satisfactory compared to their overwhelmingly positive pregnancy experiences [15]. Arguably, we should not directly compare the mothers' experiences at different stages of childbearing. However, this disparity indeed implies a significantly higher demand for effective communication during labor and birth. Unlike consultations with healthcare providers in hospitals or clinics during pregnancy, which were carried out and repeated over an extended period, mothers need to accurately convey their rapidly changing personal situation to birth attendants during a relatively short but intense childbirth process. Given that direct and individualized communication with healthcare providers throughout the labor and birth process is essential for promptly addressing mothers' needs or adjusting any treatment or intervention plan, strategies to overcome communication barriers between migrant mothers and healthcare providers need to be implemented. Healthcare professionals and social service providers should closely collaborate to ensure migrant populations have equal access to quality care. For example, childbirth preparation courses should be expanded to all migrant women in Switzerland, ranging from essential knowledge about pregnancy, childbirth, and breastfeeding to a focus on the maternity care system and available support resources, with the assistance of community interpreter services [42]. Previous research on migrant women's maternity care services in Switzerland has often overlooked their postpartum experiences during the hospital stay, but mainly focused on barriers to access maternity care during the pregnancy period, antenatal services, and pregnancy outcomes [5,9,43,44]. Our study investigated the experiences of Chinese mothers during postpartum hospitalization, with a particular focus on their need for support from their healthcare providers. Chinese mothers frequently expressed concerns about inadequate support for infant care and breastfeeding initiation, as well as a lack of support and attention to mothers who had caesarean sections. A few mothers and fathers felt their healthcare providers were sometimes critical rather than helpful. Given that postpartum care and support during hospitalization are primarily provided by family members in China, our study reveals a significant gap in understanding Chinese mothers' needs and expectations between themselves and their healthcare providers. Since migrant women and their families frequently lack knowledge of the routines, practices, and personnel of the Swiss maternity care system, assisting them in understanding the coverage of postpartum support and services in the Swiss healthcare system before labor admission should be promoted. This will enable them to be prepared even in the absence of family support and to explore strategies for coping with the challenges that come with childbirth during their postpartum hospital stay. Finally, it is interesting to observe how Chinese mothers and their family members attempted to adapt to the local practices of the host culture while retaining some traditional cultural practices that were still meaningful and accessible to them. For instance, they demonstrated flexibility in breaking hygiene and cleanliness restrictions imposed by their traditional postpartum customs while adhering to the Yin -Yang concept in Chinese traditional medicine by eating and drinking warm [45]. Healthcare workers should encourage migrant women to express their concerns and specific needs related to their cultural practices. --- Strengths and limitations The strengths of our study include the diverse perspectives of Chinese mothers, their partners, parents, and in-laws. The interviews were conducted within seven months following childbirth on average when the mothers and their family members had fresh memories of their previous birth experiences. The findings of this study apply directly to healthy migrant mothers of various parities and economic and educational backgrounds. However, migrant women with specific conditions, such as severe prenatal comorbidities or postpartum complications, or who had a newborn baby with health issues, were not included in this study. Migrant mothers falling into this group may have different concerns and expectations when it comes to their maternity care needs. Future studies are required to build a comprehensive understanding of the maternity care needs of migrant populations in Switzerland. In our study, almost all mother participants have a high educational background and above-average socioeconomic status due to the fact that immigration for non-EU citizens to Switzerland is limited to higher education, skilled work, or cross-cultural marriage. Care should be taken when applying the knowledge and experiences gained in this study to other migrant groups where a broader socioeconomic status could exist. --- Conclusions To our knowledge, this is the first qualitative study on the childbirth experiences of Chinese mothers, a small but steadily growing minority group, in Switzerland. In general, our study showed that the childbirth experiences of Chinese mothers in Switzerland were less satisfactory compared with their pregnancy experiences. Language barriers and cultural differences hinder the establishment of a trusted relationship and effective communication between the mothers and their healthcare providers. Our study provides pivotal information for further follow-up studies to address migrant mothers' needs and expectations and minimize the use of invasive medical interventions, which could potentially increase the risks of traumatic birth experiences. To achieve optimal childbirth outcomes while maximizing the efficiency of medical resource utilization within the healthcare system, it is imperative for healthcare providers and policymakers to realize Chinese mothers' perspectives on their sense of labor security, as revealed by our study to be unique compared with native Swiss mothers. They highly valued labor pain control and birth companionship, both in contrast with previous studies on the same minority group in other foreign countries. This should be viewed in the context of rapid social-economic development in their home country in the past decades, which resulted in strengthened self-esteem and self-awareness of their own physical and psychological comfort during labor and birth. Postpartum care for migrant mothers and their babies during their hospital stay was overlooked in previous studies. Our research followed up with Chinese mothers and indicated inadequate support for them during postpartum hospitalization to initiate breastfeeding or formula feeding. Chinese mothers, in particular those with caesarean sections, experienced difficulties caring for the babies and themselves in the absence of family members and an unfamiliar healthcare system. When caring for migrant populations, healthcare providers need to be aware and knowledgeable of their specific cultural practices relating to pregnancy and childbirth in order to provide safe and respectful care. Further research is recommended from the perspective of maternity care providers to investigate their experiences and perceptions of caring for migrant women or those other minority groups within the Swiss maternity system, particularly those of non-European origins with very different cultural backgrounds. --- --- Authors' contributions DC contributed to the conception and design as well as the participant recruitment, data collection, transcription, validation of the English translation, analysis, and manuscript writing. PV took part in the transcription, data analysis, and manuscript writing review. HL provided comments on the final manuscript. BZ participated in critical comments on the manuscript writing process as well as the validation of translated English quotations from Chinese. AH conceptualized the study and supervised the work of DC and PV. AH also critically commented on the data analysis and writing of the manuscript. All authors read and approved the final manuscript. --- --- --- Competing interests AH is a board member of the management committee of CA18211. The other authors declare that they have no competing interests. ---

Background In Switzerland, foreigners account for 25.3% of the permanent resident population, and the fertility rate of migrant women is higher than that of Swiss women. However, migrant women from non-European countries are more likely to report having negative childbirth experiences than Swiss women. For example, during pregnancy, Chinese migrant mothers often felt dissatisfied with the follow-up pregnancy complications and lacked information on medical procedures and prenatal courses. In this paper, we explored their childbirth experiences in Swiss hospitals and how Swiss healthcare providers supported them. Method A qualitative study employing in-depth, semi-structured interviews was conducted with 14 Chinese mothers and 13 family members. All interviews were audio-recorded, transcribed, and translated into English for data analysis. Thematic analysis was employed to generate a detailed description.Three main themes were extracted from the transcripts: (1) Sense of security, (2) Intrapartum care, and (3) Postpartum needs.Our study shows Chinese migrant mothers prioritized giving birth in a physically and psychologically safe environment, with pain control and practical and emotional support from their intimate partners. They desired a physiological labor and birth with minimal obstetric interventions. Our research also reveals their postpartum needs, emphasizing the importance of postpartum support and obtaining culturally sensitive care during their postpartum hospital stay. The study adds new knowledge of specific migrant studies in Switzerland, as called for by the Swiss Federal Office of Public Health. The results call for the transcultural care skills training of Swiss healthcare providers to enable migrant women to have a more positive childbirth experience.

Introduction Dynamical processes involving populations of individuals constitute paradigmatic examples of complex systems. From epidemic outbreaks to opinion formation and behavioral dynamics [1][2][3][4][5][6][7][8][9], the impact of the underlying web of ties in the overall behavior of the population is well known. In this context, Evolutionary Games [10,11] provide one of the most sophisticated examples of complex dynamics in which the role of the underlying network topology proves ubiquitous. For instance, when cooperation is modeled as a prisoner's dilemma game, cooperation may emerge depending on how the population is networked . Up to now, it has been hard to characterize in detail the global dynamics by which local self-regarding actions lead to a collective cooperative scenario, relating it to the network topology. Indeed, most network studies have been focused on the analysis of the evolutionary outcome of cooperation [14] either by means of the numerical analysis of steady states [12,16,18,19,24,25,29,32,34,35,[38][39][40] or by means of the analytical determination of the conditions for fixation in the population or by means of the determination of positive inclusive fitness effects for particular homogeneous network interaction structures and low intensities of selection [17,20,22] -without characterizing the self-organization process by which one of the strategies outcompetes the other. Here we show how networked individuals, engaging in a prisoner's dilemma of cooperation, give rise to a global, population wide, behavioral dynamics which deviates strongly from the original PD, depending sensitively on the underlying network of contacts: Homogeneous networks promote a coexistence dynamics between cooperators and defectors -akin to the Chicken or Snowdrift game [11,38,[41][42][43] -whereas heterogeneous networks, from broad scale to scalefree [4,44], favor the coordination between them, similar to the Stag-hunt game [45]. To this end we define a time-dependent variable -that we call the average gradient of selection -and use it to track the self-organization of cooperators when co-evolving with defectors under network reciprocity. Similar to existing analytical approaches [43,46], the AGoS is able to provide a measure of the change in time of the frequency of cooperative traits under selection. The AGoS can be computed for arbitrary intensity of selection , arbitrary population structure and arbitrary game parameterization. We further prove that the global games are not fixed: they change in time, co-evolving with the motifs of cooperators in the population. The evolutionary outcome of such a self-organization process will depend sensitively on this coevolution, which can be followed using a time-dependent AGoS. --- Dynamical Model Let us consider pairwise interactions between individuals who can behave either as a Cooperator or Defector . Whenever cheated by a D, a C receives a payoff S , while the D receives T . Mutual cooperation provides R to each player, while mutual defection provides P . One obtains the prisoner's dilemma -the most famous metaphor of cooperation -whenever T.R.P.S. We formalize the dilemma in terms of a single parameter B by defining T = B.1, R = 1, S = 1-B and P = 0. The results remain quantitatively unaltered if one adopts the more popular parameterization T = b, R = b-c, S = 2c and P = 0. In the framework of evolutionary game theory, we adopt a stochastic update rule of social learning, where in each time step a random individual i imitates the strategy of a randomly selected neighbor j with a probability that increases with the fitness difference between them [14,24,[47][48][49][50] . In the limit of well-mixed populations of size N [10,11], the frequency j/N of Cs will increase in time depending on whether the gradient of selection [51,52] G~T z {T { is positive , where T + [47] represent the probabilities to increase and decrease the number of Cs in the population by one. For the PD, Gv0 for all j and, as a result, cooperation will most probably die out. The same scenario is obtained when NR', where we recover the scenarios described by the famous replicator dynamics [11,[47][48][49][50]. The elegance of this result is best appreciated when we realize that the population ends up adopting the Nash-equilibrium of a PD game interaction between two individuals: everybody defects. Consequently, there is no difference in the outcome of the game, from an individual or from a population wide perspective, a feature that, as discussed below, will not remain true in structured populations. --- Results The previous analysis assumes finite yet structureless populations, a feature which is seldom observed in practice, with strong implications in many natural phenomena. A homogeneous network of size N represents the simplest case of a structured population, where all individuals engage in the same number of games k with their first neighbors, also imitating their behavior. Let us consider a homogeneous random network -also called a regular random graph -in which all links are randomly connected, while all nodes have each the same number of links [4,53,54]. In this case, individuals with the same strategy no longer share the same fitness: fitness becomes context-dependent. The same happens to G, becoming hard to define it analytically. Consequently, we define the AGoS -denoting it by G A as the average i) over all possible transitions taking place in every node of the network throughout evolution, and ii) over a large number of networked evolutions . This AGoS, which must be computed numerically, becomes therefore network dependent but context independent, as it recovers its population averaged, or mean-field, character. Hence, the AGoS may constitute a powerful tool to understand dynamical processes at a population-wide scale stemming from individually defined, but often seemingly unrelated, rules. The results for G A on homogeneous networks of size N = 10 3 , k = 4 and different values of B are shown in Fig. 1a. Unlike wellmixed populations where cooperation has no chance, homogeneous networks can sustain cooperation [12,14,15,24,54]. The shape of G A no longer pictures a defection dominance dilemma typical of a PD, but a gradient of selection similar to what one observes under co-existence dilemmas in well-mixed populations [43]. In other words, even though every individual engages in a PD, from a global, population-wide perspective, homogeneous networks are able to create an emerging collective dynamics promoting the co-existence between Cs and Ds. As we show below, the emergence of an unanticipated global dynamics from a distinct individual dynamics pervades throughout all evolutionary dynamical processes in structured populations studied here. The co-existence point x R is associated with the internal root (x R [ 0,1 ) of G A inexistent in well-mixed populations -whose location decreases with increasing B. Together with x R one obtains a coordination root of G A since, in the absence of cooperative partners, Cs will always be disadvantageous. However, the impact of x L is minor, as shown in Fig. 1b . Remarkably, this characterization remains valid for other types of homogeneous networks, such as lattices and regular rings whereas differences in the positions of the stable root in of G A and their dependence on B correlate perfectly with results obtained previously [14,18,24,[54][55][56][57][58], where steady-states of evolution were analyzed for such structures . Fig. 1a shows that, as we change focus from an individual to a population wide perspective, one witnesses the emergence of an effective game transformation, as evidenced by G A , which brings along important consequences: For instance, the fixation time -the time required for cooperators to invade the entire population -becomes much larger in homogeneous networks when compared to well-mixed populations, as the population spends a large period of time in the vicinity of the x R , mainly when selection is strong . This, in turn, is responsible for computer simulations to spend arbitrary amounts of time in the same configuration, even when, in the absence of mutations , the only absorbing states are associated with monomorphic configurations of the population, that is, with configurations comprising cooperators-only or defectors-only. The In line with the AGoS in a), the population spends most of the time in the vicinity of the stable-like root x R of G A . When j/N<0, Cs become disadvantageous, giving rise to an unstable-like root x L of G A which, however, plays a minor role as shown . Homogeneous random networks were obtained by repeatedly swapping the ends of pairs of randomly chosen links of a regular lattice [54]. doi:10.1371/journal.pone.0032114.g001 stationary distribution , which represents the pervasiveness of each fraction of Cs in time, confirms the scenario portrayed by G A in Fig. 1a, stressing the similarities with the evolutionary dynamics in well-mixed populations under co-existence dilemmas [50,59,60], and putting in evidence the marked difference between individual preferences and the populationwide dynamics. In Fig. 1, our analysis was limited to G A that is, we averaged over the entire time span of all runs. However, the AGoS itself evolves in time -G A as detailed in Methods. Let us explore this time dependence of the AGoS. If, at the beginning of each simulated evolution, Cs and Ds are randomly spread in the network, the occurrence of clusters of the same strategy will not occur in general. Hence, for the PD we have that G A v0 in general. As populations evolve, Cs breed Cs in their neighborhood, promoting the assortment of strategies, with implications both on the fitness of each player and on the shape of G A . In Fig. 2a we plot G A for three particular generations, whereas Fig. 2b portrays the time evolution of the internal roots of G A , on which we superimposed two evolutionary runs starting with 50% of Cs randomly placed in the population. As G A v0, the fraction of cooperators will start decreasing . However, with time, strategy assortment leads to the emergence of a co-existence root, toward which the fraction cooperators converges. The ensuing coexistence between Cs and Ds, entirely described by the shape of G A , steams from the self-organization of Cs and Ds in the network, defining a global dynamics which is impossible to foresee solely from the nature of the local interactions. It is now generally accepted, however, that homogeneous networks provide a simplified picture of real interaction networks [5,44,61,62]. Most social structures share a marked heterogeneity, where a few nodes exhibit a large number of connections, whereas most nodes comprise just a few. The fingerprint of this heterogeneity is provided by the associated degree distributions, which exhibit a broad-scale shape, often resembling a power-law [4,44,61]. In the following we use G A to understand how heterogeneity shifts the internal roots in Fig. 1 to the right, thereby transforming a co-existence scenario into a coordination one. To this end, we compute G A employing scale-free networks of Baraba ´si and Albert [61]. Fig. 3a shows G A for BA networks of N = 10 3 nodes and an average degree k = 4, whereas the circles in Fig. 3b portray the time evolution of the internal roots of G A . Heterogeneous networks lead to a global dynamics dominated by a coordination threshold, originating the appearance of two basins of attraction split by an unstable root x L of G A , analogous to the evolutionary dynamics under 2-person and N-person Stag-hunt dilemmas in unstructured populations [45,51,63,64]. This unstable root represents the critical fraction of Cs above which they are able to assort, thereby invading highly connected nodes, rendering cooperation an advantageous strategy, as Cs acquire a higher probability of being imitated than Ds. On SF networks the requirement to reach the hubs, which ensures the formation of cooperative star-like clusters [39,52], makes invasion harder for isolated Cs. This moves the unstable root located close to j/N<0 in homogeneous networks to higher fractions of Cs. Yet, once this coordination is overcome, Cs benefit from the strong influence of hubs to rapidly spread in the population, eventually leading to fixation. As a result, the stable internal root which characterizes G A in homogeneous networks collapses into the vicinity of j = N on heterogeneous structures, promoting the evolution towards fully cooperative populations. Naturally, the location of the unstable root of G A is an increasing function of B . It is noteworthy that our results remain qualitatively valid for other update rules, such as the discrete analogue of the replicator dynamics on graphs, used in many references, e.g., [16,27,38,65]. In fact, the AGoS is capable of identifying particular features of such dynamics: For instance, the partially deterministic nature of such update rule may lead to evolutionary deadlocks in heterogeneous networks, creating stationary states close to full cooperation [16,27]. In such situation, the AGoS will reflect the occurrence of these stationary configurations by shifting to the left-hand side the stable equilibrium, which may no longer coincide with j = N, remaining, however, in its vicinity. The existence of a coordination barrier for Cs in heterogeneous networks, which must first occupy the hubs before outcompeting Ds, leads to an intricate interplay between the time-dependent decline of x L and the global fraction of Cs. In Fig. 3b we show, with full lines, two evolutions in BA networks : One, which fixates in full cooperation and another, which fixates in full defection. Whenever the fraction of cooperators j/N remains sizeable for long enough, x L will eventually decrease to values satisfying j/N.x L , such that the global coordination barrier is overcome and the population will fixate into full cooperation . Otherwise, j/N may remain always below x L with the population fixating into full defection . Clearly, heterogeneous networks lead to the emergence of a global coordination barrier and associated basins of attraction that evolve in time, in a way which is well described by the time-dependent AGoS. --- Discussion To establish the link between individual and collective behavior constitutes, undeniably, one of the main goals of the analysis of any complex multi-particle or multi-component system [66]. Here we establish such a link showing how it depends on the underlying network topology. Our study shows that behavioral dynamics of individuals facing a cooperation dilemma in social networks can be understood as though the network structure is absent but individuals face a different dilemma: The structural organization of a population of self-regarding individuals helps circumventing the Nash equilibrium of a cooperation dilemma by creating a new dynamical system that can be globally characterized by two internal fixed points, x L and x R . While a single defector will be always advantageous , a single cooperator will be always disadvantageous . As cooperators assort into stable clusters, they may also become advantageous above a certain critical fraction of cooperators and below another critical fraction of cooperators x R , above which defectors will be able to ripe again the benefits of exploiting the many surrounding cooperators. Whereas in homogeneous networks the stable equilibria dictate the overall dynamics -as in co-existence dilemmas -heterogeneous networks create a global dynamics mainly dominated by the unstable equilibria, creating a coordination problem. Strictly speaking, such a dynamical system resulting from individuals interacting via a two-person game, cannot be mapped onto a two-person evolutionary game in a well-mixed population, since the latter can only comply with a maximum of one internal fixed point [43]. On the contrary, such dynamics resembles that from, e.g., N-person dilemmas [67,68] in the presence of coordination thresholds [51,64,69]. It is as if the global dynamics of a 2-person dilemma in structured populations can be properly described as a time-dependent N-person dilemma, in which the coordination or co-existence features emerge from the population structure itself, with different network topologies emphasizing differently this co-existence/coordination dichotomy. It is worth emphasizing that the approach developed here in the context of the two-person PD may be useful _ and immediately applicable _ in understanding the evolutionary dynamics of other game interactions, as well as in understanding other aspects of human sociality that extend beyond cooperation. From human behaviors and ideas, to diseases spreading or to individual preferences, most have been modeled as a person-to-person spreading process embedded in a social network [5,8,62]. In such frameworks, the identification and categorization of the global, population-wide dynamics which emerges from the apparently unrelated nature of the local interactions may enable one to anticipate the emergent outcomes of such complex biological and social systems. --- Methods Evolution is modelled via a stochastic birth-death process [47,70,71]. Each individual x adopts the strategy of a randomly selected neighbour y with probability given by the Fermi function p: 1ze {b  à {1 [14,47], where f x stands for the accumulate payoff of x and b controls the intensity of selection. In structured populations, the difference of the probabilities to increase and decrease the number of Cs ~T z {T { ) becomes context dependent, but can be computed numerically. For each individual i we compute the probability of changing behavior at time t, T i ~1 ki P n ni m~1 1ze {b fm{fi ð Þ Â Ã {1 , where k i stands for the degree of node i and n n i for the number of neighbors of i having a strategy different from that of i. The time-dependent AGoS at a given time t of simulation p, where we have j Cs in the population of size N, is defined as G p ~T z A {T { A , where T + A ~1 N P AllDs AllCs i~1 T i . For a given network type, we run V = 2610 7 simulations starting from all possible initial fractions j/N of cooperators. Each configuration of the population is defined here by the fraction j/N of cooperators. Evolutions run for L = 10 5 time steps. Hence, the overall, time-independent, AGoS is given by the average G A ~1 VL P L t~1 P V p~1 G p over all simulations and time-steps. The time-dependent gradients G A for a particular generation t 0 were computed averaging over the configurations occurr- P V p~1 G p . The stationary distributions pictured in Fig. 1b were obtained computing the fraction of time the population spent in each overall configuration j/N. In some specific limits -in particular, for weak selection or well-mixed populations -our numerical approach will provide results analogous to those obtained with other methods . Homogeneous random networks were obtained by repeatedly swapping the ends of pairs of randomly chosen links of a regular network [54]. BA networks were obtained combining growth and preferential attachment, following the model proposed by Baraba ´si and Albert [4,61]. All networks used have N = 10 3 nodes and an average degree k = 4. --- Supporting Information Text S1 Supporting Text on the analysis of the evolutionary steady states in homogeneous networks. ---

Social networks affect in such a fundamental way the dynamics of the population they support that the global, populationwide behavior that one observes often bears no relation to the individual processes it stems from. Up to now, linking the global networked dynamics to such individual mechanisms has remained elusive. Here we study the evolution of cooperation in networked populations and let individuals interact via a 2-person Prisoner's Dilemma -a characteristic defection dominant social dilemma of cooperation. We show how homogeneous networks transform a Prisoner's Dilemma into a population-wide evolutionary dynamics that promotes the coexistence between cooperators and defectors, while heterogeneous networks promote their coordination. To this end, we define a dynamic variable that allows us to track the self-organization of cooperators when co-evolving with defectors in networked populations. Using the same variable, we show how the global dynamics -and effective dilemma -co-evolves with the motifs of cooperators in the population, the overall emergence of cooperation depending sensitively on this co-evolution.

✙ ✷✿✻ billion modeled associations. Following the notion of Granger causality 17 , links in the network are retained as they predict events at the target better than the target can predict itself. More details on the on problem characteristics and performance are provided in Tab. I and II respectively. For Chicago, we make predictions separately for violent and property crimes, individually within spatial tiles roughly ✶✵✵✵❢t across and time windows of ✶ day approximately a week in advance with AUCs ranging from ✽✵ ✾✾✪ across the city. We summarize our prediction results in Fig. 2, where panels a and b illustrate the geospatial scatter of AUC obtained for different spatial tiles and types of crime, and c shows the distribution of AUCs achieved. Out-of-sample predictive performance remains stable over time; our predictions on successive years The remaining panels in the same figure illustrate how the Fourier coefficients match up, showing that we are able to capture periodicities at the weekly and bi-weekly scales, and beyond. Unlike previous efforts [1][2][3][4][5] , we do not impose pre-defined spatial constraints. In contrast to contiguous diffusion phenomena encountered in physical systems, crime may spread across the complex landscape of a modern city unevenly, with regions hyperlinked by transportation networks, socio-demographic similarity, or historical collocation. Rather than assuming that events far off across the city will have a weaker influence compared with those physically near in space or time, we probe the topological structure emergent in the inferred dependencies to estimate the shape, size and organization of neighborhoods that best predict events at each location. The results illustrated in Fig. 2d ande show that the situation is complex with the locally predictive neighborhoods varying widely in geometry and size, implying that restricting analysis to relatively small local communities within the city is sub-optimal for crime prediction and enforcement analysis. In order to analyze if the effect of reported criminal infractions diffuse outward in space and time, we simply calculate temporal-spatial distances of influences, then average across all neighborhoods in the city, revealing the rapid decay with time delay in diffusion rates shown in Fig. 2f. Interestingly we find the property and violent crimes differ in their rates of influence diffusion ; while the effect of property crimes decays rapidly in days, violent reported events shape the dynamics for weeks to come. Forecasting crime via analyzing historical patterns has been attempted before 18,19 . These approaches use state of the art machine deep learning tools based on recurrent and convolutional neural networks . In the first article 18 , the authors train a NN model to predict next-day events for ✻✵❀✸✹✽ sample points in Chicago. The We break city into small spatial tiles approximately 1.5 times the size of an average city block, and compute models that capture multi-scale dependencies between the sequential event streams recorded at distinct tiles. In this paper, we treat violent and property crimes separately, and show that these categories have intriguing cross-dependencies. Plate d illustrates our modeling approach. For example, to predict property crimes at some spatial tile r, we proceed as follows: Step 1) we infer the probabilistic transducers that estimate event sequence at r by using as input the sequences of recorded infractions at potentially all remote locations , where this predictive influence might transpire over different time delays . Step 2) Combine these weak estimators linearly to minimize zero-one loss. The inferred transducers can be thought of as inferred local activation rules, which are then linearly composed, reversing the approach of linearly combining input and then passing through fixed activation functions in standard neural net architectures. The connected network of nodes with probabilistic transducers on the edges comprises the Granger Network. If we are able to infer a model that is predicts event dynamics at a specific spatial tile using observations from a source tile ✁ days in future, then we say the source tile is within the influencing neighborhood for the target location with a delay of ✁. d illustrates the spatial radius of influence for 0.5, 1, 2 and 3 weeks, for violent and property crimes . Note that the influencing neighborhoods, as defined by our model, are large and approach a radius of ✻ miles. Given the geometry of the City of Chicago, this maps to a substantial percentage of the total area of urban space under consideration, demonstrating that crime manifests demonstrable long-range and almost city-wide influence. e illustrates the extent of a few inferred neighborhoods at time delay of at most ✸ days. f illustrates the average rate of influence diffusion measured by number of predictive models inferred that transduce influence as we consider longer and longer time delays. Note that the rate of influence diffusion falls rapidly for property crimes, dropping to zero in about a week, whereas for violent crimes, the influence continues to diffuse even after three weeks. negative change in arrest rate reveals a strong preference favoring wealthy locations. If neighborhoods are doing better socio-economically, increased crime predicts increased arrests. A strong converse trend is observed in predictions for poor and disadvantaged neighborhoods, suggesting that under stress, wealthier neighborhoods drain resources from their disadvantaged counterparts. b illustrates this more directly via a multi-variable regression, where hardship index is seen to make a strong negative contribution. model is trained on crime statistics, demographic makeup, meteorological data, and Google street view images to track graffiti, achieving an out-of-sample AUC of ✽✸✿✸✪. Our AUC is demonstrably higher , and we predict with significantly less data , and ✼ days into future . Additionally, the use of demographic and graffiti is problematic with the possibility of introducing racial and socio-economic bias, with dubious causal value. In the second article 19 , the authors combine convolutional and recurrent neural networks with weather, socio-economic, transportation, and crime data, to predict the next-day count of crime in Chicago. As spatial tiles, the authors use standard police beats, which break up Chicago into ✷✼✹ regions. Police beats reflect the classical notion of social neighborhoods, and measure approximately 1 sq. mile on average 20 4. Prediction of property and violent crimes across major US cities and dependence of perturbation response on socio-economic status of local neighborhoods. Panels a-f illustrate the AUCs achieved in six major US cities. These cities were chosen on the basis of the availability of detailed event logs in the public domain. All of these cities show comparably high predictive performance. Panel g illustrates the results obtained by regressing crime rate and perturbation response against SES variables . We note that while crime rate typically goes up with increasing poverty, the number of events observed one week after a positive perturbation of 5-10% increase in crime rate is predicted to fall with increasing poverty. We suggest that this decrease is explainable by reallocation of enforcement resources disproportionately, away from disadvantaged neighborhoods in response to increased event rates, which leads to smaller number of reported crimes. model achieves a classification accuracy of ✼✺✿✻✪ for Chicago, which compares against our accuracy of ❃ ✾✵✪ . While this competing model tracks more crime categories, it is limited to next-day predictions with significantly coarser spatial resolution. We also compare the predictive ability of naive autoregressive baseline models , which perform poorly, but provide a yardstick to meaningfully compare our claimed performance estimates. With our precise predictive apparatus in place, we run a series of computational experiments that perturb the rates of violent and property crimes, and log the resulting alterations in future event rates across the city. By inspecting the effect of socio-economic status on the perturbation response, we investigate whether enforcement and policy biases modulate outcomes. The inferred stress response of the city suggests the presence of socio-economic bias . Wealthier neighborhoods away from the inner city respond to elevated crime rates with increased arrests, while arrest rates in disadvantaged neighborhoods drop, but the converse does not occur . Resource constraints on law enforcement, combined with biased prioritization to wealthier neighborhoods, result in reduced enforcement across the remainder of the city. This provides evidence for enforcement bias within U.S. cities that parallels widely discussed notions of suburban bias in wealthy suburbs 21,22 . While self-evident at the scale of countries and regions, the existence of unequal resource allocation in cities, where political power and influence concentrates in selective, wealthy neighborhoods, has been widely suspected [23][24][25][26][27] . Our analysis provides direct support for this contention, which shows up robustly for all years analyzed, going back over one and half decades in Chicago. Figs 6 and7 show that these patterns are stable over time, at least in recent years. Additionally, Fig. 5 show the effect of perturbations across all variables, suggesting that crime reduction from perturbations seems to be most effective in regions with high crime rates, with SES confounders. Beyond Chicago, we analyze criminal event logs available in the public domain for six additional major US cities: Detroit, Philadelphia, Atlanta, Austin, San Francisco and Los Angeles. In all these cities we obtain comparably high performance in predicting violent and property crimes, with average AUC ranging between ✽✻-✾✵✪ . In addition, our observed pattern of perturbation responses in Chicago, which suggests de-allocation of policing resources from disadvantaged neighborhoods to advantaged ones, is replicated in all these cities. While crime rate increases with degrading SES status of local neighborhoods, the number of predicted events a week after a positive 5-10% increase in crime rate is predicted to go down. Thus increasing the crime rate leads to a smaller number of reported crimes, a pattern holding more often in poorer neighborhoods. Our analysis also sheds light on the continuing debate over the choice for neighborhood boundaries in urban crime modeling [28][29][30][31] . In Fig. 2d-f, we demonstrate that despite apparent natural boundaries, influence is often communicated over large distances and decays slowly, especially for violent crimes. More importantly, this study reveals how the "correct" choice of spatial scale should not be a major issue in sophisticated learning algorithms where optimal scales can be inferred automatically. We find that there exists a skeleton set of spatial tiles, which have strong influence on the overall event patterns . These induce a cellular decomposition of the city that identifies functional neighborhoods, where the cell-size adapts automatically to the local event dynamics. --- LIMITATIONS & CONCLUSION To our knowledge, this is the first analysis exploring perturbations of predictive data-driven models to probe the social dynamics of crime and its enforcement. Our ability to probe for the extent of enforcement bias is limited by our dataset; since inference of crime patterns are easily skewed by arrest rates. Disproportionate police response in Black communities can contribute to biases in event logs, which might propagate into inferred models. This has resulted in significant pushback from diverse communities against predictive policing 32 . Our approach is free from manual encoding of features , but biases arising from disproportionate surveillance might still remain. Even with its current limitations, however, this new addition to the toolbox of computational social science enables validation of complex theory from observed event incidence, supplementing the use of measurable proxies and potential biases in questionnaire-based data collection strategies. While classical approaches 33-36 broaden our understanding of the societal forces shaping both urban and regional landscapes, these approaches have neither successfully attempted to forecast individual infraction reports, nor reveal how these predictive patterns manifest systematic enforcement bias. In this study, we show how the ability of Granger Networks to predict such events not only opens new doors for precise intervention, but also advances the diagnosis and explanation of complex social patterns. We acknowledge the danger that powerful predictive tools place in the hands of overzealous states in the name of civilian protection, but here we demonstrate their unprecedented ability to audit enforcement biases and hold states accountable in ways inconceivable in the past. We encourage widespread debate regarding how these technologies are used to augment state action in public life, and call for transparency that allows for continuous evaluation, reconsideration and critique. --- MATERIALS AND METHODS In this study we use historical geolocated incidence data of criminal infractions to model and predict future events in Chicago, Philadelphia, San Francisco, Austin, Los Angeles, Detroit and Atlanta. Each of the cities considered have a specific temporal and spatial resolution, which are optimized to maximize predictive performance . The predictive performance obtained in these cities are enumerated in Table II --- Data Source The sources of the crime incidence data used in this study for the different US cities are enumerated in Table I. Theses logs include spatio-temporal event localization along with the nature, category, and a brief description of the recorded incident. For the City of Chicago, we also have access to the number of arrests made during or as a result of each event. For Chicago, the log is updated daily, keeping current with a lag of ✼ days, and we make predictions for each of the years 2014-2017 for the prediction results shown in Figure 1. The evolving nature of the urban scenescape 37 necessitates that we restrict the modeling window to a few years at a time. The length of this window is decided by trading off loss of performance from shorter data streams to that the evolution of underlying generative processes for longer streams. The training and testing periods of the other cities is tabulated in Table I. In this study, we consider two broad categories of criminal infractions: violent crimes consisting of homicides, assault, battery etc., and property crimes consisting of burglary, theft, motor vehicle theft etc. Drug crimes are excluded from our consideration due to the possibility of ambiguity in the use of violence in such events. For the City of Chicago, the number of individuals arrested during each recorded event is considered a separate variable to be modeled and predicted, which allows us to investigate the possibility of enforcement biases in subsequent perturbation analyses. We also use data on socio-economic variables available at the portal corresponding to Chicago community areas and census tracts, including ✪ of population living in crowded housing, those residing below the poverty line, those unemployed at various age groups, per capita income, and the urban hardship index 38 . Such data is also obtained from the City of Chicago data portal. Additionally, we use data on poverty estimates for the other cities, which are obtained https://www.census.gov. --- Spatial and Temporal Discretization & Event Quantization Event logs are processed to obtain time-series of relevant events, stratified by occurrence locations. This is accomplished by choosing a spatial discretization, and focusing on one individual spatial tile at a time, which allows us to represent the event log as a collection of sequential event streams . Additionally, we discretize time, and consider the sum total of events recorded within each time window. Coarseness of these discretizations reflects a trade-off between computational complexity and event localization in space and time. Spatial and temporal discretizations are not independently chosen; a finer spatial discretization dictates a coarser temporal quantization, and vice versa to prevent long no-event stretches and long periods of contiguous event records, both of which reduce our ability to obtain reliable predictors. For the City of Chicago, we fix the temporal quantization to ✶ day, and choose a spatial quantization such that we have high empirical entropy rates for the time series obtained. This results in spatial tiles measuring ✵✿✵✵✷✼✻°✂ ✵✿✵✵✸✺°in latitude and longitude respectively, which is approximately ✶✵✵✵ ✵ across, roughly corresponding to an area of under ✷✂✷ city blocks. Thus, any two points within our spatial tile are at worst in neighboring city blocks. We dropped from our analysis the tiles that have too low a crime rate to reduce computational complexity, resulting in an ◆ ❂ ✷✷✵✺ of spatial tiles in the city of Chicago. The temporal and spatial resolution is adjusted in a similar manner for the other cities . Thus, we end up with three different integer-valued time series at each spatial tile: 1) violent crime , 2) property crime and 3) number of arrests in the City of Chicago. For other cities, we have only the first two categories, since information on arrests was not available. We ignore the magnitude of the observations, and treat them as Boolean variables. Thus, our models simply predict the presence or absence of a particular event type in a discrete spatial tile within a neighboring city block and observation window, ✐✿❡✿, within the temporal resolution chosen, which is ✶ day except for Atlanta, where is it is chosen to be ✷ days . --- Inferring Generators of Spatio-temporal Cross-dependence Let ▲ ❂ ❢❵ ✶ ❀ ✁ ✁ ✁ ❀ ❵ ◆ ❣ be the set of spatial tiles, and ❊ ❂ ❢✉❀ ✈❀ ✇❣ be the set of event categories as described in the last section. At location ❵ ✷ ▲ for variable ❡ ✷ ❊, at time t, we have ✭❵❀ ❡✮ t ✷ ❢✵❀ ✶❣, with ✶ indicating the presence of at least one event. The set of all such combined variables is denoted as ❙, i.e., ❙ ❂ ▲ ✂ ❊. Let ❚ ❂ ❢✵❀ ✁ ✁ ✁ ❀ ▼ ✶❣ denote the training period consisting of ▼ time steps. Because for any time t, ✭❵❀ ❡✮ t is a random variable, our goal here is to learn its dependency relationships with its own past, and with other variables in ❙ to accurately estimate its future distribution for t ❃ ❚ . To infer the structure of our predictive model, we learn a finite state probabilistic transducer 16 for each possible source-target pair s❀ r ✷ ❙. Given a sequence of events at the source, these inferred transducers estimate the distribution of events at target r for some future point in time. Ability to estimate such a non-trivial distribution indicates the presence of causal influence. Here we assume that causal influence from the source to the target manifests as the source being able to predict events occurring at the target, better than the target can do by itself. This interpretation follows from Granger's eponymous approach to statistical causality 39 . Importantly, we do not assume that the underlying processes are iid, or that the model has any particular linear structure. Additionally, such influence is not restricted to be instantaneous. The source events might impact the target with a time delay, ✐✿❡✿, a specific model between the source and target might predict events delayed by an a priori determined number of steps ✁ ♠❛① ≧ ✁ ≧ ✵ specific to the model. Here we model the influence structure for each integer-valued delay separately. Thus, for source s and target t, we can have ✁ ♠❛① ✰ ✶ transducers each modeling the influence for a specific delay in ❢✵❀ ✁ ♠❛① ❣. The maximum number of steps in time delay ✁ ♠❛① is chosen a priori, based on the problem at hand. While these influences or dependencies may differ for different delays, they need not be symmetric between source and target pairs. The complete set, comprising at most •❙• ✷ ✭✁ ♠❛① ✰ ✶✮ models, represents a predictive framework for asymmetric multi-scale spatio-temporal phenomena. Note that the number of possible models increase quickly. For example, for the City of Chicago, for ✁ ♠❛① ❂ ✻✵ with ✷✷✵✺ spatial tiles and three event categories, the number of inferred models is bounded above by ✙ ✷✿✻ billion. ur approach consists of inferring XPFSAs in two key steps : First, we infer XPFSA models for all source-target pairs and all delays up to ✁ ♠❛① . In the second step, we learn a linear combination of these transducers to maximize predictive performance. Denoting the observed event sequence in time interval ✭✶❀ t( at source s as s ✶ t , the XPFSA H s r❀❦ estimates the distribution of events for target r at time step t ✰ ❦. This is accomplished by learning an equivalence relation on the historical event sequences observed at source s, such that equivalent histories induce an approximately identical future event distribution at target r, ❦ steps in the future. Thus, for example, the XPFSA shown in Fig. 1d has four states, indicating that there are ✹ such equivalence classes of observations that induce the distinct output probabilities shown from each state. Often this estimate is not very precise due to the possibility for multi-scale and multisource influence, e.g., when target r is influenced by multiple sources with different time delays. In the second step, we employ a standard gradient boosting regressor for each target, to optimize the linear combination of inferred transducers and learn the scalar weights ✦ s r❀❦ for source s, target r and delay ❦. Detailed pseudocode of the inference algorithms are provided in the SI-section 1. To compare with a standard neural net architecture, these probabilistic transducers may be viewed as local non-linear activation functions. With neural networks we repeatedly compute affine combination of inputs and apply fixed non-linear activation to the combined input and finally optimize affine combination weights via backpropagation, but here we first learn the local non-linear activations, and then optimize the linear or affine combination of weak estimators. Optimizing the weights is a significantly simpler, local operation and may be done with any standard regressor. In contrast to recurrent neural nets , the role of hidden layer neurons is partially accounted for by states of the XPFSA, which are a priori undetermined both with respect to their multiplicity and their transition connectivity structure. --- Computational & Model Complexity We assume the maximum time delay in the influence propagation to be ✻✵ days for all cities, which for the City of Chicago results in at most ✷❀✻✻✾❀✷✺✶❀✼✷✺ inferred models, of which ✻✶❀✻✺✵❀✵✵✵ are useful with ✌ ≧ ✵✿✵✶. Model inference in this case consumed approximately ✷✵✵❑ core-hours on ✷✽ core Intel Broadwell processors, when carried out with incidence data over the period Jan 1, 2014 to December 31, 2016. Computational cost for other time-periods and other cities are comparable and roughly scale with the square of the number of spatial tiles, and linearly with the length of time-quantized data-streams considered as input to the inference algorithm. --- Crime Prediction Metrics For each spatial location, the inferred Granger Net maps event histories to a raw risk score as a function of time. The higher this value, the higher the probability of an event of target type occurring at that location, within the specified time window. To make crisp predictions, however, we must choose a decision threshold for this raw score. Conceptually identical to the notion of Type 1 and Type 2 errors in classical statistical analyses, the choice of a threshold trades off false positives for false negatives . Choosing a small threshold results in predicting a larger fraction of future events correctly, i.e., have a high true positive rate , while simultaneously suffering from a higher false positive rate , and vice versa. The receiver operating characteristic curve is the plot of the FPR vs the TPR, as we vary this decision threshold. If our predictor is good, we will consistently achieve high TPR with small FPR resulting in a large area under the ROC curve denoted as the AUC. Importantly, AUC measures intrinsic performance, independent of the threshold choice. Thus, the AUC is immune to class imbalance . An AUC of ✺✵✪ indicates that the predictor does no better than random, and an AUC of ✶✵✵✪ implies that we can achieve perfect prediction of future events, with zero false positives. We use a flexible approach in evaluating AUC; a positive prediction is treated as correct if there is at least one event recorded in ✝✶ time steps in the target spatial tile. --- Predictability Analysis In the City of Chicago, we can predict events approximately a week in advance at the spatial resolution of ✝✶ city blocks with a temporal resolution of ✝✶ day, with a false positive rate of less than ✷✵✪ and a median true positive rate of ✼✽✪. The predictive performance in the other cities is enumerated in Table II. While not directly modeled in the frequency domain, we found that the event forecasts produce very similar signatures in the frequency domain , when compared over the first ✶✺✵ days of each out-of-sample period . --- Spatial Neighborhoods The degree of causal influence exerted by one variable on another is quantified by the coefficient of causal dependence . Identifying the source-target pairs for which the coefficient of causality is high , we note that there exists a sparse set of spatial tiles which exert nearly all of the influence in the entire set of observed variables. Thus, observing these variables alone would enable us to make good event forecasts. These tiles span the expanse of the city, and a Voronoi decomposition based on the centers of these tiles in shown in Fig 10b . Such a decomposition demonstrates an algorithmic approach to choosing optimal neighborhoods for urban analysis. --- Perturbation Analysis We experimented with positive and negative perturbations to both violent and property crime rates ranging from ✶ to ✶✵✪ of observed rates. Response to perturbed crime rates was measured as the relative change from nominal baseline in estimated time-average for the predicted event frequencies ✶ week in the future, corresponding to violent and property crimes and number of arrests. Results from our perturbation experiments shed light both on the stability characteristics of crime in Chicago, and further allowed us to look for evidence of biased police enforcement responses under stress. Under stress, well-off neighborhoods tend to drain resources disproportionately from disadvantaged locales . For economically well-off neighborhoods in the bottom ✷✺✪ of the hardship index are much more likely to see a near -proportional increase in law enforcement response, measured by the number or predicted arrests on a ✶✵✪ increase in crime rates , while the rest of the city see a drop in predicted response of about twice the magnitude . Increased crimes causes enforcement resources to be drained from disadvantaged neighborhoods to support their better socioeconomic counterparts. We performed multi-variable linear regression analysis to evaluate the question in another way. Here we regressed violent and property crime rates, independently, on the variables listed in , including a slope intercept variable in each model. In both models, the hardship index exhibits a strong, negative influence on changes in arrest rate from perturbations that increase violent and the property crime rates, which contradicts what might be expected in the absence of bias. We also carried out similar perturbation analyses for the other cities, and observed that with increasing poverty we have expected increase of observed crime rates, but an unexpected decrease in violent and property crimes after a 5-10% simulated uptick in either category of crimes . ✸✸✿✻✺ ✍ N, ✸✸✿✽✻ ✍ N, ✽✹✿✺✹ ✍ W, ✽✹✿✸✶ ✍ W ✸✵✿✶✹ ✍ N, ✸✵✿✹✽ ✍ N, ✾✼✿✽✾ ✍ W, ✾✼✿✻✸ ✍ W ✹✷✿✸✵ ✍ N, ✹✷✿✹✺ ✍ N, ✽✸✿✷✽ ✍ W, ✽✷✿✾✶ ✍ W ✸✸✿✼✶ ✍ N, ✸✹✿✸✸ ✍ N, ✶✶✽✿✻✺ ✍ W, ✶✶✽✿✶✻ ✍ W ✸✾✿✽✽ ✍ N, ✹✵✿✶✷ ✍ N, ✼✺✿✷✼ ✍ W, ✼✹✿✾✻ ✍ W ✸✼✿✼✶ ✍ N, ✸✼✿✽✶ ✍ N, ✶✷✷✿✺✶ ✍ W, ✶✷✷✿✸✻ ✍ W ✹✶✿✻✹ ✍ N, ✹✷✿✵✻ ✍ N, ✽✼✿✽✽ ✍ W, ✽✼✿✺✷ ✍ W spatial resolution ✾✽✸ ✵ ✂ ✾✽✸ ✵ ✾✽✸ ✵ ✂ ✾✽✸ ✵ ✾✽✸ ✵ ✂ ✾✽✸ ✵ ✾✽✸ ✵ ✂ ✾✽✸ ✵ ✾✽✸ ✵ ✂ ✾✽✸ ✵ ✾✽✸ ✵ ✂ ✾✽✸ ✵ ✾✺✶ ✵ ✂ ✶✵✵✻ ✵ --- Naive Baselines: Autoregressive Integrated Moving Average Models To explore the predictive ability of naive baseline models on our datasets, we consider four ARIMA ? configurations with lag orders ♣ ❂ ✺ and ✶✵, numbers of differencing ❞ ❂ ✶ and ✷, and the window of moving average q ❂ ✵. Let ② t be the series we want to model and ② ✵ t be ② t differenced by ❞ times, the ARIMA✭♣❀ ❞❀ q✮ models series ② ✵ t by where ✣ ✶ ❀ ✿ ✿ ✿ ❀ ✣ ♣ and ✒ ✶ ❀ ✿ ✿ ✿ ❀ ✒ q are the coefficients to be fitted. In Eq. , ② ✵ t ❦ s are the historical values of ② ✵ t whose inclusion models the influence of past values on the current value , and ✧ t ❦ s are the white noise terms whose inclusion models the dependence of current value against current and previous white noise error terms or random shocks . Specifically, we use the following four models for the earthquake and the crime datasets ② ✵ t ❂ ❝ ✰ ✣ ✶ ② ✵ t ✶ ✰ ✁ ✁ ✁ ✰ ✣ ♣ ② ✵ t ♣ ✰ ✒ ✶ ✧ t ✶ ✰ ✁ ✁ ✁ ✰ ✒ q ✧ t q ✰ ✧ t ② ✭✶✮ t ❂ ❝ ✰ ✣ ✶ ② ✵ t ✶ ✰ ✁ ✁ ✁ ✰ ✣ ✺ ② ✵ t ✺ ② ✭✶✮ t ❂ ❝ ✰ ✣ ✶ ② ✵ t ✶ ✰ ✁ ✁ ✁ ✰ ✣ ✺ ② ✵ t ✶✵ ② ✭✷✮ t ❂ ❝ ✰ ✣ ✶ ② ✵ t ✶ ✰ ✁ ✁ ✁ ✰ ✣ ✺ ② ✵ t ✺ ② ✭✷✮ t ❂ ❝ ✰ ✣ ✶ ② ✵ t ✶ ✰ ✁ ✁ ✁ ✰ ✣ ✺ ② ✵ t ✶✵ where ② ✭❞✮ t is ② t different by ❞ times . For simple benckmarking, we apply the ARIMA model to each individual time series, which means the predictive model is trained without exogenous variables. For the implementation, we use the Python statsmodels package ? , and the result is shown in Tab. III. The inadequate performance of ARIMA may be due to 1) the use of a single data stream limits the ability of ARIMA to capture the interplay between co-evoluting processes, and 2) a pre-determined lag order fails to capture the possibly varying temporal memory of individual processes. --- L. Distribution of Increased --- Violent Crimes from Increased Arrests Fig. 5. Perturbation Effects Across Variables. We see that the decrease of violent crimes from increase of property crimes are localized in disadvantaged neighborhoods . Similarly, the decrease of property crimes from increase of violent crimes is also localized to disadvantaged neighborhoods , as well as the decreased violent crimes from increased arrests . We see a weaker localization for the corresponding increases in crime rates under similar perturbations. Looking at other pairs of variables under perturbation , we generally do not see a very prominent correspondence with the distribution of socio-economic indicators. It seems crimes are easier to dampen in lcales with high existing crime rates, which is desirable result. But such conclusions are currently confounded by SES variables, and futher work is needed to investigate these effects more thoroughly. [4] Mohler, G. O., Short, M. B., Brantingham, P. J., Schoenberg, F. P. & Tita, G. E. Self-exciting point process modeling of crime. Journal of the American Statistical Association 106, 100-108 . g én érales du calcul des probabiliti és . PLoS ONE 12, e0176244 . Fig. 8. Out of Sample Predictive Performance over the Years. We show that the predictive performance is very stable, and variation in mean AUC is limited to the third place of decimal, at least when analyzing the last few years . [24] Duany, A., Plater-Zyberk, E. & Speck, J. Suburban nation: The rise of sprawl and the decline of the American dream . [25] Logan, J. R. d andf show that the Fourier coefficients match up pretty well as well. More importantly, while our models do not explicitly incorporate any periodic elements that are being tuned, we still manage to capture the weekly, biweekly and longer periodic regularities. --- THEORY OF PROBABILISTIC AUTOMATA Granger Net is assembled from local models which are, in general, crossed probabilistic automata . The construction of a Granger Net consists of two steps: 1) local model generation and network pruning and 2) local model aggregation for comprehensive prediction. Event prediction is accomplished by aggregating these local activations via a local regressor. No global optimization of these aggregation function is acrried out. The model generation step of Granger Net is accomplished by the algorithms GenESeSS and xGenESeSS . xGenESeSS produces XPFSA models that captures how the history of a source process influences the future of a target process. The Granger Net construction is described in Algorithm 1, and takes as input a set ❢① s ✿ s ✷ ❙❣ of length-♥ time series, hyperparameters ✧ and ♥ ✵ ❁ ♥ for local model inference, ✁ max for maximum time delay, and ✌ ✵ for thresholding admissible models. For each target sequence ① r , Granger Net outputs a set of admissible models ▼ r with a scalar weight for each model in ▼ r via model inference and pruning and training of the aggregation weights . Step --- 1: Model inference and pruning The Granger Net framework models the influence from a source time series ① s on a target time series ① r at a particular time delay ✁ by an XPFSA ❍ s r❀✁ . Thus, we infer •❙•✁ max XPFSA models for each ① r which yields •❙• ✷ ✁ max models in total. Since the number of XPFSA models increases quadratically with the number of time series and strength of the links may vary, pruning low-performing models early is important for parsimony. Granger Net rejects models by thresholding on the coefficient of causal dependence ✌ s r❀✁ of model ❍ s r❀✁ , which measures the strength of dependence of the output sequence on the input one. More specifically, we have ✌ s r❀✁ ❂ ✶ uncertainty of the next output in ① r with observation of ① s uncertainty of the next output in ① r ✌ can be evaluated from the synchronous composition of the PFSA that models the input process and the XPFSA that models the causal influence. Granger Net retains the model ❍ s r❀✁ if and only if ✌ s r❀✁ is greater than a pre-specified threshold ✌ ✵ . At the conclusion of Step 1, Granger Net returns an admissible set of models ▼ r ❂ ✟ ❍ s r❀✁ ✿ ✌ s r❀✁ ❃ ✌ ✵ ✠ for each r ✷ ❙. --- Step 2: Train linear weights In this step, we integrate the local models in ① r 's admissible set for forecasting events in ① r . To do this, Granger Net trains a linear coefficient ✦ s r❀✁ for each ❍ s r❀✁ ✷ ▼ r so that the final prediction for ① r at time step ❤ is equal to ❳ ❍ s r❀✁ ✷▼r ✦ s t❀✁ ❍ s r❀✁ ✏ ✭① s ✮ ❤ ✁ ✑ ❀ where ✭① s ✮ ❤ ✁ is the truncation of ① s at ❤ ✁. To compute the coefficients, we solve a regression problem Reg✭❳❀ y✮ for each r ✷ ❙ with the predictor variables being predictions x t ❬s❀✁, and the outcome variable being ① r ❬♥ ✵ ✰ t. Hence, the ❳ matrix is the ✭♥ ♥ ✵ ✮ ✂ •▼ r • matrix with the entry indexed by t❀ ✭s❀✁✮ given by x t ❬s❀✁( and y, the ✭♥ ♥ ✵ ✮-dimensional vector with the entry indexed by t given by ① r ❬♥ ✵ ✰ t(. We can solve for the linear weights with any standard regressor. --- Inference Algorithms On line 6 and 7 of Algorithm 1, Granger Net calls subroutine xGenESeSS, which infers XPFSA as models of crossdependencies between processes. Here, we establish the correctness of GenESeSS. The inference algorithm for PFSA is called GenESeSS for Generator Extraction Using Self-similar Semantics. The PFSA model is based on the concept of the causal state. A dynamical system reaches the same causal state via distinct paths if the futures are statistically indistinguishable. More precisely, each process over an alphabet ✝ of size ♠ gives rise naturally to an ♠-ary tree with the nodes at level ❞ being sequences of length ❞, and the edge from the node ① to ①✛, ✛ ✷ ✝, labeled by P r✭✛•①✮ -the probability of observing ✛ as the next output after ①. By the definition of causal state, if two subtrees are identical with respect to edge labels, then their roots are sequences that lead the system to the same causal state. Identifying all the roots of identical subtrees induces a finite automaton structure whose unique strongly connected component is the generating model of the process. Definition 1 ). A PFSA • is a quadruple ✭◗❀✝❀ ✍❀ ❡ ✙✮, where ◗ is a finite set, ✝ is a finite alphabet, ✍ ✿ ◗ ✂ ✝ ✦ ✝ is called the transition map, and ❡ ✙ ✿ ◗ ✦ P ✝ , where P ✝ is the space of probability distributions over ✝, is called the transition probability. Step 2 of Algorithm 2 is an implementation this subtree "stitching" approach under finiteness of input data. Note that the criterion for "stitching" two subtrees with roots ① and ① ✵ is that their edge labels are identical for all depths, which translates to ♣✭②•①✮ ❂ ♣✭②•① ✵ ✮ for sequence ② of all lengths. The criterion is not verifiable with finite data, and hence GenESeSS identifies two subtrees if they agree on depth one. Defining symbolic derivative ✣ ① to be the vector with the entry indexed by ✛ given by ♣✭✛•①✮, GenESeSS identifies ① and ① ✵ if ✣ ① ❂ ✣ ① ✵ . This approach works well under the assumption that the target PFSA is in general position, meaning that different causal states have distinct symbolic derivatives. In practice, GenESeSS uses empirical symbolic derivative defined below to approximate ✣ ① . Let ① be an input sequence of finite length, the empirical symbolic derivative ) ✣ ① ② of a sub-sequence ② of ① is a probability vector with the entry indexed by ✛ given by ) ✣ ① ② ✭✛✮ ❂ number of ②✛ in ① number of ② in ① GenESeSS identifies two sequences if their empirical symbolic derivatives are within an ✧-neighborhood of each other for certain ✧ ❃ ✵. For simplicity, we first illustrate how GenESeSS solves the transition structure of the target PFSA from a sample path ① generated from a process of Markov order ❦. Assuming the ① ✵ produced by Step 1 is ✕, the empty sequence, GenESeSS starts by calculating ) ✣ ① ✕ , ✐✿❡✿, the empirical distribution on ✝, and records ✕ as the identifier of the first state. Then, GenESeSS appends ✕ with each ✛ ✷ ✝, and calculates ) ✣ ① ✛ . By the general position assumption and assuming ① is long enough, with high probability, no ) ✣ ① ③ can be within an ✧-neighborhood of each other given long enough input ①. In this case, GenESeSS identifies the state represented by ② with that of ③. In fact, GenESeSS will identify all states represented by sequences of length ❦ ✰ ✶ to some previously-stored states. And since no new states can be found, GenESeSS exits the loop on line 8 after iteration ❦ ✰✶. Taking the strongly connected component on line 19, GenESeSS gets the correct transition structure. ✣ ① ✛ is within an ✧-neighborhood of ) ✣ ① ✛ ✵ for ✛ ✛ ✵ , However, not all processes generated by PFSA have finite Markov order. For such cases, Step 2 of GenESeSS will never exit in theory, since there exists no ♥ ✷ N such that every causal state is visited for sequences with length ✔ ♥. And if we implement an artificial exit criterion, the model inferred might be unnecessarily large, and have hard-to-model approximations. We address this issue via the notion of synchronization -the ability to identify that we are localized or synchronized to a particular state despite being uncertain of the initial state. In Step 1 of Algorithm 2 , GenESeSS finds an almost synchronizing sequence, which allows GenESeSS to distill a structure that is similar to that of the finite Markov order cases, and thus carry out the subtree "stitching" procedure described before. A sequence ① is synchronizing if all sequences that end with the suffix ① terminates on the same causal state. A process is synchronizable if it has a synchronizing sequence, and a PFSA is synchronizable if the process it generates is synchronizable. The structure of the "graph" of a perfectly synchronizable PFSA is that of a co-final automata 1 . A sequence ① is ✧-synchronizing 2 to the state q if the distribution ⑥ ① on the state set ◗ induced by ① satisfies ❦⑥ ① e q ❦ ✶ ❁ ✧, where e q is the base vector with ✶ on the entry indexed by q and ✵ elsewhere. The importance of ✧synchronizing sequence is twofold: 1) since ✣ ❚ ① ❂ ⑥ ❚ ① ❡ ✆, where ❡ ✆ is the •◗•✂•✝• matrix with the row indexed by q given by ❡ ✙✭q✮, a ⑥ ① close to e q give rise to a ✣ ① close to ❡ ✙✭q✮. And 2) although sequences prefixed by an ✧-synchronizing sequence to a state q may not remain ✧-synchronizing to state q, they are close to q on average. To find an almost synchronizing sequence algorithmically 2 , GenESeSS first calculates the convex hull of symbolic derivatives of subsequences of ① up to length ▲ , and then selects a sequence ① ✵ whose symbolic derivative is a vertex of the convex hull . Since the convex hull of ✟ ✣ ① ✿ ① ✷ ✝ ▲ ✠ is a linear projection of the convex hull ✟ ⑥ • ✭①✮ ✿ ① ✷ ✝ ▲ ✠ via ❡ ✆, we can expect sequence ① with ✣ ① being a vertex of the convex hull of ✟ ✣ ① ✿ ① ✷ ✝ ▲ ✠ to be a good candidate for an almost synchronizing sequence. The corresponding inference algorithm for XPFSA is called xGenESeSS, which takes as input two sequences ① in , ① out , and a hyperparameter ✧, and outputs an XPFSA in a manner very similar to the inference algorithm of PFSA. While a PFSA models how the past of a time series influences its own future, a XPFSA models how the past of an input time series influences the future of an output time series. Hence, while in the SSC algorithm of PFSA, we identify sequences if they lead to futures that are statistically indistinguishable, in the SSC algorithm of XPFSA, we identify sequences if they lead to the same future distribution of the output. Definition 2 ). A crossed probabilistic finite-state automaton is specified by a quintuple ✭✝ in ❀ |❀ ✑❀ ✝ out ❀ ✤✮, where ✝ in is a finite input alphabet, | is a finite state set, ✑ is a partial function from | ✂ ✝ in to | called transition map, ✝ out is a finite output alphabet, and ✤ is a function from | to P ✝out called output probability map, where P ✝out is the space of probability distributions over ✝ out . In particular, ✤✭r❀ ✜ ✮ is the probability of generating ✜ ✷ ✝ out from a state r ✷ |. Note that a XPFSA has no transition probabilities defined between states as a PFSA does. The XPFSA in the example has a binary input alphabet and an output alphabet of size ✸. The bar charts next to the ✹ states of the XPFSA indicate the output probability distributions. To generate a sample path, an XPFSA requires an input sequence over its input alphabet. Similar to the PFSA construction approach, here we compute the cross symbolic derivative, which is the ordered tuple P r✭✜ •①✮, with ✜ ✷ ✝ out and a sequence ① over ✝ in . We compute the empirical approximation of the cross symbolic derivative from sequences ① in and ① out as: ) ✣ ①in ❀①out ② ✭✜✮ ❂ number of ✜ in ① out after ② transpires in ① in number of sub-sequence ② in ① in Thus, xGenESeSS is almost identical to GenESeSS except that, in Step 1, xGenESeSS finds an almost synchronizing sequence based on cross symbolic derivatives, and in Step 2, identifies the transition structure based on the similarity between cross symbolic derivatives. Arguments for establishing the effectiveness of GenESeSS carry over to xGenESeSS with empirical symbolic derivative replaced by empirical cross symbolic derivative. --- Figures Figure 1 Crime Data & Modeling Approach. a and b show the recorded infractions within the 2 week period between April 1 and 15 in 2017. Plate c illustrates our modeling approach: We break city into small spatial tiles approximately 1.5 times the size of an average city block, and compute models that capture multi-scale dependencies between the sequential event streams recorded at distinct tiles. In this paper, we treat violent and property crimes separately, and show that these categories have intriguing crossdependencies. Plate d illustrates our modeling approach. For example, to predict property crimes at some spatial tile r, we proceed as follows: Step 1) we infer the probabilistic transducers that estimate event sequence at r by using as input the sequences of recorded infractions at potentially all remote locations , where this predictive in uence might transpire over different time delays . Step 2) Combine these weak estimators linearly to minimize zero-one loss. The inferred transducers can be thought of as inferred local activation rules, which are then linearly composed, reversing the approach of linearly combining input and then passing through xed activation functions in standard neural net architectures. The connected network of nodes with probabilistic transducers on the edges comprises the Granger Network. Space. If we are able to infer a model that is predicts event dynamics at a speci c spatial tile using observations from a source tile + days in future, then we say the source tile is within the in uencing neighborhood for the target location with a delay of D. d illustrates the spatial radius of in uence for 0.5, 1, 2 and 3 weeks, for violent and property crimes . Note that the in uencing neighborhoods, as de ned by our model, are large and approach a radius of 6 miles. Given the geometry of the City of Chicago, this maps to a substantial percentage of the total area of urban space under consideration, demonstrating that crime manifests demonstrable long-range and almost city-wide in uence. e illustrates the extent of a few inferred neighborhoods at time delay of at most 3 days. f illustrates the average rate of in uence diffusion measured by number of predictive models inferred that transduce in uence as we consider longer and longer time delays. Note that the rate of in uence diffusion falls rapidly for property crimes, dropping to zero in about a week, whereas for violent crimes, the in uence continues to diffuse even after three weeks. Figure 3 Estimating Bias. a illustrates the distribution of hardship index . c, d, e, and f suggest biased response to perturbations in crime rates. With a 10% increase in violent or property crime rates, we see an approximately a 30% decrease in arrests when averaged over the city. The spatial distribution of locations that experience a positive vs. negative change in arrest rate reveals a strong preference favoring wealthy locations. If neighborhoods are doing better socio-economically, increased crime predicts increased arrests. A strong converse trend is observed in predictions for poor and disadvantaged neighborhoods, suggesting that under stress, wealthier neighborhoods drain resources from their disadvantaged counterparts. b illustrates this more directly via a multi-variable regression, where hardship index is seen to make a strong negative contribution. Figure 4 Prediction of property and violent crimes across major US cities and dependence of perturbation response on socio-economic status of local neighborhoods. Panels a-f illustrate the AUCs achieved in six major US cities. These cities were chosen on the basis of the availability of detailed event logs in the public domain. All of these cities show comparably high predictive performance. Panel g illustrates the results obtained by regressing crime rate and perturbation response against SES variables . We note that while crime rate typically goes up with increasing poverty, the number of events observed one week after a positive perturbation of 5-10% increase in crime rate is predicted to fall with increasing poverty. We suggest that this decrease is explainable by reallocation of enforcement resources disproportionately, away from disadvantaged neighborhoods in response to increased event rates, which leads to smaller number of reported crimes. Figure 5 Perturbation Effects Across Variables. We see that the decrease of violent crimes from increase of property crimes are localized in disadvantaged neighborhoods . Similarly, the decrease of property crimes from increase of violent crimes is also localized to disadvantaged neighborhoods , as well as the decreased violent crimes from increased arrests . We see a weaker localization for the corresponding increases in crime rates under similar perturbations. Looking at other pairs of variables under perturbation , we generally do not see a very prominent correspondence with the distribution of socio-economic indicators. It seems crimes are easier to dampen in lcales with high existing crime rates, which is desirable result. But such conclusions are currently confounded by SES variables, and futher work is needed to investigate these effects more thoroughly. Out of Sample Predictive Performance over the Years. We show that the predictive performance is very stable, and variation in mean AUC is limited to the third place of decimal, at least when analyzing the last few years . as well. More importantly, while our models do not explicitly incorporate any periodic elements that are being tuned, we still manage to capture the weekly, biweekly and longer periodic regularities. Figure 10 Automatic Neighborhood Decomposition Using Event Predictability Computing a biclustering on the source-vs-target in uence matrix isolates a set of spatial tiles that are, on average, good predictors for all other tiles. Using this set, we use a Voronoi decomposition of the city , which realizes an automatic spatial decomposition of the urban space, driven by event predictability. --- Let ◗new ❂ ❀ be the set of new states;

Policing efforts to thwart urban crime often rely on detailed reports of criminal infractions. However, crime rates do not document the distribution of crime in isolation, but rather its complex relationship with policing and society. Several results attempting to predict future crime now exist, with varying degrees of predictive efficacy. However, the very idea of predictive policing has stirred controversy, with the algorithms being largely black boxes producing little to no insight into the social system of crime, and its rules of organization. The issue of how enforcement interacts with, modulates, and reinforces crime has been rarely addressed in the context of precise event predictions. In this study, we demonstrate that while predictive tools have often been designed to enhance state power through surveillance, they also enable the tracing of systemic biases in urban enforcement-surveillance of the state. We introduce a novel stochastic inference algorithm as a new forecasting approach that learns spatio-temporal dependencies from individual event reports with demonstrated performance far surpassing past results (e.g., average AUC of ✙ ✾✵✪ in the City of Chicago for property and violent crimes predicted a week in advance within spatial tiles ✙ ✶✵✵✵ ft across). These precise predictions enable equally precise evaluation of inequities in law enforcement, discovering that response to increased crime rates is biased by the socio-economic status of neighborhoods, draining policy resources to wealthy areas with disproportionately negative impacts for the inner city, as demonstrated in Chicago and six other major U.S. metropolitan areas. While the emergence of powerful predictive tools raise concerns regarding the unprecedented power they place in the hands of over-zealous states in the name of civilian protection, our approach demonstrates how sophisticated algorithms enable us to audit enforcement biases, and hold states accountable in ways previously inconceivable. T HE emergence of large-scale data and ubiquitous data-driven modeling has sparked widespread government interest in the possibility of predictive policing [1][2][3][4][5] : predicting crime before it happens to enable anticipatory enforcement. Such efforts, however, do not document the distribution of crime in isolation, but rather its complex relationship with policing and society. In this study, we reconceptualize the process of crime prediction, build novel methods to improve it, and use it to diagnose both the distribution of reported crime and biases in its enforcement. The history of statistics has co-evolved with the history of criminal prediction, but also with the history of enforcement critique. Sim éon Poisson published the Poisson distribution and his theory of probability in an analysis of the number of wrongful convictions in a given country 6 . Andrey Markov introduced Markov processes to show that dependencies between outcomes could still obey the central limit theorem to counter Pavel Nekrasov's argument that because Russian crime reports obeyed the law of large numbers, "decisions made by criminals to commit crimes must all be independent acts of free will" 7 . In this study, we conceptualize the prediction of criminal reports as that of modeling and predicting a system of spatio-temporal point processes unfolding in social context. We report a fundamentally new approach to predict urban crime at the level of individual events, with predictive accuracy far greater than has been achieved in past. Rather than simply increasing the power of states by predicting the when and where of anticipated crime, our new tools allow us to audit them for enforcement biases, and garner deep insight into the nature of the dynamical processes through which policing and crime co-evolve in urban spaces. Classical investigations into the mechanics of crime [8][9][10] have recently given way to event-level crime predictions that have enticed police forces to deploy them preemptively and stage interventions targeted at lowering crime rates. These efforts have generated multi-variate models of time-invariant hotspots [11][12][13] , and estimate both long and short term dynamic risks 1-3 . One of the earliest approaches to predictive policing is based on the use of epidemic-type aftershock sequences (ETAS) 4,5 , originally developed to model seismic phenomena. While these approaches have suggested the possibility of predictive policing, many achieve only limited out-of-sample performance 4,5 . More recently, deep learning architectures have yielded better results 14 . Machine learning systems, however, are often black boxes producing little insight regarding the social system of crime and its rules of organization. Moreover, the issue of how enforcement interacts with, modulates and reinforces crime has been rarely addressed in the context of precise event predictions.Here we show that urban crime may be predicted reliably one or more weeks in advance, enabling modelbased simulations that reveal both the pattern of reported infractions and the pattern of corresponding police enforcement. We learn from recorded historical event logs, and validate on events in the following year beyond those in the training sample. Using incidence data from the City of Chicago, our novel spatio-temporal network inference algorithm infers patterns of past event occurrences, and constructs a communicating network (the Granger Network) of local estimators to predict future infractions. In this study, we consider two broad categories of reported criminal infractions: violent crimes consisting of homicides, assault, and battery, and property crimes consisting of burglary, theft and motor-vehicle thefts. The number of individuals arrested during each recorded event is separately modeled and allows us to investigate the possibility and pattern of enforcement bias. We begin by processing event logs to obtain time-series of relevant events, stratified by location and discretized by time, yielding sequential event streams for 1) violent crime (✈), 2) property crime (✉) and 3) number of arrests (✇), as shown in Fig. 1, panels a, b and c. To infer the structure of the Granger Net, we learn a finite state probabilistic transducer 15,16 for each possible source-target pair s❀ r and time lag ✁ (Fig. 1d), yielding

Introduction An important challenge for children is developing friendship relations [1,2], and failing to gain a sense of belonging in the peer group has been associated with poor psychosocial outcomes [3]. Being overweight is one aesthetic feature that can hinder the establishment of friendship relations [4]. Children who experience weight-based stigma have been found to be at an increased risk of social isolation, loneliness, depression, low self-esteem, and reduced quality of life [4]. Moreover, despite current high rates of childhood obesity [5] potentially normalizing excess body weight, obese children seem to be stigmatized even more so than when rates were relatively low [6]. The impact of weight-based stigma on young peoples' friendships has been well documented. Overweight children are less attractive as friends, receiving fewer friendship nominations than their non-overweight peers [7][8][9][10][11]. As a result of this social exclusion they tend to be found at the periphery of their peer networks [12] despite nominating as many, or more, friends than their non-overweight peers [7,9,12]. Moreover, overweight children are likely to select each other as friends [7][8][9][10], and may subsequently influence and reinforce each others' overweight status, because studies have shown that children and their close social ties become more similar in body weight over time [8,[13][14][15]. Together, these findings highlight the importance of excess body weight as a social marker for the establishment of friendships, and the negative implications of the resulting marginalization of overweight youth by their peers. Yet, to date, studies of weight-based marginalization have solely focused on exclusion from friendships, indicative of a passive form of social exclusion. It is not known if overweight children are also overtly rejected by their peers, whereby peers clearly express dislike of overweight children. Traditionally, rejection has been measured as the extent to which children were nominated as being disliked by their peers [3,16], however recently it has been conceptualized as a dyadic phenomenon, where it is defined as a negative relationship between pairs of children which can either be unilateral or mutual [17,18]. Involvement in these relationships of rejection-so-called "antipathies"-has been found to negatively impact the social development of children in ways that are distinct from passive marginalization [19][20][21]. For example, research on neglected and rejected children have shown overtly marginalized youth are especially at risk for maladjustment [19]. So, overweight might lead to marginalization by having fewer friends, but its impact on youths' psychosocial development is likely to depend on whether marginalization is also the result of being overtly rejected by peers. A more complete assessment of the marginalization of overweight youth should therefore assess friendship relations as well as examine the extent to which overweight is related to having antipathies in the peer group. The aim of this study is to examine how overweight is related to both friendship relations and antipathies in children's peer network. Overweight children are expected to receive fewer friendship nominations, as well as more dislike nominations, relative to their non-overweight peers, with the latter indicative of a more overt process of rejection and marginalization. Consequently, we also expect that overweight children have more friendships with each other. As gender is an important determinant steering friendship preferences [22], this attribute is controlled for. Statistical models for social networks , were used to test for associations between individual and relational variables using a modeling framework that accounts for the complex structure of friendship and antipathy networks and the inherent dependencies within them [23][24][25]. --- Materials and methods --- Sample A peer nomination subsample was used from the first wave of the TRacking Adolescents' Individual Lives Survey cohort, collected between March 2001 and July 2002 [26][27][28]. The TRAILS sample consisted of preadolescents living in five municipalities in the north of the Netherlands. The 'peer subsample' comprised of school classes with at least ten TRAILS participants, in which peer relations among classmates were assessed. Children in special education or in small schools, and children who repeated or skipped a grade were excluded from this peer subsample [28]. Written consent to participate in the study was obtained from both the parent and child. The study protocols were approved by the Netherland's national ethics committee 'Central Committee on Research Involving Human Subjects' . For the present study, classes with participation rates under 60% were excluded to ensure reliable estimates of friendship and antipathy patterns. This yielded a target sample of 28 school classes in the last year of primary education, with information on friendships and antipathies received of 714 children, including TRAILS participants and the non-participating classmates that they nominated . Information on nominations received by non-respondents was retained to gain a more complete representation of the network structure, however the inclusion of non-respondents in our classroom social network data means that their outgoing nominations are coded as missing. This limits our ability to distinguish between unidirectional vs. mutual friendship and dislike relations, and so our analyses focus on overweight children's involvement in any friendship and dislike relation. Information on all other variables was only available for respondents, resulting in a target sample of 504 children . There were no significant differences between the analytic sub sample included in this study and other TRAILS participants in terms of body mass index or the proportion of participants who were overweight or obese. --- Measures Friendships. Participants nominated an unlimited number of their best friends in their school class. Friends could be any gender and could include classmates not participating in TRAILS. In the Netherlands, children take courses with a fixed group of classmates, therefore classroom-based friendships are likely to be an important segment of their broader friendship networks. Best friend nominations were used to define a directed adjacency matrix representing the friendship network within each school class, where for each directed pair of students x ij = 1 if student i nominated student j as a friend. Antipathies. Participants nominated peers they disliked in their class, and similar to friendship relations, the number and gender of nominations was not restricted. Again, dislike nominations were used to define a directed adjacency matrix representing the antipathy network within each school class, where for each directed pair of students x ij = 1 if student i nominated student j as someone they dislike. Anthropometry. Height and weight were measured individually by trained research assistants using a SECA 208 stadiometer and a SECA 761 mechanical scale, and used to calculate participants' BMI . Internationally validated age and gender specific BMI cut-off points [29] were used to classify participants as non-overweight or overweight . Demographics. Participants reported on their gender and age. --- Statistical analyses A two-stage multilevel procedure was applied for the analyses. In the first stage, each network was analyzed separately using ERGMs [23,25]. The unit of the analysis is the ordered pair of students in a classroom , and the dependent variable is the observed value of a friendship or dislike tie . As a consequence of the binary nature of the dependent variable, ERGMs are logistic models to predict the probability that a tie exists. As pairs of students are clearly not independent, dependencies in the data are explicitly modeled as structural effects in ERGMs . These structural tendencies were estimated alongside effects testing the hypothesized associations between weight status and friendship or antipathy relations and control-attribute effects, to identify those most likely to explain the structure of the observed networks [24,30]; that is, the particular configurations of ties that occur more or less than would be expected at chance levels, given the number of nodes and density of the network . ERGMs were fit separately for the friendship and antipathy networks using PNet [31], and a Markov Chain Monte Carlo approach was used to estimate model parameter and standard errors. To assess relationships between participants' attributes and their friendship and antipathy relations, three types of effects were included in the models. A "sender effect" represents the association between an attribute and the number of nominations given by participants. A "receiver effect" represents the association between an attribute and the number of nominations received. A "similarity effect" tests whether the probability of a network tie is greater among dyads that have the same score on an attribute. These participant attribute effects were included for weight status and gender . To control for potential differences in the network position of classmates who did not participate in the survey, we included a receiver parameter for nonrespondents to model their incoming nominations. The ERGM analyses yield a set of parameter estimates and associated standard errors for each of the classes. In the second stage, these findings were combined in a meta-analysis [32,33], where class coefficients were split into an average coefficient and a class-dependent deviation. To differentiate between true and error variance, and thus to obtain more precise estimators for the average effects and the variance of the effects across classes, we accounted for the differences in standard errors between classes, such that classes with large standard errors have less influence on the average effect size. Additionally, classes with very large standard errors on a parameter were removed from the analysis of the average effect and variance of that parameter, as in these cases the regression coefficient was usually very high as well. The program MLwiN [34] was used for an iterated estimation of the weighted least squares. Average effects across classes are deemed significant at a .05 level when the ratio of the parameter estimate to the standard error exceeds 1.96 [30]. Significant positive parameter estimates indicate the effect is more prevalent than would be expected by chance, given other effects in the model, and the reverse is true for negative estimates. In addition to the average effects across classes, we also report on variance of the effect, which indicates whether classes differ in the extent to which the effect occurs. --- Results --- Descriptive statistics and visualizations Descriptive statistics for both types of relationships are presented in Table 1. There was an average of 25 participating students per classroom . The mean number of friendship nominations given and received was 6.91 and 4.81 , respectively, whereas the mean number of dislike nominations given and received was 3.78 and 2.15 . Approximately 15 percent of the participants were overweight, and there was an average of 2 to 3 overweight children per classroom, with a range of 5.0% to 50.0% of participants that were overweight in each classroom. As an example, --- Statistical models for friendship networks Results from the ERGMs fit to the friendship networks show that the observed networks were explained by several processes. The significant structural network effects indicate that there was a strong tendency for friendship ties to be reciprocal . The negative alternating-in-star and marginally significant positive 2-in-star parameters indicate that participants who received a low number of friendship nominations were most common, and the negative alternating-out-star and positive 2-out-star effects show a tendency for most students to nominate a small number of friends. Additionally, the positive transitive closure and negative multiple connectivity effects showed that "friends of friends" tended to be friends, particularly when there were multiple shared friendships. Participants were also more likely to nominate same-gender peers as friends, although this effect varied considerably between classes. Non-respondents also tended to receive more friendship nominations than respondents. Over and above these effects, overweight children were less likely to be nominated as friends than their non-overweight classmates . This effect was negative in 70% of the classes, and significant across classes, indicating its consistency. Weight status was not associated with the number of friendship nominations given , and there was no evidence that friends tended to be alike in weight status, as shown by the non-significant overweight similarity effect. --- Statistical models for antipathy networks Results of the ERGMs fit to the antipathy networks showed a tendency for dislike nominations to be reciprocated, for some participants to receive a higher number of nominations than others , and for some participants to make more dislike nominations than others . There was no evidence of triadic closure in antipathy networks, whereby node A disliked node B, node B disliked node C, and node A also disliked node C. This is not surprising, as it indicates that classmates did not have a tendency to cluster in triads or small groups of classmates who all disliked each other. The gender effects revealed that relative to girls, boys were more likely to nominate peers that they disliked, and to be nominated as someone who is disliked. The negative gender similarity parameter indicates that antipathy relations were more likely among cross-gender peers. Additionally, non-respondents tended to receive more antipathy nominations than respondents. Finally, overweight children were more likely to receive dislike nominations than their non-overweight classmates. They were also more likely to nominate peers that they disliked. Specifically, the odds that a dislike tie was present versus absent was 1.65 greater when the receiving student was overweight, and 1.15 greater when the nominator was overweight. The receiver effect was positive in 86% of the classes , and significantly positive https://doi.org/10.1371/journal.pone.0178130.g001 in 4 of these classes. The sender effect was positive in 12 classes, and significantly positive in 4 of these classes. There was no tendency for peers who shared an antipathy relationship to be similar in weight status, as shown by the non-significant overweight similarity effect. --- Discussion Building on previous findings that overweight youth receive fewer friendship nominations [7][8][9][10], this study shows that overweight children are not only passively marginalized by their peers by receiving fewer friendship nominations, but they are also overtly rejected by being disliked by more of their peers. Overweight youth were also found to nominate as many friends as their non-overweight peers; a finding which is in line with longitudinal research showing that marginalization is predominantly driven by overweight youth being excluded by peers who do not reciprocate their extensions of friendship, rather than a result of their own withdrawal [10]. Additionally, this study revealed that overweight children were more likely than their non-overweight peers to dislike their classmates. The combined tendencies for overweight youth to dislike more of their peers, for overweight youth to receive more dislike nominations, and for dislike relationships to be reciprocal, indicates that overweight children are generally more involved in unidirectional and mutual antipathies. This social environment, characterized by fewer friendships and greater antipathies, is likely to put overweight youth at increased risk for psychosocial maladjustment [4]. The resulting social isolation may also promote unhealthy behaviors, such as excessive food intake and decreased participation in sports and physical activities [35], which can lead to further weight gain and thus a cycle of poor physical and social outcomes. We did not find support for the hypothesis that overweight children would be more likely to befriend other overweight children, which has been shown in other studies [10][11][12]. Our results may differ because these other studies examined friendship relations among a larger set of peers within grade cohorts or entire schools, while our study examined friendships among a smaller set of peers within classrooms. Hence, opportunities for overweight youth to befriend one another were few with just 2 to 3 overweight students per classroom, on average. It may be that socially marginalized overweight youth seek out friendships with overweight peers who are outside of their class. --- Limitations Due to the cross-sectional nature of the data, we examined associations between overweight and the structure of these social networks, but could not test for factors driving the selection of network partners or influences of the network on participants; processes which may lead to these observed associations. For example, although we did not find evidence of similarity in weigh status among peers who shared friendship relations or antipathy relations, the emergence of weight similarity amongst adolescent and adult friends found in other studies [8,13] could result from longer-term influence processes, whereby weight status assimilates as a result of shared friendship, potentially via similar engagement in obesity-related behaviors [36][37][38] or shared weight norms [39][40][41]. Studies that follow youth through these developmental stages are needed to understand the emergence and timing of these processes. It would also be valuable for future longitudinal research to identify social processes and broader structural features of peer networks that give rise to mutual vs. unidirectional friendships and antipathies among overweight youth. For example, mutual antipathies may be established to achieve structural balance within particular triadic network structures, whereby youth dislike the peers that their friends' dislike; a process that may be socially adaptive and not associated with psychosocial risk for overweight youth [42]. A further limitation of our study is that peer relations were only assessed within classrooms. It is possible that findings may differ with other types of peers . --- Conclusions Stigmatizing and marginalizing overweight children does not serve to discourage overweight, but rather contributes to their increased risk of negative physical and mental health outcomes [43]. These negative psychological outcomes are not surprising given our findings that overweight children actively seek out friendships, but are marginalized as a result of being passively and overtly rejected by their non-overweight peers. This overt rejection entailed being the target of a greater number of antipathies from their peers, as well as the originator of a greater number of antipathies directed towards their peers. Addressing weight-based stigma, especially amongst non-overweight children, should be a standard component of obesity prevention efforts, with the aim of improving social integration and overall quality of life of overweight children as well as their physical health. --- Data are from the TRacking Adolescents' Individual Lives Survey . Readers may contact [email protected] to request the data. Information about the specific conditions under which the data are available can be found on www.trails.nl/en/hoofdmenu/data/ data-use. For this study, JKD and RS accessed the data by requesting it from [email protected]. Other interested researchers can access the data using the same procedure. RS was affiliated with the TRAILS study as a consulting obesity expert, via the Department of Epidemiology of the University Medical Center Groningen. No other authors were affiliated with the TRAILS study. The authors of this study did not collect any data on behalf of TRAILS. --- --- Data curation: JKD RS. Formal analysis: KdlH JKD MJL LvR. Funding acquisition: JKD RS. --- Methodology: KdlH JKD MJL RS. Writing -original draft: KdlH JKD. --- Writing -review & editing: KdlH JKD MJL LvR RS.

Weight-based stigma compromises the social networks of overweight children. To date, research on the position of overweight children in their peer network has focused only on friendship relations, and not on negative relationship dimensions. This study examined how overweight was associated with relations of friendship and dislike (antipathies) in the peer group. Exponential random graph models (ERGM) were used to examine friendship and antipathy relations among overweight children and their classmates, using a sub-sample from the TRacking Adolescents' Individual Lives Survey (N = 504, M age 11.4). Findings showed that overweight children were less likely to receive friendship nominations, and were more likely to receive dislike nominations. Overweight children were also more likely than their non-overweight peers to nominate classmates that they disliked. Together, the results indicate that positive and negative peer relations are impacted by children's weight status, and are relevant to addressing the social marginalization of overweight children.

[5] [6] [7] [8] . Although well-enforced smokefree legislations protect people from SHS in public places, people often remain exposed to SHS inside their homes [9,10]. Inhome SHS exposure increases the risk of cardiovascular and respiratory diseases, lung cancer and asthma, [11][12][13][14] and is associated with poorer mental health outcomes in adults [15] and children [16]. It has also been associated with depression, [17][18][19][20][21][22] stress, [23,24] and anxiety [22,25]. Psychosocial issues related to home SHS exposure have also been reported in the literature. These include family strife and unhappiness, [26] the need to navigate social and relational norms, [27] heightened distress and lowered sense of agency among caregivers of higher risk children, [28] a sense that one's privacy has been invaded and the ability to enjoy one's home has been undermined [29], and the tension residents face between sympathizing with the neighbour's smoking addiction and the belief in a collective responsibility to refrain from actions detrimental to fellow residents [30]. As there is no risk-free exposure to SHS, and air purifiers and ventilation are ineffective protection mechanisms, [31][32][33] the only way to protect people from SHS is to eliminate smoking where others are exposed [34]. However, households often face challenges in agreeing on a smoking ban and end up compromising on less effective strategies such as restricting smoking to specific parts of the home or smoking out of a window [35][36][37]. Even non-smoking households are exposed to SHS from neighbouring homes, especially those living in multiunit housing [38][39][40][41]. A Hong Kong study found that, among non-smoking adolescents, those exposed to SHS from neighbours were more likely to report symptoms of respiratory disease [42]. Although no country has banned smoking inside homes, patchwork legislations exist in the United States covering public multiunit housing, [43] multiunit housing in 67 Californian municipalities, [44] and apartment complexes with voluntarily adopted smokefree measures [45,46]. In Singapore, a city-state in Southeast Asia with a multi-ethnic population, the issue of SHS exposure in homes has been the subject of Parliamentary debates since 2017 due to high volumes of complaints about SHS drifting in from neighbouring homes, especially following the COVID-19 lockdown measures [47,48]. In a 2020 survey, 85% of Singapore residents supported a proposal to ban smoking near a window or balcony in multiunit housing [49]. With 95% of Singapore residents living in multiunit housing, including condomiums and public housing estates [50], SHS drift into others' homes appears to be widespread despite a low adult smoking prevalence at 11% [51]. Comprehensive smokefree legislations cover many public places and shared residential spaces such as common corridors, stairwells and void decks [52,53]. Although the Singapore Government has not articulated any formal plans to regulate smoking inside homes, public and Parliamentary debates were ongoing as at September 2021 [54]. Little is known on the nuances of how residential SHS affects individuals, families and neighbours living in densely populated, multi-ethnic urban settings such as Singapore. This study aims to understand the psychosocial impacts, views and experiences of residential SHS exposure in a densely populated urban setting. --- Methods In March-August 2021, we conducted in-depth interviews with 18 key informants who had been involved in public discourse, policy, advocacy, or handling complaints related to residential SHS and 30 Singapore residents who smoked in their home or were exposed to SHS in their home . We recruited key informants via email invitation, and residents via email flyers, social media and snowball recruitment. Prior to joining the study, residents provided information on their sociodemographics, smoking habits and history of residential SHS exposure to enable sample balancing in terms of age, gender, ethnicity, housing type, smoking status, and experiences with in-home SHS. Interviews with key informants were one-on-one while interviews with residents were one-on-one or dyadic, in cases where two household members preferred to be interviewed together. Interviews lasted 40-70 min each, were conducted in English, the most widely spoken language in Singapore, and were done either face to face or online using Zoom conferencing. Interview questions followed an open-ended format . We reimbursed each resident or dyad with S$50 cash. Interviews were audio recorded, transcribed verbatim, and imported into NVivo. We developed an initial codebook with deductive codes originating from a priori topics in the interview guides, and subsequently modified the codebook to include inductive codes upon multiple reading of the transcripts. Finally, all transcripts were double coded and compared among the researchers working independently to ensure coding consistency. Similar codes were combined and new codes were added to the codebook during the coding process. Discrepancies were reviewed and discussed by the researchers until consensus was reached. Codes were then organised into categories, sub-categories and overarching themes . Although data for all participants were coded together, we distinguished between key informants, smokers, and non-smoking residents to identify differences in the themes emerging from each group. The study was approved by the National University of Singapore Institutional Review Board . Participants were informed of the study procedures and risks and provided written informed consent prior to the interview. --- Results Participants discussed four themes: their perceptions and experiences with residential SHS; strategies used to minimise SHS exposure from neighbours; strategies used to minimise SHS exposure from smokers in the household; and views on what it means to smoke in a socially responsible manner. In what follows, we discuss findings from key informants, smokers and non-smoking residents, with quotes to illustrate our points where relevant. --- Perceptions and experiences with secondhand smoke --- Harm perceptions of secondhand smoke Participants unanimously agreed that SHS is a serious health hazard, especially to vulnerable people such as children, associating it with lung cancer, asthma and cardiovascular diseases. Some participants knew someone who had died from SHS exposure. However, participants appeared to have various misperceptions on the relative harms of SHS. A few, including smokers, described SHS as more harmful than active smoking, while others believed that SHS is only harmful if exposure is heavy or prolonged: That kind of low level, I guess is OK, I guess it's not so bad. Because your lungs… rest and doesn't collect all these particles all the time. -Resident Others, especially key informants, were unsure and highlighted a need for more evidence on the doseresponse effect of SHS exposure: --- There is a need to show in concrete terms the harm that SHS poses… exactly when, and at what levels and how, does exposure to SHS become harmful? -Public Officer Several participants believed that the harmfulness of SHS depends on other factors such as genetics or preexisting conditions: …someone with asthma or some respiratory issues, it would be different. Smaller amounts could be an immediate reaction and all of that, but like seeing regular, healthy able-bodied, and stuff, I don't imagine it's that bad or maybe I'm just being hopeful. -Resident --- In-home secondhand smoke from neighbours Key informants described SHS incursion from neighbours as a longstanding and common issue, affecting many residents due to the high density in which people in Singapore live. Key informants and residents also described a building's layout, airflow, and proximity to areas where people smoke , as reasons why some homes may be more affected than others: My room is at the back side, so if they go to the back Participants described the SHS from neighbours as difficult or impossible to escape: --- I've lived in three different condos in Singapore… In all three, I had smoker neighbours. It's not like moving house can solve this problem. Right now it's worse in this current condo that I'm staying in. We were sandwiched among three smokers, upstairs, downstairs and next-door. -Advocate --- Secondhand smoke in common residential areas Participants commonly recalled experiences with people smoking in common residential spaces where smoking is prohibited such as void decks, common corridors and stairwells. Most smokers admitted to smoking in these areas, out of convenience or a perceived lack of enforcement: --- Let's just put it down to pure laziness, so that's why I smoke at the common corridor and staircase landing. -Resident As long as you don't get caught, it's not a problem. So I have seen people smoking at the void decks, I have seen people smoking in the corridors… we do it, but we do it discreetly. -Resident Smokers who had observed enforcement of this smoking ban felt that this was an effective deterrent, while others were unsure of the rules: --- I generally follow it because I also don't want to just randomly get fined $300 by plainclothes NEA officer. -Resident …at our void deck, where the rubbish bin is, there is the smoke thing [rubbish bin with ashtray]. So I think it's okay to smoke there. -Resident --- Secondhand smoke exposure following the COVID-19 pandemic Both key informants and residents felt that residential SHS had increased following the COVID-19 pandemic, primarily due to people spending more time at home. However, a few residents living with smokers felt that the pandemic had either improved or not changed their exposure to SHS, in cases where habits of the smokers they lived with remained the same or the lockdowns resulted in them not living together with the smoker. --- Personal impacts of secondhand smoke When asked about how SHS exposure has affected them, participants reported a wide range of medical conditions. Those exposed to SHS by people smoking inside their home recalled experiences with lung cancer, breast cancer, asthma and eye conditions. Even those whose family members only smoked in confined or outdoor parts of the home reported sinus and respiratory issues such as chest pains and breathlessness. Participants from nonsmoking households reported respiratory symptoms, worsening of their asthma, allergic reactions, headaches and migraines following SHS incursion from neighbours. They also described medical conditions in children, notably respiratory issues, sinus issues and eye irritation, as being caused or aggravated by a neighbour's SHS: The poor toddler has been suffering from chronic bronchitis and even pneumonia due to her downstairs chain-smoker neighbour. -Advocate Participants also reported negative impacts on their mental wellbeing, with SHS described as a source of stress, anxiety, negative moods, and sleeping disorders. Those exposed to SHS from neighbouring units commonly indicated these as being tied to a sense of frustration, hopelessness, and constant worrying about the health effects: Participants highlighted the inconvenience and frustration of constantly having to close windows to block SHS from neighbouring homes. One participant, whose mother was a cancer patient, described how continually having to open and close windows was disruptive to her recovery: …she wants fresh air. Open window then got smoke, to her is troublesome also because when she's resting, after 5 minutes while she is lying in the bed so comfortable, she have to get up to close the window. My -Resident Participants described how having to keep their windows closed made the home stuffy and unhygienic in Singapore's hot and humid climate: --- My toilet floor is always wet and because when it's wet and it's moist, it affects my walls and windows with mould and mildew. -Resident I can't even smell fresh air in my own personal space. -Resident . Participants reported feeling nauseous, irritated, or frustrated by the SHS smell. This was a theme even among most of the smokers: As a smoker, I hate secondhand smoke. I don't like the smell. -Resident . I like to smoke my one stick, I don't want to smell the smoke of everybody else… the smell of cigarettes that's not the one that you're inhaling is different… you're getting like the ash at that point, rather than the nice nicotine. -Resident Participants also highlighted the inconvenience and financial burden of having to re-wash laundry that had been exposed to SHS or ash dropping from neighbouring units and having to run airconditioning instead of opening a window. --- Minimising SHS from neighbours --- Strategies to minimise secondhand smoke from neighbours When faced with SHS from neighbours, most affected participants reported that they close their windows or doors to block out the SHS. This was generally considered the most effective strategy, although it came at the cost of forfeiting fresh air and ventilation in their homes: We don't have aircon in the house, so we depend a lot on fan and we do need ventilation some way, somehow. -Resident Some participants also reported using an air purifier or fan, or moving into another room to avoid the SHS: I' --- Confronting neighbours about secondhand smoke A few participants had confronted their neighbours about SHS, with approaches ranging from friendly to antagonistic. Some had approached neighbours with gifts, a polite note on the door, or a friendly conversation emphasizing the impact of SHS on their children's health: --- I thought in the first place, we could address it quickly in a sense that I pay him a visit and still talk to him nicely, saying that can you close the window and not allow your smoke [to] drift into my place? -Advocate More antagonistic approaches included leaving notes in common areas singling out units suspected as the SHS source, scolding the neighbours, or spraying insecticide on them: She will go upstairs and like scold the person and be like 'look at my clean clothes now. ' Most participants, however, were reluctant to confront their neighbour about SHS as they felt anxious it would lead to conflict or believed that, with no regulations, these efforts would be futile: --- I don't want to actually confront them because that would put me in a difficult position… what if the person gets aggressive? -Resident …with no regulation for smoking in the house, there's technically nothing that we as neighbours who are non-smokers can do about it. -Resident --- Smokers' responses to neighbour confrontation Regardless of the approach, most attempts to settle the issue directly with neighbours were described as unsuccessful. The smoking neighbour's responses ranged from avoidant to hostile, while others responded amicably but took no action to reduce SHS: --- They kind of just nod their head and then walk away. -Resident . [The] downstairs smoker refused to open the door on multiple occasions. The upstairs smoker insisted that it's his right to smoke at home, because it's not against the law and told us to mind our own business. Then the next-door neighbour turned aggressive. -Advocate Smokers or their family members, when asked how they would respond to a neighbour's request to reduce SHS, gave a range of responses ranging from reluctant to willing to compromise. Those who were reluctant believed that their SHS was unlikely to affect others or that they were entitled to smoke in their home: --- I will get a bit defensive because this is my house. -Resident . I will tell them to close their own windows. Because I actually smoke in the middle of my living room so, I have no idea how my secondhand smoke will actu-ally affect them. -Resident Those who were willing to compromise generally had more awareness of the health effects of SHS and expressed a stronger interest in keeping a good relationship with neighbours: I would apologize first because I would feel really, really bad about it since I'm very cautious about this kind of stuff, honestly. -Resident …we don't want any trouble with our neighbours. We have a very good relationship with our neighbours. -Resident Other smokers indicated that their response depended on the neighbour. Those perceived as inconsiderate or unfriendly were more likely to be met with reluctance. A few smokers indicated they might be more sympathetic towards those experiencing health issues: --- I will only stop if you stop, stop stomping and moving furniture in the middle of the night. -Resident It really depends on how belligerent they are about it, to be honest… If someone's bringing up health complications, then I would be a lot more understanding. -Resident For smokers, the main reason they did not smoke inside their home with windows closed was to minimise SHS exposure to their family members, especially children: --- He'll run to the kitchen window and smoke, which I think is not nice to the other neighbours but you can see he's trying to be considerate for his grandchildren and his guests. -Resident While most smokers were reluctant to smoke outside their home due to the inconvenience, in one case it was more challenging as the smoker had a mobility issue: --- He's got some mobility issues, some health condi-tion… for him specifically to go down and smoke and taking our time , I think it's a bit, it's pretty hard for us. That's why we allow him to smoke in the house instead. -Resident --- Minimising secondhand smoke from smokers in the household --- Smokers' strategies to minimise secondhand smoke The strategy most commonly taken by smokers to minimize SHS in the home was to limit where in the home they smoke, usually to an outdoor area or enclosed space within the home . The majority also closed doors to minimize SHS incursion from these spaces into other parts of the home. Another commonly reported strategy was to smoke near a window or ventilation system, such as an air filter or fan, to blow out the smoke. Two participants also avoided smoking inside the home unless their family members were out. --- Non-smokers' strategies to minimise secondhand smoke Strategies most reported by non-smokers to minimize in-home SHS were closing doors and using fans or air purifiers. They generally described these strategies as ineffective: --- Ultimately I still can smell it. Like no matter what, if he's smoking I can smell it. -Resident A few participants reported avoiding the areas where household members smoke: --- I got fed up, I spend all my time in the [bed]room. -Resident . --- Confronting smokers about secondhand smoke in the home Non-smokers described various approaches they had used to persuade a family member to reduce in-home SHS. Some participants simply set a no-smoking house rule or asked the family member to restrict their smoking to specific areas within the home. Others had attempted to talk to the smoker about quitting, but with little success. A minority had approached the matter from a health perspective, but reported that they had little success with this unless they were able to make a personal appeal: The facts are useless in a scenario like this… I can say to him now, 'you know my brother has cancer right? So you probably shouldn't smoke around him. ' Then he'll get it because then he has a personal connection to the matter. -Resident Others simply expressed their discomfort or disapproval when the family member smoked inside the home, in various ways: …being sarcastic, like cough in front of them when they smoke. --- -Resident I'm very angry, I scream at him. -Resident . A few smokers were described as willing to change their habits after pressure from family members, while others were described as reluctant to change. This reluctance was often expressed as an unwillingness to listen: It was quite clear that he was not receptive. Kind of, in one ear, out the other. -Resident I always say, 'can you at least do it outside?' But he'll just be like, 'yah lah, ya lah, ya lah' . But he doesn't do it. -Resident --- Issues in addressing in-home secondhand smoke The inability to resolve the in-home SHS issue was described as a source of conflict within some families and guilt among smokers: When asked about barriers to getting smokers to reduce in-home SHS, participants described how, since smoking had become a deeply ingrained habit, it was difficult to change their smoking routine. They also cited the inconvenience of having to move outside to smoke. Traditional patriarchal norms made it difficult for some participants to confront family members, especially in cases where the smoker was their father: I You don't own the house. You do not dictate to him. I mean, he has, you know, Asian society, he's still the senior. -Resident --- Views on socially responsible smoking While participants, especially smokers, held the view that people are entitled to smoke or do what they like inside their own property, they also strongly felt that smokers don't have a right to smoke in their homes if it affects others. These views were echoed equally among smokers, non-smokers and key informants: --- What about the right to throw garbage out of the house? What about the right to pour water out of your house… do we allow for those rights? -Academic Neighbours or people who live in their house and say, 'it's my house, my own problem, I smoke, my own problem. I don't disturb you' , which is a very wrong concept they have, because definitely they're bothering somebody. -Resident Key informants, non-smokers and some smokers generally described smokers as being on a spectrum, ranging from those who voluntarily try to smoke in a socially responsible manner to those who appear to be indifferent or unaware of their impact on others: I got a handful of smoker friends, they are very con-siderate… but there are also [an] inconsiderate group that we are facing. -Advocate Several participants felt that Singapore's culture of entitlement compounds the issue by encouraging selfishness and disregard for others among some smokers: We're just generally very entitled people. We don't really care about other people. -Resident Most smokers described themselves as taking steps to smoke in a socially responsible manner. However, they appeared to have differing views on what that entails in practice. While some smokers went to great lengths to avoid smoking near people, especially children, others simply complied with no-smoking rules. In general, younger and female participants perceived SHS as more harmful and expressed a greater desire to smoke in a socially responsible manner. A few participants described specific actions that had been taken to minimize SHS disturbances to their neighbours, including seeking consent from the neighbours to smoke or voluntarily smoking in an area downstairs, away from the building. --- Discussion This is, to our knowledge, the first study detailing the psychosocial aspects of residential SHS in Singapore, and the first in Southeast Asia to also explore SHS incursion from neighbouring homes and include the perspectives of smokers and other stakeholders. Residential SHS has wide-reaching impacts on physical and psychosocial wellbeing, even if smokers try to minimise SHS or if the SHS is from a neighbouring home. This is consistent with literature demonstrating the adverse health impacts of low levels of SHS exposure, [55][56][57][58] as well as evidence associating SHS exposure with mental health conditions such as depression, [17][18][19][20][21][22] stress, [23,24] and anxiety [22,25]. Our findings indicate that these negative mental impacts are tied to a sense of entrapment or discomfort in one's personal space, a perceived lack of control over the situation, resentment or frustration towards smokers, and constant worrying about the health effects, especially on children. For smokers, inability to resolve the issue was a source of guilt and family conflict. Residential SHS may have more of a psychosocial impact as it encroaches into the private space and is often tied to interpersonal relationships with family members or neighbours [59]. Family, community, and cultural dynamics add further complexity to the problem. In our Singapore households, as well as those in other studies, creating a smokefree norm at home posed interpersonal, structural and cultural challenges, [60,61] influenced by knowledge and risk perceptions of SHS, one's sense of agency, interpersonal relationships, and wider community norms [27]. Traditional patriarchal households, as often found in Asian societies, may face additional barriers if the smoker, often a male adult, resists influence from family members. Some of our participants who lived with a smoking husband, brother or father reported this problem, expressing that it would be inappropriate to objecting to the 'elder' , 'head of the household' or the homeowner from doing as he pleases in the home. This has similarly been reported in other studies [27]. Studies from countries with similar patriarchal norms suggest that equipping non-smokers with skills to influence the smoker may help to break down such cultural barriers [62,63]. Patriarchal norms may also be used to facilitate the creation of smoke-free homes. An emphasis on the role of men as protectors of the family/community, coupled with education on the harms of SHS and smoke-free norms at the societal level, frames the establishment of a smoke-free home as an act of male responsibility in protecting and caring for women and children [64,65]. This is consistent with our results where male smokers report feeling guilt over exposing their family to SHS knowing that it harms their family members. Participants described SHS incursion from neighbours as a widespread issue, with building layout, airflow, a unit's location, and lifestyle factors, such as spending more time at home following the COVID-19 pandemic, cited as factors affecting SHS concentrations. An increase in cigarette-use or SHS exposure at home during COVID-19 lockdown measures have also been reported in other countries [66][67][68][69][70], and the substantial impact on children in smoking households who spent less time in smoke-free places and more time at home has been highlighted [71]. Singapore has a high population density, with 95% of residents living in multiunit housing [50]. Even with a low smoking prevalence, SHS may be more concentrated in crowded urban settings such as Singapore's, as high rates of in-home SHS have been reported in other densely populated cities including New York City, [72] Los Angeles, [73] and Seoul [74]. In such settings, factors such as building layout, airflow and social distancing measures may have more tangible impacts on SHS levels inside homes. To address the issue of residential SHS, smoking has been banned in public multiunit housing in the United States, [43] and public and private multiunit housing in California [44] and Canada [46,75,76] although non-compliance and weak enforcement presented a challenge [77][78][79][80]. In Singapore, debates have been ongoing over whether to implement a similar ban, [47,48,54] and interventions such as designated smoking points [81,82] and public education campaigns [83] are being considered. Due to Singapore's hot and humid climate, closing the window to block out a neighbour's SHS was considered unfeasible, leaving participants with neighbourly confrontation as the only recourse. Most were reluctant to do so, expressing a fear of conflict or sense of futility in the absence of regulations. The overall reluctance to approach neighbours may reflect Singapore's culture, which tends to favour top-down regulation and social harmony over direct confrontation. As in other conflict-averse societies, the ability to resolve the issue amicably may depend on the neighbourly relationship [84]. Our findings suggest that a smoker's willingness to compromise may also be influenced by their harm perception of SHS and view on what it means to smoke in a socially responsible manner. For some, this meant avoiding exposing others as much as possible while for others it simply meant not breaking the law. While smokers took measures to reduce SHS, these were described as ineffective by non-smokers. This suggests that public education campaigns may be an effective intervention, if they emphasise that even low SHS levels are harmful and that the only way to smoke responsibly is to completely avoid exposure to others. An approach akin to that of a communitybased intervention that was successfully implemented in India [85] and Indonesia [86] might also prove promising for Singapore. The intervention messaged the importance of smokefree environments as a women and children's health issue, and established smokefree homes as a norm at the community-level [85,86]. Such an approach may be an effective intervention for protecting people against SHS in their homes in a context where neighbours' SHS is viewed more as a nuisance than a health threat and addressing SHS incursion at the individual level is too confrontational and daunting. --- Strengths and limitations Our study design gave participants an opportunity to share freely and surface themes that might not have been apparent a priori. While our findings may be informative for contexts similar to Singapore's , they may be less generalizable to contexts where knowledge of the harms of SHS is better, smoking is still a norm, or where people are more comfortable with asserting their individual rights. As our study was conducted during the COVID-19 pandemic, our sample may under-represent certain groups, such as participants with limited online access. --- Conclusion Residential SHS has wide-reaching negative impact on psychosocial wellbeing, especially in densely crowded settings where SHS is difficult to avoid. With no regulations covering smoking inside homes, neighbours are left to resolve the issue amongst themselves, often unsuccessfully due to various interpersonal, structural and cultural barriers. --- --- --- --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: People remain exposed to secondhand smoke, a serious health hazard, inside their home as households face challenges in setting no-smoking rules or are exposed to secondhand smoke drifting in from neighbouring homes. This study explores the psychosocial impacts, views, and experiences with residential secondhand smoke in a densely populated urban setting.In-depth online or face to face interviews with 18 key informants who had been involved in public discourse, policy, advocacy or handling complaints related to residential secondhand smoke, 14 smokers, and 16 nonsmokers exposed to secondhand smoke inside their home. All participants were residents of Singapore, a densely populated, multi-ethnic city-state. Interview transcripts were coded in NVivo using a deductive and inductive coding process. Findings: Secondhand smoke has wide-reaching impacts on physical and psychosocial wellbeing, even if smokers tried to minimise secondhand smoke. Feelings of anxiety and stress are generally tied to feeling discomfort in one's personal space, a perceived lack of control over the situation, resentment towards smokers, and concerns over the health effects. Family, community, and cultural dynamics add complexities to tackling the issue, especially in patriarchal households. Secondhand smoke exposure from neighbours is considered a widespread issue, exacerbated by structural factors such as building layout and the COVID-19 pandemic. Resolving the issue amicably is considered challenging due to the absence of regulations and a reluctance to stir up conflict with neighbours. While smokers took measures to reduce secondhand smoke, these were described as ineffective by other participants. Smokers appeared to have contrasting views from other participants on what it means to smoke in a socially responsible manner.Given the wide-reaching psychosocial impacts of residential secondhand smoke, there is a case for stronger interventions, especially in densely populated urban settings where it is more difficult to avoid.

Introduction Interdisciplinary efforts in network science have considerably deepened our understanding about the structure and dynamics of the society [1,2]. This active development of the field is largely attributed to the huge amount of digital information that has become available due to rapid development of the information and communication technology . Many empirical studies have been conducted on social networks including those based on email [3], mobile phone call [4][5][6], short-message communication, social networking services [7,8], and scientific collaborations [9]. Among these, MPC data sets play a special role because mobile phones are frequently used in our daily life and the coverage of the service is almost 100% for adults in large part of the world. As a matter of fact, an analysis of the MPC data set [4,5] has been successful in validating the "strength of weak ties" hypothesis proposed by Granovetter [10], stating that if the tie between two persons is strong, then the overlap between their neighbors will be large. This has far-reaching consequences for the overall structure of the social network: It suggests that the network consists of strongly wired communities connected by weak ties. This structure was proven at a societal level by applying percolation analysis to a social network, in which the duration of a phone call between two individuals serves as a proxy for their social tie strength. Hereafter we will refer to these networks as having Granovetterian structure. This led to the attempts to constructing a model of social network formation reflecting the observed structures [11,12]. More recently, this model was generalized to reflect the multiplex character of social networks [13]. In the original model by Kumpula et al. [11] two main mechanisms for the link-formation process were introduced, namely cyclic and focal closures, as they have been empirically observed to be the fundamental mechanisms for creating social ties [3]. Here the cyclic closure describes the formation of a link with the neighbor of one's network neighbor or, in other words, with a friend of a friend. The focal closure refers to making a link with the one sharing an attribute independently of the local network topology or geodesic distance between the individuals. In addition to these two mechanisms, a link reinforcement was introduced to correspond to the general observation that social ties get stronger when they are used. While in the previous studies the cyclic and focal closure mechanisms have been shown to play an important role in generating a Granovetterian structure, much less is known about the role of the process in which a relationship becomes terminated. In the original model [11] nodes were removed with a small probability or with a slow rate and new nodes without any links were added to the system to replace them. This way the asymptotic stationarity of the system with some average degree could be maintained. Deletion of a node corresponds to the death of a person or giving up a service or moving far away so that all contacts break. In real human society, however, the termination of a relationship may occur in various ways and there is no empirical support that the deletion of a node would be the main mechanism for terminating links. Even without a removal of a person, a relationship may end abruptly, for example, when a couple in intimate relationship suddenly break up. According to an empirical study, an individual drops out one member of her relationships every 7.2 months on average [14]. Apart from these rather sudden and drastic changes in human relationships, there is also a more gradual fading out of a friendship, which is typically seen when old but aging friends make less and less contact by time. It is not clear whether such difference in tie terminations affects the emergent network properties. In this paper, we investigate the effects of various link termination mechanisms and show that the link aging may lead to a significantly different network from the one with abrupt link termination or link deletion. The comparison between the models with different link termination mechanisms indicates that the link deletion reproduces the empirical findings from the MPC data set better than other mechanisms, implying that a link deletion plays a prominent role in the evolution of social networks among the link termination mechanisms. This paper is organized as follows. First we define the models with three different link termination mechanisms, including the original node deletion mechanism. Then we demonstrate that modular structures are more enhanced under the link aging than other mechanisms. Various local network features, such as degree distributions, link weight distributions, and link overlap, are investigated and compared with the MPC data set. Finally, we present summary and discussion in the last section. --- Methods First we briefly review the original weighted social network model proposed in [11]. It considers an undirected weighted network of N nodes, with links between them being updated by the following three mechanisms. The first mechanism is local attachment , in which node i chooses one of its neighbors j with probability proportional to w ij that stands for the weight of the link between nodes i and j. Then node j chooses one of its neighbors but i, say k, randomly with probability proportional to w jk and if nodes i and k are not connected, they are connected with probability p Δ with a link of weight w 0 . In addition all the involved links increase the weights by δ, whether a new link is created or not. The second mechanism is global attachment , in which a node with no links or otherwise with probability p r makes a new link to a randomly chosen node with weight w 0 . The third mechanism is node deletion , in which a node loses all its links with probability p nd , equivalently to replacing the node with a new isolated node. These three processes, i.e., LA, GA, and ND, are applied to all nodes at each time step. The initial condition is set to be a network without any links and the network reaches a statistically stationary state after a sufficient number of updates. This model turned out to show the Granovetterian structure of the society with strongly wired communities connected by weak ties. In this paper we extend the original model such that instead of ND, we consider two different link termination mechanisms: link deletion and aging. In case of link deletion we assume an abrupt termination of links, such that each link is removed from the system with probability p ld at each time step. Here we assume this probability to be independent of the link weight such that even a strong link can be suddenly terminated. In contrast, for the link aging mechanism we assume a gradual degradation of the link weight, such that at each time step, weights of all the links are multiplied by an aging factor f , where f controls the speed of aging. If the link weight becomes less than a threshold value, w th , the link is removed from the system. From now on, models with ND, LD, and aging mechanisms will be called the ND model, LD model, and aging model, respectively. --- Results --- Global modular structures In Fig 1 we show for comparison typical snapshots of the networks for three different link termination mechanisms with similar average degree. All the networks show modular structures, where the links connecting communities are typically weaker than the intra-community links, thus indicating the existence of the Granovetterian structure. However, the strength of the modularity varies significantly. The aging model shows clearly the strongest modularity among three networks while the LD model shows less modular structure than the others. Another interesting finding for the aging model is that the link weights within a community are more homogeneous. Furthermore, typical link weight in a community is inversely correlated with the community size. This is contrasting to the other models, where each community has both weak and strong links. In the following, we investigate these observations quantitatively. Here we performed simulations by setting N = 10 4 , p Δ = 0.05, p r = 0.0005, w 0 = 1, and δ = 1. For the ND model with p nd = 0.001 a typical Granovetterian structure is observed [11] such that the average degree is 11.0. In order to compare the results with the LD and aging cases, we control the parameters for the LD and aging models so that average degrees are comparable with that of the ND model. We adopted p ld = 0.0035 for LD model, and f = 0.9, w th = 0.01, and p r = 0.005 for aging model. Note that a higher p r is used for the aging model, otherwise the fragmentation of the network occurs. The results were obtained by running the simulations up to 25000 time steps and averaging over 50 realizations. The average degrees and other quantities are summarized in Table 1. We note that qualitatively similar results are observed for reasonable ranges of parameters other than those specified above. One of the most prominent differences between these models is found in their community structures. Although all models show community structure, the aging model has the most modular structure while the LD model has a lower modularity. In Table 1 we find that the Parameters are controlled such that these three networks have comparable average degrees . Color of links denotes the link weight. Strong and weak links are depicted in red and light blue, respectively. Note that different color scales are used for visibility. For the aging mechanism, community structures are clearly visible and the link weights in each community are more homogeneous. In the networks using the ND and LD mechanisms, strong and weak links coexist in each community. doi:10.1371/journal.pone.0133005.g001 aging model has the highest clustering coefficient and maximum modularity among three models. For calculating modularity Q we used the Louvain method [15] for binary graphs, a fast and efficient method for detecting non-overlapping communities. We find that the aging model has a slightly larger modularity than the ND model, while the LD model has a much smaller modularity than the ND model. It should be noted that the difference between the aging and ND models is more remarkable than one might think because ten times larger p r value is used for the aging model. If in contrast we use the same p r for the aging model, we get Q > 0.99 and the network is fragmented into communities most of which are cliques. This indicates that the link aging makes the network more modular than ND or LD. We have obtained the modularity using the Louvain method for weighted graphs to find less difference in modularities due to the fact that inter-community links are typically weaker than intra-community links. The higher modularity for the aging model is closely related to the Granovetterian structure. Since only weak links may be removed under the aging mechanism, inter-community links, which are typically weak, are removed more frequently. On the other hand, links within a community do not disappear easily as their weights are strong and once the Granovetterian community structure is generated, the modular structure is enhanced even more by aging such that the communities are more persistent. In contrast, since the ND and LD models remove links irrespective of the link weights, the intra-community strong links may be removed by chance, which results in less modular structure for the ND and LD models. We also note the difference between the ND and LD models. Although these two models assume an abrupt termination of the links, the ND model has higher values of clustering coefficient and modularity. This is because LD tends to reduce the fraction of intra-community links, which becomes evident by considering a clique of size n that after an ND event has n -1 links removed and a clique of size n -1 remains, meaning that the network has still a high modularity. However, if we remove n -1 links randomly by link deletion, the remaining network is no longer a clique, and we get a lower modularity because the number of links in a community is decreased while the number of nodes remains same. This implies that the community structure is in general more robustly maintained for the ND model than the LD model. Another notable difference of the aging model is the homogeneity of the link weights within each community. The link weights in a community are approximately similar for the aging model whereas the networks for the ND and LD models contain various weights of links in each community. In order to quantify this homogeneity, we introduced the relative fluctuation of intra-community link weights, which is defined as follows. First we identified communities then, for each community, calculated the relative fluctuation σ c /hwi c , where σ c and hwi c are the standard deviation and the average of the intra-community link weights, respectively. In order to distinguish the inter-community and the intra-community links, we needed to identify the non-overlapping communities. As shown in the Table 1, the aging model has a σ c /hwi c value less than one, indicating a smaller variation in link weights than observed in the other models. It is also notable that the average link weight of a community hwi c is inversely proportional to the community size. Fig 2 shows the average link weight as a function of the community sizes for a single run of the aging model. It clearly shows that the average link weight decreases as the community size increases, and it is well fitted by a function proportional to 1/. This relation between the average link weights and the community size for the aging model can be explained using an approximation that a community is a clique of size n having a uniform link weight hwi. This is a reasonable approximation because the modularity of the simulated network is close to one. The link weight hwi is estimated by the balance between the reinforcement by LA and the loss by aging. Here, we note that the increase of the link weights by GA is negligibly small compared to the reinforcements by LA. The probability p that a link l is included in one LA event is given by the number of possible triangles including l divided by the number of possible triangles in the clique, p ¼ n À 2 n 3 ¼ 6 nðn À 1Þ :ð1Þ At each time step, the expected increase of a link weight is pn because LA happens n times in that clique each time step. On the other hand, the decrease of a link weight is hwi. Comparison of the increase and the decrease gives hwi ¼ 6 ð1 À f Þðn À 1Þ ;ð2Þ which is in a quite good agreement with Fig 2. Aging is a multiplicative process, which dominates strong links, while reinforcement is additive with major impact on weak links. The link weights in a community tend to converge to their equilibrium value hwi. Thus, the dynamics of w is stable around its equilibrium value. A more detailed linear stability analysis is given in S1 Appendix. The differences between the community structures are also observed in the percolation analysis. Provided that links are sorted according to their weight, removing the links in the ascending order results in a sharp transition at a relatively early stage indicating the fragmentation of the society. In the opposite case, when links were eliminated in the descending order, the percolation threshold is significantly higher than that for the ascending order because strong links are within the communities. Fig 3 shows that all models have the Granovetterian structure that a difference between ascending and descending order percolation thresholds, Δf c , is observed. Although they show the Granovetterian structure similarly, the amount of Δf c shows a large variation. The aging model shows a much lower percolation threshold for the ascending order compared to the ND and LD models, while the percolation thresholds for the descending order are comparable. This is another indicator of the high modularity for the aging model. Since the fraction of the inter-community links to the intra-community links is low when the network is modular, removal of smaller number of weak links may lead to the fragmentation of the network. Furthermore, as we have seen, weak links are present in the communities for the ND and LD models, which makes the percolation threshold for the ascending order even higher. Thus, the aging model shows a faster fragmentation when weak links are removed first, while the LD model remains connected up to a higher percolation threshold. --- Local network properties Not only the global modular structures but also several local network properties show qualitative differences between the three models. In this Section, we obtain several local network properties and compare them with those of the MPC data set. Degree distributions for three models are shown in Fig 4. While the LD and aging models show Gaussian-like distributions, the ND model has an exponential tail for large k values. This is because a node in the ND model has monotonically increasing degree for most of the time until it is deleted. On the other hand, in the other models degrees of nodes can increase and decrease. Thus, the degree fluctuates around its mean value and the distribution gets similar to a Gaussian distribution due to the central limit theorem. In Fig 4 we depict the distributions of the node strength, defined as the sum of the link weights a node has. As in the case of degree distributions we find that the LD and aging models show peaks while the ND model shows a monotonically decaying behavior. In comparison, the MPC and many other empirical data sets show a monotonically decreasing degree and strength distributions [4]. Hence, the ND model seems to reflect best the empirical findings on that dataset. However, the monotonically decaying behavior might not correspond to common sense that in reality a person usually has a typical number of friends and typical amount of communication [17]. Thus we would expect that the monotonically decaying behavior found in the data set is due to its incompleteness and would change if other channels of communication could be taken into account. For this reason we speculate that the LD and aging models might capture the reality better than the ND model. Fig 4 shows the link weight distributions, where we find clear differences between the models. Both the ND and LD models show a monotonically decaying behavior which is approximated by a power law with an exponent close to -1. It indicates the coexistence of strong and weak links. In contrast, the aging model shows an initial power law decay for w < 1, but has a peak at around w % 5 and then decays quickly. As the MPC data set shows monotonically decaying behavior, we conclude that the ND and LD models fit better with the empirical data than the aging model. The initial power law decay in the link weight distribution for the ND and LD models is explained as follows. A link weight increases by the reinforcements of LA. Assuming that the probability that a link is selected by LA is proportional to its link weight, then the dynamics of the link weight w is approximated as dw/dt = cw, where c is a constant. Hence the expected weight of a link at age t is w = exp since w = 1, where τ r is a constant characterizing the speed of the reinforcements of the link weight. Since a link is removed randomly with a constant rate, the age distribution of links is an exponential distribution P = exp/τ d , where τ d is a time constant of the removal process. Using these equations, P / w - is obtained and its exponent is close to -1 since τ r ( τ d , i.e., the reinforcements occur much more frequently than the removal of a link. This is consistent with the simulation results when w is small. The deviation from power-law behavior is observed for large w because of natural cutoffs due to node and link deletions. On the other hand the initial power law decay in the link weight distribution for the aging model is explained by the decaying dynamics of w. Most of the links whose weight is less than one consist of the links which have not been reinforced after they are generated. The time evolution of the weight of such a link is given by w = exp, where τ a = 1/ is a time constant for aging. Assuming a new link is generated by a constant rate, the distribution of the link age becomes uniform. With this assumption we get the link weight distribution proportional to w -1 . Although the initially decaying behavior of the distribution is similar to those for the ND and LD models, the underlying mechanism is different. The node strength for the aging model does not show a dependence on the degree in Fig 4 , as the strength is given as s = hwi = 6/. This is clearly different from results of the ND and LD models and the MPC data set, where a node of a higher degree has a higher node strength. The observations from MPC and other social network data sets show assortative mixing, i.e., people having many friends tend to be connected to those who also have many friends. This is measured by the correlation of degrees between neighboring nodes. Fig 5 shows the average degree of the nearest neighbors of a node with degree k. We find assortative mixing for all three models although the degree of the correlation is different. The strong positive correlation for the aging model arises from the strong modular structure. Since a node and its neighbor often belong to the same community and the communities are often cliques, nodes in the same community have similar degrees, which is determined by the size of the community. The networks for the ND and LD models are less modular, thus the positive correlation is not strong as the aging model. One can ascertain the Granovetterian structure also by correlating the link weight and its overlap measure O ij . The topological overlap of the neighborhood of two connected nodes i and j is defined by the fraction of common neighbors [5]: O ij ¼ n ij ðk i À 1Þ þ ðk j À 1Þ À n ij ;ð3Þ where n ij is the number of neighbors common to both nodes i and j. Networks having the Granovetterian structure show an increasing behavior of O ij according to w. As shown in Fig 5 , positive correlations are found for the ND, LD, and aging models, while for the aging model it is found for w < 1. This indicates the existence of the Granovetterian structure. This positive correlation supports the robustness of the Granovetterian structure with respect to the modifications of link termination mechanism. The negative correlation found in the overlap for the aging model for w > 1 is explained by the negative relation between the community size and the link weight. As we have seen in the previous section, most links with weight larger than one are within communities and the link weight is inversely related to the community size. Links within larger communities usually have larger overlap. This is because the number of common neighbors is approximately proportional to the community size whereas the number of inter-community links do not depend on the community size since most of them are created by GA mechanism, which is independent of the network topology. Therefore, weaker intra-community links, which often belong to a larger community, tend to have a larger overlap. Even though O ij and w are negatively correlated for intra-community links, the Granovetterian picture that communities with strong ties are connected by weak ties is still valid. Empirical MPC data set has also shown a non-monotonic O as observed in the aging model. While most of the links show a positive correlation between w and O, the top five percent links of link weights show a negative correlation [4]. However, the underlying mechanism of MPC data set is not explained by the aging model. This is because the negative correlation for the aging model is originated from the negative correlation between the average degree and the link weights. In the MPC data set, however, the correlation between weight and degree product hw ij jk i k j i is negligible [4], which is not present in the aging model. In the case of the MPC data, the change in the trend in the O ij vs w ij plot could be traced back to a different origin: The extremely high values of weights were results of calls between persons, who almost exclusively talked to each other [5]. Fig 5 displays the relation between the local clustering coefficient and the degree. While the LD and ND models show a smoothly decreasing behavior, the aging model has a plateau around at k = 10. Not only MPC but also other social network data sets [18] often show negative correlations and the curve is often fitted by a power law decay with exponent close to -1. Therefore, the aging model does not on its own explain the empirical data. The reason could be that the aging model produces a rather narrow degree distribution where the asymptotics cannot develop. The community size distribution is yet another interesting statistics. Ahn et al. have demonstrated that the MPC data set shows a power law with exponent -3 in the size distribution of link communities [19]. In order to compare this empirical analysis with our models, we calculate the community size distribution using the link community detection method. In Fig 5 we show the distributions of the number of nodes in a link community, where we have adopted the method for a weighted graph. We observe that the ND and LD models show approximate power-law decaying behavior, which is qualitatively consistent with the observation from the MPC data set. However, the aging model has a clearly different profile as it shows an initial power law decay and a peak indicating the existence of a characteristic size. If we call the modules shown in Fig 1 as "node community", the former region generally corresponds to the links connecting node communities and the latter corresponds to the links within node communities. This distinction between inside and outside node communities is detectable only for the aging model. We speculate that this is because the modular structure for the aging model is so strongly emphasized that the distinction becomes visible. The ND and LD models as well as the MPC data set do not show distinct regions, implying that the aging model over-emphasizes the community structure. The statistics investigated above together with the MPC data set observations are summarized in Table 2. In the table, the expected behavior which matches to our common sense is also shown since we suspect some of the MPC data set does not reflect the real society. This is not only because of the limited observation period but because of other communication channels often used as well as MPC in our daily life. As discussed in [17], we expect there is a typical degree or amount of communications therefore degree distribution or node strength distribution should have a peak. We also expect the link overlap generally have a positive relationship with link weights [10]. These expectations are shown in the last column of the table although these are debatable. We find that the ND and LD models reproduce reasonably well most of the general trends of the MPC data set and the common sense while the aging model, at least on its own, does not do that. Hence we conclude that overall the MPC data are quite successfully reproduced by the ND or LD models. --- Discussion In this paper, we have investigated how the mechanism of link termination affects the emerging network structures by introducing and comparing three different models, namely node deletion , link deletion , and aging models. First of all, we would like to emphasize that the Granovetterian structure is robustly observed for all three models. It supports the observation by the previous studies that cyclic and focal closures play key roles in generating the Granovetterian structure. On the other hand, there are significant differences between the ND, LD and aging models. We find that aging promotes modular structure in networks having the Granovetterian structure. This is because inter-community weak links are often deleted while strong links existing within communities need a longer time to be removed, thus making the communities more persistent under the aging than under ND or LD. It is also notable that the aging makes link weights in each community more homogeneous. In contrast, with link deletions assumed in the ND or LD models communities are less modular and links have variable weights. With the LD model it is even harder to maintain high modularity than with the ND model because the density of links in a community tends to decrease more rapidly with LD. Local network properties for these three models, such as degree distribution, node strength distribution, and link overlaps, have also been investigated and compared with the MPC data set observations. The results show that the ND and LD models reproduce the MPC data set while the aging model shows different profiles due to its highly modular structure with homogeneous links. This implies that abrupt link deletion is the major mechanism for deleting ties in real society, while link aging, being presumably present, plays a minor role. The difference between the ND and LD models are less clear, but we think that the LD model describes the reality better than the ND model. Degree and node strength distributions for the LD model do not show a monotonically decaying behavior but have peaks at characteristic scales. This matches better to our common sense that each person has some amount of resources devoted to communication although we do not have conclusive evidence from empirical data yet. Furthermore, an empirical analysis has shown that the relationships terminate with a significant rate even with individuals who have had a long term relationship [14]. Thus, the LD model would be a good starting point for a simple model of social networks. In this paper, we have focused on non-overlapping communities but in reality, the communities may be overlapping as discussed in [19]. In order to realize such overlapping communities, we have recently proposed a multilayer model, where the network in each layer is generated in a similar way to the original weighted social network model [13]. In that case the node deletion takes place in each layer independently, which corresponds to an event that a node leaves one of its communities. This may also sound a reasonable assumption since we sometimes lose links not only to a person but collectively to his or her close friends. Although we expect most of the properties investigated in this paper to hold for the multilayer model, further studies on multilayer models would be interesting. In reality, the link termination process happens in more complicated and various ways and our model is at best a simplification. Although we assume that the link does not recover once it is removed, more than half of the real conflicts are reconciled [14]. Furthermore, the probability of link deletion should also depend on the age of the friendship [20]. In Twitter data set, it was observed that "unfollowing" behavior depends on the local network topology such as reciprocity or existence of common friends [21]. As these aspects are missing in our models the future investigations on these factors are expected to make the model even more realistic. Two aspects should be emphasized finally. First, when we compare our models with data, we have to restrict ourselves to the available sources, which means single channel communication records. This is an obvious source of bias, as multiple channels are crucial in human communication, which may change some of the statistics qualitatively. Second, the three different mechanisms separately treated here in disjunct models are simultaneously present in a real human society. The advantage of the present study is that we have gotten insight about their role in the formation of the social network. --- All relevant data are within the paper and its Supporting Information files.

In social networks of human individuals, social relationships do not necessarily last forever as they can either fade gradually with time, resulting in "link aging," or terminate abruptly, causing "link deletion," as even old friendships may cease. In this paper, we study a social network formation model where we introduce several ways by which a link termination takes place. If we adopt the link aging, we get a more modular structure with more homogeneously distributed link weights within communities than when link deletion is used. By investigating distributions and relations of various network characteristics, we find that the empirical findings are better reproduced with the link deletion model. This indicates that link deletion plays a more prominent role in organizing social networks than link aging.

Introduction and background The same event can be described in many different ways, according to who reports on it, and the choices they make. They can opt for some words rather than others, for example, or they can use a passive rather than an active construction, or more widely, they can -consciously or not -provide the reader with a specific perspective over what happened. Such choices do not just pertain to the realm of stylistic subtleties; rather, they can have substan-Figure 1: "Cyclist slams into car door" Figure 1: "Car driver opens door and hits cyclist" Figure 1: "Cyclist injured in road accident on 5th Street" Figure 1: "Collision between bike and car" We use alternative captions to illustrate how the same event can be described from alternative perspectives, which can evoke different perceptions in the attribution of responsibility to the actors involved. tial consequences on how we think of -or frameevents. Indeed, it is known that the way a piece of news is written, especially in terms of perspectivetaking, heavily influences the way readers perceive attribution of responsibility in the events described . Figure 1 1 illustrates how the same event can be reported on from different viewpoints, in ways that do affect the perception of the participants' responsibilities. We are interested in unpacking responsibility attribution using NLP tools in the context of a socially relevant phenomenon, namely gender-based violence . Violence against women is worryingly common and therefore often reported in the news. A re-1 Drawing inspired by the illustration in https: //www.outsideonline.com/culture/opinion/ look-you-open-your-car-door/ . --- arXiv:2209.12030v1 [cs.CL] 24 Sep 2022 port by the European parliament details an estimate of 87,000 women intentionally killed in 2017. While Italy is listed in this report as one of the European countries with the lowest number of femicides, they are still too frequent and have been constant in the last 25 years . Most discouragingly, a report from November 2018 by two Italian research institutes points out that the stereotype of a shared responsibility between the violence victim and its perpetrator is still widespread among young generations: "56.8% of boys and 38.8% of girls believe that the female is at least partly responsible for the violence she has suffered" . Working on Italian news, Pinelli and Zanchi observe that in descriptions of femicides, the use of syntactic constructions with varying levels of transitivity -from transitive active constructions on one side of the spectrum, via passives and anticausatives to nominalization constructions on the other side -corresponds to various degrees of responsibility attributed to the perpetrator. For example, while "he killed her" makes the involvement of an active agent fully explicit, with "she was killed " the event is accessed via the patient shifting attention away from the agent, and expressions such as "the murder" or even "the event" moves both participants to the background. In a related contribution, Meluzzi et al. investigate the impact of argument structure constructions on responsibility attributions by means of a survey on artificially-constructed GBV reports in Italian. Their results further confirm the findings of Pinelli and Zanchi on the effects of readers' perception on the agentivity and responsibility of the perpetrators and the victims. The outcomes of both studies is in line with previous work in psycholinguistics showing that in events involving violence , the linguistic backgrounding of agents hinders their responsibility and promotes victim blaming . Based on such framing choices, how will the general reader perceive the described event? Can we model such perceptions automatically? In this paper we aim to answer these questions, still focusing on descriptions of femicides in Italian news, and exploiting frame semantics as a theoretical and practical tool, as well as most recent NLP approaches. Using specific pre-selected semantic frames, automatically extracted using a state-of-the-art semantic parser , we identify descriptions of GBV events from Italian newspapers. On these descriptions we collect human judgements through a large-scale survey where we ask participants to read the texts and ascribe a degree of perceived responsibility to the perpetrator, the victim, or to some more abstract concept . More details are provided in §2. Next, we model perception of responsibility automatically by developing a battery of regression models exploiting a variety of linguistic cues which range from surface to framebased features. The training objective of such models is the prediction of the human perception scores. We achieve a strong correlation with a transformerbased model. The fine-grained character both of the survey and the result analysis that we conducted also allows us to observe differences in prediction complexity for the various aspects that we consider. Modeling and evaluation are discussed in §3. The results we obtain show that different linguistic choices do indeed trigger different perceptions of responsibility, and that such perceptions can be modelled automatically. This finding not only confirms previous research which was conducted on a much smaller scale, but also opens up the possibility to conduct large-scale analyses of texts exposing to both producers and consumers of texts which perspectivization strategies are at play and their effects.2 --- Femicide perception dataset We designed an online questionnaire study in which participants were presented with sentences extracted from the RAI Femicides Corpus , a collection of 2,734 news articles covering 937 confirmed femicide cases perpetrated in Italy in 2015-2017, and asked to rate the level of agentivity and responsibility expressed in each sentence. The results of the questionnaire demonstrate a clear effect of semantic frames and syntactic constructions on the perception of descriptions of femicides. --- Question formulation The level of responsibility ascribed to event participants can be expressed in multiple ways triggering different perceptions in the readers. Since responsibility is a complex concept, we break it down into three dimensions in order to make it more understandable for our participants, and to get a more nuanced picture of readers' perceptions. The three dimensions are: Table 1 shows hypothesized ratings on these dimensions for a number of artificial examples, demonstrating that the three dimensions are closely related, but do not always match: for example, the first and second sentences both focus on the role of the murderer and describe his actions as the cause, but the second sentence arguably attributes less blame to the murderer by describing him as 'blinded' by jealousy, implying that he does not bear full responsibility to his actions. Note that the ratings presented in the table merely represent a hypothesis about how the sentences are likely to be perceived; perception is inherently subjective and these examples should not be taken as a 'gold standard' of any kind. To put the amount of responsibility attributed to the murderer in perspective, we also asked readers about the perceived level of focus, causation, and blame placed on the victim, an object , a concept or emotion , or on nothing at all. For a given sentence, participants were asked to give ratings on a 5-point Likert scale to each of these categories. Participants also had the option to indicate that the sentence was irrelevant and skip answering it. The full set of questions is given in Table 2. Note that, taking into account preliminary results from a pilot study, the categories have been adapted slightly to each individual question: for example, we omitted the 'none' category for the focus dimension , and in the 'cause' dimension we made the descriptions of each category slightly more elaborate. --- Sentence selection Relevant sentences were extracted from the corpus following a two-step process: First, occurrences of semantic frames were automatically extracted using the LOME parser . This information was combined with an automatic dependency parse using SpaCy to classify syntactic constructions. For example, he murdered her would be classified as "KILLING:active" , she died as "DEATH:intransitive", and the tragedy as "CATASTROPHE:nonverbal". 3In a second step, we selected typical frames that encode possible ways of expressing the murder event with various degrees of emphasis on the various participants, and randomly sampled sentences containing at least one of these frames. Typical frames were selected by manually annotating the example sentences from Pinelli and Zanchi with FrameNet frames, and selecting the frames evoked by words that refer to the event of the death of the victim . This yielded the set of frames {KILLING, DEATH, DEAD_OR_ALIVE, EVENT, CATASTROPHE }, all of which can be used to describe exactly the same event but with different levels of dynamism , agentivity , and generality . We excluded frames that refer to events that are related to but distinct from the murder itself, such as CAUSE_HARM and USE_FIREARM , or OFFENSES . We then sampled sentences from our corpus in such a way that we created a corpus with an equal num- ber of examples of each frame-construction pair, and equal numbers of headlines and body-text sentences. --- Practical implementation Given the considerable cognitive load of analyzing sentences as well as the emotional load of reading text about a heavy and distressing topic, participants were asked to provide ratings on only one dimension, for a set of 50 sentences. Furthermore, attempting to find a balance between the depth and breath of our annotations, we decided to set a target of 10 participants for each sentence and each dimension, meaning that 30 participants are needed to fully annotate each block of 50 sentences. In order to distribute participants evenly across sentence sets and dimensions, without knowing the response rate in advance, we created 60 groups and assigned participants to groups on a rolling basis: one group was open at a time, and once the required number of participants was reached, it was automatically closed and the next group was opened. Once a group was full, we manually inspected the responses for completeness and quality. Due to the subjective nature of the task, there are no 'wrong' responses per se, but we considered responses to be of low quality if they met at least one of the following three criteria: implausibly fast completion of the questionnaire,4 suspicious patterns of marking sentences as irrelevant and skipping them , or suspicious response pat- terns . The link to the survey platform5 was distributed amongst university students enrolled in bachelor's and master's degrees in different programs at several universities in Italy. Responses were collected anonymously, but participants were asked to state their gender, age, and profession. --- Results Our final dataset covers 400 sentences with ratings from 240 participants in total . In Table 3, a summary of the perception scores aggregated across sentences is given. We give both the mean score , averaged over all participants and all sentences, and the standard deviation of averaged scores across sentences. Overall, the attributes corresponding to the perpetrator tend to have higher average scores but also more variance than the other attributes . More details about the distribution of scores per question and attribute are given in the Appendix. Due to the inherently subjective nature of the task, and in line with previous studies on perceptual norms , we did not calculate inter-annotator agreement scores. Table 4 shows average scores for the focus question, split by typical frame and construction. This shows significant effects: sentences containing the KILLING frame tend to put higher focus on the murderer, and substantially more so when using an active construction. Meanwhile, the use of the CATAS-TROPHE, DEAD_OR_ALIVE, and DEATH frames, as well as the KILLING frame used in an active or passive construction increases the focus on the victim. On the other hand, there were no significant differences in focus scores for the object, and significant but smaller differences in focus on a concept or emotion. In each of these cases, the findings correspond to what we expected based on linguistic theory: if an event participant is lexically encoded in the predicate and syntactically required to be expressed, it is more likely that this participant will be perceived as being under focus. More focus on the murderer and the victim was also expected, both based on the content of the sentences, and on the fact that several frames lexically encode the presence of a victim and/or a killer, but not necessarily that of an inanimate concept or emotion . --- Perception score prediction In this section, we introduce models for automatically predicting femicide perception scores, as well as a suite of evaluation measures for evaluating these models. We model our task as a multi-output regression task: given a sentence S, we want to predict a perception vector p, in which every entry p i represents the value of a particular Likert dimension from the questionnaire . --- Participant aggregation In order to train a single model that generalizes over individual participants, we first z-score the perception values for each sentence and each participant and then take the average value across participants. Z-scores are calculated separately for each Likert dimension and participant to account for two types of variability: i) within-dimension score intensity preference and ii) between-dimension preference. Type refers to different participants making different use of the score range: depending on confidence levels and other factors, participants might choose to make heavy use of the extremities of the range or concentrate most of their scores in a particular part of the range . Type refers to the possibility of participants having a tendency to always assign higher or lower scores to particular dimensions. For example, some participants may always give a higher score to 'blame on the murderer' vs. 'blame on the victim'. By performing regression towards zscored perception values, we force our models to predict between-sentence variability: we are most interested in predicting how each sentence is perceived relative to other sentences and less in absolute scores since these are highly subjective and depend on many individual biases. --- Metrics We evaluate our multi-output regression problem from several angles. First, we use Root Mean Squared Error to measure error rates. This is complemented by R 2 , which estimates the proportion of variation in the perception scores that is explained by the regression models. R 2 is defined both for each dimension and as an average over dimensions. Next, Cosine measures the cosine similarity between the gold and predicted vectors of perception values and provides an estimate of how well the relations between the dimensions are preserved in the mapping. An alternative interpretation is the Most Salient Attribute metric: we evaluate regression as accuracy on the classification task of predicting which Likert dimension has the highest perception value for each question . For example, if for a particular sentence, "concept" is the highest-scoring dimension for the blame question, this means that "blame on a concept" is more salient in this sentence compared to other sentences. Note that the fact that z-scores were computed individually for each dimension makes a major difference here: the dimension with the highest z-scored value does not necessarily also have the highest absolute value. Similarly to the risks of assigning higher or lower scores to particular dimensions, in this case participants may give more points to "murderer" on the blame question than to "concept", even in sentences where "concept" is very salient. In such cases, "concept" would always have a lower absolute value than "murderer", but might have have a higher z-scored value in sentences where a relatively high score was given to "concept" and a relatively low one to "murderer". --- Models We compare two types of models: ridge regression models trained on different types of input features, and a selection of relevant pre-trained transformer models, fine-tuned for multi-output regression. For reference, we also run a 'dummy' baseline model that always predicts the training set mean for each variable. 6Features For the ridge models, we use a series of feature representations with increasing levels of richness. By comparing models trained on different representations, we gain insights into what kind of information is useful for predicting perception scores. Features are divided into three categories: Surface features represent the lexical content of the input sentences, either with simple bag-of-words vectors, or with pre-trained FastText embeddings .7 By contrast, Frames features are based on the frame semantic parses of the sentence. The first variant, f1, is similar to a bagof-words, but using counts of any frame instances and semantic role instances present in the sentence instead of unigram counts. Variant f2 is similar but includes only mentions of our pre-defined frames-of-interest . Moreover, f1+ and f2+ are versions of f1 and f2 that concatenate the bag-of-frame features to the unigram features from bow. Finally, Sentence features are transformer-derived sentence-level representations. SentenceBERT uses representations derived from XLM-R ;8 BERT-IT Mean and XLM-R Mean use last-layer representations, averaged over tokens, from Italian BERT XXL and XLM-R, respectively. Transformers We also implement a neural regression model that consists of a simple linear layer on top of a pre-trained transformer encoder. 9 We experiment with several variants of BERT with different pretraining corpora and model sizes. Italian BERT XXL Base is a base-size monolingual BERT model trained on the Italian Wikipedia and the OPUS corpus; BERTino is a distilled version of this model. We compare these with Multilingual BERT Base and Multilingual DistilBERT , trained on concatenated Wikipedia dumps for 104 language, and XLM-RoBERTa Base , trained on CommonCrawl data for 100 languages. We use cased models in all cases. Implementation Ridge regression models were implemented using scikit-learn . Transformer models were implemented using Huggingface Transformers . We split the dataset into 75% training and 25% test data. We used 6-fold cross-validation within the training set to search for hyperparameters : α for ridge regression; initial Adam learning rate and weight decay for transformers. The parameters with best performance across folds were then used for training the final model. --- Results Table 5 shows the main results on the test set for the RMSE, COS and R 2 metrics. Strongest results are obtained with the fine-tuned monolingual BERT models across all measures, with an overall R 2 scores around 0.45, meaning that these models explain almost half of the observed variance in perception scores. The multilingual BERT models perform consistently worse, with an average R 2 of 0.38 or below. Interestingly, we observe a drop in performance between the fullsize and distilled models for mBERT, but not for the monolingual Italian BERT, where BERTino even performs slightly better than the original model. Drops in R 2 do not always align with drops in cosine scores: for example, XLM-R scores 0.06 R 2 points lower than BERT-IT/base, but the cosine score drops by only 0.01, while mBERT/dist loses 0.10 points on R 2 and 0.09 on COS. Thus, it appears that some models are less accurate at predicting the exact magnitude of perception scores but relatively good at capturing the overall score pattern across dimensions. While the ridge regression models perform substantially worse than the transformer models, comparing the results between different feature representations is insightful for understanding what information is needed to predict perception: the Surface and Frames models all perform similarly with R 2 scores around 0.20 , while the models with Neural features perform better . Simple counts of unigrams and frames give very similar overall scores; concatenating these features leads to a small improvement . This suggests that frames are useful for summarizing relevant lexical material , but that the additional information about semantic and syntactic structure that is provided by role and construction labels does not lead to substantial gains. Using FastText embeddings instead of unigrams does not lead to gains, either. Meanwhile, comparing ridge models trained on transformer-derived features, we find best results with mean last layer representations from Italian BERT , with slightly lower scores for the two models based on XLM-R ; surprisingly, Sentence-BERT does not seem to have an advantage over averaged last-layer representations . Comparing R 2 scores across different questions and attributes reveals large differences in difficulty of prediction: for example, blame on murderer gets good scores across models, while blame on victim has relatively poor scores even for the strongest models , and at-baseline scores for the weaker models -notably, distilled mBERT, which performs decently on other attributes. Caused by no-one is even harder to predict, with no model scoring above 0.10. The Focus question has the overall best and most consistent performance, especially for the Italian BERT-based models, which achieve decent performance for each of the four attributes. This pattern is also reflected in MSA : for Focus, it is substantially easier to predict the dimension with the highest score than for Blame and Cause. However, all models perform well above chance level for each of the questions, with the strongest overall scores for BERTino . The gain in performance achieved by the BERTbased models with respect to the surface feature models varies substantially between attributes. For example, the bow model has a surprisingly high score for blame on murderer , with only moderate gains from the BERT-IT and BERTino models . By contrast, bow scores poorly on focus on concept , whereas BERT-IT and BERTino have good scores . To get additional insight into the differences between models, we performed a feature attribution analysis. For the bow and f1+ ridge regression models, we simply extracted the feature weights with the lowest and highest absolute values; for transformers, we applied the integrated gradients interpretation method 10 to obtain token-based attribution values for all sentences in the test set, and used the averaged values for tokens above a frequency threshold as an approximation of the overall feature importance. The results for blame on murderer and focus on concept are shown in Table 7. For blame on murderer, all three models seem to focus on similar lexical items: for example, "uccide" kills') has a high positive attribution value in both the bow ridge regression and the finetuned BERTino model, and in f1+ we find a positive score for the KILLING frame, which is an abstraction over killing-related words. We also find that personal relationships get positive attributions in all three models. By contrast, we find negative attribution values for "accaduto" and the corresponding EVENT frame in bow and f1, which maps neatly onto our observations discussed in §2.4. For focus on concept, no insightful differences between the three models are immediately obvious. We do find several intuitively relevant features in each model: "passionale" and "femminicidio" could to examples of concepts that sentences could give focus to, whereas "omicida" and "killer" could be seen as emphasizing the role of a human agent rather than an abstract concept. --- Conclusion & Future Work This paper has presented a detailed analysis of human perceptions of responsibility in Italian news reporting on GBV. The judgments we collected confirm the findings of previous work on the impact of specific grammatical constructions and semantic frames, and the perceptions they trigger in readers. On the basis of the results of our survey, we have investigated to what extent different NLP architectures can predict the human perception judgements. The results of our experiments indicate that finetuning monolingual transformers leads to the best results across multiple evaluation measures. This opens up the possibility of integrating systems able to identify potential perception effects as support tools for media professionals. In the future, we plan to run a more detailed analysis of the data considering differences along individual and demographic dimensions of the respondents. In addition to this, natural follow-up experiments will focus on the application of the approach to other languages and cultural contexts both targeting GBV as well as other socially relevant topics, e.g. car crashes . --- Ethics Statement Limitations This work has a strong connection with multiple theoretical frameworks: Frame Semantics, Construction Grammar, and Critical Discourse Analysis. The way we have structured the questionnaire aimed at collecting data from human participants with respect to different sentenceswhich in different ways contained variations in syntactic structures and semantic frames that could be linked to findings and claims about the "perception" and its effects in the interpretation of sentences. The use of state-of-the-art NLP tools to identify these properties in a large collection of data represents both an advantage and a limit . While representing an unicum in the language resource panorama, since there are no previous comparable and available corpora, the number of available sentences used to train the models is somewhat limited. The final corpus, however, represents an optimal compromise between number of judgements needed to obtain a solid representations of perceptions by users and number of data points that could be used by stochastic NLP architectures to learn from the data. Finally, the outcome of the perception judgements can be generalized to the population of Italian young adults attending universities . This is a limitation of the data collection process. We tried to minimize this by reaching out to students in multiple universities and at different faculties and disciplines . --- Data collection The questionnaire was conducted using the Qualtrics XM platform. Participation to the questionnaire was on a voluntarily basis. Participants could interrupt their participation in any moment. Only fully completed questionnaires have been retained. Participants received compensation -upon completion of the questionnaire. Participants have been recruited mainly among undergraduate students at different universities in Italy. Participation was fully anonymous: 1) partici-pants could access the questionnaire via a unique special access token that could be obtained by filling in a form; 2) no personal information other than the participants' email address was stored; 3) IP addresses were not stored or tracked; 4) the special access token and the participants' email were decoupled. Participants could receive their compensation only by providing the unique access token. --- Dual use The experiments we have run investigate to what extent models are able to predict human perceptions along three dimensions with respect to GBV. The very nature of the task limits the potential misuse by malevolent agents. At the same time, malevolent agents can purposefully misrepresent the results to minimize the negative aspects associated to the reporting of the phenomenon by media. By making the models and the data publicly available, together with a detailed explanation of how the models work and how results should be interpreted in a correct way, we mitigate these risks. Intended use As it is the case for supervised models, sensitivity to the training material is high. At the moment, we have not tested the portability of the models to other topics. We do recommend to use these models only on data compatible with the phenomenon we have taken into account, i.e., GBV against women. Although the application of the models to any other type of texts reporting violence and killing against other targets may still give some valid results, we discourage its use since risks of unforeseen behaviors are high, with potential harmful consequences for the victims of violence.

Different linguistic expressions can conceptualize the same event from different viewpoints by emphasizing certain participants over others. Here, we investigate a case where this has social consequences: how do linguistic expressions of gender-based violence (GBV) influence who we perceive as responsible? We build on previous psycholinguistic research in this area and conduct a large-scale perception survey of GBV descriptions automatically extracted from a corpus of Italian newspapers. We then train regression models that predict the salience of GBV participants with respect to different dimensions of perceived responsibility. Our best model (fine-tuned BERT) shows solid overall performance, with large differences between dimensions and participants: salient focus is more predictable than salient blame, and perpetrators' salience is more predictable than victims' salience. Experiments with ridge regression models using different representations show that features based on linguistic theory perform similarly to word-based features. Overall, we show that different linguistic choices do trigger different perceptions of responsibility, and that such perceptions can be modelled automatically. This work can be a core instrument to raise awareness of the consequences of different perspectivizations in the general public and in news producers alike.

For half a century, health communication campaigns have been at the forefront of efforts to reduce the burden of tobaccorelated diseases . Although tobacco use remains the most preventable cause of death in the United States , cigarette use has steadily declined over recent decades with evidence pointing to effective communication campaigns as a substantial contributor to that success . Despite this accomplishment, the US is far from free of the health and economic costs associated with the tobacco epidemic. New research suggests that decades of permissive indoor smoking norms have left a legacy of persistent toxic pollutants . This neglected form of tobacco exposure, called thirdhand smoke , perpetuates public risk of tobacco-related disease and death. THS is the toxic chemical residue left in indoor environments after the smoking stops . As scientific understanding of THS grows, calls for THS prevention and educational materials also increase . Health communication campaigns are needed to bridge gaps in public awareness of THS-related risks in order to promote behavior change and implement policies to reduce exposures. Building on past tobacco prevention campaigns and employing the tenants of McGuire's input-output model, this study tested the effects of social media-based campaign messages on improving THS-related knowledge, attitude, efficacy, and behavior in California adults. Changing these perceptions is central to reducing exposure to tobacco-related pollutants in indoor environments, thereby reducing the burden of tobaccorelated disease and death. --- Thirdhand smoke exposure THS is the toxic residue left behind from the smoke of cigarettes and cigars, . THS sticks to surfaces, accumulates in dust, and becomes embedded in everything from carpets and furniture to toys and pillows ). The residue, which contains a mixture of toxic chemicals, including several known to cause cancer and asthma , persists in indoor environments for months to years after smoking stops ). Humans are unintentionally exposed to these chemicals through skin absorption, breathing, and ingestion . Studies have found exposure to THS directly damages DNA, induces oxidative stress, and changes reproductive cell function . Children are most at risk for negative health effects . Despite the growing evidence of the harms of THS exposure, social scientific research has found that the public is unaware of the harms of THS exposure. People recognize THS by the odor of stale tobacco smoke or tobacco stains on walls. However, most people do not make the connection between these experiences and the term THS or the potential negative health effects of exposure. In focus group discussions of smoke-free homes, Escoffery et al. found that most of their participants had not heard of the term THS. Similarly, in a survey of over 1,400 US adults, Winickoff et al. found that only 65.2% of nonsmokers and 43.3% of smokers believed that THS could harm children. Given these gaps in public knowledge and the potential for harm, public health education efforts, such as health communication campaigns, are needed to increase public awareness of the presence of, and risks of exposure to, THS. --- Tobacco prevention campaigns A large body of research demonstrates the success of health communication campaigns in reducing smoking-related death and disease . More specifically, research has credited health communication campaigns with helping to lower the US smoking rate , reduce youth initiation , increase quit attempts among smokers , change beliefs about smoking , reduce the risk of smoking relapse , and increase tobacco-free policy compliance . After many decades of successful print and traditional media-based approaches, such as television and radio spots, tobacco prevention campaign approaches have moved to online and social media platforms. Research has supported the effectiveness of online tobacco prevention campaigns for increasing cessation-related information seeking , outreach and dissemination efforts , access for hard-to-reach populations , and perceived risk of smoking . Thus, online platforms can be effective campaign channels for reaching target audiences. As for what messages to disseminate online, best practice calls for developing persuasive approaches that follow theoretical and evidence-based guidelines . --- Theoretical approaches to communication campaigns McGuire's input-output model was proposed as a theoretical explanation of the persuasive process surrounding mass media campaigns. The model follows a stepwise approach positing that evidence-based decisions of five key inputs will trigger ten outputs as products of the persuasive process . More specifically, the model proposes that campaigns are most effective when evidence-based decisions guide [1] identification of a focused and tangible goal, [2] use of credible information, [3] careful message creation with the target audience, [4] audience-based channel selection, and [5] audience segmentation . Through these five central strategies, sequential persuasive processes occur beginning with audience exposure and leading to increases in attention, interest, and knowledge. Following the occurrence of these processes, changes in attitude occur and influence outputs: memory storage and retrieval, behavior, and continued behavioral performance . In determining which outputs to assess, researchers typically turn to other theories or key findings in the literature to identify which outputs are most appropriate for their persuasive goals . In health behavior change research, the most common constructs examined to evaluate behavior change are knowledge, attitude, and efficacy. Consistent with the early presence of knowledge in McGuire's output list, Maibach and Cotton articulate the requirement of a sufficient amount of knowledge to make perception-based and behavioral changes. Logically, improving knowledge will be most important for less-known health topics, such as THS. In addition to knowledge as a key persuasive construct, theoretical reviews consistently find the constructs of attitude and efficacy are among the most common in theories and models focused on the processes surrounding health-related behavior change . The theoretical emphasis on the constructs of knowledge, attitude, and efficacy is reinforced across decades of applied research. Recent tobacco prevention research echoes the theoretical assumptions of knowledge, attitude, and efficacy as key factors in behavior change. Studies have shown knowledge of smoking-related harms is higher among nonsmokers than smokers with knowledge positively associated with increased cessation and reduced intention to smoke . Similarly, studies have shown attitude is a determinant of intention to smoke and active smoking and efficacy is a determinant of cessation and intention to smoke . Such consistent findings surrounding the role of knowledge, attitude, and efficacy suggest that these variables will be key determinants of THS-related behavior. Thus, this THS awareness campaign sought to change the outputs of knowledge, attitude, and efficacy to promote positive behavior change. --- Purpose The purpose of this quasi-experimental study was to evaluate the effectiveness of a public health campaign at increasing THS awareness, as measured by knowledge, attitude, efficacy, and behavior. Two hypotheses were tested. The first hypothesis sought to evaluate the general extent to which the primary outcomes changed from the beginning until the end of the campaign. H 1 : Knowledge, attitude, efficacy, and behavioral intention related to thirdhand smoke will increase from baseline to the final posttest. Assuming at least part of H 1 would be supported, this study also posed a research question to explore the extent to which particular campaign message were persuasive. RQ 1 : Which campaign messages will be most positively associated with changes in primary outcomes? Finally, building on past research and theoretical expectations, the second hypothesis sought to test the role of campaign recall in the behavior change process. The tested model with hypothesized paths is displayed in Figure 1. --- Method --- Procedures & message development The goal of the campaign was to increase THS awareness among California adults. Given the lack of awareness surrounding the science of THS, the target audience can be understood as an uninvolved audience that is not actively seeking information and is unlikely to be aware of the personal relevance of the topic . Louis and Sutton suggest three persuasive approaches to engage an uninvolved audience. The most applicable to the context of THS is the presentation of content that shows a discrepancy between expectations and reality . This discrepancy was emphasized by highlighting the unexpected reality of THS risk factors, exposure routes, and impacts on people, pets, and property through an emphasis on children, traveling, pets, and enclosed environments, such as cars, apartments, and homes. Because the majority of US adults use social media, and 69% of those use Facebook , campaign messages were disseminated over a sevenmonth period as advertisements on the Facebook platform. Message development followed the guidelines outlined in Atkin and Freimuth . Following a thorough review of the literature on effective tobacco prevention strategies for the target audience, we developed fifteen potential campaign messages that were iteratively reviewed by content and persuasion experts. The final draft messages were focus group tested with members of the target audience in the Los Angeles area; participants received a 50 USD gift card in appreciation for their time. Based on focus group feedback, the fifteen messages were reduced to seven, which can be seen in Figure 2. These messages were formatted to fit the look and advertising requirements of Facebook. --- --- Intervention At the conclusion of recruitment, the seven campaign messages were posted as Facebook advertisements daily for four months using the same targeting terms as the recruitment advertisement . The advertisements were run at the same rate in Facebook, meaning each message was pushed equally by its algorithm. The budget for the campaign messages was 10,500. USD After four months, the messages recorded 1.89 million impressions, reached 679,553 users, and had 15,752 advertisement clicks. The conversion rate of advertisement clicks from reach was 2.3%. When clicked, the campaign messages directed users to pre-determined pages on the Thirdhand Smoke Resource Center website with relevant content . Each month, the 1,087 participants received an invitation to complete the next survey. --- Measures All seven surveys included items to assess social media use, primary outcomes , and reported THS exposure. During the campaign, surveys also included items to assess message recall. Demographic characteristics were collected at baseline. Table 1 presents demographics and primary outcomes for each wave. --- Knowledge The development of the adapted eight-item knowledge measure is described in detail elsewhere . Response options were on a 5-point Likert-type scale from strongly disagree to strongly agree. Example items include: Thirdhand smoke contains dangerous chemicals; Thirdhand smoke can linger in hotel rooms where guests have smoked; and Thirdhand smoke can make kids sick. The scale total score means and standard deviations ranged from m = 4.38 to 4.50 and SD = 0.58 to 0.67 across the seven measurement waves. --- Attitude To assess attitude, seven attitude items were adapted from established SHS and THS measures . Example items are: Hospitals should hire only nonsmokers; Sellers should be required to disclose if someone has smoked in their home; and Childcare providers should be nonsmokers. Response options were on a 5-point Likert-type scale from strongly disagree to strongly agree. Across the seven measurement waves, the scale total score --- Efficacy Following Bandura , efficacy was conceptualized as the ability to successfully avoid THS exposure. Four items were adapted from Sherer et al. . Response options were on a 5-point Likert-type scale from strongly disagree to strongly agree. All items responded to the stem, I am able to. Example items include: avoid exposure to thirdhand smoke; determine if a place is smokefree; and ask people not to smoke around me. Principal component factor analyses revealed high first-factor saturation at six of the seven waves, supporting single-factor loadings and explained variance above 50% . Similarly, across all seven waves, Cronbach's Alpha supported low, but acceptable measure reliability . Scale score distributions were approximately normal, and the scale total score means and standard deviations ranged from m = 3.57 to 3.77 and SD = 0.85 to 0.91 across the seven measurement waves. --- Behavior Five items to assess behavior were adapted from established measures of preventative behaviors related to SHS . This adaptation included items that assess both behavioral intention and actual behavior. Examples include: I would buy furniture from a smoker; I would buy a car that has been smoked in; and In general, I avoid places where people have smoked. Response options were on a 5-point Likert-type scale . Across the seven measurement waves, the scale total score means and standard deviations ranged from m = 3.78 to 3.86 and SD = 0.94 to 1.01. --- Campaign recall The approach to assessing campaign recall follows Record et al. . Participants were shown all seven campaign images and asked to select which they recalled seeing over the last month and, for those they recalled seeing, how frequently they recalled having seen them from once or twice to every day . Participants who did not report that they recalled seeing the images were automatically coded as "never" seeing the advertisement. Thus, the final measure was a five-point scale ranging from 0 to 4 with individual message recall ranging from m = .02 to m = .15 ; see Figure 2 for individual recall for each message. --- Data analysis Participants with six or more missing item responses across the measures of knowledge, attitude, efficacy, and behavior were excluded from the within-wave analysis. Using this criterion, six participants had insufficient data and were excluded from a within-wave analysis. For participants with fewer than six missing responses, hot-deck imputation in Stata was used, replacing missing values via variable matching based on within wave gender and smoking status . The approach replaces the missing data with a randomly select response from the matched options. Reported regression analyses were completed in Stata; reported path analyses were run in AMOS ; all other analyses were performed in IBM SPSS . To explore change over time across the primary outcomes, a mixed linear random effects model, where subjects were the random factor and time was a fixed factor, was run with all seven time points controlling for demographic factors. To account for change over time, covariance structures were modeled with an autoregressive residual structure. For all regressions in which the omnibus effect of time was significant , polynomial contrasts testing linear, quadratic, and cubic relationships were tested to probe for nonlinear associations. D-effect sizes were calculated comparing the means of waves 2 through 7 to wave 1 using the square root of the pooled variance across all 7 waves. Overall change in knowledge, attitude, efficacy, and behavior, was assessed by subtracting composite mean scores for the primary outcome variables at time 1 from time 7, with positive scores indicating positive change. To assess message exposure, participant recall for each message was summed across all seven timepoints. Finally, path analysis following maximum likelihood estimations was modeled to explore theoretically expected persuasive routes. --- Results Among the 1,087 participants, 124 completed all seven waves of data collection with follow-up participation ranging from 238 to 326 participants per wave. Preliminary analyses found that prior to the study, 63.5% of participants had not heard of the term THS. Compared to those without prior awareness, independent samples t-test did not find individuals with prior awareness of THS to differ regarding changes in attitude, knowledge, efficacy, and behavior. Select demographic data are presented in Table 1; perceptions of THS and personal exposure in the past 30 days are presented in Table 2. Regarding social media use, at each wave, participants logged into Facebook the majority of the last 30 days , with most participants believing they were either just as active or more active than their peers on Facebook . Pearson r correlations showed a weak nonsignificant association between Facebook use and recall of campaign messages. The first hypothesis expected knowledge, attitude, efficacy, and behavioral intentions related to THS exposure to improve from the first wave to the last wave. For each outcome variable, a mixed linear regression model was run. As shown in Table 3, smoking history, prior awareness of THS, and age were significantly predictive across the four outcomes with race and gender predictive of efficacy and behavior . As shown in Figure 3, analyses found a significant omnibus effect of time on knowledge following a cubic relationship , on attitude following a positive, although nonsignificant, linear relationship , and on efficacy linearly . Behaviors did not significantly change over time. Thus, H1 a-c were supported and H1 d was not supported. To explore the extent to which each campaign message was individually persuasive, associations between individual message recall and change in the primary outcomes were examined. Recall varied between messages from m = .02 to m = .15 . Pearson r correlations found changes in knowledge to be weakly but positively associated with exposure to Another Cold , Pets , and Dust . Similarly, changes in attitude were weakly but positively associated with Wave sample sizes are provided in Table 1. Questions asked each wave; *Response options on a 5-point scale from strongly disagree to strongly agree; + Response options yes or no; a Data collected when the COVID-19 stay-at-home orders were in effect. exposure to the messages Another Cold , Pets , and Dust . Changes in efficacy and behavior were not positively associated with exposure to any particular message. Thus, messages featuring THS impacts on the frequency of colds in children, negative health effects for pets, and the presence of THS in dust were more persuasive in promoting knowledge and attitude change than the other four messages. Finally, the second hypothesis tested a model that expected campaign recall to predict changes in knowledge; changes in knowledge would predict changes in attitude and efficacy; changes in attitude and efficacy would predict changes in behavior. To accurately test expected persuasive routes, campaign recall was used for the three most persuasive advertisements . Five fit indices were evaluated following Kline . Three fit indices supported satisfactory model fit compared to a worst fitting or no model with expected covariance residuals . The overall chi-square model fit statistic was statistically significant indicating poor model fit with respect to a best possible fitting model. Similarly, the RMSEA suggest poor fit . Parameter estimates found all expected relationships to be statistically significant except that efficacy did not predict behavior. Thus, H2 a-d were supported and H2 e was not supported. --- Discussion This quasi-experimental study evaluated the effectiveness of a public health campaign on improving THS awareness, as measured by knowledge, attitude, efficacy, and behavior in a sample of 1087 California adults. The campaign design built on past tobacco prevention materials and was evaluated following theoretical expectations of the persuasive process. Results demonstrated significant change over time in THS-related knowledge, attitude, and efficacy but did not show a significant change in behavior. However, behavior change, which included measures of intention, was indirectly associated with participant recall of campaign messages: recall of the most persuasive messages predicted change in knowledge, which in turn predicted change in attitude and efficacy, and change in attitude ultimately predicted change in behavior. The intervention producing positive change in knowledge is consistent with past meta-analytic findings of the effectiveness of mass-media health campaigns; however, the positive change in efficacy and finding a lack of significant change in behavior is counter to the findings from the meta-analysis . The difference in these findings could be due to their analysis including all health-related campaign topics. The findings from the present investigation have numerous implications for practice and theory. Tobacco control advocates and researchers, particularly those in areas with lower tobacco use prevalence such as California, are working toward the "tobacco endgame," which envisions a world free of the commercial tobacco epidemic . Bringing smoking prevalence to near-zero levels and removing all forms of exposure to toxic tobacco residue will require substantial behavior change and intensified regulations. This would include expanding existing tobacco control policies that fail to protect the public from THS exposure in private settings associated with real estate transactions, rental contracts, used car sales, and other personal property transactions. Implementation of new policies that extend to such settings will necessarily require widespread public knowledge of the persistent risks of tobacco use and high levels of public support for regulatory action. In this context, communication campaigns attempting to change public awareness of THS are a critical area of tobacco research. Findings from this study suggest that progressive tobacco prevention can benefit from using strategic communication approaches from past tobacco prevention campaigns , especially those that leverage the persuasive topics of children's health, pets, and dust. As technology evolves, campaign approaches need to adapt to trends in mediated communication platforms . Consistent with recent approaches to tobacco prevention campaigns , the present campaign leveraged the modern social media platform of Facebook, and the results support Facebook as a successful dissemination platform. Campaign recall was generally low for participants, which is not unexpected in massmedia campaigns. Although low campaign recall suggests that the Facebook algorithms did not perfectly reach the recruitment group, targeting was efficient enough to support persuasive outcomes. Relying on corporate social media algorithms can be risky, especially in unstable political climates where advertising regulations can change with little notice. However, flexible campaign designs should continue to engage social media platforms as a means to effectively reach large target audiences. Consistent with theoretical expectations for health-related behavior change and past research , findings from this study support that changing knowledge and attitude is a key factor in changing behavior. Interestingly, efficacy was not found to be significantly related to THS prevention behaviors. Although past tobacco prevention research has found efficacy to be a central construct in tobacco control , the context of THS has a lower degree of personal agency. For instance, deciding to stop smoking is in our personal control, whereas avoiding environments that have been formerly smoked in might feel out of the control of individuals. Thus, the lack of support for efficacy in the persuasive process is more likely a product of the THS context than a misguided theoretical expectation. However, future campaigns that promote greater awareness of personal and public steps that could be taken to reduce exposure may shift perceptions of efficacy and increase support for THS control policies. Such personal behaviors could include washing hands and changing potentially contaminated clothes before holding a baby, while policies could include mandatory disclosure of THS to prospective buyers or tenants within real estate law or expansion of housing code violations to include THS. As a final implication, current tobacco control toolkits, such as those available from the CDC, lack THS-related resources. Given the results of this study, tobacco control toolkits should include THS educational components that leverage the negative impacts of THS exposure, in particular, persuasive messages that emphasize the impact of THS on children and pets as well as the prevalence of THS in dust. Comprehensive tobacco control toolkits that include THS educational materials help move public perceptions in the direction of the tobacco endgame. --- Limitations A few limitations need to be acknowledged. First, the dissemination channel relied heavily on Facebook algorithms. According to Facebook's advertising and algorithm guides, the participants recruited from baseline would be priority targets for future advertisements from the Thirdhand Smoke Resource Center. However, there was no way for the Center to ensure that participants were being prioritized with the advertised messages. Recall measures supported that the messages were appearing on many participant's newsfeeds, although much lower than initially assumed. For participants who reported no recall, it is possible that the messages never appeared on their newsfeed. Second, findings reflect perspectives and behaviors of California adults. As a state with low smoking rates and high tobacco control efforts, these findings might not be the same in other states or regions of the world where smoking rates are higher and tobacco prevention is more challenging. Finally, because participants engaged in monthly testing, the attribution of the observed changes to the media campaign could be threatened via a testing effect. Although this is important to note, the risk is deemed low as only a small portion of the sample completed all seven surveys and were at risk of such an effect. --- Conclusion Future research should build on the findings reported here to continue to educate the public on THS risks, both on and off social media platforms. As technology changes, platforms that allow greater control over audience targeting will strengthen the validity of future intervention findings. Tobacco prevention advocates should consider including information on THS risks and exposure routes in their educational materials as well as provide resources for preventing and removing THS. Finally, advocates working toward the tobacco endgame of eliminating the commercial tobacco epidemic should include THS educational materials in their campaigns, promotional materials, and briefs. Preparing the public to recognize and remove THS pollutants in their personal environments will be a last important step in creating a world free of toxic tobacco toxicants.

Despite a growing body of research outlining the harms of thirdhand smoke (THS), the public remains generally unaware of risks and exposure routes. This project built on past tobacco prevention campaigns and the tenants of McGuire's input-output model to implement and evaluate a seven-month Facebookdisseminated campaign seeking to improve THS awareness among California adults (n = 1087). Multilinear regression showed that THS-related knowledge (χ 2 [6] = 19.31, p < .01), attitude (χ 2 [6] = 13.88, p < .05), and efficacy (χ 2 [6] = 13.81, p < .05) significantly increased by the campaign's end, with messages highlighting children's health (r = .110, p < .05), pets (r = .145, p < .01), and dust reservoirs (r = .144, p < .01) as the most persuasive. Path analysis modeling found campaign recall to be associated with changes in knowledge (β = .161, p < .01), which predicated attitude change (β = .614, p < .001) and, in turn, behavior change (β = .149, p < .05). Findings suggest social media campaigns should continue to educate diverse populations about new tobacco risks and that tobacco control advocates should consider integrating educational THS messages. CONTACT Rachael A.

INTRODUCTION Library and Information Science Education as a field and area of study is indispensable in this competitive era of knowledge and information management and the technological advancement in the area of ICT has made significant changes in the profession towards achieving the desired objective and to meet up with the modern trends of effective information services delivery. Education in Library and Information Science has taken a new turn in the face of far-reaching developments sparked off by the trend towards globalization. Although the challenges for libraries in Africa have been enormous, they are not insurmountable given our human capacity to adapt to change. It is largely true that libraries in Africa are adapting to change particularly on innovations brought about by the application of ICT. Library and information science professionals are not only having to adapt to change in library services but also require in-depth and structured education and training programmes for occupation engagements in all library types. However, one important aspect of such professional discourses derive from the search for suitable, relevant academic and professional programmes is one that should be made available in our library schools in response to societal needs based on the various library types . is the application of library science which comprises the practical services rendered by librarians in their day-to-day attempts to meet the needs of library patrons or users . Information and communication technology involves the handling of information and communication processes using technology such as computers, internet, multimedia and voice over internet protocol . According to Ekoja information and communication technology is the equipment used for capturing, processing, storing, transmitting and accessing information, which has offered librarians and other information professionals' tremendous opportunities in information handling. Ebijuwa on the other hand defined ICT as tools used for collection, processing, storage, transmission, and dissemination of information. In the view of Achebe , Information and Communication Technology is defined as the type of technology that links the computer to the global telecommunications network to make it possible for users to acquire process, compare, store and disseminate oral, printed and pictorial information. They are the electronic means of capturing, processing, storing and disseminating information. Prior to this, there had been rudimentary training programmes for Library Assistants coordinated by some libraries. There was also Associateship Library Association certificate programme organized by University of London. This afforded some Nigerians, among other Africans, to enroll for the programme which was of two parts. The examination for the first part took place in Nigeria while the second part was at Great Britain . --- EMERGENCE OF LIBRARY AND INFORMATION SCIENCE EDUCATION IN NIGERIA Similarly, Sharr compiled a report in 1963 on the library needs of the Northern Nigeria. Part of the report's recommendations led to the establishment of the library school in Zaria in 1968. By the mid 1970's, both the Ibadan and Zaria library schools could no longer cope with the growing demands for library personnel's training needs of the country, owing to the increasing number of academic and research libraries in the then newly-established universities and similar tertiary institutions including research institutes. Then came a period proliferated with the establishment of library schools across the country, owing to the growing number of higher institutions granted operational license in Nigeria, which propelled a greater need for more personnel. According to Nkanu, Iyishu & Ogar , the strength of the philosophy of Nigeria's educational system, the quality of education at all levels has been oriented towards inculcating appropriate educational aims to which the philosophy is linked. The attainment of these educational aims can be viewed from a broader perspective in the expectation that, for Nigeria education system to fully utilize available opportunities arising from the use of emerging technologies in the information age, the utilization of information and communication technologies must be given top priority. Information and Communication Technology must be seen as one of the most important tools that can positively influence education and all sectors of the society. However, acquisition of information literacy skills will assist library and information science professionals to effectively apply ICT-based technologies to the provision of library and information services. --- Impact of ICT on Library and Information Science Education in Nigeria. The impact of Information and Communication Technology to people generally and students in particular cannot be overemphasized . This is true because ICTs facilitate quick and easy access to a wide range of information/information resources worldwide. In fact, it is now difficult to imagine a world without information technology. In this digital age, tertiary institutions strive to be up-to-date in their curricula. The provision and use of ICT is part and parcel of the entire system, to both the students and the institutions. It is one thing to recognize the importance of ICTs and another to know if they are effectively used by students and academics. If ICTs are put to effective use, the essence of acquiring them is to a large extent justified viceversa. ICT provides libraries with capabilities for the location of information, storage and retrieval of information, and dissemination of information. Internet access enables libraries to locate information stored in other computers around the world. With online search facilities, information stored at different locations can be easily retrieved. Through the use of web pages, e-mail, and CD-ROM, libraries disseminate information. Digitization of library information resources, which converts print resources into electronic form, means that such information can be accessed from homes, offices, or any workstation connected to the Internet. Furthermore, Issa, Amusan, Aliyu & Ladan stressed that librarianship has always depended on the current technology to function effectively. With the increasing use of modern technology for information management, librarianship, like other professions, has never been the same again. There is no doubt that in the face of information explosion, modern librarianship heavily relies on the use of ICT to function. Therefore, in order to keep pace with the global trend in the profession and to enhance job performance, training and education of library personnel should be designed and adjusted to reflect the trends in terms of technological advancement. The influence of Information and Communication Technologies is now manifested in every sphere of human endeavour including the field of Library and Information Science . Abubakar on his part observes that we now live in a changing society where information technology has taken the centre stage and is accompanied by massive increase in knowledge which leads to a new paradigm in the field of Library and information Science. Mohammed in his view noted that the advances in the sciences and in the application of Information and Communication Technologies particularly the Internet, intranet and other network technologies have continued to impact positively on the methodologies of library and information service delivery, education and training of information professionals as well as the area of information seeking behaviour and use. --- DEPRESSED ECONOMY Depressed economy may also mean the same thing with economic recession. The National Bureau of Economic Research , defined recession as a significant decline in economic activity spread across the economy lasting more than a few months, which is normally visible in real domestic product, real income, employment, industrial production and wholesale-retail sales. Depression is regarded as a more severe version of recession. According to Mankiw , depression is a repeated period during which real GDP falls. Many factors contribute to economic depression but inflation is the major cause. This could be as a result of increased production cost, high energy cost and national debt, . In Nigeria today, government has already accepted that the country in undergoing recession owing to drops in government revenue, this is largely due to a drop in oil price in the international market. --- EFFECTS OF DEPRESSED ECONOMY ON LIBRARY AND INFORMATION SCIENCE EDUCATION The consequence of global depressed economy on Nigeria educational system is very devastating. Nigeria being a mono-cultural economy relies mainly on crude petroleum exports. With the declining revenue accruable from crude oil sales at the international market, the survivor of educational sector is threatened. The facilities necessary for effective teaching of library and information science education are not adequately available in most library schools in Nigeria due to the state of the economy of the nation. The problem is aggravated by high cost of these facilities poor internet connectivity, inadequate funding of library schools resulting in poor budgets and their inability to run ICT compliant programmes. --- AREAS OF APPLICATION OF ICT IN LIS EDUCATION Library and information science Education in Nigeria today cannot be relevant without effective preparation of new generation of librarians to effectively use the new information technology. Application of ICT in library education is not new, but the level of application and success is worrisome. These applications can be categorized as follows: Library Automation Library automation is the application of computer s and other related devices to the technical processes, public services and other "housekeeping" functions in libraries. This has to do with acquisitions, organization of materials , reference, indexing and abstracting, compilation of bibliographies, circulation system, among others. Both students and educators will need to have the skills of applying ICT to all library operations and activities in other to be relevant in this information age. And also, this will enhance efficiency in information acquisition, processing and dissemination. --- Operation of Online Databases As part of the electronic processes of information dissemination to user communities, it has become every important that library schools should subscribe to some online data bases to teach their students their operations and importance in getting very relevant and current information in different subject areas. These databases can be in broad or in narrow subject areas such as HINARI in medicine; AGORA in Agriculture; ERIC in Education, . --- CHALLENGES OF ICT ON LIBRARY AND INFORMATION SCIENCE EDUCATION IN NIGERIA The situation of Information and Communication Technologies and Library and Information Science Education in Nigeria entails the prevailing conditions of the faculty staff, the students, practitioners and LIS schools themselves in the formulation of effective curriculum that will provide effective competencies in rendering services to users. However, Library and Information science schools still face many challenges in trying to provide appropriate and sustainable solutions for improving curriculum and skills that will support development in all ramifications . This scenario has been attested by Diso that, Nigeria is characterized by underdeveloped economy, unstable political culture and chaotic socio-cultural environment, with low productivity and low capacity utilization, low industrialization, poor infrastructure, unemployment and underemployment and pervasive poverty. Other such characteristics include high rates of illiteracy, insecurity, popular apathy, traditionalism, poor reading culture, poor information environment and at the state level, poor policy formulation and implementation. All these problems have undermined the development of library and Information science schools in Nigeria especially with regards to the deployment of ICTs. In the view of Saka , there are no adequate facilities provided, and, as such, the acquisition of ICT skills is low at undergraduate levels. This is because of the inadequate computer laboratories, which has led to low level of ICT skills/practices. It is paramount to note that the traditional mode of teaching in our library schools today cannot withstand the modern trends as computers and other ICT facilities are needed in training the prospective LIS graduates. Regrettably, the number of computers in the ICT laboratories in Nigerian library schools cannot cater for the teeming population of students. One of the major obstacles on the use of ICT in Library and Information Science Education is financial cost. These cost includes among other, provision of computer hardware and software, cost relates to internet connectivity and cost of training personnel that will handle the ICT equipment. Another challenge is the lack of technical expertise by the library educators to teach ICT related courses to students. In their observations, Nwokocha, Onwubiko & Unagha said that some of the library schools that have ICT facilities cannot practically impact the knowledge to their students effectively. --- CONCLUSION The emergence of Information and Communication Technologies in all facets of human life has always been a welcome development and its introduction into library and information science education in Nigeria leaves much to be desired. Not only has ICT transformed the information world, it has made the search for information much easier. In the library profession, the application of ICT has brought about quick dissemination of information and effective services delivery, hence its sustenance. --- RECOMMENDATIONS  The following recommendations will enhance library education and the challenges of ICT in Nigeria.  There should be constant training of Library and Information Science Educators to meet up with the currents trends in the profession.  Adequate ICT facilities should be acquired in all library schools in Nigeria to enhance easy processing, storage, management and dissemination of information.  A comprehensive curriculum on library science education should be formulated in Nigeria and reviewed periodically to meet up with the constant changes in the information era.  The government and other well-meaning individuals should provide financial support for the development of Library science education in Nigeria through the establishment of library schools and provision of basic infrastructures.  There is the need for information networking among library schools to share their resources. --- AUTHORS BIOGRAPHY Abdulrazaq

According to Etim (2002), Library and Information Science Education is the study of organized collections of books and other nonbook materials. Through this study, a person acquires approved skills, standard or level of attainment for maximum utilization of information resources in skills with which a learner, student, pupils, researcher of information must possess before being considered information literate. Library and information science can also be seen as the study of issues related to libraries and information services. This includes academic studies regarding how library resources are used, how people interact with library systems, and the organization of knowledge for efficient retrieval of relevant information. Librarianship

‫اﻟﺠﺰاﺋﺮ‬ ‫ﻓﻲ‬ ‫اﻟﻌﺮﺑﻴﺔ‬ ‫اﻟﻠﻐﺔ‬ ‫واﻗﻊ‬ --- Abstract Algeria faces a critical reality embodied in a linguistic struggle among three parties: classical Arabic, colloquial Arabic and the French language. This struggle is a remnant of the French colonization which has worked hard to fight the Arabic language, marginalize it and replace

it with French. In order to maintain their Arabic-Muslim identity, Algerians were forced to use slang Arabic. Although the French colonization is over, this struggle continues. The discussion around the reality of language in Algeria is confronted with a substantial problem related to the negligence of classical Arabic, the creeping of colloquial Arabic into the educational and cultural institutions and the media, in addition to the French language, which works to rival Arabic in the field of education (scientific and technical), and in some administrative transactions, and in daily communication among members of the educated class. This research aims at shedding light on the manifestations of this struggle in public life and in the fields of education, the media, and administration. It also seeks to expose the historical reasons for this struggle, the arguments of those who advocate the use of the colloquial Arabic and the dangers associated with such arguments. In addition, this research presents the argument of those who advocate the use of classical Arabic as the official language, and the language that should prevail in the daily, official and educational communications. The argument of those who support the use of French is also included. The research concludes with the most important recommendations and suggestions to promote Arabic to take the lead in this country and to restore its prestigious position. ‫ﻧ‬ ‫ﺼﻴﺮة‬ ‫زﻳﺘﻮﻧﻲ‬ ‫ـــ‬ ‫ــــــــــــــــــــــ‬ ‫ـــــــــ‬ ‫ـــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ‬ ‫ــــــــــــــــــــــــــــــــــــــ‬ ٢١٥٩ ‫ـــــــــــــــــــــــــــــــــــ‬ ‫ـــــــــــــــــــــ‬ ‫ﻟﻸﺑﺤﺎث‬ ‫اﻟﻨﺠﺎح‬ ‫ﺟﺎﻣﻌﺔ‬ ‫ﻣﺠﻠﺔ‬ ‫ـــــ‬ ) ‫اﻻﻧﺴﺎﻧﻴﺔ‬ ‫اﻟﻌﻠﻮم‬ ( ‫اﻟﻤﺠﻠﺪ‬ ٢٧ ) ١٠ ( ، ٢٠١٣

Introduction The advent of antiretroviral treatment in poor countries promises to dramatically change the nature of HIV care, opening up new possibilities and narratives for healthcare workers. In the light of changes brought about by ART, this paper examines the sources of hope and frustration among healthcare workers who care for people with HIV in poor rural Zimbabwean health centres. AIDS is the single biggest killer in sub-Saharan Africa . HIV infection rates are highest among economically productive adults of childbearing age, taking a tremendous toll on families, communities and regional economies. In 2008 AIDS accounted for 1.4 million deaths -14.7% of total deaths -in the region . --- Healthcare worker motivation in resource-poor settings Healthcare worker resource shortages have been identified as the greatest potential limitation to developing accessible HIV treatment in sub-Saharan Africa . Many papers have examined how to retain healthcare workers in resourcepoor environments and reverse the trend of out-migration to better paying jobs in middle-or high-income countries . A systematic review of the literature on healthcare worker motivation in poor countries identified seven employment preference themes: financial rewards, career development, continuing education, hospital infrastructure, resource availability, hospital management and recognition/appreciation. Nurses were usually willing to forego pay increases for other improvements in their employment conditions , a finding that encourages further research into nonmonetary sources of inspiration and motivation for healthcare professionals in financially strapped healthcare systems. --- Antiretroviral treatment as a motivating force for HIV healthcare workers These studies provide valuable information for the design of effective healthcare worker retention plans, suggesting that financial and resource-oriented limitations are not insurmountable. However, one area that is not widely explored is the importance of medical regime changes and treatment breakthroughs in motivating nurses and other health professionals. This paper seeks to contribute to this area by examining nurse's motivation and frustration in the context of the roll-out of ART in Zimbabwe. The positive changes brought about by life-prolonging HIV treatments has been explored in high-income settings such as the USA, Canada and UK . For example, in New Jersey, USA where "care has shifted from preparing patients to die to preparing patients to live", many healthcare providers reported a sense of hope about their patients' futures . Few studies have examined the changes ART has brought about to the healthcare profession in low-resource settings. A notable exception is Stein, Lewin, and Fairall's article "Hope is the pillar of the universe", which reported the renewed sense of hope and motivation engendered by ART roll-out in Free State, South Africa. Nurses in the study reported a strong commitment to providing this lifesaving treatment, despite the demands it made on their time and energy. The nurses drew significant motivation to deliver highquality ART despite resource limitations from their difficult professional experiences of HIV nursing before ART. Our article seeks to expand upon these findings. While the nurses in the Stein, Lewin and Fairall study were interviewed at the beginning of ART roll-out, we interviewed healthcare workers who had been administering ART for over a year. --- HIV and antiretroviral treatment in Zimbabwe Zimbabwe is one of the first African countries to show a declining HIV rate with the adult prevalence rate falling from 24.6% in 2003 to 14.3% at most recent measure . While there have been antiretroviral drugs in Zimbabwe since the early 2000s, they were in extreme short supply and prohibitively expensive for most people. In the late 2000s the government and partner aid organisations succeeded in a massive scale-up of ART. By December 2009, 218,589 people, about half of those in need, were on free ART through the public health sector . --- Methods We draw on qualitative data gathered by four researchers over six weeks in late 2009 from three healthcare sites in rural Zimbabwe that had been providing free ART for over a year. Site details have been anonymised. The research arose from a collaboration between two UK universities and a Zimbabwean public health research institute. Ethical approval was granted by the Medical Research Council of Zimbabwe and the Imperial College Research Ethics Committee . --- Context All three sites studied had infrequent electricity. One site, a rural clinic, used water from a nearby hand pump, while the other two, both large rural hospitals, had running water, albeit irregularly. The hospitals were each staffed with a doctor or clinical officer,1 and around 40 nurses and 40 additional staff . At both sites the doctor/clinical officer was frequently away visiting smaller clinics and attending meetings and workshops in the cities. Nearly all HIV patient care was thus provided by nurses and counsellors. The clinic was visited monthly by a doctor and was otherwise staffed by three nurses and five additional staff. --- Research design Data were gathered through 18 in-depth interviews with health professionals , one focus group and over 100 hours of ethnographic observation at the healthcare centres. Permission to conduct research was received from senior administrators at each site. Staff members were then approached individually and invited to participate; all agreed. The interviews and focus group used semistructured guides to explore people's views of how HIV and ART had impacted on their work, sources of support and frustration and the challenges of providing ART. Data were collected by three Shona-speaking fieldworkers and a fourth researcher working with an interpreter. Audio files were transcribed and translated into English from Shona. To thank informants, focus group participants were given soap, and interviewees a t-shirt promoting HIV testing. --- Thematic analysis and coding frame Data were analysed using Attride-Sterling's thematic network technique. Text segments pertaining to healthcare worker motivation and frustration were coded for their central idea, such as patient illness, patient improvement, shortages of drugs and patient enthusiasm. These codes were clustered into more inclusive themes. These themes provide the headings for the sub-sections in section "Findings". --- Findings Healthcare workers were overwhelmingly positive about their capacity to provide highquality ART. They told compelling stories of dramatic patient recoveries and shared memories of friends and family members suffering from HIV/AIDS. However, staff frequently expressed frustration with the ongoing difficulties of providing HIV care in a remote and poor area. They did not discuss the hardships of HIV work before ART, instead focusing on the current challenges that they faced. --- Dramatic physical patient recoveries as source of satisfaction Almost every staff member cited the inspiring nature of dramatic patient recoveries. They described the remarkable improvements brought about by ARVs, which reverse weight loss and drastically reduce infections. These stories had a similar format: a patient arrived at the clinic looking extremely thin and unhealthy, often too weak to walk. Through ART they regained complete physical health: Initially, patients are so sick with HIV. However, with ARVs we can control it; the [viral] levels remain low. Even for some patients who had stopped working due to HIV, they are now back on their former jobs, caring for their families. They come to get medical reports that they are fit and can go back to work -it really motivates you. A lot of patients come here in wheelbarrows. They would have become so thin. But when they start receiving treatments they recover to such an extent that we can't recognise some of them when they regain weight. I remember one guy who had to explain he first came here in a wheelbarrow but he was now walking and looking very fit. These are some of the satisfying things when working with HIV sufferers. Playing a role in returning patients to health has a strong positive impact on health professionals. However, such recovery narratives may become less common as ART is increasingly well-established. WHO guidelines state: "The optimum time to commence ART is before patients become unwell or present with their first opportunistic infection". Recovery stories thus draw from the time when ART had just become available or from more recent cases of problematic late initiation. As ART enables patients to remain in good health without first exhibiting dramatic symptoms of AIDS, these recuperation narratives will become distant memories. The future impact of this change on nurse motivation will need to be researched. --- " The person now believes that there is still life after HIV": patient emotional improvement as source of motivation Similar to physical recoveries were stories of improvements in patients' emotional states through counselling and treatment. These improvements motivated staff to provide a hopeful and inspiring environment. An HIV counsellor explained that "it is so good when you help someone and see them picking up the pieces and moving on with life again". A nurse discussed how patient acceptance and happiness affected him: Suddenly the patient starts to improve. He becomes happy just like anyone else. He comes to accept what he now is. You see that his life is back again, he is no longer viewing himself in a negative way. That's what makes me happy. He is like anyone else, he has accepted the situation … and the person now believes that there is still life after HIV. When patients accept their status and embrace the idea of "life after HIV", enthusiasm for and dedication to treatment generally follows. Another nurse explains that patients who accept their status "… come on the [ART review] date without fail, you do not have to look for them, they commit to it within their hearts that they are supposed to come back and they come". Patient commitment to ART, which was closely linked to acceptance of HIV status, was cited as a further source of motivation. --- Personal experiences with HIV: strong motivating factor Every healthcare worker interviewed had an experience with HIV/AIDS in his or her personal life. Primarily, healthcare workers recounted stories of close family members dying or nearly dying of AIDS although in one case a counsellor discussed his own journey of coming to terms with his HIV-positive status. Alongside the sorrow and loss, many staff reported how these experiences influenced their professional lives in a positive way. In some cases, it was the experience of HIV/AIDS among their family that inspired them to take up HIV work. In all cases, these experiences enabled healthcare workers to treat HIV-positive patients with greater understanding and commitment. Personal experiences inspired some nurses to be more compassionate on a general level: [The death of my brother] has actually been a source of inspiration for me…I felt really sorry for him and it became the basis for me to try and offer all that I could to these HIV sufferers. No wonder I like working with HIV positive people…my brother's illness gave that inspiration, and I have developed this love for HIV sufferers. These HIV sufferers need love and they need to be accepted as they are. I give them a handshake and they feel comfortable and loved. Personal experiences with a friend or relative suffering from AIDS helped staff approach HIV in a non-stigmatising, loving way. Some staff drew specific lessons from their experiences. For example, in the following quotation a nurse discusses how the death of her brother and his wife influenced her to make a strong professional effort to encourage counselling: The most painful part is that they wasted time and resources consulting traditional healers…It was this that taught me the importance of counselling. I now really feel so attached to these ART patients and I clearly understand how it can be so hard for them to comply. So I always try to make sure that I give correct and enough information to them, and try to make follow up counselling until I am convinced the person has accepted their HIV results. Healthcare workers made clear linkages between their personal experiences with HIV and their professional efforts to deliver high-quality HIV care. Although HIV rates are dropping in Zimbabwe and more HIV-positive people are living in good health, the type of personal experiences with HIV reported by these healthcare workers will likely remain a strong motivating force in the healthcare sector for many years. --- Wide range of shortages: major source of frustration Our study sites suffered from many severe shortages: there were not enough doctors, medicines frequently ran out, key machines such as CD4 counters and oxygen tanks were in short supply or broke down for days at a time, salaries were often late or too low and common supplies were often out of stock. Healthcare workers frequently referenced these shortages as extremely frustrating and disheartening. While ARVs were generally stocked, they occasionally ran out or had to be carefully rationed, as discussed below: It's painful when we do not have ARVs for the patients. You will be expecting the patient the following day and you do not have the tablets. Yet you were the one who told that patient that he must take these tablets without fail. Then he comes here there are no tablets. What do you expect that person to do? How would you feel if it was you? Many staff members were extremely distressed by being blamed for problems they could not control: We may not have the medication, so when they come there expecting to get the drug they end up thinking that you do not want to give it to them. If you ask them to go and look for it from other pharmacies, they ask you where you expect them to get the money to purchase the drug. People might hate the pharmacy tech, say if they are told the drugs they have been prescribed are out of stock. Maybe they think I am hiding these drugs somewhere. In other cases, staff distress stemmed not from patients' misdirected anger but from a sense that they were letting patients down, for example, by turning them away because the CD4 blood cell count machine was out of order. Several staff members described feelings of helplessness and inadequacy surrounding death or suffering that could have been prevented by simple supplies, such as a blood transfusion system. In the following, a nurse discusses the emotional toil of these shortages: I always feel inadequate when a patient who is beyond what we can handle here comes seeking assistance. In such cases I feel like 'if only we had a doctor here", or "if only we had oxygen here we could have helped" …So when we face situations that call for a doctor and other equipment that we do not have one feels so bad and so helpless. Some healthcare workers expressed pride in their centre's services, suggesting patients preferred their hospital and received superior care. While this pride was a source of motivation to continue delivering good quality care, it also led to healthcare worker strain when resource shortages compromised the capacity of the staff to provide good care: There are cases when patients have to be seen by the doctor because their needs are beyond what we can do as nurses. However, the doctor is always not there, and we really feel helpless in front of patients. We really feel bad when we see a patient deteriorating … There is the issue of the CD4 machine. It's down. I will be turning back quite a good number [of patients] so that's one other factor that discourages patients …The truck that brings the ARVs from Harare might delay. Like today we were supposed to give patients two months supply of ARVs, but we are only giving them two weeks supply so that everybody may get something. So the patients are no longer confident that we will get constant supplies and we are also not sure. Losing the confidence of patients in the hospital's capacity to provide good care was a major concern for staff. --- Memories of the professional challenges before ART apparently not a source of motivation The healthcare workers did not refer to the challenges of nursing before ART when discussing their motivation. In fact, they never spoke of the hardships of providing healthcare for HIV patients before ART was available, despite the fact that almost all of them had worked with HIV patients for years before ART was available. This finding was surprising in light of the sources of motivation cited by South African nurses in Stein et al.'s paper. Those nurses discussed ART as something new and hopeful that could drastically change the horrible reality of HIV nursing before ART was available. Their willingness to stretch already-thin resources and to work extremely hard in order to administer ART was directly related to their recent professional experiences of the horrors of HIV nursing before ART. The nurses in Stein, Lewin and Fairall's study spoke of the previous trauma of having to cope with large numbers of deaths among their patients before ART. A key difference between the Stein, Lewin and Fairall study and ours is the amount of time elapsed between the roll-out of ART and the research. While the participants in our study had been administering free ART for over a year, the nurses in Stein, Lewin and Fairall's study were interviewed on the cusp of ART roll-out. It appears that in as short a period as one year, ART can become a fairly normal component of nursing. The absence of comparisons to the time before ART suggests to us that the healthcare workers in our study had rapidly moved from seeing ART as an exciting drug capable of transforming their professional lives to focusing on the challenges associated with providing ART care, such as late supplies of drugs or CD4 count machine breakdowns. --- Conclusion This paper has sought to contribute to the research literature on sources of healthcare worker motivation and the changing nature of HIV nursing brought about by ART. It specifically sought to build on the findings of Stein et al. who examined healthcare providers' experiences of delivering ART in primary healthcare clinics at the initiation phase. We found that dramatic physical recoveries of AIDS patients because of ART were key source of inspiration and motivation. The types of deterioration and recovery featured in these motivation narratives are, thankfully, likely to become less common as ART provision enables HIV-positive people to remain healthy without first exhibiting signs of AIDS. Being a part of or observing a patient's emotional improvement, from feeling disheartened to seeing the possibilities of "life after HIV", brought healthcare workers a sense of joy and purpose. Further, personal memories of close friends or family members suffering from AIDS were another key source of nurse's motivation, especially surrounding the motivation to provide compassionate and non-stigmatising care. The healthcare workers in this study did not link their motivation to provide high-quality ART to their memories of working with HIV patients before ART was available. This finding was surprising because the healthcare workers in Stein et al.'s study cited the difficulties of HIV nursing before ART as a key source of motivation to provide highquality ART. This difference in sources of motivation is likely linked to the fact that the healthcare workers in the Stein, Lewin and Fairall study were interviewed at the very beginning of ART roll-out, while our study examined ART over a year after the drug regime had been introduced. ART roll-out across sub-Saharan Africa presents unique opportunities and challenges for the healthcare profession, enabling HIV care to evolve from a primarily opportunistic infection and palliative approach to a long-term chronic illness management approach. However, ART, like all other medical interventions, is subject to the daily frustrations of providing healthcare in resource-poor settings. The findings in this study provide a basis for optimism about the future of ART nursing care and HIV management in sub-Saharan Africa since healthcare workers were found to draw strong motivation to provide good care from their personal experiences with AIDS and the dramatic recoveries made possible by ART. However, our findings also highlight areas of frustration surrounding severe resource shortages that must be addressed in order to enable nurses to continue delivering top-level HIV/ART care to all those in need in sub-Saharan Africa.

The roll-out of accessible and affordable antiretroviral (ARV) drugs for people living with HIV in low-income countries is drastically changing the nature of HIV-related healthcare. The Zimbabwean Ministry of Health has renewed efforts to make antiretroviral treatment (ART) for HIV free and publically available across the country. This paper describes the findings from a multi-method qualitative study including interviews and a focus group with healthcare workers (mostly nurses), totalling 25 participants, and field notes from over 100 hours of ethnographic observation in three rural Zimbabwean health centres. These health centres began providing free ARV drugs to HIV-positive people over one year prior to the research period. We examined sources of motivation and frustration among nurses administering ART in these resource-poor health centres. The findings suggest that healthcare workers administering ART in challenging circumstances are adept at drawing strength from the dramatic physical and emotional recoveries made possible by ART and from their personal memories of the suffering caused by HIV/AIDS among close friends or family. However, healthcare staff grappled with extreme resource shortages, which led to exhaustion and frustration. Surprisingly, only one year into ART provision, healthcare workers did not reference the professional challenges of their HIV work before ART became available, suggesting that medical breakthroughs such as ART rapidly come to be seen as a standard element of nursing. Our findings provide a basis for optimism that medical breakthroughs such as ART can reinvigorate healthcare workers in the short term. However, we caution that the daily challenges of nursing in poor environments, especially administering an ongoing and resource-intensive regime such as ART, must be addressed to enable nurses to continue delivering high-quality ART in sub-Saharan Africa.

Introduction Globally, the numbers of persons living with dementia is projected to increase from 58 million in 2020 to 152 million in 2050 . Much of this increase is expected to occur in low-and middle-income countries . In India, approximately 3.8 million people were estimated to be living with dementia in 2019. . However, despite this relatively high burden, only 1 in 10 persons with dementia receive any diagnosis, treatment or care in the country . This considerable treatment gap can be attributed to low awareness of the condition, poor accessibility to health care services, shortage of specialists to diagnose and manage dementia, out-of-pocket [OOP] costs associated with care seeking, the absence of both a strong referral system and a formal long term care system, which are vital to the effective management of dementia. The Indian healthcare system is complex , comprising a mixture of public, private and informal providers that vary substantially across areas of the country . Partly as a result of this heterogeneity in service availability, health seeking behaviors are also considerably diverse. In comparison to rural areas, urban areas have a large concentration of private care providers , which allows for more choice with respect to care services. Decisions regarding utilization of type of healthcare services are influenced considerably by socioeconomic factors . OOP expenditures are a major contributor to total health expenditure in India ). However, the government is working towards providing greater social protection for the socioeconomically disadvantaged in addition to improved accessibility to essential services through the Ayushman Bharat [AB] programme and Ayushman Bharat-Health and Wellness Centres . While these are positive steps, formal long-term care support services for dementia continue to remain minimal or absent. A few day-care centres and residential care facilities exist only in metropolitan cities and are associated with significant OOP costs . As a result, care provided by families in their own homes remains the cornerstone of dementia care in India. The study reported here was conducted as part of the wider Strengthening Responses to Dementia in Developing Countries international project led by the London School of Economics and Political Science. STRiDE has been conducted across seven middle-income countries with the aim of building research capacity and bridging gaps in evidence with respect to dementia. In the first phase of the project, we drafted an in-depth desk review of the dementia care landscape in India ), which provided us with an understanding of the diversity in support for people with dementia and their families in the Indian context. Informed by this desk review, we attempted to take a systems wide approach by compiling relevant aspects of the Indian system for dementia. We utilized professional experiences of the multi-disciplinary team to highlight the complexity and differences in care experiences of people with dementia and their families in India. Very few studies have examined pathways to care for people with dementia in India and none has specifically examined pathways to care from receiving a diagnosis to long-term support. This paper aims to fill this gap. It contributes to generating evidence on how people with dementia and families currently navigate care and support services, understanding what factors shape care access and use, and investigating the current treatment gap for dementia by highlighting existing challenges that need to be addressed. The paper uses a case vignette approach that gathers inputs from a multi-disciplinary group of professionals with expertise in dementia and is informed by knowledge gathered through the in-depth review of available evidence on the dementia care landscape in India . --- Methods We used a case vignette approach to illustrate pathways to care for people with dementia. Case vignettes are short stories based on hypothetical persons or scenarios, often used to support data collection in qualitative research . In vignette-based studies, participants are asked to draw upon their own experience to predict how the characters will behave or feel, or how the scenario will evolve . As a result, vignettes represent a structured tool to elicit participant's reasoning and to gather participant's experiential evidence and tacit knowledge, which can then be utilized to complement formal evidence . This approach has been used to determine pathways to care for people with dementia and found to be particularly effective in facilitating comparisons in care systems across countries . Prior to developing the India specific case vignettes, a comprehensive desk review of the dementia care landscape was drafted involving answering series of questions related to the health system, long-term care system, dementia policy context, social protection schemes and diversity in services or support available for people with dementia and their families in India. This in-depth desk review and expert inputs gathered from a multidisciplinary group of professionals that comprised the STRiDE India team contributed to the development of the case vignettes. It is important to highlight that as part of the wider STRiDE project primary research interviews with various stakeholders on topic areas such as stigma and awareness of dementia had also been conducted and this further informed expert input in the vignette development. The case vignettes were validated by six members of the project's national advisory group and analysed using an approach that builds on framework analysis to develop scenarios that 1) identify how people living with dementia and their caregivers navigate care and support by accounting for the diverse and heterogeneous factors that influence care; 2) highlight gaps and challenges associated with existing models of dementia care services in the country; 3) provide recommendations to overcome challenges identified. Ethical approval was not required as only the research team was involved in this study. The overall project received ethical clearance from the institutes involved in the study. --- Approach As part of the STRiDE project, base case vignette scenarios were developed collaboratively by researchers and NGO partners from the seven STRiDE country teams and facilitated by the UK based LSE team during a workshop in March 2019. As part of the workshop, the STRiDE team worked in small groups and identified and ranked a set of variables likely affecting care and support people with dementia and their caregivers receive . Focusing on these variables, five core scenarios were developed, that briefly introduce people with dementia and their families across different socio-economic groups and living environments . Each STRiDE country team then provided feedback as to whether these scenarios were applicable in their country and suggestions about how the vignettes could be improved . The base vignettes were finalised once no further feedback was received. A detailed description of the different steps involved in completing and validating the base vignettes can be found in the STRiDE vignette guidance document . --- Development of India specific vignettes The five base case vignettes were initially reviewed by the STRiDE India country team. The majority of the team had been working in service provision for people with dementia and their families for over 10+ years in India. Based on individual expertise of the team and the STRiDE India desk review of the dementia care landscape the country team identified gaps in these five vignettes that were pertinent to the Indian context. Group consensus on modifications and addition of new vignettes were achieved over multiple meetings through which team members shared their insights and comments until a collective decision was reached. Taking into consideration the heterogenous situations that people with dementia and their families find themselves in India, the country team modified one base case vignette and further developed three additional case vignette scenarios. One base case vignette was revised to include a case of vascular dementia, as stroke has been identified to be a leading cause of non-communicable neurological disease burden in India and a major cause of dementia in LMICs including India . Three additional vignettes were subsequently developed to capture specific features of the Indian context. Firstly, approximately 90% of families receive no diagnosis, treatment or care and a case vignette representing undiagnosed dementia was included. Secondly, the proportion of young-onset dementia has been found to be higher in India than in developed countries, which may be due to the younger age of the Indian population . Young onset dementia is accompanied with significant economic burden and is often associated with delays in diagnosis or even misdiagnosis . To reflect this, a case vignette on young-onset dementia was developed. Lastly, there was a need for the vignettes to highlight health insurance in the Indian context. About 41% of households have "any usual member" covered by a health insurance scheme , 2021, pp.3). We however attempted to reflect the substantially smaller proportion of households covered by health insurance schemes that cover some outpatient expenses, a case vignette representing a government health insurance scheme was included and a private health insurance scheme was added to the young onset dementia vignette. Therefore, a total of eight case vignettes were created to represent the dementia care context in India. For each vignette, the team then articulated how the socio-demographic characteristics and personal and family circumstances of each character were likely to shape aspects such as health-seeking behaviors, time to diagnosis, stigma experienced, and costs associated with care. These aspects were identified in the STRiDE guidance document and also by expert inputs from country team. Comments shared by the team members about individual case vignette scenarios were consolidated and subsequent discussions involved listing of ideas and seeking clarifications on how each vignette scenario would evolve until consensus was achieved. Hypothetical names were added to the vignettes to make them more personable; they do not refer to any existing people with dementia or caregivers. These vignettes are published and freely available for viewing on the STRiDE website . After exploring the individual care pathways via the case vignette approach, we then used a framework approach to identify patterns as well as differences --- Validation of vignettes Our study is reflective of the professional experiences and expertise of team members. We attempted to ensure that we are not portraying just our professional views by having our vignettes critically reviewed by experts from the STRiDE India National Advisory Group . The INAG was formulated at the beginning of the STRiDE project and comprises of 21 members whose role was developed to hold the research team accountable for the activities of the project and included professionals with diverse experience in health and/or long-term care in India in addition to members of the general public such as former carers. The INAG was invited via email to validate the vignettes considering whether the vignettes represented the diverse situations that people with dementia and their families find themselves in and accurately described health seeking behaviors and care trajectories. They were also asked to suggest any other scenarios for inclusion. Six INAG members provided feedback on the vignettes via email. This was consolidated and discussed by the core team. --- Analysis of vignettes An approach based on the framework method et al., 2021). The India team also added a section on pre-diagnosis and awareness to a column as this was an important topic in the Indian context. Subsequently, we charted data on gaps and challenges to be overcome for each vignette scenario by reflecting on the data captured for each of the previously highlighted questions. Using these pre-determined categories, the country team grouped them together to form three major topic areas. Pre-diagnosis and awareness were grouped together to form topic 1: pre-diagnosis; where a family would go for memory concerns/behavioral disturbances formed topic 2: diagnosis; and support available to carers/people with dementia were grouped to form topic 3: post-diagnostic support and long-term care. Gaps and challenges to access care were reflected upon and included in each of the three topic areas above. --- Results Table 1 provides a brief overview of key characteristics from each of the 8 vignette scenarios developed. We set out to explicitly explore the diverse situations that people with dementia and their families find themselves in India and highlight patterns we identified. The vignettes characterize the potential pathways to care for people with dementia that reside in rural, semi-urban or urban areas, from different socioeconomic groups and relying on different levels of support from the family. Across all the scenarios, socioeconomic and demographic related factors typically play a critical role in influencing utilization and access to services. Other factors [cultural and/or societal beliefs; knowledge of dementia; severity of dementia; age of dementia onset; availability of resources; social protection; living arrangements; support and relationship with family members] were also illustrated as determining helpseeking behaviors and care services sought by people with dementia. Utilizing key topic areas, we highlight how such factors shape specific phases of the dementia care journey. [ that receiving a diagnosis is an issue that people with a range of characteristics share. --- Diagnosis The diagnosis seeking phase is likely to vary substantially based on availability of services within a given area and individual socioeconomic status. Accessing a specialist is often key in receiving a diagnosis for dementia across the case vignettes. In rural areas, families from a lower-socioeconomic group have been shown to have heterogenous pathways to care . People with dementia or their families may seek a traditional medical practitioner, or directly communicate concerns to a community health worker who may direct them to a primary health centre , or the family may not seek any formal care services at all. Even if a person with dementia or their family sought a GP at a PHC, it was established in the vignettes that the GP would be unlikely to recognize symptoms of dementia; possibly recommending a visit to a specialist in the city. In --- Post-diagnostic support services and long-term care After receiving a diagnosis from a specialist, the vignettes highlight the absence of a system to provide coordinated care and support services for people with dementia or their families. Day care centres, residential care centres, paid attender agencies and psychosocial support services for caregiving are discerned to be few in the entire country; primarily concentrated in metropolitan cities . These services were depicted to be associated with substantial OOP costs thereby more affordable to those from a higher socioeconomic group. Although even among higher socioeconomic groups, stigma associated with the utilization of institutional services such as residential care may possibly influence service uptake or sustained use. Notably, while the vignettes indicate that domestic helpers may aid in caregiving among more middle and higher socioeconomic groups , their role was highlighted as particularly relevant in situations where there is no family support available . Overall, within families the vignettes have emphasized that it is the women who primarily provide care with respect to instrumental activities of daily living for people with dementia with limited or no psychosocial support and while managing multiple households and/or professional responsibilities in addition to their caregiving responsibilities . As a result, they have been shown to experience emotional distress and negative consequences for their physical health . Certain exceptions have also been highlighted in the vignettes with respect to access to post-diagnostic support, such as paid attenders/residential care centres, among high socioeconomic groups residing in metropolitan areas. --- Discussion To our knowledge this is the first attempt to comprehensively illustrate the care pathways of people living with dementia in India using input from a multi-disciplinary group of experts using a case vignette approach. We organized the care pathways in three topic areas and identified the challenges across each phase. Low awareness of dementia among the general public and medical professionals is depicted to contribute to delays in dementia diagnosis across case vignette scenarios. Specialists are highlighted to be key to providing a dementia diagnosis, but large urban-rural disparities in distribution of specialist services were recognized. After receiving a diagnosis, post-diagnostic support is emphasized to be limited, and it is the families that are described to primarily provide long-term care until the persons with dementia die. Overall, these findings highlight challenges that lie within the current provision of dementia care services in India. A summary of the challenges identified are provided in Table 2. --- [Insert Table 2 here] Prior to dementia diagnosis, awareness of dementia among the general population and medical professionals are illustrated as a crucial factor influencing whether, when and how a family receives a formal diagnosis, if at all. Awareness of dementia as a medical condition is low in the Indian context. Studies have found that memory problems are often thought of as part of natural ageing and that the behavioral symptoms associated with the condition are stigmatized, perceived in certain communities to be deliberate . However, this lack of awareness is not limited to general public. A study conducted in a medical college in the state of Gujarat reported that only 49% of undergraduate medical students were aware of dementia symptoms . This is more concerning as the mean age at presentation of dementia is also lower in the Indian context and youngonset dementia in particular is more likely to be associated with delays in diagnosis as also demonstrated in one vignette scenario ) . Therefore, poor awareness among GPs further contributes to diagnostic delays as highlighted in several case vignettes; this can be attributed to the significant lack of emphasis of dementia in the undergraduate medical curriculum . Outside of seeking a GP, in the diagnosis phase people with dementia and their families may access several other services as well. Differing beliefs and variability in availability of services are likely to influence care services sought; studies have found utilization of services from non-registered practitioner , unlicensed medical practitioners to be prevalent among elderly residing in rural India. Seeking such services can contribute to further diagnostic delays. It is important to recognize that India has a strong traditional system of medicine and efforts have been made by the government through the Ministry of AYUSH to support and promote traditional medicine systems . There are opportunities to integrate traditional medical practitioners into models of dementia care service delivery possibly through collaboration with trained general practitioners or dementia specialized professionals. Currently, diagnosis of dementia is primarily provided by specialists . This is problematic because the number of neurologists and psychiatrists available in the country to meet population needs is grossly inadequate ; with a particularly high shortage experienced in rural areas where about 70% of the population resides . This was also highlighted in the case vignettes, which further emphasized the challenges experienced by persons living in rural or semi-urban areas, who have limited access to specialists and have to travel significant distances to seek care often incurring multiple direct and indirect costs . Household OOP expenditure was identified as the main way to access services during the diagnosis phase and for ongoing post-diagnostic support. OOP expenditure by households has been identified as a major contributor towards total health expenditure in India . As a result, socioeconomic status plays a crucial role in determining access to care in the Indian context. Studies have demonstrated that persons from higher income groups are more likely to choose private hospitals associated with more costly care compared to persons below poverty line who are likely to choose government services . However, this varies, and public health infrastructure of individual states greatly influences choices in care services . Coverage by a health insurance scheme in association with other factors ) also influences healthcare utilization. The vignettes depict a very small proportion of the population covered by health insurance schemes with some outpatient coverage, but importantly highlight that predominantly these schemes do not provide coverage for long-term care services such as day care, residential care and paid attender services for dementia. Long term care irrespective of a diagnosis is primarily provided by families in their own homes with limited psychosocial support as emphasized by the vignettes. A gender disparity with respect to this family-based care provision however exists, with women predominantly providing care, as highlighted in the vignettes. This has been consistently reported by studies on unpaid care for elderly in India, with women experiencing worsening mental health as a result of their caring role . While families have had to take on this caregiving role with limited support as a result of societal norms and the absence of a system or services to provide continuity in care , there are a few long-term care services that are available in the country to select groups . These services are predominantly in large metropolitan cities and associated with high OOP costs that make such services inaccessible to the majority of the population . In addition, there is inadequate regulation or monitoring of services such as paid attender agencies, residential care centres or day care centres, so the quality and costs of these care services are highly variable . However, well-developed states such as Kerala have a larger number of old age homes as well as a palliative care facilities to support those with terminal illnesses, the latter, however, primarily for persons with cancer . The strengths of this study are that we relied on the considerable direct professional experiences of the research team. The majority of the team members were providing actual care services to people with dementia and their caregivers in India. Our methodological approach can further be considered as an innovative way to illustrate pathways to care for people living with dementia in the Indian context and doing so in a cost-effective way. The developed vignettes help to provide information about the support available to people with dementia and their families in the light of limited published evidence in this area. In addition, we also ensured that our results were checked for face validity with key stakeholders in the field of dementia in the country. While this work provides an important first step towards a more systematic illustration of dementia care pathways in India, we nevertheless acknowledge several limitations . The case vignette approach while useful in describing a general overview of pathways to care for people with dementia in India, could not account for the substantial state-wide differences that are likely to influence these care pathways. Also, we are aware that the vignettes may be shaped by the experiences of the team. We aimed to mitigate this risk of providing a biased or partial account of dementia care in India by working as an interprofessional and multidisciplinary team, constantly discussing and challenging the results of our own work among us. We also sought feedback from stakeholders as well as anchored the vignettes in the existing literature about dementia in India. In addition, we acknowledge that people living with dementia and their caregivers did not contribute to the development of the vignettes. While the professional inputs of majority of team members are informed from their day-to-day work with people with dementia and/or their families as well as other aspects of the STRiDE project including some empirical work , we recognize the urgent need for primary research on pathways to care for dementia care in India. --- Conclusion and Recommendations This paper has highlighted that low awareness, stigma, heterogeneity in health seeking behaviors, poor infrastructure and human resource availability, costs associated with seeking care and an absence of a formal long term care system considerably influence care journeys of people with dementia and their families in the Indian context. A number of recommendations have been recognized as crucial to address these challenges and have been identified in part through the in-depth desk review : • Raising awareness among the general public in order to reduce stigma and so that people with dementia are able to recognize symptoms early on and seek the appropriate help is critical. Development and implementation of awareness programmes that take into consideration context specific factors influencing perceptions of dementia, with a strong focus as well on prevention through behavioral change communication strategies is essential . • Placing greater emphasis on dementia in the undergraduate medical curriculum in order to raise awareness among primary care physicians to recognize, diagnose and appropriately refer people with the condition. Community health workers such as Accredited Social Health Activists workers should also be trained to identify dementia. Such task shifting interventions are vital in settings wherein specialist shortages exist. Such interventions have been trialed and found to be effective for mental health conditions in rural areas of India . • Improving financial protection, with programmes such as Ayushman Bharat having considerable scope of reducing OOP expenditure. It is critical that dementia stakeholders actively engage in continued operationalization of the programme to ensure services for people with dementia are well integrated with efficient delivery. In addition, development of schemes that cover costs associated with informal care provision is critical. For example, the state of Kerala provides a nominal remuneration per month to female unpaid carers of persons with mental or physical illness . • Developing evidence-based community interventions to support informal care provision. For example, a randomized controlled trial in Goa utilized home care advisors, recruited locally and trained for a week on dementia, to support caregivers at their homes through fortnightly visits for a period of 6 months, supervised by specialists . The authors found that this intervention led to improvements in caregiver mental health. There is a need for larger studies that trial such community-based interventions in different parts of the country using locally available resources. --- Next Steps To our knowledge this is the first study in India to systematically illustrate care scenarios for people with dementia and their families from pre-diagnosis to post-diagnostic and long-term care support utilizing a case vignette approach. It has highlighted situations that people with dementia and their families may experience and provided possible recommendations to address challenges informed by an in-depth desk review of the dementia care landscape . By describing potential care situations of people with dementia and their families, this paper provides a base for future larger empirical work with persons with dementia and their families that examine care pathways for dementia in the Indian context. This could include primary qualitative research with people with dementia and/or carers wherein the vignettes may be used to elicit their responses on pathways to care for dementia in the Indian context. --- Acknowledgements The authors would like to thank the experts who aided in validating our vignettes and the wider STRiDE team. --- Funding This work was supported by the UK Research and Innovation Global Challenges Research Fund .

Background: Limited evidence exists on how people living with dementia and their family/unpaid carers navigate care and support in India. Aim: This study used case vignettes to illustrate likely pathways to care for dementia, from receiving a diagnosis to long-term support, in India and to highlight gaps and challenges associated with current care provision for persons living with dementia. Methods: As part of the Strengthening Responses to Dementia in Developing Countries (STRiDE) project, and to contribute to an analysis of dementia care policies and systems in India, case vignettes were used to illustrate the diverse situations that people with dementia and their families may experience when seeking care in the Indian context. Eight hypothetical, but realistic cases of people with dementia were created by a multi-disciplinary team with experience in dementia care in India, to map out the likely care journeys of each case. Results: Investigating eight diverse care trajectories of people living with dementia highlighted important patterns relevant to the Indian context. We identified delays in dementia diagnosis to be attributed to low awareness of dementia among the general public and medical professionals in addition to a critical shortage of specialist services involved in facilitating dementia diagnosis. Post-diagnosis, support was recognized as limited and associated with considerable out-of-pocket (OOP) costs. Families primarily provide long-term care for people with dementia till end of life.Several steps need to be taken in order to improve dementia care in India. Increasing dementia awareness among both medical professionals and general public is essential. Shortages in dementia specialists can be addressed in part through appropriate task shifting. Lastly, more research is needed to develop evidence-based community interventions to support informal care provision for persons with dementia in India.

Introduction Increasingly there are calls from the research sector that postgraduate scholarships are insufficient for the increased cost of living in Aotearoa New Zealand. Through open letters , petitions , surveys , and the news media researchers are consistently highlighting the inadequacy of scholarships and the urgent need to build stable career paths for emerging researchers. This paper provides a context for understanding the postgraduate research scholarship landscape in Aotearoa, including how postgraduate research scholarships are understood in relation to dominant neoliberal framings of higher education and persistent inequities within the sector. The progression to postgraduate research studies is often presented as linear: a Summer scholarship and an Honours and/or research Masters progresses into a Doctor of Philosophy , which is configured as the 'gateway' into a research career. Each of these stages can be supported by scholarships. All eight universities in Aotearoa award summer scholarships, research Masters degrees, and PhDs. Summer scholarships are small, funded, and supervised projects that introduce scholars to primary research over university holidays. Research Masters degrees often involve some combination of coursework and a significant self-directed research component and represent a first step into postgraduate research. PhDs in Aotearoa are similar in structure to those offered in the United Kingdom: supervised independent research towards the completion of a thesis, with coursework seldom required. All universities offer contestable scholarship support from internal funding, and some scholarships are also offered through external research funding schemes, including the Marsden Fund, the Endeavour Fund, or the National Science Challenges. Because the scholarships are contestable, not all postgraduate students receive scholarships. These scholarships typically encompass payment of university fees as well as a contribution to living costs and any other support, e.g. to attend conferences or pay for research costs. In this paper, we use the term 'scholarship' throughout, but our discussion is limited to the living costs portion . While presented as linear, in practice the postgraduate research landscape is increasingly complex and fractured, and postgraduate student scholarships provide a pinch point that sits at the intersection of many persistent issues. Financial precarity among students and researchers has become an increasingly urgent issue over past decades, with the massification of higher education and increases in student debt , the expansion of the 'gig economy' within higher education , and rising costs of living . Overlaying these developments are questions about the purpose and future of the research sector and workforce, including the potential for postgraduate research pathways to address persistent inequities within the sector . These challenges have been amplified with the Covid-19 pandemic, --- 3 New Zealand Journal of Educational Studies 57:505-523 which has seen an acceleration of living costs as well as a growth in precarity in academic workforces . Within this complex terrain, we argue that there is a vital equity dimension that needs to be brought to the fore: current student scholarships have the potential to limit who can realistically complete a postgraduate research degree. This paper provides much-needed evidence that postgraduate student scholarships are inadequate: they do not reflect the increased living costs and are not aligned with Minimum Wage increases. Further, the length of scholarship awards often falls short of the median time it takes to complete a graduate degree. This data provides persuasive evidence of the urgent need to lift the value and length of scholarships in line with increasing living costs. While we focus on equity, it is important to acknowledge there may also be implications for recruitment. We make comparisons between scholarship payments and the minimum wage to demonstrate their insufficiency, but people considering postgraduate study will likely make their decision based on comparison with what they could earn in full-time employment on a graduate wage. For the purposes of this study, our focus is on domestic fees, while acknowledging there are significant and particular challenges affecting international students, with associated wellbeing implications. In presenting this evidence, at times we consciously employ a language of 'investment', mirroring the logic used in academia to justify low scholarships, so we can demonstrate the shortcomings of this logic. We do so cautiously, recognising that such use risks further embedding and reinforcing neoliberal ways of thinking. As authors, we fundamentally oppose the neoliberal and managerial logics that structure, commodify, and marketise education, while being used as a justification to suppress wages. The title of this paper recognises the neoliberal understanding of education as a private good, a framing that has come to define much of the debate about postgraduate student scholarships. Much like a business, if you have no money you are expected to provide 'sweat equity'-to invest time and energy through work, as part of a cost-sharing relationship, to earn any personal benefit or profit. The response to calls for increased scholarships so far can be understood in these terms: universities provide the qualification and in return postgraduates provide the sweat-they do the work. Students with fewer financial means must supplement their income, compromising academic performance, reducing extra-curricular opportunities, risking burnout, and embedding existing inequities. The status-quo of low scholarships, supplemented by postgraduate 'sweat', excludes people from participating in postgraduate education, preventing them and their communities from realising the public benefits that such an education can produce. --- Context Scholarships sit at the intersection of multiple debates about the value of postgraduate research, which reflect wider shifts and contradictions within the higher education sector. It is clear, for example, that a PhD is no longer simply a mechanism for preparing one for a career in academia. The academic job market is saturated, leaving many early career researchers in fragmented and casual contracts for extended periods . In some cases, key research institutions have argued that there is a "mismatch between the relatively narrow skills" of PhD graduates and the "skills needed in employment roles" that would enable successful careers "of value to New Zealand" , implying that a PhD also inadequately prepares postgraduate researchers for a career outside academia. Yet postgraduate research is nevertheless valuable to universities and the wider research sector as research outputs, commoditised to stimulate the knowledge economy: the work of postgraduate researchers are included in productivity metrics like publications per million dollars spent . Moreover, postgraduate research is increasingly positioned as an important site to address wider societal equity commitments, as highlighted in discussion documents like Te Apārangi Royal Society's Early Career Researchers in Aotearoa: Safeguarding and strengthening opportunity after COVID-19 . Although often overlooked, postgraduate research can produce material benefits in and of itself-particularly when embedded in communities and attuned to their needs . Many of the tensions within the conceptualization of postgraduate research are rooted in neoliberal shifts in higher education in the 1980s. This period instigated a shift towards education as a commodity and individual investment, while maintaining some language about education as a public good. Alongside increasingly corporate management and branding of public services , structural shifts in university funding models saw universities shift away from exclusive, well-supported degree programmes where government funding kept tuition low and living allowances relatively universal, to a model that increased access by passing significant costs on to students . The 1988 Hawke report advocated for a shift towards seeing higher education as a personal 'investment' with increased personal benefits which justified pushing costs onto individualised private citizens and away from the public purse . To increase access to university without significantly increasing government spending, the Labour Government of the late 1980s abolished free fees for tertiary education, with the subsequent National Government of the 1990s introducing a student loan scheme and student allowances for students under 25 based on parental income . With these 'cost-sharing' reforms, postgraduate students saw increases in tuition fees, as well as an associated rise in student debt. These issues were brought to a head with the removal of the postgraduate student allowance in 2013 by the National-led Government. In justifying these reforms, the Ministry of Education noted the "value of postgraduate tertiary education" in terms of "social and economic benefits", but also framed postgraduate study as a matter of individual investment, noting that "masters graduates earn more than bachelor graduates" and therefore "expected student loan repayment times are shorter on average for postgraduate borrowers than for other borrowers" . Yet the logic of personal investment-that students pay for education in exchange for presumed subsequent benefits-obfuscates processes of 'lean reproduction' and exploitation that can underpin postgraduate study . Particularly for PhD researchers, the tertiary workforce is increasingly characterised by 1 3 New Zealand Journal of Educational Studies 57:505-523 precarious short-term employment, competition with colleagues and unequal pay as university management attempts to produce ever more with less . Student debt is also at unprecedented levels, with the removal of postgraduate allowances further amplifying reliance on student loans as well as other forms of debt . While many Doctoral graduates move into a variety of jobs outside of academia, this transition is often fraught and involves extended periods of fragmented and precarious work, while also often repaying significant debt . Studies have sought to explore the human costs of the financial conditions of graduate students. In a study of Otago University doctoral students during their first 2 years of candidature, Cornwall et al. identify how financial strain is interwoven with other stressors, which had a deleterious impact on students' wellbeing and performance. Their research showed that many students juggled employment, in addition to their full-time studies, to meet their monthly financial obligations, regardless of whether they received a scholarship. These demands of employment together with the personal demands of family and care work left many students with time-related stress and guilt about allowing job demands to take time away from their study. Some students also connected their financial constraints to a sense of isolation, for instance making it difficult to see family. In addition, both international and domestic students expressed financial worries about maintaining personal finances once scholarship funding finished, especially given uncertainty about the PhD structure and timely completion . These pressures on doctoral students are likely to last throughout the rest of their candidature and may become especially amplified in the months before submission. Other studies of doctoral students internationally have identified that inadequate funding can be a hurdle to successful graduate student completion . There are vital equity issues embedded in questions of financial strain on postgraduate students. For instance, Acker and Haque remind us of the many students who do not have the benefit of "financial cushioning" to enable them to pursue their postgraduate study unencumbered by debt and worry. There are also potential flow-on effects. Inequity and exclusion are persistent issues within New Zealand's research sector, despite commitments from the government and universities to value diversity and honour their obligations to equity under te Tiriti o Waitangi. Māori and Pacific scholars can experience systemic structural disadvantage and racism in universities, including the day-to-day reality of feeling isolated and the devaluing of Māori and Pacific knowledge, which can lead to the exclusion of Māori and Pacific bodies from universities . Research by McAllister et al. and Naepi on the current numbers of Māori and Pacific academics show severe underrepresentation of Māori and Pacific scholars in New Zealand's eight universities. Māori and Pacific comprise 16.5% and 7.5% respectively of the total population of Aotearoa, and yet comprise 4.8% and 1.7% respectively of academics. Naepi et al. have examined the period following a PhD for Māori and Pacific scholars. They identify a chronic under-representation of Māori and Pacific staff in both fixed-term contracts and permanent positions, suggesting that the pipeline into academia for Māori and Pacific scholars is fundamentally broken. Yet as the authors note, the idea of a pipeline alone is inadequate, and instead situate the experiences of PhD scholars within a wider context of navigating intersecting storms of racism, sexism, neoliberalism and neocolonialism, within the foundational whiteness of universities . In New Zealand, fewer women enter the post-degree academic career pipeline than men and even once they do their work in academia is undervalued. Women, whose research career trajectories otherwise resemble men's in terms of research performance, are promoted more slowly and have smaller incomes, resulting in a lifetime gender pay gap of ~ NZ$400,000 . This means that the financial strain experienced by women as postgraduate students does not result in equitable outcomes in terms of academic opportunity, promotion, and pay. This situation is exacerbated globally by new managerialism in academia, which promotes precarity with gendered consequences . There is no available data on the participation of people with disabilities in the New Zealand academy, but research in other countries shows that they are significantly underrepresented in postgraduate study, even compared to participation in undergraduate study . Academic spaces can be particularly contributors to poor mental health, with low pay and precarity being significant factors . LGBTQIA + scholars, particularly transgender scholars, have also been shown to have more precarious and stressful existences in academic spaces . For all these groups, financial stress during postgraduate study represents an additional barrier to academic participation. --- Methods There is currently a paucity of papers that explore inequities in scholarship values and how they have changed over time. Here, we have collated data about the value and tenure of scholarships from each of New Zealand's eight universities. All scholarships are tax-free. We present data on the value and tenure of PhD, Masters and Summer scholarships. We ask three questions of the data: how has the value of scholarships changed over time? is there a mismatch between the time it takes to complete a PhD and the length of PhD scholarships?; and how does the living cost contribution of the average doctoral scholarship compare to the Minimum Wage and the Living Wage? A request for scholarship data was sent to each individual university in July 2021. Universities are not individually named in the data we present as we felt that anonymity would increase their willingness to contribute data. We requested three sets of data: the value and duration of Summer, Masters and Doctoral scholarships from 2000 to 2020, 2) the length of scholarships and whether fees were included, 1 3 New Zealand Journal of Educational Studies 57:505-523 and the length of time to complete a PhD broken down by ethnicity and subject area. Doctoral scholarships and PhD completion data includes both Doctor of Philosophy and Doctor of Education. Our request was discussed at a Universities New Zealand meeting and UNZ then took it upon themselves to collate and provide the data request. UNZ acquired data for PhD completions through the IDI . The IDI is a large, research database which houses deidentified data about people and households. 1 Information on requirements attached to doctoral scholarships was obtained from University regulations published on their respective websites in October 2021 and presented in Table 2. 2 Where restrictions on the amount of paid work were specified on a per annum basis in university documents, we divided by 52 weeks to calculate the average hours allowed per week. Some universities were unable to provide the full temporal data set we requested due to historical data not being recorded. Data for Summer scholarships was particularly patchy, with one university not providing any data despite offering Summer scholarships. Limited scholarship data was provided as a range of minimum and maximum values, rather than a single value. This variation and range are presented in Fig. 1, and we present averages only in Figure S1. Additionally, we are aware that individual researchers occasionally top-up PhD scholarships, our data does not account for this and it is unknown how widespread this practice is. PhD completion data obtained through the IDI was based on EFTS and excluded suspensions. The median EFTS provided in the data abstract for fulltime students was then divided by 12 to calculate an approximate duration of PhDs. --- Results --- PhD Scholarships The average PhD scholarship value, which are tax-free, increased steadily from 2000-2010, increasing from $14,000 to $24,476 respectively . However, from 2011 to 2019, the average value of PhD scholarships remained stagnant. Over this period the highest and lowest scholarship values both changed by less than $1000. There was also large variation in scholarship values among universities . For example, in 2020, the largest PhD scholarship was $32,369 and the smallest was only $21,000 . In 2014, the Minimum Wage exceeded the average value of PhD scholarships by $364 and this gap has continued to increase over time. In 2019, the average PhD scholarship was $5417 less than the Minimum Wage and $11,238 less than the Living Wage . All PhD scholarships are 3 years in length. At one university, however, recipients have, for the past 15 years, been able to apply for a six-month extension on their scholarship. Most PhD scholarships include paying tuition fees, on top of living costs. One university only began including fees in 2019 . --- Masters Scholarships Masters scholarships have persistently been below the Minimum Wage, with the exception of one university in 2003 only . The average value of Masters scholarships increased from 2000 to 2008, but stagnated between 2008 to 2018. Interestingly, the average value of Masters scholarships has decreased in the past 5 years due to some universities introducing scholarships of lower values . The maximum Masters scholarship value has remained at $15,000 from 2003 to 2020 . The difference between the average Masters scholarship and the Minimum Wage in 2020 was $21,129.65 . In 2020, the minimum Masters scholarship offered was $5500 . All Masters scholarships are one year in length. Two out of eight universities did not include fees in their scholarships. For a single year, one university offered Masters students with disabilities a scholarship with a two-year tenure, but subsequently ceased offering this arrangement . --- Summer Scholarships Although there is some variation among universities, the value of Summer scholarships has remained persistently low in the past 18 years . The average value varied from $4000 in 2002 to $6000 in 2020 . Summer scholarships are usually for 10 weeks of full-time work. --- PhD Completions The median number of years to complete a PhD persistently exceeds the length of the three-year PhD scholarships regardless of the field study, by at least one year . The median number of years was highest in the Social Sciences and Management at 4.75 years and lowest in Engineering and Technology and the Natural Sciences at 4 years . --- Restrictions on Doctoral Scholarships Only three of the eight universities in New Zealand include compulsory student service fees or levies in their doctoral scholarships. Doctoral researchers at the remaining five must pay compulsory annual fees ranging from $610.00 to $1017.60 to maintain full-time enrolment and access to their scholarship income. Payment of doctoral scholarships are always contingent on full-time study, or prorated for part-time study where necessary. As such, recipients must agree to apply themselves to their studies full-time. Only three institutions specify an hourly expectation for full-time doctoral study. Victoria University of Wellington and Auckland University of Technology specify that a recipient must work on their research for at least 35 h per week . Massey University specifies that students are expected to engage in 40-50 h of study per week . Many institutions allow recipients to take reasonable holiday leave on an annual basis without it affecting their entitlement. No New Zealand university, however, formally stipulates a recipient's entitlement to annual, parental, bereavement, or sick leave. Every university, except the University of Canterbury , places explicit restrictions on the hours a recipient may work in addition to their research. Across these institutions scholars are limited to approximately 10-12 h of additional paid work per week, with some institutions specifying a weekly limit and others an annual limit of 500-600 h work within a calendar year. Such work is often subject to approval of a scholarship committee . Most universities place restrictions on co-tenure of their doctoral scholarships with other awards. In general, if a recipient receives other approved scholarships their university provided doctoral scholarship will begin to be attenuated once if it exceeds a specified amount. Most institutions allow a recipient to receive either an additional 30% or 50% of their scholarship value from another award before their entitlement is reduced. Massey allows co-tenure up to an additional $25,000; Lincoln caps recipient's income from awards and scholarships at $35,000 total. None of the New Zealand university generic doctoral scholarships impose conditions on a doctoral researcher's intellectual property -this generally resides with the researcher. --- Discussion Our findings have highlighted the inadequacies of research scholarships in Aotearoa. Not only is there a significant and growing gap between scholarships and Minimum/ Living Wage, but they are also insufficient in terms of the time that it takes to complete a research degree. The neoliberal logics that currently guide our tertiary sector provide the justification for paying postgraduate researchers below-living wage. The cost-sharing approach to postgraduate research, where universities provide financial support while scholars provide sweat equity, is skewed in favour of universities, who have kept their financial contribution low relative to inflation and cost of living such that it is now incommensurate with the skilled labour of researchers. The understanding, by governments and universities, that tertiary education is both a public and private good, paid for in a cost-sharing relationship, hides the reality of living on a subsistence-scholarship that simultaneously restricts earning power through contractual limitations on hours worked. In this discussion, we consider the potential implications of this growing gap for developing a more inclusive and equitable research sector. The gap between graduate scholarships and the cost of living raises concerns about who is able to engage in and complete postgraduate studies. There has been growing interest in developing supportive 'pipelines' into research to help address persistent inequities within the sector . Disengagement from postgraduate studies is, however, evident in the decrease of progression onto postgraduate studies from 2006 to 2019, where the progression to a higher degree within 3 years from Honours/Postgraduate Diploma decreased from 24 to 18% and from 7 to 6% for Masters to a higher degree among domestic students . The number of students transitioning to postgraduate studies began to drop-off in 2014, a year after postgraduate living allowances were cancelled. The data on rates of progression to higher postgraduate degrees suggests that the government's justification for continuing to delay the reinstatement of postgraduate allowances is faulty. These enrolments do not indicate an overall increase in the rate of participation. These concerns are especially apparent for Māori and Pacific scholars. While 1 3 New Zealand Journal of Educational Studies 57:505-523 there are higher or similar rates of progression to graduate studies for Māori and Pacific communities, there are lower completion rates for Māori and Pacific students in bachelor or higher qualifications . There is a lack of data on Masters and PhD scholars that would add the much needed in-depth understanding of how equity operates within the progression and completion rates, and how low scholarship values contribute to the ongoing under-serving of Māori and Pacific communities. The gap between postgraduate scholarships and the cost of living also suggests a hidden toll for many of those who do engage in postgraduate research. With no allowance, the gap between scholarships and living costs needs to be met through other means. This resonates with other studies that have highlighted the juggling of multiple sources of financial support . Those with the capacity to work in addition to their studies can make up some of the shortfalls; however, taking on the maximum allowed paid work in addition to full-time study results in a workload in excess 40 h per week which is nevertheless unlikely to pay commensurately with a living wage. Remaining shortfalls could be met through increased student loans , private debt and/or family contributions, if available. While there is a general increase in earnings for those who complete graduate degrees, these are field dependent and arguably not significant enough to justify high levels of debt or years of financial hardship. For instance, overall, the increase in earnings between a PhD and Masters 11 years after completing a degree is only $11,535 per annum, which arguably does not justify the increased debt taken on in fees and living costs. This costing does not take into account ethnic pay gaps which are shown to exist at a Bachelors degree or higher . There is also the issue of the length of scholarships. The higher education sector values doctoral completion rates, as evident in the Research Degree component of the Tertiary Education Commission's Performance-Based Research Funding which makes up 25% of the total yearly $321 million fund . This funding is calculated on a three-year rolling average of "research-based postgraduate degree completions, weighted by research volume, relative costs of the subject areas, and ethnicity and completions in Te Reo" . This creates a timeboundary for PhDs, as institutions can only claim up to 4 EFTS per doctorate per student, with any study exceeding that reported as "non-funded delivery" . Yet these incentives to complete PhDs within 4 years are only met by Engineering and Technology and the Natural Sciences, indicating that there is a disconnect between the length of time it takes to complete a PhD and funding incentives. While scholarship deadlines of 3 years are frequently described as a 'carrot' to encourage students to complete, completions across all fields of study are consistently at least a year longer, leaving students to attempt to make up this financial shortfall. There is an important debate to have about the length of PhD stipends and the incentives and restrictions these provide for students and for supervisors. The concerns about equity we raise in this paper should be centred in these discussions. Although it has been beyond the scope of this paper, the disconnect between the length of a PhD and scholarships has particular implications for international students, who are also working within visa requirements. Together, these shortfalls in scholarships mean that postgraduates who are not independently wealthy must accept sub-Minimum Wage incomes and a consequently reduced quality of life to engage in postgraduate study. The excessive workload necessary to meet the basic cost of living materially impacts the quality of research, as postgraduate students have less time to focus on their research, to rest, to think, and to engage in university life. It contributes to a culture of overwork and prevents working postgraduate students from accessing the opportunities necessary to cultivate a research career. This requirement of overwork and underpayment systematically disadvantages those with disabilities, caring or familial responsibilities, and those without access to independent wealth who need a reliable, liveable income. Postgraduates who supplement their income through additional work may need to transition to part-time study to meet expenses, creating additional years of income lost compared to other work. The high cost of accommodation also means students could be limited, for extended periods of time, to cheaper, poor-quality accommodation, often involving significant commutes to city campuses, with associated demands on their time and potentially detrimental effects to their health and wellbeing. These disadvantages are exacerbated for Māori and Pacific scholars, who on average earn less than their Pākehā counterparts, are less likely to have access to familial wealth, and have been subject to colonial brutality and disempowerment . These systemic factors exclude people from participating in postgraduate education, preventing them from realising the public benefits that such an education can produce. There are, however, relatively straightforward steps that could be taken to make a big difference for postgraduate students: 1. Raise Summer, Masters and PhD scholarships to match the Living Wage 2. Extend award tenure of PhD scholarships and national funding levers so scholarships reflect the actual time it takes to complete a postgraduate degree, including recognition of students with disabilities 3. Reinstitute the postgraduate living allowance As a result of emerging researcher advocacy, some of these actions are already underway. A promising example in Aotearoa is the raising of scholarships by the Cancer Society to $40,000 per year, plus $10,000 per year in tuition fees, and a tikanga contribution of $10,000 across the course of the award Awards 2021, 2021). A more ambitious and comprehensive approach to addressing these issues could follow international precedent and treat PhD scholars as fixed-term employees for the duration of their studies. This would give scholars access to a wage that appropriately reflects their skill and expertise, leave allowances, professional development opportunities, 1 3 New Zealand Journal of Educational Studies 57:505-523 and equal access to a union which could advocate for their working conditions . This approach to scholarships could work well as part of wider discussions about more varied PhD programmes and industry linkages. Related options could involve restructuring scholarships to account for housing cost and location, childcare, and other cost of living factors, even extending to providing affordable housing close to campus. Such approaches would aim to improve the educational experience and create a less inequitable environment for post-graduates by providing an immersive environment where they live and work on campus, with realistic living costs accounted for, without the need for supplementary employment. Greater use of submission scholarships may also be an important mechanism by which to support students through an often financially precarious time. By continuing to offer scholarships below Living Wage for highly skilled, home-grown researchers with the capability to work on world-leading research in Aotearoa, we devalue the knowledge that we develop and produce. At this point, it needs to be asked: who would-or is able to-sign up for a postgraduate degree with little income and low career prospects? It is increasingly urgent that we understand how universities have justified the levelling out of graduate scholarships so that we can challenge what is contributing to a decrease in graduate progression, inequity in our universities and growing frustration from graduate students about their 'stolen wages' . ---

This paper responds to calls from past and present students to increase the value of postgraduate scholarships in Aotearoa New Zealand. Here we provide context for understanding the scholarship landscape in Aotearoa, including how scholarships are understood in relation to dominant neoliberal framings of higher education and persistent inequities within the sector. We present data which provides insight into the current inequities in Summer, Masters and PhD scholarship values. The average value of PhD scholarships has remained stagnant between 2011 and 2019 resulting in the average being $11,238 less than the Living Wage in 2019. We show that the average length of time full-time PhD students take to complete their doctorates exceeds the three-year tenure of scholarships. We argue the status-quo of low scholarships, supplemented by postgraduate 'sweat', excludes people from participating in postgraduate education, preventing them and their communities from realising the public benefits that such an education can produce. We suggest that these inadequacies could be addressed through (1) raising Summer, Masters and PhD scholarships to the living wage; (2) extending tenure of PhD scholarships; and (3) reinstating the postgraduate student allowance.

Introduction Parents or also known as families as the first educators who shape the attitudes and personality of a teenager in the family and are responsible for the education of their adolescents both formally and informally as well as religious education as a foundation for their children. To form a reliable attitude and personality in facing future challenges. Families or parents are the main and first educators for their children, therefore it is from them that children first receive education, thus the first place of education for children is in family life. Parents are the closest relatives who have services and also provide love, attention, and protection all the time to their children because parents are the first and main educators for their children. Adolescents are a national asset that is not merely the object of development itself but also the subject of development. Adolescents have the right to obtain family education, especially those related to sex, which aims to prevent adolescents from falling into negative or deviant sexual behavior such as promiscuity. Adolescence is a phase of development between childhood and adulthood, Adolescence in psychological science is also introduced with other terms, such as puberties, adolescence, and youth. In Indonesia, it is often associated with puberty. Zakiah Derajat suggests that adolescents are children who are in the transition period between childhood and adulthood, where children experience rapid changes in all fields. Adolescents are no longer children, both in body shape, attitude, and way of thinking and acting, but also not mature adults. Entering adolescence, a teenager will consider himself not a child anymore, so adolescents are more difficult to manage, have a curiosity about something very high, love adventure, and always want to try risky challenges, without prior consideration. When they do something, they only think about short-term risks. Sex education is an education that is no less important than other education to be given to children. The role of parents in providing sex education to children is very large because they are the closest people to children, so it is hoped that parents will be able to become the first source of information for children. Sex education is also better done as early as possible so that it can be a prevention of sexual harm. Especially in this era of globalization, the development of the times is increasingly rapid, the flow of information spreads so fast that everyone can have access to any source of information, as well as increasingly diverse associations in the community, especially the behavior of teenage children. In addition, at this time there are very rampant cases of sexual violence against children or adolescents committed by the closest people including family. One of the causes is that children do not have the provision of knowledge that can make children or adolescents anticipate the possibility of bad treatment for sex problems. For this reason, it is very necessary to have proper sex education for teenagers so that they can face various possibilities that occur. --- Method The author uses a type of qualitative research with a library research approach that describes. Library research is an activity to gather information that is relevant to the topic or problem that is the object of the research. This research. Various information obtained from some of the literature is then synthesized which eventually becomes a conclusion. The conclusions are then arranged logically and systematically descriptively to answer the main problems in this study. --- Results and Discussion Adolescence is a period of high curiosity, parents as the main factor in providing the most important education for children, providing sex education and knowledge about adolescence is the main step to avoid children from the influence of promiscuity, providing knowledge about sex is a good provision, for educators and parents as well as adolescents themselves need to know adolescent development Thus, educators and parents can overcome the problems of these adolescent children --- Definition of sex education Sexual education according to Sarwono quoted by Reny Safita is something that happens information about human sexuality issues that are clear and correct including the process of conception, pregnancy until birth, and sexual behavior that is given should be related to the norms that apply in society. Sex education is one of the sciences that are given to children about everything related to sex including everything related to sex how the development of male and female sex, menstruation, wet dreams, and hormonal changes, including marriage, childbirth, and others related to sex. Gunaso said that the purpose of sexual education is to help a child's emotional attitude toward a mature and responsible life towards his sexual life. Stanton and Brenna argue that sexual education is character building, sexual education is not only providing information but also providing insight to children so that children can build personal values in their growth so that children have a broad view of the world providing provisions to children for what they need to instill in children to make good decisions they make and form a good personality towards children with religious values. The Urgency of Sex Education Sex education to children is the responsibility of parents, especially during adolescence with sex education to teenagers, making them avoid promiscuity and free sex whose risks can destroy the future. The role of parents in the growth and development of children is very important, parents as the first education, as the most pending in the growth of children, in the age of globalization parents must place themselves as friends. The relationship between parents and children must be well established, with parents as the main figure in the child's life. The establishment of good communication between parents and children will create openness between parents and children. --- The Purpose of Sex Education According to Lilik , there are several objectives of sexual education in adolescents including: 1. For adolescents to obtain correct, clear, and accurate knowledge about sexual life such as reproductive organs and their functions and care, sexually transmitted diseases , healthy sexual behavior, and so on. 2. So that adolescents can manage their sexual urges appropriately 3. To have healthy behavior related to their sexual life 4. To be able to practice religious law correctly about their sexual life 5. Does not fall into promiscuity that abuses his/her sexual life 6. Can avoid deviant sexual behavior such as masturbation/onanism, sodomy, incest 7. Avoid sinful or adulterous acts. --- Teenagers 1. Definition of teenagers Adolescence is a transitional period from childhood to adulthood which generally starts at the age of 12 or 13 years and ends in the late teens or early 20 years . Meanwhile, according to Gunarsa . Adolescence comes from the Latin puberty or puberty and adolescence which means youth between 17 and 30. So in conclusion, adolescence is a transition period from children to adulthood starting at the age of 12 years to the age of 25 years --- Characteristics of Adolescents An adolescent is at the transitional boundary between child and adult life. Even though his body looks "grown up", when it comes to acting like an adult he fails to show his maturity. They have not had much experience with the adult world, so the following things are often seen in them. a) Restlessness, a state of unease prevails in adolescents because they have many desires that cannot always be fulfilled. b) Full of contradictions. The contradictions that occur within themselves also confuse them, both for themselves and for financial matters. c) Great desire to try everything he doesn't know yet. d) The desire to explore the surrounding nature in adolescence is broader e) Daydreaming and fantasizing. http://dx.doi.org/10.30983/bicc B I C C P R O C E E D I N G Aulia Muharani, etc 152 The role of parents in providing f) Group activities. Between one desire and another, challenges often arise . --- Adolescent Sexual Development Hurlock suggests the tasks of adolescent development in terms of sexuality in full include : a. Acquire correct sexual knowledge. b. Developing a pleasant attitude towards the opposite sex. c. Developing mature relationships with the opposite sex. --- Parents' Understanding of Sex The success of learning activities is related to the quality and skills of educators . Parents are the central educators in providing sex education to children, so understanding what will be taught is a big factor in determining success. From the results of the study, it was found that the definition of sex can be categorized into: sex is related to biological and physical sex is related to psychological matters sex is related to culture and morals sex is related to social issues Sex is also related to biological and physical issues. In addition, sex is also seen as something related to psychological issues. Sex is also related to culture and morals. This was revealed by an interview with one of the subjects and sex is also related to social issues, this can be seen from the interview excerpt on one of the subjects. --- Important Supervision in Sex Education --- Supervision in Behaviour and Speech Supervision of the essential foundation of paternal and maternal shyness in all matters of family life can create security and safety for the lives and education of adolescents. To prevent teenagers' excessive attention to sexual matters, parents should guard their speech and avoid using words that have to do with sex; refrain from telling stories, jokes, and banter related to sexual matters; and not allow their marital relations to be seen by teenagers. In some families that do not pay attention to the upbringing of their children, fathers and mothers along with young children and teenagers sit together watching films that show scenes of people making love and special relationships between men and women. Such programs will stimulate and cause perversion in children and adolescents, and lead to excessive attention by adolescents to sexual matters. Parents should be careful in their behavior towards adolescent girls at the age of puberty, avoiding jokes, sing-songs, smiles, and gestures related to sex, because adolescent girls are more perceptive than parents imagine. Scientifically speaking, the absence of supervision in this matter will lead to faster sexual development, the emergence of untimely tendencies, and the filling of the minds of adolescents with disturbing things. Even a wife and husband sleeping in the same bed in the presence of a young girl will disturb her mind. --- Bed supervision Sexual tendencies will be more apparent in quiet times, especially during sleep. When going to bed, teenagers must be tired, because staying in the guarded condition above ends badly can lead to deviation, both in thought and action. Children should be taught from a young age to sleep in an unlocked room, not to sleep face down, and not to wear cramped nightgowns. It would be better to tell them to go to the bathroom before going to bed and put their hands outside their clothes. --- Clothing Supervision What needs to be considered in this matter is the way parents and young women dress. In dressing, mothers should try not to be seen with their bodies exposed in front of their children. Similarly, fathers should not wear inappropriate clothing in the home environment and leave parts of their bodies exposed. Young women should avoid wearing men's clothes, as well as short, tight, delicate, and thin clothes, as this will have undesirable effects. Tight clothes, for example, not only interfere with the development of the body but also reveal the curves of the body. This can be a medium for sexual perversion. --- Health Supervision Maintaining a healthy body is an obligation and a must for everyone, especially teenage girls. Maintaining the health of reproductive organs, especially on menstrual days and before, has a very important role. A mother, especially in the first year of life, bathes her daughter, cleans her, and so on. --- Conclusion Based on the results of the study, the researchers concluded that the role of parents in providing sex education to adolescents is very important in providing learning and being a guide for adolescents because parents are the closest people in adolescence. Parents must be careful in behaving towards adolescent girls in the period of puberty, avoiding jokes, singing Gungan, smiles, and gestures related to sex, because adolescent girls are more responsive than parents imagine, sexual education is helping a child's emotional attitude towards a mature and responsible life towards his sexual life.

This research was conducted to explore and know the role of parents in providing sex education to adolescents and preventing premarital sex. The research method used is a type of qualitative research with a library research approach to collect various information and obtain in-depth information. The data collection technique used is library research. the object of research and documentation while the data analysis technique used is by reducing data, presenting, and verifying data. The results of the study show that the role of parents in providing sex education from an early age has been carried out but has not been optimal. The roles parents have carried out include setting an example of loving their bodies, watching their children while watching television, teaching them not to undress in front of the opposite sex, and providing clothes according to the sex of the child. While some things have not been done optimally, such as giving an example of protecting oneself from sexual violence, here we also see the role of parents in educating their children to protect themselves from deviant acts.

Introduction Currently, access to the Internet through smartphones has led to their functions going beyond purely communicative ones, allowing the management of massive, instantaneous, and easily accessible information. More and more young people are acquiring a smartphone at an early age . Home confinement by COVID-19 imposed the need to adapt to the new reality using technology to ensure that private and working life could continue under health restrictions. Smartphones, more than ever, have been an excellent tool for communication and information management, which has boosted the use of these devices in all areas of daily life . Smartphones give individuals responsibility for their health, education, security, home, etc., assuming the position of basic goods that can improve daily routines by downloading Apps and, at the same time, help to cope with the emotional burden of the pandemic . Thus, they promote better use of resources by housing multiple functions in a single device . However, the advantages offered by using smartphones can lead to the need for and pathological addiction to this device, which can lead to a pathology known as nomophobia. Nomophobia disorder manifests as an irrational fear or feeling of anxiety at not having one's phone, not using it, running out of battery, or not having or losing an internet connection, which deprives individuals of constant communication . Although nomophobia is not marked in the Diagnostic and Statistical Manual of Mental Disorders , it is commonly perceived with the phobia criteria outlined in the DSM-V such as excessive and unreasonable fear or anxiety associated with an unintended object or situation, which causes immediate anxiety. Despite the person recognizing that her fear is disproportionate and avoidance of the feared situations, at the same time, we can observe disruption of routines and relationships due to the phobia. In this case, in 2014, Bragazzi and Del Puente, proposed to consider the inclusion of nomophobia in the DSM-V given its clinical relevance, epidemiological characteristics, and the importance of this condition shortly, as evidenced . The possibility offered by smartphones to obtain easily accessible communication over time encourages people to be permanently connected and can lead to an addiction problem Young people use smartphones much more intensively than other sectors of the population, as social and interpersonal relationship needs take on central importance. This population group is therefore much more prone and vulnerable to smartphone-related problems . In addition, social networks aggravate problematic use. According to the latest research , there are differences between men and women in the time spent using smartphones, which is predominantly intense and constant. In addition, the exceptional situation brought about by the arrival of COVID-19 led to new forms of smartphone use by young university students who were in a situation of home confinement . Age and gender have also been indicated in numerous studies as conditioning factors in nomophobia. Likewise, young women score higher on the scales for this disorder, due in part to their greater use of social media platforms. However, smartphone use is becoming increasingly homogeneous across individuals, which casts aside the classic gender roles set by the collective imagination . Regarding this last point, it is worth noting that the scientific literature and studies on technology use by young people have been lacking over the last decade in comparison to the burgeoning current research . There is currently much controversy regarding gender as a modulating factor in nomophobia. There are previous studies that show differences between men and women in nomophobia, with women being more likely to have this disorder . Similarly, there are studies in which no differences were found according to gender . In this sense, it is interesting to know whether Health Sciences students suffer from nomophobia and whether there are differences based on gender. The study aims to describe the use of Mobil devices and to analyze the presence of nomophobia in first and second-year university students in the Faculty of Health Sciences at the University of Zaragoza, with stratification of the results by gender, degree, and population nucleus where they usually live. --- Materials and methods --- Study design We performed an observational, descriptive, cross-sectional study during April 2020. The study complies with the STROBE checklist for cross-sectional studies. --- Population and scope of the study The study was conducted at the FHS of the University of Zaragoza with university students enrolled in their first and second year at the faculty who fully and correctly answered the questionnaire. No probabilistic purposive sampling by the investigators was used. The results were obtained during the questionnaire carried out in April 2020. Of the total number of students that made up the study population , the sample calculation for a confidence level of 95% and a margin of error of 5% was 237 students. The final sample consisted of 318 participants. Inclusion criteria refer to the student in the faculty of health sciences of the University of Zaragoza, with informed consent signed, and exclusion criteria were those participants with incomplete or incorrectly completed questionnaires which, therefore, were not useful for the study. --- Data collection instrument or procedure Due to the lockdown ordered by the Government of Spain as a consequence COVID-19 outbreak, information was gathered using an online form on the Google Apps For Education platform. It is a package of Google tools and services designed for educational centers and allows collaboration and learning safely and securely. In the present investigation, this tool was accepted by the Research Ethics Committee of the Community of Aragón and the University of Zaragoza, Spain. The data were protected and each student could only access the questionnaire once. Participants received an email containing the survey to their university email account, which remained freely open for participants throughout April 2020. The email contained a link that directed them to the questionnaire on the GAFE platform. When they accessed the questionnaire, participants were asked to sign their consent to participate in the study and were given a brief explanation of the guidelines to continue and access it. The questionnaires were registered and coded with random numbers and letters. Nomophobia was assessed using the Spanish version of the Nomophobia Questionnaire scale . With an ordinal alpha of 0.95 for the total score, this scale has 20 items, with 7 answer options . Scores range from 20 to 140, with which higher scores are linked to higher levels of nomophobia. The 15th percentile pertains to occasional users or not nomophobia , the risk zone is from the 80th percentile , whilst problematic smartphone use starts at the 95th percentile . The Nomophobia Questionnaire in its Spanish version confirms the factorial structure of the Spanish version of the NMP-Q of Gonzalez-Cabrera et al. and the original four-factor model of Yildirimy and Correia conducted through a non-probabilistic sampling on 5,012 participants, showing a hierarchical model of 4 factors with good psychometric fit. --- Study variables Primary variables included questions on student sociodemographic characteristics , academic education , and their smartphone . The second variable includes questions about NMP. --- Statistical procedures Descriptive statistics . The distribution of categorical variables is shown by absolute and relative frequencies. Distribution normality was verified using the Shapiro-Wilk test. The accepted significance value throughout the entire study was 0.05, assuming a confidence interval of 95%. Accordingly, values below 0.05 were considered significant. A univariate and bivariate analysis was performed. In the bivariate analysis, the relationship between the quantitative and qualitative variables was studied. The mean difference of the continuous variables was compared. In the cases where at least one quantitative variable was compared, after rejecting the normality of the groups, a nonparametric equivalence Mann-Whitney U test was conducted, which yielded the value of p for the compared groups. Where the surveyed population was stratified into three groups, first the normality and equality of variances were tested. Next, an ANOVA test was run for equal variances, and a Kruskal-Wallis test was to determine if there were variations from the normality. The analysis was performed using the RCommander program . --- Ethical considerations This study was developed following the Data Protection Regulation 2016/679 of the European Parliament and the Organic Law 3/2018 on data protection. The information was treated confidentially and anonymously as the data were dissociated. The study was approved by the ethics committee of Aragon . --- Results --- Sociodemographic characteristics The sociodemographic characteristics of the population under study are shown in Table 1. They indicated that 81.8% of the surveyed students were women and 18.2% were men. 79.56% of the students were from a city, whilst 20.44% had their normal residence in a village. The mean age was 20.4 years . There was a clear predominance of female students at a general level, which was maintained in all the degrees based in the Faculty of Health Sciences. However, there were notable differences between the physiotherapy degree, with a masculinity ratio of 0.4, and the occupational therapy or nursing degrees where the masculinity ratios were between 0.16 and 0.14, respectively. --- Smartphone use and risk of nomophobia The students used their smartphones for approximately 4 h per day and checked them on average every 35 min as shown in Table 2. The analyses showed that men spent a little over 3 and a half hours per day on their smartphone, whilst women did so for a longer average period of 4 h and 35 min as well as Occupational Therapy students who reported using their smartphone an average of 4.75 h per day. On the other hand, 86.48% of students stated they were able to spend 24 h without their smartphone, although almost the same percentage stated they would not be able to go without it indefinitely. Nevertheless, a higher percentage of female students than male students reported daily dependence on their smartphones . On the other hand, they were not found by a degree or population nucleus . According to the rating of NMP-Q, it was observed that rating between 34-72 there is no nomophobia, 73-94 risk of nomophobia, and 100-140 nomophobia. After analyzing the student scores on the NMP-Q scale, it was observed that 45.28% of the population in the study were not homophobic, 37.42% were in the risk zone and 14.78% of the students had nomophobia . --- Relationship between gender, degree, and population nucleus Table 5 shows the results of the analysis of the nomophobia variable in the NMP-Q scale. The results are broken down by student gender, university degree, and population nucleus where they usually live. Regarding gender, the mean score was higher in women than in their male peers, with a maximum score of 132 points compared to a maximum of 119 points in men. However, no significant differences Frontiers in Public Health 04 frontiersin.org were identified in nomophobia scores based on student sex . Related to the university degrees that were present in the study were degrees in nursing , physiotherapy , and occupational therapy . Nursing students presented a mean score of 73.66 points in the NMP-Q, future occupational therapists had 70.42 points, and physiotherapists had 70.21 points. The data shows that no significant differences were found for nomophobia scores according to the degree . According to the type of population nucleus in which the students usually resided, the results showed no significant differences in their score on the NMP-Q scale . --- Discussion The COVID-19 confinement provided an opportunity to investigate the prevalence of nomophobia among health science students during a period of health crisis. The smartphone is currently the device of choice, as it allows information to be obtained and communicated continuously and instantaneously. This communicative function contributes to the satisfaction of individuals' need for social relationships . Especially during late adolescence and later youth, interpersonal relationships acquire greater relevance, as they favor the feeling of acceptance by the social group . Young people use more intensive the smartphone, which is reflected in the hours of use of the device and the constant checking of it . Along the same lines as the present investigation, several studies showed high results regarding the time spent using smartphones by young people . The COVID-19 pandemic highlighted the need to use technology to maintain communication with the rest of society. University students manage their daily activities autonomously and rely mainly on their smartphones for this purpose . As in the present study, the time that young people use their smartphones daily exceeds 3 h in most studies and even exceeds 5 h . Most studies agree that women tend to use their smartphones more daily for socializing . In addition, there is an underestimation of smartphone usage time among students, since, as indicated by Cabrera García-Ochoa university students claim to use the smartphone between 2 and 4 h per day and, however, they claim to feel annoyed for not being able to check the device in class. Checking the smartphone is done routinely and automatically, creating the false perception of less exposure time to the device . In this way, the constant use of the device and its main function in the lives of young people can lead to its dependence and the development of affectations such as nomophobia . Analysis of the prevalence of nomophobia during the month which coincided with the strictest home confinement measures in Spain, provides insight into the prevalence of this disorder at a time when information and communication technologies were a key element in being able to maintain educational, work, and personal activity . Coinciding with the above results, the works of Daei et al. , Cain et al. , and Moreno Guerrero et al. showed moderate nomophobia rates very similar to those of the present investigation. In some studies, this proportion tended to increase . The elevated risk of nomophobia among young people prevails over milder forms of dependence. It has been observed that the nomophobia score increases especially when there are situations that make interpersonal relationships impossible, such as the impossibility of contacting or being contacted by family and friends . The need for social relationships that the smartphone allows satisfying through its various functionalities is the factor that drives young people to make intensive use of this technology to expand their relational sphere . According to Braña Sánchez and Moral Jiménez , greater use of the phone can predict an increased risk of nomophobia, as well as generalized anxiety. Mainly, the authors relate these factors to dependence on social networks, which leads individuals to need their smartphones excessively, in turn generating disorders such as nomophobia. As in several studies , even though the highest nomophobia scores on the Nomophobia Questionnaire scale were recorded by women, no significant differences were found Frontiers in Public Health 05 frontiersin.org with their male peers. The results of the present investigation contribute to the existing literature on nomophobia, which presents discrepancies in terms of gender . In Spain, the majority of university Health Sciences studies are attended by women, with rates of around 80%. This fact implies that in the present research, a population sample with homogeneous characteristics was obtained . Regarding the different university degrees studied, the future nurses obtained a mean score of nomophobia that exceeded the risk percentile on the NMP-Q scale without any significant differences with the Occupational Therapy and Physiotherapy students. Similarly, nomophobia in Health Sciences studies is high, especially in the Nursing degree . Berdida and Grande , in their research with nursing students, found a relationship between their nomophobia and a lower level of attention and motivation. Nomophobia represented a danger to their health, but also to their patients. One of the main functions of future health professionals is vigilance and clinical safety, which was altered when attention was reduced by the urgent need to consult the smartphone . The COVID-19 pandemic made it impossible for students acquisition of practical skills, to attend physical classes and contributed to the transition to the implementation of all their educational activities through distance learning technologies . During this health crisis, in which information on the evolution of COVID-19 was mainly obtained through the smartphone, reinforces the results obtained in the present investigation of the extensive time spent using the smartphone. COVID-19 signified a risk of suffering from nomophobia in students. The results obtained showed a moderate nomophobia score as in similar studies on the university population . Reduced face-to-face social interactions generated a context of high anxiety and intense smartphone use that contributed to the development of nomophobia during this period . --- Strengths and limitations One of the limitations was that almost 82% of the participants were women in a study stratified by gender, which is an enormous limitation. Although multiple comparisons analysis provides confidence in the results of the present study, future research should include a more heterogeneous sample so that comparisons by gender have more statistical power. In addition, although there were no age differences among the participants, it would be interesting to conduct future studies that include larger age ranges. However, due to the exploratory nature of this study, these results are considered a preliminary test for future research to address the problem of nomophobia among the young university population. Furthermore, it offers a reading of a current social disorder at a time of epidemiological crisis. Specifically, future work comparing the results before and after the COVID-19 health crisis in similar populations would be beneficial. --- Conclusion The study indicated an intensive use of the smartphone that can be classified as dysfunctional, with a large proportion of health science students showing a risk of suffering nomophobia. On the other hand, it should be noted that the students of the different degrees in the research did not use their smartphones in the same way, with daily use being higher in occupational therapy students. In any case, concerning the rates of nomophobia presented by the students, no significant differences were found by gender. Likewise, no relationship was found between nomophobia and the degree and the type of population center to which the students' normal residence belonged. Increasingly, relationships between individuals and between individuals and institutions are mediated by electronic devices. In this context, it is necessary to know whether this relationship with technology is healthy or could turn into a health problem. For this reason, the research focuses on future health professionals, who will be in charge of following the trajectory of nomophobia and establishing measures or action plans according to the current situation. To this end, future research related to this disorder is required, including a more heterogeneous population with a larger sample size. --- Implication for practice This fact reveals a unifying break and identifies smartphone use, as popular as it is necessary for young people, as a common, homogenous trait. These findings may contribute to a better understanding of the use of mobile phones, especially when the time spent on and commitment to them may increase. Intense and widespread use of smartphones by young people contributes to the need to improve the technological education of this population group, orienting the guidelines to the healthy use of smartphones. Prevention of nomophobia and health promotion are key for educational projects to contribute to reinforcing healthy Classification of nomophobia levels according to the Nomophobia Questionnaire Spanish version. --- Data availability statement The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. --- Ethics statement The studies involving humans were approved by The ethics committee of Aragon . The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study. --- --- --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Introduction: Currently, access to the Internet through smartphones has led to their functions going beyond purely communicative ones, allowing the management of massive, instantaneous, and easily accessible information. This research analyzed the differences in smartphone use and the prevalence of nomophobia, mainly according to gender and university degree of Health Sciences students at the University of Zaragoza during the COVID-19 confinement in Spain. Methods: A descriptive cross-sectional study was carried out on a sample of 318 first and second-grade students, who completed an online questionnaire sent to their institutional email, which included sociodemographic questions, other questions about smartphone use, and the Nomophobia Questionnaire (NMP-Q) scale. Results: Compared to men (n = 58), women (n = 260) were more likely to use their smartphones more intensively daily, as were occupational therapy students compared to the other degree programs studied. The prevalence of nomophobia was moderate, being around the risk of suffering from it. No significant differences in scores for nomophobia among students were found according to gender, university degree, or population nucleus for the nomophobia scores of the students. Discussion: The present study extends the existing literature on nomophobia by providing results of interest in terms of gender and the exceptional healthcare context of COVID-19. The results suggest that despite intense daily smartphone use young people did not reach severe nomophobia figures. This fact underlines the need for appropriate and healthy technology education. Understanding the characteristics of the populations that use the smartphone the most may help to analyze nomophobia rates and the massive use of the device.

Plain English summary The nonuse of contraceptive methods by women who need contraception may impact the lives of these women and the planning of their families. One example is unintended pregnancy and its possible implications. Research has identified several reasons for the nonuse of contraceptive methods. However, studies have mostly only focused on women who are married or who live with a partner. In this study, we included all sexually active women in each country; doing so allowed a more comprehensive overview of the reasons for why women whose demands for contraception are not satisfied do not use contraceptive methods. We presented the differences between the reasons for nonuse of contraceptives reported by partnered and unpartnered women. Data from 47 low and middle-income countries were obtained from the Demographic and Health Survey Program for analysis. The results showed that, on average, about four out of ten women in need of contraception were not using any contraceptive methods. There was great variation among countries in the reasons reported. Overall, the most common reasons for nonuse were "health concerns" and "infrequent sex." Nonuse due to "opposition from others" was more common among married women. It was noteworthy that wealthier women often mentioned "health concerns," "infrequent sex," and "method-related" as reasons for not using contraceptives. The present research is important in helping professionals from various fields understand why women who need contraception do not use contraceptive methods so as to offer guidance and provide adequate care. --- Background The nonuse of contraceptive methods by people who need contraception has been associated with potential implications at the individual, familial, community, and global levels. Sustainable Development Goals [1], as well as the proposal of maternal and child health indicators [2], have been essential in increasing the visibility of this topic and highlighting points for improvement. In addition, several agencies and organizations around the world have engaged in global efforts to finance actions that promote family planning, especially in low and middle-income countries [3][4][5]. However, despite these efforts, the sexual and reproductive rights of women are not always respected. Although there is evidence of progressive increases in the use of contraceptive methods [6], many women still face various barriers to contraceptive use [7]. It is estimated that 214 million women in LMIC in need of contraception do not use any modern contraceptive methods [8]. Considering that all women, as advocated by the Sustainable Development Goals, should have access to sexual and reproductive health services, understanding the reasons behind why this need remains unfulfilled is essential for action planning. An investigation performed by Sedgh and Hussain [9] on 51 LMIC showed that among married women, the most frequent reasons for nonuse of contraceptives were low frequencies of sexual relations, and fears of side effects and potential health risks. However, a major limitation of that study is the absence of unmarried, sexually active women in the study sample. The reasons for nonuse are likely very different between married and unmarried women, and this is possibly linked to social barriers and other difficulties related to access to contraceptives that unmarried women face. However, literature concerning family planning lacks information regarding reasons, reported by the women themselves, for not using contraception. As discussed by Sedgh and Hussain [9], the analysis of the reasons for nonuse in different population subgroups may provide relevant information to support the design of specific initiatives. Added to that, there is increasing evidence in scientific literature of the importance of using measures that are not only expressed as aggregate data [10], that is, of evaluating the specificities of each country and the possible inequality contexts. Considering the situation, the present work aimed to address this gap in knowledge. In addition to evaluating the reasons for nonuse of contraceptive methods by sexually active women with demand for contraception not satisfied in LMIC, subgroup analyses of the reasons for nonuse were performed to investigate possible barriers and to propose strategic priorities to improve women's sexual and reproductive health. Furthermore, the analyses of inequalities might identify what strategies may be more suitable for each subgroup. --- Methods We analyzed data from the Demographic and Health Surveys Program for 47 countries. The study focused on the most recent surveys, with data collected in and after 2010, to provide a current overview of the topic investigated. The Multiple Indicator Cluster Surveys were not included, since the data of interest could not be retrieved from these surveys in a manner that allowed for comparison. DHS are standardized surveys that use a cross-sectional design to collect data from LMIC. All surveys use similar questionnaires, methodology, and sampling strategies, which ensures the comparability of results. In general, the areas of interest are initially identified through censuses or other procedures. After that, households are selected based on preestablished methodological steps [11,12]. The present analysis included the latest available DHS with information on demands for family planning and covered all sexually active women of reproductive age in each country. Six surveys that did not contain information on sexually active women living without a partner were excluded . --- Definitions and outcomes The study's primary outcomes were the reasons reported by sexually active women of reproductive age for not using contraceptive methods. The indicator Demand for Contraception Not Satisfied utilizes the number of women who are not using any contraceptive methods in its numerator and the number of women of reproductive age in need of contraception in its denominator. In other words, the study included women who were in need of contraception but did not use any contraceptive methods. Women in need of contraception were defined as those who were sexually active and fertile, but did not intend to become pregnant in the next 2 years or did not know if or when they intended to become pregnant [2,11,13]. Sexually active women were defined as those who were married or lived with a partner and those who were not married but had sexual intercourse in the 30 days prior to the survey. Infertile women excluded from the study were defined as those who: 1) had never had a period or were amenorrhoeic in the 6 months before the survey even though they were not in the postpartum period, 2) were hysterectomized, and 3) had been married or in a stable relationship for 5 years or more but did not become pregnant despite a lack of contraceptive use [11,14]. Although pregnant women who considered their pregnancies untimely or had not desired their pregnancies are generally classified as in need of contraception [11], they were also excluded from this study because being pregnant is one of the filters for the questions regarding the reasons for nonuse of contraception. The reasons for nonuse were evaluated using the following question: "You have said that you do not want any children. Can you tell me why you are not using a method to prevent pregnancy? Any other reason?" The reasons for not using contraceptive methods were grouped into eight broad groups of reasons : 1) "respondent opposed" , 2) "other opposed," 3) "lack of knowledge," 4) "health concerns," 5) "lack of access," 6) "method-related" , 7) "fatalistic" , and 8) "infrequent sex." Because women could report more than one reason, each response category was analyzed separately. --- Statistical analysis Descriptive analyses were used to identify the main reasons for the nonuse of contraceptive methods among women whose demands for contraception were not satisfied in the various LMIC. When analyzing the reasons for nonuse, the countries were ranked according to the level of demand for contraception not satisfied: less than 30.0%, 30.0% to 50.0%, and more than 50.0%. Considering these coverage levels, the reasons for nonuse were also evaluated according to the following stratification variables: marital status ; woman's age ; area of residence ; woman's education ; parity defined as the number of live births ; wealth index . The wealth index is calculated using principal component analyses which take into account characteristics of the household, ownership of selected assets, and educational attainment of the head of the family [15,16]. In addition, inequality analyses were performed using the Slope Index of Inequality . Instead of calculating the simple absolute difference in the prevalence of the outcome of interest in the wealthiest vs. poorest quintile, the SII measures the difference in predicted coverage using a statistical model for these quintiles. In other words, the SII takes into account the prevalence of the outcome of interest in the five wealth quintiles to estimate the absolute difference between the wealthiest and the poorest quintile [15,16]. All statistical analyses were performed using the STATA 13.1 statistical package [17] and took into account the sample weights of surveys through the command "svy." --- Results Considering the 47 countries analyzed, the mean prevalence of demand for contraception not satisfied was 40.9% . Seventeen countries had a prevalence of demand for contraception not satisfied higher than 50%, and in five of these countries , the prevalence was 70% or higher. The Republic of Chad had the highest prevalence . The two countries with the lowest prevalence of demand for contraception not satisfied were Colombia and Honduras , both located in Latin America and the Caribbean. "Health concerns" was the top reason for nonuse in 22 out of the 47 countries. In 18 countries , "infrequent sex" was the most prevalent reason . In Gabon, the same prevalence was recorded for both "health concerns" and "infrequent sex," and the two were also the most common reasons for nonuse in this country. "Other opposed" was the most prevalent reason in three countries . In Gambia, "respondent opposed" was the most prevalent reason for nonuse whereas in Ethiopia and Niger, "fatalistic" was the most common reason . The prevalence of the reasons reported were also different according to the level of demand for contraception not satisfied . On average, "health concern" and "infrequent sex" were the most common reasons for all levels of demand not satisfied, but the prevalence of these reasons was higher among countries with demand not satisfied below 30%. All the other reasons tended to be more prevalent among countries with higher demand not satisfied. It is important to note that the most common reasons for nonuse of contraception varied greatly between countries. In the five countries with demand not satisfied of 70% or more, the most prevalent reasons for nonuse were: Angola -"health concerns" , Mali -"other opposed" , Gambia -"respondent opposed" , Guinea -"other opposed" , and Chad -"infrequent sex" . Stratification provided an overview of the situation in the countries according to different subgroups of women. These analyses are presented by level of demand for contraception not satisfied: countries with demand not satisfied below 30% , from 30% to 50% , and above 50% . We also provide the stratified estimates for each reason by country in the supplementary material . These analyses reveal important differences on why some women do not use contraception. Nonuse due to "health concerns" was more frequently reported among women who were older, had higher levels of education, had more wealth, were married or living with a partner, had at least one child, and among those living in urban areas. "Infrequent sex," the second most reported reason, was higher among those who had higher levels of education, had more wealth, were unmarried, and had no children. In all the coverage levels assessed, a difference close to eight percentage points was observed for the comparison between married and unmarried women in "infrequent sex," with the latter being more likely to mention this reason (Figs. 1, --- and 3). There were no marked inequalities in the reporting of nonuse due to "respondent opposed." On the other hand, nonuse as a result of "other opposed" occurred mostly among married women rather than unmarried women in all coverage levels assessed. However, the prevalence of this reason was higher among countries with higher demand not satisfied. For those with demand not satisfied of > 50%, "opposition by others" was reported by 15.1% of married women and 8.5% of unmarried women while for countries with demand not satisfied of < 30%, this reason was reported by 7.8% of married women and 3.6% of unmarried women . In countries with demand not satisfied between 30 and 50%, there were higher inequalities observed in terms of wealth and education; poorer and less educated women were more likely to report this reason . Nonuse due to "fatalistic," "lack of access," and "lack of knowledge" were more frequently reported by women in the poorer groups, those with lower education levels, and those who lived in rural areas . "Fatalistic" as the reason for nonuse was also systematically higher among older women and those with higher parity. "Lack of knowledge" was also about twice higher in rural areas than in urban areas for all levels of demand for contraception not satisfied evaluated. Nonuse due to "lack of access" was also higher in rural areas than in urban areas for all levels of demand for contraception not satisfied , corresponding to, respectively, 5.3%, 3.5%, and 4.6% in urban areas, and 6.3%, 6.1%, and 8.0% in rural areas . Regarding the inequality among the reasons for not using contraceptive methods by country, determined by the SII, the results clearly show that nonuse due to "health concerns," "infrequent sex," and "method-related" showed pro-rich inequality patterns in most countries, which means that these reasons were more prevalent among the richest women . On the other hand, the reasons "other opposed," "fatalistic," "lack of access," and "lack of knowledge" were linked to patterns of pro-poor inequality , which means that these reasons were more prevalent among the poorest women. --- Discussion This study evaluated the reasons for nonuse of contraception among women with demand for contraception not satisfied in 47 LMIC. The mean prevalence of demand for contraception not satisfied in this group of countries was 40.9% . Regarding the reasons for nonuse by women with demand for contraception not satisfied, it is noteworthy that "health concerns" and "infrequent sex" were the most prevalent reasons in many countries. Five countries, all of which are located in Africa , presented prevalence of demand not satisfied of > 70.0%. In West and Central Africa, where many of the lowincome countries analyzed are located, a study identified inequalities and a higher probability of long-term contraceptive use by the wealthier women in this region than by the poorer women [18]. This underscores the need to empower women and the importance of learning about the unique characteristics of each region, for example. Conversely, Colombia and Honduras , located in Latin American and the Caribbean, had the lowest prevalence of demand for contraception not satisfied. The literature on Latin America and the Caribbean describes a decline in fertility rate over time [19] as well as other demographic changes, such as aging populations [20]. In this context, reproductive health policies, among other strategies, play an important role in family planning [21]. Some advances also may be attributed to the availability of a variety of contraceptive methods [20]. It is possible that these various processes together may explain the low levels of demand for contraception not satisfied identified in Colombia and Honduras. However, it is important to highlight that the low prevalence of demand for contraception not satisfied in these countries do not reflect an ideal scenario, although some of these countries have low levels of demand not satisfied, the optimal prevalence for this indicator would be zero. All women in need of contraceptives should have access to them. Practicing contraception does have side effects, and nonuse due to fear of these consequences was addressed in the reason "health concerns." Conversely, one possible result of not using any contraceptive methods is unintended pregnancy [22]. It should be noted that, besides the associated stigma, unintended pregnancy may have other consequences for the woman and her family of which the general population is often unaware. Such consequences may include negative health outcomes for the woman and the child as well as social costs [23][24][25]. However, it should be noted that unintended pregnancies are not always regarded as negative outcomes by women and their families, which shows the importance of contextual variables [26]. Nevertheless, studies also suggest that the possible side effects of contraceptive use are few compared to the possible risks resulting from some types of pregnancies [27][28][29]. Women in the highest wealth quintile had a higher mean prevalence of nonuse due to "health concern." This raises the question of whether these women have access to a wide range of contraceptive methods or whether this access is restricted to only a few contraceptive options that may not meet individual demands. In this study, we showed that some participants mentioned "method-related" reasons for not using contraceptives which may indicate dissatisfaction with the existing methods of contraception; this supports the need for a wider range of contraceptive options, as pointed out by Barot [30]. In addition, health care professionals need to provide evidence-based information about the methods Fig. 1 Reasons for nonuse of contraceptive methods according to stratifiers in countries with demand for contraception not satisfied < 30.0% Fig. 2 Reasons for nonuse of contraceptive methods according to stratifiers in countries with demand for contraception not satisfied between 30.0 and 50.0% ( of contraception and listen to women's beliefs and opinions so that women can make informed decisions about the most appropriate contraceptive methods for them. Women who reported "infrequent sex" as a reason for nonuse may consider themselves less likely to get pregnant, as suggested by Sedgh and Hussain [9] and Barot [30]. Over time, women have taken on new roles in society, and there is evidence of declining unmet contraceptive needs [6,31]. Work opportunities may sometimes lead to geographical relocation [9,32,33], which may then cause couples to live apart. In this situation, the professional advantage obtained may come at a personal cost with regards to women's reproductive health decisions. These aspects may explain, at least in part, some of our results and had more schooling had a higher mean prevalence of nonuse due to "infrequent sex"). It is important to provide these women information on the functioning of the reproductive system and the importance of contraceptive use, and offer them support for reproductive health decisions. Long-acting reversible contraceptives may be a good alternative for these women. Some women avoid using contraceptives due to "other opposition" which can be their partners, other people from their families or communities, or even their religions. Women sometimes face obstacles in their relationships with their partners , and this can be inferred from the stratified analysis since in general, the mean prevalence of this reason was higher among married women than among unmarried women. This is striking, as all women should have the right to make their own reproductive decisions. This also highlights the need to empower women, as doing so will allow them to have autonomy over their own bodies and lives and make informed decisions regarding contraception [34]. It is possible that cultural and personal issues, which were not captured in the health surveys, may also be involved in women's decisions to use or not use contraceptive methods. Some women are engaged in homoaffective relationships, which is another aspect that is not evaluated in the surveys. In this case, there would be no concerns regarding using contraception to prevent pregnancy, although the use of condoms, for example, would be essential in preventing sexually transmitted infections. In regard to women who reported a "fatalistic" or "respondent opposed" reason, it is possible that this involves beliefs or other personal issues that require further investigation. In countries where nonuse due to "lack of access" is high, investments made to increase the availability of contraceptive methods is essential. Beyond access, it is Fig. 3 Reasons for nonuse of contraceptive methods according to stratifiers in countries with demand for contraception not satisfied > 50.0% also imperative that health care professionals are prepared to provide information on the specific characteristics of existing contraceptive methods to ensure that people can choose the method that best fits their individual needs [23,35,36]. However, it is noteworthy that some women face a variety of barriers [7,37] when attempting to fulfill their reproductive health needs. Some limitations of the present study must be addressed. First, since countries without information on unmarried, sexually active women were excluded, some regions may have been underrepresented. In other words, the group of countries that were investigated may not represent the entire region or all LMIC. In addition, in presenting means, the data may differ from individual results, as substantial variations were identified. In this sense, it is relevant to pay attention to the data from each individual country to understand what these data reveal about the local reproductive health situation. Being currently pregnant was a filter for the DHS questions regarding reasons for not using contraceptives. However, it is important to understand whether these women had a contraceptive failure or whether they were not using any contraceptive methods . Another limitation is that, by grouping reasons such as nonuse due to religious concerns and nonuse due to partner opposition similarly under "other opposed" , it is impossible to detect whether religious constraints predominated in some countries over other persons' opposition. As well, we cannot rule out the possibility of differences in individuals' understandings and interpretations of the survey questions, and thus in the way the themes under investigation were reported. Nevertheless, the present study also has many strengths, such as the use of recent data for a large number of countries and the inclusion of information about both married and unmarried women. This provides a comprehensive overview of the reasons for nonuse of contraception, and is a step forward in relation to previous studies [9,11]. In addition, the present study only focused on, for the calculation of the indicator, women in need of contraception and provided knowledge on reasons why contraceptive methods are not used, with stratification by subgroups. By directly pinpointing groups in need of interventions, we can better fulfill the contraceptive needs of women. Understanding the reasons behind why women are not using contraceptives has many potential implications for reproductive health policies. Identifying the opinions of women themselves on the topic and detecting possible difficulties they face may serve as a tool to help professionals from diverse fields keep in mind the importance of increasing access to reproductive health services, empower women to make their own reproductive decisions, and provide necessary information and support to ensure that women are capable of overcoming barriers. --- Conclusions "Health concerns" and "infrequent sex" were the most frequent reasons for nonuse of contraception in many LMIC, though the prevalence of each reason varied across different countries and subgroups of women. To decrease the prevalence of demand for contraception not satisfied, it is necessary to design, test and implement contextualized interventions. Generally, it is necessary to listen to women's beliefs and opinions, provide women with more information on the importance of using contraceptives to avoid unwanted pregnancies and sexually transmitted infections, inform them of the contraceptive methods available and their side effects, and increase their access to a wide range of contraceptive methods. the project, and participated in all subsequent steps. All authors read and approved the final manuscript. --- --- Supplementary information Supplementary information accompanies this paper at https://doi.org/10. 1186/s12978-019-0805-7. Additional file 1. Reason respondent opposed according to stratifiers in women with demand for contraception not satisfied. Authors' contributions LRM and FE reviewed the literature, analyzed and interpreted the data obtained from female participants, and wrote the manuscript. FE and AJDB contributed to data interpretation, substantially revised the manuscript, and produced the final figures and tables. MFS conceived the work, supervised --- --- --- Competing interests The authors declare that they have no competing interests. ---

Background: Nonuse of contraceptive methods by women in need of contraception may impact their sexual and reproductive health. The aim of this study was to describe the reasons for nonuse of contraception among women with demand for contraception not satisfied in low and middle-income countries (considering both overall countries and various subgroups of women). Methods: We used the latest Demographic and Health Survey data from 47 countries. A descriptive analysis of the reasons for nonuse of contraceptive methods was performed among sexually active women with demand for contraception not satisfied. The prevalence of each reported reason was also evaluated according to marital status, woman's age and schooling, area of residence, wealth index, and parity. Wealth-related absolute inequality for each reason was also evaluated using the Slope Index of Inequality. A pro-rich inequality pattern means that the reason is more prevalent among the richest women while a pro-poor means the reason is more common among the poorest ones. Results: On average, 40.9% of women in need of contraception were not using any contraceptive methods to avoid pregnancy. Overall, the most prevalent reasons for nonuse of contraceptives were "health concerns" and "infrequent sex," but the prevalence of each reason varied substantially across countries. Nonuse due to "opposition from others" was higher among married than unmarried women; in turn, the prevalence of nonuse due to "lack of access" or "lack of knowledge" was about two times higher in rural areas than in urban areas. Women with less schooling more often reported nonuse due to "lack of access." Pro-rich inequality was detected for reasons "health concerns," "infrequent sex," and "method-related", while the reasons "other opposed," "fatalistic," "lack of access," and "lack of knowledge" were linked to patterns of pro-poor inequality. Conclusions: Family planning promotion policies must take into account the different reasons for the nonuse of contraceptive methods identified in each country as well as the contextual differences regarding women of reproductive age (such as social norms and barriers that prevent women from accessing and using contraceptives).

INTRODUCTION The knowledge economy of the 21st century requires skills such as creativity, critical thinking, problem solving, communication and collaboration -skills that cannot easily be learnt from books, but rather through learning-by-doing and social interaction. Big ideas and disruptive innovation often result from collaboration between individuals from diverse backgrounds and areas of expertise. Public libraries, organisations and coworking spaces have been continuously seeking for means to facilitate social encounters and peer collaboration to nurture crosspollination of skills, creativity and innovation. The more diverse the people that an institution hosts, the more potential there is for social and collaborative learninghowever, the social atmosphere appears to be more public and less familial. This controversy is illustrated by the social space and user interactions usually found in public libraries. The library is one of the few remaining "truly" public places that is frequented by people from a broad cross-section of society with a high diversity of socio-cultural backgrounds and areas of expertise. As such it has a high potential for mutual inspiration and cross-pollination of skills, knowledge and experiences among library users. However, the library also appears to be perceived as a typical "third place" in the public realm where users usually regard each other as strangers. People mostly work within their "individual bubbles" , many even weaving "an individual net around themselves that does not invite communication with others" , for example, by marking their work space with coats, bags, notebooks, and other possessions . In general, library users perform their individual activities next to each other, without taking advantage of the social capital and knowledge networks within the community of other users. Whilst isolated work should not be discouraged -in fact, the library as a place for individual study and rejuvenation is highly appreciated by users and needs to be preserved as such -there is untapped potential for serendipitous social learning, inspirations and the creation of social capital . Especially in today's new economy with an increasing amount of independent, self-employed, and project based workers, the role of public libraries, as socially inclusive spaces for coworking, is more important than ever before. How can public libraries cater for the social needs of coworkers , such as opportunities through social interactions and ability to share knowledge with other coworkers? This paper contributes new insights and knowledge to the question how digital technology can support the design of collaborative interactive spaces. We present the design and evaluation of 'Gelatine,' an ambient media system we developed in the context of a case study at The Edge -a dedicated space for social and collaborative learning at the State Library of Queensland. Gelatine is a real-time user checkin-system that makes visible the invisible social aspects of the library as a 'place,' in particular, by displaying currently physically present users' backgrounds, skills, and interests, on a public screen inside The Edge. --- THEORETICAL FRAMEWORK: SPACE, PLACE, AND SENSE OF PLACE Scholars distinguish between the concepts of space and place . While the term 'space' primarily refers to geometrical and physical configurations of infrastructure, 'place' covers a broader concept embracing a social layer of meanings that people attach to a space. People attach social attributes such as meanings, memories and experiences through their everyday practices, activities and interactions with and in a space, thus render space into 'place.' The same location might have a different meaning to different people, depending on their subjective and emotional relationship to a place. Space is designed and built by architects and spatial designers, while place is a social construct -created by the way people make sense and use of it. Placemaking, then, as practiced for example by urban planners or architects, is an attempt to design spaces that accommodate and invite particular activities and spatial behaviours. However, as Harrison and Dourish point out, the question if and how a place is actually produced, is not in the hands of a designer, but subject to how users appropriate and make use of that space. "Placeness can be designed for, but it can't be designed in […] The best that the designers can do is to put the tools into their hands. Trying to do moretrying to build places -is not our job." De Certeau's concept of 'tactical practices' , Levebvre's distinction between perceived, conceived and lived space, and Soja's notion of counterspaces further describe the gulf between how designers often imagine a space being utilised and perceived, and how people actually interpret, make sense of and use space. The different interpretations and meanings that people attach to a place collectively shape a sense of place , and shape the behavioural norm of people at this place. What does this mean for the design of collaboration and learning spaces? Library designers, for example, put much effort into the physical design of lounge areas, meeting rooms, cafés and other 'open' spaces that invite social interaction, collaborative work, peer-to-peer learning, meetings and social hangouts . However, in practice, there seems to remain a social barrier to peer interactions and focused encounters between most library users. The behavioural norm is to work in isolation from other users . Physical infrastructure that accommodates and invites social learning is not sufficient to turn a social learning space into a place where social learning is actually being practiced and experienced. We identified such a gulf between designed space and lived place in our previous fieldwork at The Edge . The motivation to bridge this gulf marked the point of departure for this study. Can a public screen application that highlights particular social aspects and use patterns of a place, re-inforce and amplify a particular 'sense of place' that users would not be able to perceive otherwise? The following section discusses previous work on public screens, before we discuss the design rationale for Gelatine. Struppek provides a summary of projects demonstrating the 'social potential' of urban screens, i.e. screens that display cultural content and support the development of a local community around those screens. The common denominator of most such public screen applications is their aim to foster social behaviour between people with weak ties or no ties, in order to counteract the natural behavioural patterns of 'civil inattention' between unacquainted people. Previous studies have explored the impact of public screens in various settings, e.g., organisation and workplaces , third places or urban outdoor places . Results of those studies report that public screens can serve as an icebreaker for conversations , increase awareness amongst colleagues as well as social capital and participation among coworkers . In contrast to other social networking systems that foster relationships between people with already existing strong ties , situated social software applications, such as CoCollage , help create links between co-located people in the same space, i.e. "help people who are in the same physical context become friends, or at least become more familiar strangers" . Such links do not necessarily have to involve direct faceto-face interactions, but can be indirect, or what Goffman refers to as 'unfocused interactions' . Willis et al. introduces 'shared encounters' as a term to describe interactions between people who share the same physical context, i.e. interactions "between two people or within a group where a sense of performative co-presence is experienced and which is characterised by a mutual recognition of spatial or social proximity" . 'Digital encounters' as defined by Fatah et al. , then, are shared encounters that are mediated by digital technology, such as a mobile phone or public display. Fatah et al. show that digital encounters can create new forms of situated interactions that would not evolve without the use of technology. However, as Konomi et al. state, "digitally augmented settings may not effectively support encounters, unless it is integrated with human interaction processes and social conventions" , Digital systems that are designed to facilitate shared encounters need to be sensitive to the socio-cultural peculiarities and function of the particular place of installation. So far, only little is known about the design and potential social impact of situated social software applications displayed on public screens in library environments. Previous work on interactive public screens in libraries focused on facilitating serendipitous discoveries by providing an alternative, digital access point to the library's book archives, e.g. visualising the circulation of checked out books , or providing flexible, artistic, animated and playful interfaces to explore library archives. Gelatine has a different focus; it does not facilitate connections between users and books, but rather connections among the user community. It aims to enhance and reinforce the vision of contemporary libraries as places for social learning, participation, interaction and collaboration. The following section describes the background and design rationale behind Gelatine. --- RELATED WORK: PUBLIC SCREENS AND SHARED ENCOUNTERS --- GELATINE: SYSTEM DESIGN The design of Gelatine follows the idea of Commons 2.0 . Whilst libraries have long been perceived as 'gatekeepers' for information and knowledge , Commons 2.0 puts a strong emphasis on social constructivism , collaboration and co-creation of knowledge. Commons 2.0 recognises and promotes the library user themself as an asset and resource for information, inspiration and social learning to other, co-present users. The design rationale behind Gelatine is based on extensive previous ethnographic field research at our case study site at The Edge , and identification of five user personas with different preentry motivations, perceived barriers and needs. The results reveal a mismatch between the vision of the State Library of Queensland for The Edge as a place for social learning and collaboration, and users' actual predominant patterns of isolated work and rare interactions with other coworkers. The design was specifically targeted at two issues: We found that users lack a perceived affordance to identify other users as potential information resources. Whilst library catalogues, themed bookshelves and the labels on book covers provide easily perceivable affordances to search, browse and find printed information material, the physical environment of most libraries does not communicate much about their user community and the collective intelligence it has to offer. As a consequence, users of The Edge, as a Commons 2.0 library space that went entirely bookless, are often puzzled about what the space is for and what one can do there to begin with -they lack a sense for The Edge being a place for social learning and collaboration. Instead, perceiving many other visitors working in isolated silos reinforces a sense of place that counteracts its intended purpose. As one user stated, "when I go in there, I don't really talk to anyone that I don't know, just because they are kind of already doing stuff mostly individually, but I don't know… it feels like the wrong type of people are there." Aiming to bridge these two barriers, Gelatine was designed as a combination of a checkin-system and a real-time public screen installation that highlights the skills, interests and help requests of currently present users in the space . Rendering such invisible social resources visible, the two design goals were to enhance people's perception of other users in the space as a source for incidental as well as goal-directed social learning, and promote The Edge as a place for social learning and collaboration. Gelatine provides an online / mobile website for users to create a personal profile with keywords that describe their skills, areas of interests, as well as areas that they have a problem in or want to learn more about. This profile information is linked to their RFID membership card, which they can swipe at one of the 'checkin-points' at the entrance of The Edge or sub locations such as individual workspaces, computer lab or coffee kiosk to confirm their presence in the space. RFID cards, as a method for a seamless user checkin process, were selected in respect of the library's institutional mission as a socially inclusive space . Smart phones or other 'personal' devices could discriminate against users who do not own and cannot afford them. Each checkin point is made up of a network controlled RFID / NFC reader box, that, every time a user checks in, attaches a timestamp to the user's ID and saves it to the Gelatine system database. A custom designed visualisation on each of two public screens is updated in real-time according to the profiles of those users who checked in. The two screens are designed to answer three basic questions: How can other users help me? How can I help other users? What areas of interests can I engage in with other users? Each of the two screens displays a tag cloud visualisation of keywords that describe the skills and needs of all currently checked-in library users. Tag clouds were selected as a visualisation technique in order to facilitate browsing and serendipitous information discovery . Further, in contrast to explicit search interfaces, tag clouds are better suited for the presentation of ambient information that can be perceived in the periphery of people's attention. Figure 2 shows an installed prototype of the Gelatine screens, as well as their location in the floorplan as installed at The Edge. In order to avoid the 'screen blindness' phenomenon often found with people passing by public screens , and to represent available skills and knowledge grouped by individual users rather than in an aggregated tag-cloud form, we added an additional 'lowtech' display next to the two public screens: We installed a small, networked thermal printer that prints out a user profile ticket for each user that checks in. The user can then decide to pin his ticket on a blackboard to make their skills apparent and available to other users in the space. The two screens and the ticket blackboard provided a central display of engagement opportunities with currently co-present users . In order to encourage face-to-face encounters, users can click / tap on the respective tag to find the user behind that tag if they are curious about a particular skill. The associated user information and location are polled from the checkin-system database in real-time and displayed in a pop-up window. To provide a sense of currency, the different tags in the clouds are colour-coded according to the respective checkin-timestamp. Tags of users that checked in most recently are displayed in bigger font-size and red colour. Keywords of check-ins further in the past are coloured in green , blue and white . --- METHODOLOGY In our evaluation, we were interested in what social impact Gelatine had on users, i.e. how it impacted users' perception of and relationship to the social environment at our case study site. Previous studies that discuss the social impact of public screens pre-dominantly employ long-term, longitudinal evaluations over a period of a few months or years . However, in order for a public screen to have a social impact, it needs to embody a satisfying standard of usability, as well as attract awareness to start with. Such questions require thorough, site-specific trial and error testing of different human-computer interaction specific aspects of public screens, e.g., their position and orientation or means to raise by-passers' attention and motivation for use . Conducting the evaluation from an early stage of the development came with a tradeoff: In order to identify socio-cultural barriers and challenges relevant to the system design, thus be able to inform the re-design process from an early stage, it was crucial to evaluate the system in the 'wild' . On the other hand, a pre-mature user interface and usability issues might not attract users to naturally use the system in their everyday context. We applied a mixed methodology to tackle this challenge: As the system was yet to be approved by the State Library of Queensland for an official launch, we were only allowed to recruit a number of selected pilot study users who would create personal profiles in the system that we used as valid 'user generated' content for the evaluation of the public screens to be installed in the main visitor area at The Edge. We recruited 21 pilot study users who all form part of "Hack The Evening" -a group of technology enthusiasts who meet at The Edge every Thursday night to discuss and collaborate on projects around hacking and making things with hardware and electronics, and do-it-yourself technology. We selected "Hack The Evening" as a representative user group who uses The Edge for collaboration and social learning purposes, while remaining open to new members. The pilot study users checked in during their weekly "Hack the Evening" meetups at The Edge, as well as their casual visits at The Edge. The Gelatine screens and profile printouts were set up at the main foyer of The Edge and evaluated in a user study for a total of six days. In the user study we not only observed the 21 pilot study users who had a user card to actually perform a check-in, but were actually even more focused on other visitors who came to The Edge and encountered the screen during their everyday visits. We did observations for a period of 3 hours every day, and varied the timeslots between afternoon, lateafternoon and evening . We engaged 24 users in follow-up interviews , who we selected according to different interaction patters with the screens. Michelis and Müller identify six different phases of user interactions with a public screen, i.e. passing by, viewing and reacting, subtle interaction, direct interaction, multiple interactions, and follow-up action. They provide empirical data showing that these phases form an 'audience funnel' -only a certain percentage of users transitions from one phase to another, and eventually, only a fraction of users ends up engaging in direct or multiple interactions with the public screen. Our approach was to sit in a distant corner with a clear line of sight on the screens and covertly observe users ignoring, or going through and/or dropping out of these six phases in the audience funnel. We then approached the users after they finished their interaction, or passed by the screens, and asked them for a follow-up interview where we would ask them to reflect on their experience of viewing or interacting with the screen. Users who had only passed by ignoring the screen, we later 'took by hand' to the screen and asked them to 'think aloud' as they interacted with it. In total, we interviewed 13 users who ignored and passed by, 7 users who viewed and reacted, and 4 users who directly interacted with the screens or printouts . Interviewing users who engaged in viewing and reacting, we focused on how the screen affected them as an ambient information display . With users who directly interacted with the screen, we focused on the results and motivations behind them initiating a face-to-face encounter as a follow-up action to their interaction with the screen. Each user interview involved a 3-step process: We first asked questions about the user's general visiting and use patterns as well as their perceptions of The Edge. How often do they visit? What activities do they normally engage in when they visit? What are their relationship, perception and usual interaction patterns with other users? This first phase was to help us match the interviewee to one of the five use personas that we identified in previous extensive field work at the same case study site , and evaluate how Gelatine impacted different user groups. We then revisited the Gelatine screens with the user, and asked them to reflect on their earlier interactions as they went through the audience funnel and 'think aloud'. The questions were open ended, and targeted at shedding light on their impressions and perceived usefulness of the system and information on the screen. The third step included an open discussion about concerns and suggestions for future versions of Gelatine. Each interview went for an average of 20-30 minutes, and every user was offered a coffee voucher as a compensation for their time. All interviews were audio-recorded and transcribed. For the data analysis we borrowed grounded theory techniques, and categorised the user comments according to emerging and reoccurring reactions to Gelatine, and how it has or has not impacted their sense of place in comparison to their previous use patterns at The Edge. --- DISCUSSION OF FINDINGS The following sections discuss our interviewees' thoughts, impressions and reasons for use or non-use of the screen throughout the different phases of the audience funnel . --- Passing by Even though the screens were placed in the middle of the foyer , most people who entered during our 3 hour observation slots either completely ignored or only took short notice of them . Only 4% actually interacted with the screen. We followed-up with 13 users who ignored the screen and interviewed them on their pre-entry motivations, use patterns and attitudes towards the space. 12 out of those 13 interviewees reported to be regular users, who had been visiting The Edge for at least three months on a weekly or sometimes daily basis. They all come to The Edge with a particular pre-entry motivation, i.e. a purpose of completing a set of pre-defined tasks. Upon entering The Edge, these users follow established routines without paying much attention to signs, posters or installations in general. Their main focus of attention is to occupy a space and work on the tasks they came in for. For example, we encountered a backpacker who, for the past two months would come in every day to access a computer and search for jobs on the Internet; an Indian student and her tutor would use a work lounge to study English three times a week; another daily visitor -a young man -would come in to read online news, watch YouTube videos and download online e-lectures to avoid stressing the download limit of his Internet plan at home; a self-employed programmer uses The Edge between two to five times a week to work on his freelancing projects and avoid the distractions from home. In contrast, the few users who actually viewed and reacted or directly interacted with the screen were mostly people who were new to The Edge. We usually observed such users wandering around the space, looking at posters, brochures and signage in an attempt to find out what this place is about. Eventually, they would stumble upon the public screens and visitor profile printouts and spend a few minutes browsing through them. Some regular users would view and react to the screen during short work breaks, e.g., on their way back from getting a coffee at the internal café or the toilet. In summary, these observations show that people, who have more open and exploratory attitudes towards the space, are more likely to naturally push through the 'audience funnel' than others who follow established routines and tasks in the space. To those, who actually noticed and interacted with the screen, its nature as an ambient information display turns out to be crucial for them to perceive the displayed information in the first place. Had the information been hidden behind a website, none of these users would have actively sought or become aware of it. --- Viewing and Reacting The impressions and reactions upon viewing and reacting to the screens were mixed, and depended highly on the individual user's pre-entry motivations and attitudes towards the space and other people in the space. --- Non-use Four interviewees, after recognising the social intentions of the system, reported that they do not have any interest in socialising or meeting other users. This matched our observations of tactical practices these interviewees and many other users apply to isolate themselves from the social environment in the space; they, for example, put on headphones or close the curtains around their workspace, signaling their desire not to be disturbed or interrupted in their activity. These following statements represent typical responses we heard in the follow-up interviews: "I am here to find a job, not interested in much else to be honest." "I am here to prepare for my English test, why would I bother talking to anyone else?" --- "I don't normally communicate with people. I don't come here for social purposes […] I don't think I would use [Gelatine]. I can see how it's a good service, and how a lot of people might benefit from it, but I am more of an individual user. I have my own interests and stuff that I look at." For such users -with their own pre-defined work agenda that involves isolated work, and without intrinsic motivation for social learning or shared encounters with others -a system like Gelatine does not make a difference; they choose to remain non-users . --- Amplified sense of place For first-time visitors, who, upon entering are generally curious about the purpose of the space , the screens and user tickets provide a lens that 'amplifies' their perception of the socio-cultural environment that is embodied, but underrepresented in the place. In contrast to many posters, brochures and signs that were put up to promote library offers, such as workshops or events, the screens provided a real-time glimpse into other users' activities and interests. By making visible these invisible sociocultural aspects of a place, new people to the space perceive a sense of place that would be invisible or hard to grasp otherwise. The invitation of the screens to approach other users for providing or seeking help frames their notion of the purpose and function of The Edge as a place for collaboration and learning with others. As a first-time visitor couple, who stumbled upon the screen, stated in their follow-up interview: They were able to construct an idea of the purpose of The Edge as a space for collaboration and social learning, which they -prior to the installation -would struggle with if not explained by a staff member. Further, the nature of the screen content is perceived as constructed bottom-up, i.e. socially constructed by people, rather than imposed top-down the library. For example, in contrast to a billboard at the entrance of the space, the public screen does not tell what the space is built for from the space designers' and planners' point of view, but rather how it is actually being used by other users . --- "When you read you see someone is offering [...] you realise it's made by people; this is not like just made up by someone, you know, you kind of recognise this is sort of a social thing. Somehow you do, I don't even know how I do. You just kinda get it…" . As one interviewee pointed out, the screen provides an 'implied consent' that the users who signed up for Gelatine are happy to be approached and open to conversations: "… I guess the fact is that there is this sort of implied consent to have that happen, because they've put their stuff up there; you know it's not just I've come here to work in the space privately and now I'm gonna have ten people who are interested into what I am interested in and come and harass me" This implied consent further strengthens the notion of a place where meeting and interacting with strangers is socially accepted. Some first-time visitors that we interviewed have marked The Edge as a future destination for serendipitously meeting likeminded and skillful people. "It [looking at the screen] definitely makes me think that I should come down here to do some of my work. Just the idea of being in the same space is kind of cool […] I don't know, the potential that I could end up having a conversation instead of doing my work, that's always nice… yeah like, there's all sorts of strange people here." They remember The Edge as a place where they are likely to meet people that have the skills and interests that are displayed on the screen. perspective what the space has been built for, Gelatine displays how the space is actually being used by its users. --- Amplified sense of coworkers In our previous user study at The Edge , we identified that one of the core motivations for regular users coming to the Edge is the ability to work in a social environment among other users, and away from the distractions of home. The exponential rise of professional for-profit coworking spaces since 2006 underlines the desire of independent workers to work amongst others, as well as most coworkers' desire for social interactions , the ability to share knowledge , and the serendipitous opportunities facilitated through such interactions . In contrast to the isolated work attitudes of non-users' discussed above, users with coworking attitudes at The Edge are generally open and glad for serendipitous discoveries in their environment. They appreciate Gelatine as a virtual window that triggers curiosity, and provides inspiration and stimulation by other users. --- "Even if I wouldn't actually go and chat to anyone there, but it still looks nice; friendly; even if you don't do anything about it, if you just read it then I think it's good for you. It gives you an idea about what are the people using this space for. I think that's interesting" . Whilst our interviewees reported that they were obviously aware of their coworkers' presence, the profile information on the screens and printouts revealed new facts about their skills and backgrounds. One regular coworker that has been coming to The Edge almost every day for over a year, was surprised by the amount and diversity of skills he found in the tag cloud. After having scanned different keywords in the tag cloud, he stated: "I am actually learning new things about this place that I didn't know. I thought there was much less than what's there [on the screen]. To be honest, I really only thought it was design, video and music. I didn't think there was anything more than that […] I thought everybody who is in here is very like... sort of arty; it's interesting to see that there is a lot to offer here." . For another interviewee, who generally tends to work alone, the screen raised some awareness and thoughts of other coworkers that he did not have before. "I guess I hadn't thought about why other people come here. I am usually in my own headspace when I come here. So it's interesting to see that so many people use it for such a diverse range of uses." . This increased awareness of interests and skills of strangers in the same space, affords serendipitous encounters and inspiration. Another interviewee, for example, who saw 'scuba-diving' in the tag cloud reported that "[scuba-diving] is something that's not even actually offered here [by the library], but you end up doing it because you were here […]" He felt that, by incidentally sharing the same space with other users at the same time, he has got access to their interests and areas of expertise -"…now you can really say you are at the right place at the right time. I mean if you are here, and this person is here as well or you have got their email.". --- Direct Interaction and Follow-up Action The goal in this phase was to evaluate the system against the opportunity it provided to identify the location of and ice-break face-to-face conversations with like-minded users or users with complementary skills. We regarded the follow-up action as a user initiating a face-to-face conversation as a result of their screen interaction. Only 6 users of all people we observed during our observation timeslots made it naturally to the direct interaction phase. Out of those we experienced two successful face-to-face encounters as well as two unsuccessful attempts to initiate an encounter. We asked those users to reflect upon their experience, as well as gathering additional thoughts and feelings from other users who we manually guided through the direct interaction phase. One successful face-to-face encounter was initiated by I12 -a university student of IT and digital media, and regular user at The Edge for three years. He reported a frequent issue he faces when working on his university assignments: "…sometimes I am stuck, so I have to go and search, and search, and search, and try find specific things. If you're learning a new program, you need to go on the Internet, but I wish someone just tells you what to do" . During our pilot study, he stumbled upon a profile printout from a user at the "Hack The Evening" group who had specified skills in 3ds-Max -a software tool for 3D modeling that I22 needed for a university assignment. That day, he deliberately stayed at The Edge for longer than usual in order to join "Hack The Evening" for their meetup at 5.30pm. Meeting his target user later on, they spent the following 2 hours discussing different techniques and alternative developer tools for 3D modeling. I22 was also given a list of web links, e-books and other free sources to learn about 3D Modeling. He decided to come back for the next "Hack The Evening" meetup, as reportedly, he has learned more, and in an easier way than he would have by himself. This example shows a successful face-to-face encounter that was mediated by Gelatine, and more so, resulted in a social learning experience. Users take Gelatine as an 'implied consent' and invitation to initiate new face-to-face encounters. I14, for example, approached an electronics hobbyist who was offering his expertise on 'Arduino' , and had a chat to him about how he could apply Arduino for his hobby in creating new interfaces for electronic music instruments. Reflecting upon his follow up action, he stated: "…so what's the worst, like, they just think I am idiot and I just have to leave. I could probably deal with that, I just never come back again . I think it was doable." . For a user to initiate a conversation, a personal benefit has to arise from that conversation. "If it was particular to my exact interest, than yeah. I would go 'oh wow'; I would go and have a chat with them." . Initiating conversations just for the sake of socialising and personal chit chats without any deeper purpose does generally not appeal. "So if there is an actual thing where there's gonna be some sort of transaction then yes, but just for general chit chat probably not..." Similarly, another interviewee stated his strict interest in professional skills, rather than personal details about other users "…I am more interested in his professional skills, not the colour of his hair. Like, if I want to get something done. Tell me what he is good in" . The decision to approach another user also appears to depend on someone's general personality and openness towards other people. In contrast to I14 and I22's example above, I15 and I7, both regular coworkers at The Edge, were a bit more hesitant when we asked them to 'think aloud' while they reviewed where particular users were located. "…I don't know if I'd feel comfortable just going to a booth and… well, maybe… to be honest, I didn't feel very hard about it and probably I was just on my way to grab the coffee. I don't know, I think it's nice. It looks friendly." . This example illustrates the significance of the situated context and a user's urge to gather a particular skill for them to initiate a face-to-face conversation. Another strong motivation for use emerged by regular users who were keen to identify like-minded others in order to grow their community of interest. I3, even though he has been a daily visitor at The Edge for almost one year, works on music projects but finds it hard to identify like-minded musicians and music producers at The Edge. He does not feel well connected; "…like there is not anything much that makes musicians here interact, to be honest. I heard of so many that use this space, but, I have only met two. But I am pretty sure that there is so many more than them. Because some people actually go in the recording studio and leave, they don't stick around in the public areas..." He perceives Gelatine as a welcomed tool that will help him find other musicians. "It will be so much easier you come and see 'oh that person is actually somewhere in here.' I will definitely be a user of this, I am signing up. To meet other musicians and music producers […] Honestly, I was trying to find a way how to put it, but I mean this is what's missing. It's very vital for people to interact..." . --- Limitations, Challenges and Future Work Through the evaluation, a few challenges emerged to be considered by future versions of Gelatine or similar systems. Non tech-savvy interviewees had problems understanding the mental model behind the RFID checkin concept and the tag-cloud reflecting the profiles of all currently present people in real-time. McCarthy et al. report similar challenges in their study with users at third places, and suggest 'information flyers' as a fruitful solution. One user that interacted with the screen found a skill-tag that he was interested in, and which pointed to a person at workspace 8. Wanting to initiate a conversation, he walked over to workspace 8, but found the person occupying the space was a homeless person . He assumed that that person could not possibly be the person on the screen with the C++ skills. A profile image behind each tag might help match people in the space with their digital representations on the screen. Another participant raised the issue that the turnaround time to find a facilitator for the types of problems he usually faces might be too slow. "…if you've a problem right now, I don't think that system will be fast enough to help you" . On the other hand, users reported that they are usually too busy working on their own things to actively go and browse through the tag cloud in order to provide help for someone else. "…it's not a problem of me willing to help, it's kinda willing to help without spending a lot of time on just kinda reading or searching for whom to help […] that'll be a waste of my time" . Future versions will integrate web and mobile application modules that 'helpers' can configure to be notified if someone in the space requests help with an issue in their particular areas of expertise. A longitudinal study would need to shed light on how Gelatine would actually transform a space like The Edge in long-term. Placemaking requires providing users with a sense of ownership, and encouraging them to continuously shape and re-appropriate spatial infrastructure according to their needs and comfort. In contrast to the physical space in most libraries, Gelatine's infrastructure has been designed to embrace the user community to become co-designers and co-developers of the system. It is based on open-source software and hardware, the development code is made openly accessible on Github, and we provide an API to the Gelatine database for users to create their own visualisations of checkin-data and user profile information . Whilst the tag cloud visualisation showed its potential for serendipitous skill discoveries among users, it also triggered some alternative visualisation ideas among the interviewees, such categorisation of checked-in users according to their background or check-in timestamp . Further research needs to investigate the alternative visualisations through dedicated participatory design sessions and their respective impact on users' sense of place and ownership. Further research also needs to investigate opportunities of other smart space applications and concepts that are enabled through the Gelatine infrastructure. Gamification elements, as we discuss elsewhere , could enable users to collect reward points or unlock "badges" through providing assistance to other users, which then can be exchanged for premium library services . The evaluation also highlighted the contradictory perceptions of "non-users" and "users with coworking attitudes" which raise cultural and political questions for the space managers, such as the installation of "quiet" and "social" spaces. --- CONCLUSION Designers and placemakers cannot create a 'sense of place' on their own. Place is created by people, their intrinsic motivations for particular activities, and patterns of use. This study shows that public screens such as Gelatine, by making invisible socio-cultural aspects of a place visible, have the potential to 'amplify' users' perceived sense of place as well as sense of situated people at a place. In contrast to personal displays , Gelatine embodies digital footprints of situated users as part of the public space; hence serendipitously perceivable to all passers-by, most of whom would not actively retrieve or stumble upon such information otherwise. Further, the study shows that whether a user takes notice, ignores or interacts with the screen highly depends on that user's individual pre-entry motivations and attitudes towards the surrounding place and people at the place. In summary, Gelatine was used to fulfil needs that users have had before Gelatine was introduced , e.g., for divergent and convergent information behaviour . First-time users' browsed the screens to get a better sense of purpose and function of the space, and coworkers identified likeminded or more knowledgeable users. On the other hand, users who were not interested in engaging with their fellow coworkers prior to the installation of Gelatine, showed no interest in using Gelatine either. This underlines the design of public screens not only being a matter of interaction design, but even more so a matter of a broader placemaking strategy at the place of installation. The purpose and function of a public screen need to match with the prospective user's role and motivation of being at that place. Given such a match, public screens such as Gelatine can serve as a powerful hybrid tool for placemaking. Follow-up thoughts on hybrid placemaking are elaborated in .

Public libraries and coworking spaces seek for means to facilitate peer collaboration, peer inspiration and crosspollination of skills and creativity. However, social learning, inspiration and collaboration between coworkers do not come naturally. In particular in (semi-) public spaces, the behavioural norm among unacquainted coworkers is to work in individual silos without taking advantage of social learning or collaboration opportunities. This paper presents results from a pilot study of 'Gelatine' -a system that facilitates shared encounters between coworkers by allowing them to digitally 'check in' at a work space. Gelatine displays skills, areas of interest, and needs of currently present coworkers on a public screen. The results indicate that the system amplifies users' sense of place and awareness of other coworkers, and serves as an interface for social learning through exploratory, opportunistic and serendipitous inspirations, as well as through helping users identify like-minded peers for follow-up face-toface encounters. We discuss how Gelatine is perceived by users with different pre-entry motivations, and discuss users' challenges as well as non-use of the system.

Background The past century has witnessed a demographic transition characterized by a rapid ageing process. Population projections for the coming years show a considerable increase in the proportion of older people worldwide [1][2][3]. To effectively and efficiently respond to the growing health needs of elderly populations, it is critical to have in-depth understandings about their health conditions and related socioeconomic factors. Literature from developed countries has consistently shown that socioeconomic disparities in health existed in general populations as well as among the elderly. Individuals with lower socioeconomic positions are more likely to suffer from both morbidity and mortality [4,5]. Similarly, elders who had higher education and socioeconomic status experienced less depression [6]. However, studies from developing countries have shown inconsistent findings on the effects of socioeconomic status on elderly health. Some studies show that older adults in lower socio-economic positions generally experience worse health than those in better-off groups [7][8][9][10][11]. Some other studies showed that the strength of associations between socioeconomic status and elderly health were less clear or non-existent [12,13]. Although many studies have been conducted to examine inequalities in health status across socioeconomic groups, almost all of them have investigated different aspects of health status separately [14]. There remains lack of research on co-variation of multiple health outcomes and their socio-economic co-patterning, especially among the elderly. As "health" is defined by WHO as "a state of complete physical, mental and social wellbeing and not merely the absence of disease or infirmity" [15], it is important to explore co-morbidity of physical functioning and psychological well-being among the elderly. Vietnam, a developing country in South-East Asia, has experienced the population aging phenomenon. The proportion of people aged 50 years and over rose from 12.6% in 2000 to 14.1% in 2005 and will account for 18.9% of total population in 2015 [3]. Similar to other low-and middle-income countries, little research has been conducted in Vietnam on the issues of elderly health, particularly in regard to the co-variation of different health outcomes. In response to the necessity and urgency of having scientific evidence on the issues of socioeconomic determinants of elderly health, this paper, using a multivariate multilevel modelling approach, aims to 1) examine the effects of different socio-economic factors on physical functioning and psychological well-being among older adults in a rural community in northern Vietnam; and 2) investigate the extent to which the two outcome variables co-vary within individuals. --- Methods --- Data source We used data from the WHO/INDEPTH study on global ageing and adult health. The overall study, conducted during 2006-2007, gathered information on health state and quality of life among people aged 50 years old and over in nine demographic surveillance sites from the International Network for the continuous Demographic Evaluation of Populations and Their Health in developing countries http://www. indepth-network.net: South Africa , Viet Nam , Tanzania , Bangladesh , Kenya , Ghana , Indonesia , and India . The total sample size was over 46,000 respondents [16]. In Vietnam, this study was conducted in 2006 in a rural district in Northern Vietnam, Bavi district, within the framework of a Demographic Surveillance System called FilaBavi . Face-to-face household interviews were conducted with all the people aged 50 years old and over who lived in FilaBavi areas. The interviews were done by trained surveyors of the FilaBavi using standard WHO/ INDEPTH questionnaire-summary version. More detailed descriptions of the Bavi district and FilaBavi can be found elsewhere [17]. --- Measurements --- Outcome variables In this study, both physical functioning and psychological well-being were analysed as outcome variables. Physical functioning was measured by asking respondents about their functional difficulties in the last 30 days, including the level of difficulty 1) in standing for long periods; 2) in taking care of your household responsibilities; 3) in joining in community activities [for example, festivities, religious or other activities] in the same way as anyone else can; 4) in concentrating on doing something for 10 minutes; 5) in walking a long distance such as a kilometre; 6) in washing the whole body; 7) in getting dressed; and 8) in day to day work. For psychological well-being, the study subjects were asked: 1) "Overall in the last 30 days, how much of a problem did you have with feeling sad, low or depressed?" and 2) "Overall in the last 30 days, how much of a problem did you have with worry or anxiety?" The response set for each question was a five-point scale where 1 = None, 2 = Mild, 3 = Moderate, 4 = Severe, 5 = Extreme/cannot do. The scale reliability coefficient for physical functioning and psychological well-being questions was 0.89 and 0.86, respectively. Total score for each dimension was the sum of all the relevant question scores. Higher scores indicated a person with poorer status of physical functioning or psychological well-being. --- Independent variables We included a wide range of socioeconomic information as independent variables such as sex, age, educational level, marital status of the individual, household size, place of residence and economic status of the household. Educational level was categorized into three groups: I: No schooling; II: Less than six year of education; III: Graduated from primary school and higher. Marital status was categorized as: I: Currently in marital partnership ; II: Currently single . Place of residence was defined I: Riverside/island; II: Highland; and III: Mountainous area. Economic status of the respondent's household was measured by asset-based wealth quintiles. The wealth quintiles were constructed using a principal component analysis technique [18]. --- Statistical analysis A multivariate response model was constructed to answer our research questions. The model treats the individual as a level two unit and the multiple measurements observed within an individual as a level one unit. We developed a 2-level model of 17070 at level 1 nested within 8535 individuals at level 2 . By treating multiple outcomes within the multivariate response model, we were able to estimate the covariance between two outcomes nested within individuals, as well as the variance for each outcome in a simultaneous manner. The two-level model can be written as follows: Y B Z B Z B Z X B Z X U U ij ij ij ij j ij j j j       01 1 02 2 11 1 12 2 1 2 Where: Z if physical functioning if psychological well-bein ij 1 1 0  g g           Z 2ij = 1 -Z lij , X j = independent variables , var = σ 2 u1 = variance in physical functioning, var = σ 2 u2 = variance in psychological well-being, cov = σ u12 = covariance of physical functioning and psychological well-being The modeling was done in two steps. In Model 1 , no explanatory variable was included. Model 2 had all independent variables. The results were presented as coefficient and standard errors . A significance level of p < 0.05 was used. MLwiN software, version 2.02 http://www.cmm.bristol.ac.uk/ MLwiN, was used for the analyses. --- Ethical considerations The protocol of this study was approved by the Scientific Board of the FilaBavi. All human subjects in the study were asked for their written informed consent before collecting data, and all had complete right to withdraw from the study at any time without having any threat or disadvantage. --- Results --- Characteristics of the study populations Of the total of 8,874 people aged 50 years and over who lived in the study setting at the time of the survey, there were 8,535 who participated in the study . 4% did not respond to the survey because they were away or were not healthy enough to take part in the survey . There were no significant differences in socioeconomic characteristics between the respondents and the non-respondents. Of the final sample, 37.7% were aged 50-59 years, 36.5% 60-69 years, 24.4% 70-79 years and 11.4% 80 years and over. The characteristics of the final sample are described in Table 1. --- Levels of physical functioning and psychological wellbeing The scores of physical functioning ranged from 5 to 40. The psychological well-being levels were between 2 to 10. The means of physical functioning and psychological wellbeing levels reported by the study respondents are presented in Table 2. The variations in physical functioning and psychological well-being levels by socio-demographic variables were similar. Lower physical functioning and psychological well-being were found in 1) women; 2) older people; 3) people with lower education level; 4) people who were currently single; 5) respondents from poorer household; and 6) mountainous dwellers compared to those in other category of the same variable. --- Multilevel modelling results Table 3 shows the random part of the multivariate response model. The empty model indicates that there was significant variation in each outcome of interest. The variance in both physical functioning and psychological well-being became smaller after the socio-economic variables were included . The socioeconomic factors accounted for about 24% and 7% of variation in physical functioning and psychological well-being scores, respectively. There was a significant correlation between these two independent variables. About 14% of the covariance between these two outcomes of interest was attributable to the socioeconomic factors. However, the adjusted correlation coefficient of 0.35 in the full model indicates that physical functioning and psychological well-being did not strongly co-vary . Table 4 presents the fixed part findings of the full model. There were some similarities as well as differences in socioeconomic patterning of physical functioning and psychological well-being. Women were more likely to have poor physical functioning and low psychological well-being levels than men. While age was shown to be a negative predictor of physical functioning, it had no significant effect on psychological well-being. People with higher educational levels reported being better in both physical functioning and psychological well-being. Those who were currently in marital partnerships had better status of physical functioning and psychological well-being. Significant economic differentials were found for psychological wellbeing but not for physical functioning. Mountainous dwellers had significantly lower levels of both physical functioning and psychological well-being. Household size had no important effect on the two outcomes variables. --- Discussion In this study, adopting the WHO's definition of health, we considered both physical functioning and psychological well-being as fundamental end points of elderly health. Taking advantage of the multilevel modelling approach, we were able to investigate the co-variation in socioeconomic correlates of physical functioning and psychological well-being within individuals. We found some common significant socioeconomic predictors of both physical functioning and psychological well-being among the elderly in the study setting such as gender, education and place of residence. Independently of other factors, women were shown to suffer more from both physical and psychological problems. This is consistent with findings from previous studies in Asia which showed that women were more likely to reported poor health than men [19][20][21]. A strong positive effect of education on both physical functioning and psychological well-being was also pronounced. Studies from Japan [7], Taiwan [8] and China [9] also reported that higher educational attainment resulted in a decreased incidence of functional limitations. We found that mountainous dwellers had lower levels of physical functioning and psychological well-being compared to people living in other areas. Negative effects of disadvantage residence on elderly health were also revealed in a Chinese study [22], and a Korean-Japanese study [23]. We also observed that while physical functioning declined with advancing age, psychological well-being did not vary significantly by age. Similar observation was documented in study in the US [14] and in China [24]. In the present study, economic differentials were found for psychological well-being but not for physical functioning. The overwhelming age effect possibly diluted the influence of economic status on physical functioning. Studies from Thailand [12] and Taiwan [25] showed that income had significant independent influences on functional disorders. A study from US found strong income associations for both health and happiness [14]. It is worth noting from the multivariate response model that the socioeconomic variables of interest had stronger associations with physical functioning than psychological well-being . As a result, the two health outcomes did not co-vary to a strong degree . This finding highlights the importance of analyzing multiple dimensions of health status simultaneously in inequality investigations. Examining a single health outcome may misclassify a person as being in "good health" while in fact he/ she has another bad outcome. The single health outcome approach may also lead to underestimation of inequality problems in a population. There are several limitations we need to note from this study. Firstly, accuracy and validity of self-reported information in older people could be questionable. Low educational level and the presence of cognitive retardation in older people might have reduced the accuracy and validity of the findings. Secondly, because of the cross-sectional nature of the data, our study could not provide any interpretation of causal relations between socioeconomic status and physical functioning and psychological well-being among the elderly in the study setting. Thirdly, some possible joint effects of socioeconomic factors on the outcome variables have not been investigated. --- Conclusions The present study provides initial insight into the extent to which socioeconomic indicators are related to multiple health outcomes among older adults in a developing country setting. It shows that there exist problems of inequality in health among older adults in the study setting. The evidence should be useful for health authorities in responding to the growing health-related needs of elderly populations with limited economic resources. Investigation of health issues among older people is not simple task. This multilevel modelling approach, which offers several technical advantages, should be further utilized. More sophisticated research, such as using longitudinal study designs, is needed to examine the causal relationship between multiple health outcomes and socioeconomic conditions. --- Authors' contributions HVM, DLH, PB, SW made substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data. HVM, DLH, SW, NTKC and PB have been involved in drafting the manuscript or revising it critically for important intellectual content. All authors read and approved the final manuscript. --- Competing interests The authors declare that they have no competing interests.

Background: There remains a lack of research on co-variation of multiple health outcomes and their socioeconomic co-patterning, especially among the elderly. This papers aims to 1) examine the effects of different socioeconomic factors on physical functioning and psychological well-being among older adults in a rural community in northern Vietnam; and 2) investigate the extent to which the two outcomes variables co-vary within individuals. Methods: We analyzed the data from the WHO/INDEPTH study on global ageing and adult health conducted on 8535 people aged 50 years old and over in Bavi district of Vietnam in 2006. A multivariate response model was constructed to answer our research questions. The model treats the individual as a level two unit and the multiple measurements observed within an individual as a level one unit. Results: Lower physical functioning and psychological well-being were found in 1) women; 2) older people; 3) people with lower education level; 4) people who were currently single; 5) respondents from poorer household; and 6) mountainous dwellers compared to that in those of other category(ies) of the same variable. Socioeconomic factors accounted for about 24% and 7% of variation in physical functioning and psychological well-being scores, respectively. The adjusted correlation coefficient (0.35) indicates that physical functioning and psychological well-being did not strongly co-vary. Conclusions: The present study shows that there exist problems of inequality in health among older adults in the study setting. This finding highlights the importance of analyzing multiple dimensions of health status simultaneously in inequality investigations.

Introduction The influence of social advances and the ensuing lifestyle changes on individuals is an integral part of subjective well-being research. Since the turn of the century, the digitization of society, fueled by the popularization of the Internet and social media, has increasingly become a prominent characteristic of modern China, as well as of many other nations. Researchers, witnessing the social changes brought about by digitization, are becoming to give more thought to if, and how, digital technology, particularly social media, can have an impact on people's SWB. 1 In China, one particular social media platform is almost unanimously believed to have played the biggest role in the changes toward an increasingly digital-based way of life. This social media platform is WeChat. Launched in 2012 by Chinese tech behemoth Tencent, and gradually gaining great popularity ever since, WeChat crossed the threshold of one billion users in 2018. Now, WeChat is indisputably the biggest social media in China, with 1.24 billion monthly active users. 2 Unlike Weibo, China's equivalent of Twitter, whose major function is information dissemination, 3 WeChat has many other useful functions, in addition to instant messaging, such as video calling, paying, ride hailing, and hotel booking. It is virtually a "one-stop shop for its users", 4 "a digital life platform for the Chinese", 5 or "China's app for everything". 6 WeChat's success has attracted attentions from its international counterparts. According to a report in The Economist, Zuckerberg "seems keen to turn Facebook into a Western version of WeChat". 7 New York Times also pointed out that WeChat's success "has implications for Google, Twitter and many others, as well as Facebook." 8 Due to its powerful functionality and wide use, researchers have noticed WeChat's potential to exert influence on people's interpersonal relations, day-to-day life, work, and, ultimately, well-being. The existing literature on the relationship between WeChat use and users' SWB is primarily focused on younger groups, especially college students. Research on the elderly is scant. This may be attributed to the facts that younger samples, particularly college students, are much easier to recruit than older adults, and older adults are typically late adopters of tech products. In 2016, there were approximately 7,680,000 older WeChat users , accounting for only 1% of the user population. 9 In 2018 , the number of older users jumped to 63,000,000, but still accounting for a small proportion of its user base . 10 On the other hand, China is fast becoming an aging society. According to the newly released national census data by the China National Bureau of Statistics, as of the end of 2020, China had 264 million people aged 60 and above, accounting for 18.7% of the total population, up from 13.26% in 2010. 11 In urban areas, the percentage is even higher. For example, in Beijing, older adults of that age group account for 19.63%; in Shanghai, it is 23.38%. 11 At a time when "digitization meets population aging", it is of great importance to explore the influence of WeChat use on the SWB of the elderly. The current study chose urban older adults living alone as the research subject for three reasons: First, older adults living alone are, in the Chinese context, highly disadvantaged people to whom special attention should be paid; second, with divorce rates rising and more adult children working away from their parents, older adults living alone is very likely to become more prevalent in China; third, WeChat use was the focus of this study, and whether the elderly live with their spouse/child or not is an important factor influencing their SWB. Older adults living alone are more demographically homogeneous, making them better research subjects for the research. --- Literature Review Factors Influencing the SWB of the Elderly SWB is influenced by many factors. For the elderly, and the Chinese elderly in particular, the following factors have been mostly examined by researchers: 1. Health. Much of the fear of aging stems from health concerns. In fact, evidence has shown that healthy older people have significantly higher levels of SWB than those with serious illness. 12 In addition to the direct effect of illness-related factors, such as pain and impaired mobility, poor health may also indirectly undermine the SWB of the elderly by posing an economic burden. 13 2. Urban vs Rural. Urban/rural disparities are a prominent Chinese characteristic. Research has shown a wide gap between China's urban and rural areas in terms of economy, environment, infrastructure, and much more. The SWB of the urban elderly is much higher than that of the rural elderly to the extent that it was labeled the "difference between Heaven and Earth". 14 3. Empty Nest and Living Alone. Research shows that the physiological and emotional wellness of the empty nesters is significantly lower than that of the non-empty nesters. Moreover, the average SWB score among the empty nesters is much lower than that of the non-empty nesters. 15 Li et al compared three types of Chinese older adults: Those living with their spouse and children, those living only with their spouse, and those living alone. A total of 2225 older adults in seven Chinese cities were surveyed in the research. The results showed that the elderly living alone had the most negative emotions, and their SWB was significantly lower than that of the other two types of older adults. 16 Literature also showed that compared with those living family members, older adults living alone usually receive less attention and resources from the public or their relatives, 17 are less likely to adopt health-promoting behaviors, 18 and selfneglect is pervasive among them. 19 4. Social Support. Social support is defined as information leading the subject to believe that they are cared for and loved, esteemed, and a member of a network of mutual obligations. 20 For the Chinese elderly, support from family members, their children in particular, plays the biggest role in determining their SWB, especially when their social connections diminish with retirement. As a result, many studies singled out children's support, also known as intergenerational support, as a prominent factor contributing to Chinese older adults' SWB. Some scholars believe intergenerational support is more important for aged adults' SWB than is social support from outside the family in China, 21 and compared with western populations, family support has a greater influence on the alleviation of depression among older adults in Asia. 22 5. Education and Income. Although findings of previous studies around the world showed mixed results, there seemed to be a pattern that, when an individual is in a financially challenged situation, an increase in income can substantially increase their SWB level. When the individual overcomes such financial constraints, the correlation between income and SWB becomes weak. 13 For example, a survey on 236 older adults in Beijing showed no significant correlation between SWB and income. 23 On the contrary, Ren et al found that income is a strong predictor of SWB of the rural elderly. 24 The inconsistency might be explained by the fact that most of the Chinese elderly in rural areas have a much lower level of income, so the impact of income is significant. In contrast, seniors in urban areas enjoy a much higher income level, resulting in the effect of income being considerably less significant than rural seniors. Similar to income, the influence of education level on the SWB of Chinese seniors in urban areas is not significant. 23,24 To a large degree, this might be because the overall level of education among Chinese older adults is very low. As of 2010, 90.91% of the Chinese elderly only had junior high school education or below. 25 Even those who were lucky enough to have had access to higher education during the 1960s and 1970s often found their classes disrupted by the Cultural Revolution. 6. Self-esteem. Self-esteem, also referred to as selfworth, as its name suggests, is a person's overall subjective sense of personal worth or value, or, simply put, how much they appreciate and like themselves. In Western countries where individuality is more appreciated than in the East, self-esteem has long been an essential determinant of a person's SWB. Murrell, Meeks, and Walker found that a high level of selfesteem could ward off depression over time in older adults. 26 Alaphilippe's study suggested that selfesteem is an important aspect of the adaptive processes at all stages of life, especially in later life, positively associated with the quality of adaptation and SWB. 27 In recent years, Chinese scholars have been paying more attention to self-esteem when studying people's SWB, and findings also suggest that there is a positive connection between the two. 28,29 Impact of Social Media Use on SWB Although the majority of existing studies on SWB focus on young people, many of the findings can still provide valuable insights and perspectives for studies on the elderly. Research results were mixed in terms of the relationship between social media use and their users' SWB. Some studies showed that social media use is positively associated with their users' SWB. [30][31][32] An explanation of this positive relationship is as follows: As "social animals", having connections with other "social animals" is a basic need of human beings; 33 good interpersonal relationships are a key element affecting an individual's SWB. 34 Social media breaks the physical barriers of interpersonal communication and provides an efficient way for interpersonal interaction. It is particularly effective in reinforcing fledgling and distant relationships. As such, social media promotes the establishment and maintenance of interpersonal relationships with unprecedented conveniences, building a sense of belonging for many people. 35 While the use of social media does facilitate information exchange, some studies found that there is a dark side to it. Holton argued that there are great surpluses of irrelevant information on social media platforms, resulting in information overload and increased levels of stress and anxiety among their users. 36 Even worse, social media overuse could cause a feeling of "not having done anything meaningful", negatively affecting people's emotional state or even leading to a deterioration of mood. 37 Most people use social media for a range of activities. These activities can be broadly classified into two categories: Active and passive usage. 38 Active usage usually involves direct interaction with others or searching for specific information. Passive usage refers to the monitoring of other people's lives without engaging in direct exchanges with them. 39 For example, scrolling through news feeds or looking at others' tweets. According to the "active-passive" framework, researchers suggested that, generally speaking, active use of social media enhances SWB while passive use reduces it. 40 --- Research Questions and Hypotheses The two basic research questions of the current study were: 1. Does the use of WeChat have an impact on the SWB of the elderly living alone? 2. What is the mechanism by which this effect occurs? Examining WeChat's features and analyzing the usage pattern of older users, along with existing studies of other social media platforms' impact on their users, we proposed Hypothesis 1 to address research question 1. H1: WeChat use is positively associated with the SWB of older adults living alone. Since our sample consisted of both WeChat users and nonusers, H1 comprises two dimensions: whether there is a statistically significant difference of the SWB between the two groups. within the user group, whether there is a correlation between the older adults' proficiency in WeChat use and their SWB. Accordingly, H1 was further delineated into two sub-hypotheses: H1a: : Older adults living alone who use WeChat have a higher SWB level than those who do not use WeChat. H1b: : Among the WeChat users, the more proficient one is in using WeChat, the higher their SWB level. To address research question 2, we drew on existing literature and a pilot investigation and proposed three possible mediators: Intergenerational support, social activity, and self-esteem. As discussed above, existing studies have demonstrated that the three factors are all positively associated with older adults' SWB. It is important to determine whether WeChat use is positively associated with these factors. The corresponding hypotheses are as follows: H2: WeChat use is positively associated with the intergenerational support that older adults living alone receive. H3: WeChat use is positively associated with the social activity level of older adults living alone. H4: WeChat use is positively associated with the selfesteem of older adults living alone. Similar to H1, each of the three hypotheses was further delineated into between-and within-group sub-hypotheses as follows: H2a: Older adults living alone who use WeChat have a higher level of intergenerational support than those who do not use WeChat. H2b: Among WeChat users, the more proficient one is in using WeChat, the more intergenerational support they receive. H3a: Older adults living alone who use WeChat have a higher social activity level than those who do not use WeChat. H3b: Among WeChat users, the more proficient one is in using WeChat, the higher their social activity level. H4a: Older adults living alone who use WeChat have a higher self-esteem level than those who do not use WeChat. H4b: Among WeChat users, the more proficient one is in using WeChat, the higher their self-esteem level. These hypotheses, if supported, could be expressed as: WeChat use increases the SWB level of older adults living alone by improving their intergenerational support, social activities, and self-esteem . --- Materials and Methods --- Participants and Sampling Broadly speaking, older adults living alone include those who have not had children and those who have adult children but live separately. Since the majority of Chinese elderly, whether living alone or not, are not childless and intergenerational support is potentially a key mediator, the participants of our study were older adults living separately from their child who lives in another city. In addition, we set exclusion criteria as follows: Severe chronic illness or physical impairment; major negative life events, such as loss of a close relative; older than 80, because older adults over 80, also called "the fourth age", are classified by gerontologists as a group of "biocultural incompleteness, vulnerability, and unpredictability that is distinct from other old age group"; 41 having more than one child, because having multiple children may cause confusion when filling out the intergenerational support questionnaire. Besides, excluding those who have multiple children could make the sample more demographically homogeneous, potentially increasing the research validity. Since China implemented its one-child policy between the late 1970s and mid-2010s, most urban older adults who meet the age requirement of our survey have only one child. Excluding those having multiple children was not a difficult task. The survey was conducted between July and October 2020 in Gui Yang, the capital city of Gui Zhou Province in southwest China. A total of 415 participants were recruited by snowball sampling. Eight investigators were employed and trained to conduct the paper-pencil survey. Each participant received a gift with a value of CNY 15 as an incentive to participate. Table 1 shows the demographic data of the sample. --- Measurements Measurement of WeChat Usage The Questionnaire on WeChat Proficiency proposed by the Tencent Research Institute was used to measure how well the participants use the social media platform. 42 The original questionnaire is composed of 19 items , such as "text messaging", "video calling", "posting moments" and "WeChat payment". For each function, there are three evaluation options: "I don't use this function", "I use this function but not proficiently" and "I use this function proficiently" with scores of 0, 1, and 2, respectively. The scores for each item are added to yield an overall score, which indicates the user's proficiency in WeChat. Prior to our survey, we consulted some elderly people about the understandability of the questionnaire, and revised it to ensure that it was without any confusion for the elderly. Some items in the original questionnaire were deleted or merged. The revised questionnaire consists of 16 items. In the current study, Cronbach's α was 0.855. See Appendix A for the complete questionnaire. --- Measurement of SWB The Memorial University of Newfoundland Scale of Happiness was used to measure the SWB. The MUNSH was developed specifically for the elderly by Kozma and Stones 43 and adopted by a great number of researchers in many countries for studies on the mental health of older adults. Consisting of 24 items, scores of the MUNSH range from 0 to 48, with a higher score indicating a higher level of SWB. The MUNSH was introduced into China in the late 1990s, and it is now one of the most commonly used SWB scales in research on Chinese older adults. 44 In the current study, Cronbach's α was 0.874. --- Measurement of Intergenerational Support The Parents-Adult Children Social Support Scale 45 was used to measure the level of intergenerational support that older adults receive from their adult children. The original scale is a four-point Likert scale , consisting of 26 items. Since our research subjects were older adults who live alone, we revised the questionnaire by removing the items that only apply to those living with their adult children, for example, "My child prepares meals for me." The revised scale consists of 18 items; the average score of each item is the intergenerational support score. The higher the score, the higher the intergenerational support level. In the current study, Cronbach's α was 0.858. See Appendix B for the revised scale. --- Measurement of Social Activities Existing questionnaires assessing Chinese older adults' social activities are mostly either neighborhood basedasking respondents how often they participate in activities that are carried out within their neighborhood , or institution based-asking about activities organized by an organization . However, with the rapid social development in recent years, Chinese older adults' social activities have gone far beyond the neighborhood and have begun shifting from heavily relying on institutions for organization to organizing by themselves. In light of this, we designed an eight-item questionnaire to measure the social activity level of older adults. In the current study, Cronbach's α was 0.776. See Appendix C for details of the design of the questionnaire. --- Measurement of Self-Esteem The Rosenberg Self-Esteem Scale 46 was used to measure the level of self-esteem. The 10-item scale is by far the most widely used instrument to measure individuals' self-esteem. 47 Scores of the RSE range from 10 to 40, with a higher score indicating a higher level of self-esteem of the respondent. In the current study, Cronbach's α was 0.820. --- Ethical Considerations This study was conducted in accordance with the Declaration of Helsinki. Participation in this study was voluntary. Confidentiality and anonymity were ensured in this study. Before the research, participants were told about the purpose of the survey and that information collected would be used for research purposes only. Informed consent to participate in the study was obtained from participants. The study was approved by the School of Humanities, Shanghai Institute of Technology. --- Results --- Relationship Between WeChat Use and SWB Between-Group SPSS was used to perform statistical analyses. Among the 415 participants, 221 use WeChat, while 194 do not. The mean score of SWB of the WeChat users was 34.68 , higher than that of the non-users, which was 32.73 . The t-test showed that the difference was statistically significant . Since a higher score means a higher level of SWB, the result suggested the SWB of WeChat users is significantly higher than that of the nonusers. Thus, H1a was supported. --- Within-Group The correlation test was performed within the user group. The results showed that, of the 221 WeChat users, proficiency in WeChat use was positively associated with SWB , suggesting that the more proficiently one uses WeChat, the higher their SWB level. Thus, H1b was supported. Results showed both H1a and H1b were supported. As a result, H1 was supported, eg, WeChat use is positively associated with the SWB of older adults living alone. Relationship Between WeChat Use and Intergenerational Support, Social Activity, and Self-Esteem --- Between-Group As shown in Table 2, the user group scored significantly higher on intergenerational support than did the non-user group , suggesting that WeChat users have a higher level of intergenerational support than the nonusers. H2a was supported. The same was observed with social activity with user group scoring significantly higher than the non-users . That is, WeChat users have a higher level of social activity than the non-users. Although the user group scored slightly higher than did the non-user group on self-esteem, the difference was not statistically significant , indicating there is no statistically significant difference on self-esteem between WeChat users and non-users. Thus, H4a was rejected. --- Within-Group The correlation analysis showed that WeChat proficiency was positively correlated with intergenerational support , indicating that the more proficient in WeChat use one is, the more intergenerational support they receive, hence H2b was supported. Results also showed that WeChat proficiency was positively correlated with social activities , indicating that the more proficient in WeChat use one is, the more active they are in their social life, hence H3b was supported. The Pearson's coefficient between WeChat proficiency and self-esteem was 0.011 , suggesting that there is no correlation between WeChat proficiency and self-esteem and H4b was then rejected. In the analyses above, both H4a and H4b were rejected, hence H4 was rejected. The result suggested that the use of WeChat does not have a statistically significant effect on participants' self-esteem. We proposed self-esteem as a possible mediator because in our interviews, a number of interviewees expressed that using WeChat enables them to "catch up with the world" or "expand their views", reducing the feeling of being useless and boosting their confidence. However, the hypothesis was not supported empirically by our survey results. Even though the positive effect of selfesteem on SWB has been demonstrated by multiple studies, how an individual's self-esteem could be influenced is relatively unclear. Future research could further explore the relationship between WeChat use and older adults' self-esteem. --- Modification of the Conceptual Research Model The above analyses supported H1, H2, and H3 but rejected H4, indicating that the use of WeChat is significantly correlated with intergenerational support and social activity but not with self-esteem. As such, self-esteem should be removed from the mediation model . In the modified model, two parallel mediators, intergeneration support and social activity , serve as the pathway linking WeChat use and the SWB of older adults living alone. The modified model was tested with PROCESS . The results demonstrate that the effects of all indirect paths of influence were significant, but the direct path yielded an effect of marginal significance . While the total effect of WeChat use on SWB was evident, the influence of WeChat use was smaller with marginal significance. The mediators indirectly transmitted and amplified the influence. Of them, intergenerational support increased the effect from 0.0334 to 9.1123 . This finding shows that intergenerational support is the strongest mediator for explaining how SWB is perceived among elderly people. --- Discussion This study examined the relationship between WeChat use and the SWB of Chinese urban older adults living alone. The results showed that WeChat use is positively associated with the SWB level of the participants. The finding is consistent with He and Huang's study, 48 which is, to our knowledge, the only research exploring the relationship between WeChat use and Chinese older adults' SWB. It' worth noting that He and Huang investigated Chinese older adults in general, not specifically those who live alone, suggesting that our results may extend to seniors of other types of living status. In order to find out how WeChat use influences the SWB of the elderly, we examined the relationship between WeChat use and intergenerational support. The results showed there is a strong association between the two, providing new evidence to Huang 49 which suggested that mobile communication tools of any kind can contribute substantially to the maintenance of the relationship between parents and their distant children. In our opinion, 50 With WeChat, intergenerational support would be a lot easier to fulfill. This study also examined the relationship between WeChat use and social activities of the elderly as another mediator linking WeChat use and SWB. Evidence abounds that social activities have multiple benefits for older adults. This might be particularly true for those who live alone given that they may have less interaction with others within their homes. Chan et al found that older adults living alone are more likely to be socially isolated and may benefit more from social activities. 51 WeChat is a tool well suited for the elderly to organize social activities. Before WeChat was introduced, people relied heavily on cell phone calls and text messages to organize such activities. However, one-to-one phone calls are inefficient if the activity involves many people with different preferences. With WeChat's "group chat" function, which allows as many as 500 people to group chat in one dialogue box, it is a lot easier to organize activities, allowing older adults to be much more active than before. Besides, other functions, for example "location sharing", are of great use in facilitating social activities, potentially increasing their willingness to participate. --- Practical Implications First, for policy makers, this study provides useful information regarding the effect of WeChat use on older adults. Chinese government has realized that the lack of digital literacy not only poses challenges to the older adults, but also hinders the overall development of the society. On November 24, 2020, the General Office of the State Council issued a guideline to address the problem which stipulates that by the end of 2022, long-term mechanisms bridging the digital divide will be established. 52 It is expected that more plans to help the elderly improve digital literacy will be proposed afterwards. Our study could serve as a part of scientific basis for policy-making. Second, for the older adults, our study could be deemed as a statement: Unlike the dishwasher or digital camera which could save you some time or bring you some joy if you use them but also totally OK if you do not, WeChat, as well as some other internet-based applications, has been increasingly integrated into all aspects of the society to the extent that some scholars view WeChat as "an infrastructure for China". 53 If you do not use WeChat, it is unlikely that you can go about your daily life without encountering any trouble. Older adults must realize the necessity of including themselves into the digital world. Furthermore, they should make great effort to explore the multiple benefits of digitization to promote successful aging. Do not use age as an excuse to distance yourself from modern technologies. As Knowles and Hanson pointed out, the notion that aging per se leads to technology abandonment does not withstand scrutiny. 54 Third, the children of older adults, whether living with or away from their parents, should provide more technical support to your parents. Chinese tradition places a lot of emphasis on caring for the elderly or what we call "filial piety". However, filial piety is largely expressed by caring about parents' health conditions. In modern society, it very necessary for younger adults to expand their understanding of filial piety to include many other areas that may influence their parents' well-being. Our study showed that WeChat use is significantly associated with SWB of older adults living alone . Adult children should take teaching their parents how to use WeChat and other internet skills as their responsibility. It is of special importance for younger adults who do not live with their parents since they are not able to provide immediate support. As such, they should make good use of the time when they are together to teach their parents these important digital skills. Additionally, according to Zhou's study, the more time children spend teaching their parents digital knowledge, the more harmonious the family will be. 55 --- Theoretical Contributions As noted in the introduction, studies on the relationship between WeChat use and SWB of older adults are very scant. Against the backdrop of the co-occurrence of population aging and society digitization, our study contributed to the bridging of the literature gap. Secondly, our study focused on older adults living alone, a group which were previously studied from viewpoints that usually reveal the vulnerability of them. Our study demonstrated a possibility that living conditions of this disadvantaged group could be considerably improved through the use of WeChat, thus adding new perspective to the research of the older adults living alone. --- Limitations and Future Work As with most studies, the present study has some limitations that need to be discussed. First, survey participants were mainly selected from the investigators' own networks, resulting in a certain degree of sampling bias. Second, SWB is a highly sophisticated concept that could be influenced by many factors, even those that have yet to be identified or conceptualized by scholars. Our study proposed two mediators, but it does not mean that they are the only variables that mediate WeChat use and SWB. Different studies could propose different mediators to explain how SWB could be influenced by the use of certain social media, thus providing more perspectives on research in this field. --- Conclusion This study aimed to examine the effect of WeChat use on the SWB of Chinese urban older adults living alone and how the effect could occur. Results suggested that WeChat use is positively associated with SWB of the elderly. Findings also showed that the relationship between WeChat use and SWB is mediated by intergenerational support and social activities. That is to say, WeChat use by Chinese older adults living alone increases their intergenerational support and social activities; the increased intergenerational support and social activities then enhances their SWB. --- Disclosure The authors report no conflicts of interest in this work. --- Psychology Research and Behavior Management --- Dovepress --- Publish your work in this journal Psychology Research and Behavior Management is an international, peer-reviewed, open access journal focusing on the science of psychology and its application in behavior management to develop improved outcomes in the clinical, educational, sports and business arenas. Specific topics covered in the journal include: Neuroscience, memory and decision making; Behavior modification and management; Clinical applications; Business and sports performance management; Social and developmental studies; Animal studies. The manuscript management system is completely online and includes a very quick and fair peer-review system, which is all easy to use. Visit http://www. dovepress.com/testimonials.php to read real quotes from published authors.

Purpose: WeChat is China' biggest social media platform. In addition to instant messaging, it has a great many useful functions that can help Chinese citizens adapt to an increasingly digitized society. This study aimed to examine the effect of WeChat use on the subjective well-being (SWB) of Chinese older adults living alone, and how the effect could occur. Participants and Methods: A total of 415 Chinese older adults living alone participated in the survey. Participants filled out questionnaires (scales) measuring their SWB, proficiency in WeChat use, social activity levels and self-esteem respectively. Mean scores of SWB were used to determine the difference in SWB between the users and non-users, and correlation test was performed to find out the association between proficiency in WeChat use and SWB among the users.The mean score of WeChat users was 34.68 (SD = 7.301), significantly higher than that of the non-users, which was 32.73 (SD = 7.168) (t = 2.741, p = 0.006). Among the users, the more proficient in WeChat use one was, the higher their SWB score (r = 0.458, p < 0.01). The results combined to demonstrate that WeChat use has a positive effect on the SWB of Chinese older adults living alone. The findings also showed that the increase in SWB was obtained through the mediation of intergenerational support and social activities.This study suggests that WeChat use by Chinese older adults living alone could increase their intergenerational support and social activities; the increased intergenerational support and social activities then enhance the SWB of Chinese older adults living alone.

Introduction he main objective of this research is to understand the degree of relationship that the dimensions inclusive, equitable and quality education have. Arohuanca and Choque indicate that on January 24, the world celebrates the International Day of Inclusive Education. Without quality, inclusive, equitable education and opportunities for all throughout life, countries will not achieve gender equality or break the cycle of poverty that is leaving behind millions of children, youth and adults. The above refers to Sustainable Development Goal number four of the UN. .: "Ensure inclusive and equitable quality education and promote lifelong learning opportunities for all". Achieving inclusive and equitable quality education for all will require greater efforts, especially in developing countries and for vulnerable populations, including persons with disabilities, indigenous peoples, refugee children and poor children in rural areas. Despite considerable gains in education enrollment over the past 15 years, many children do not acquire basic reading and math skills by the end of their primary education. Equity issues within countries are also a major challenge. This goal ensures that all girls and boys complete free primary and secondary education by 2030. It also aims to provide equitable access to affordable vocational training, eliminate gender and wealth disparities, and achieve universal access to quality higher education . Efforts should be focused on re-enchanting teachers with the work of teaching and fostering an approach centered on the student as a whole person, not only in terms of academic performance, as it relates to the quality of teacher education. It is important to listen to the voices of students, both in the classroom and in their local, national and global community. This would allow for a more concrete and experience-based resolution of education-related problems. There is an urgent need for multidimensional and interconnected measures to link school completion and job entry. The Universal Declaration of Human Rights makes it clear that every child has the right to a free basic education . For many of the poorest families, school is still too expensive and children are forced to stay at home doing household chores or working themselves. To provide quality education to the children of impoverished families, investment is needed in educational scholarships, teacher training workshops, school construction, and improved access to water and electricity in schools. Education is one of the most effective ways to break the cycle of discrimination and poverty that people with disabilities often face. A combination of discrimination, lack of training in inclusive teaching methods among teachers and lack of accessible schools leaves this group particularly vulnerable to being denied their right to education. To ensure that all children enjoy their basic human rights without discrimination, disability inclusion must be incorporated into all policies and plans. Education liberates the intellect, frees the imagination and is integral to self-respect. Funding more adult education programs gives those who lack basic reading and numeracy skills an opportunity to find a job, start a career, educate their own children and live a healthier life. The present research aims to highlight the attention to diversity with love and respect; inclusive, equitable and quality education and for this we must pay attention to the following objectives: identify the factors of inclusive and equitable quality education, described in the current publication; determine the levels of inclusive and equitable education developed among rural school teachers in the province of Guayas; discuss the relationship between the dimensions of inclusive, equitable and quality education; identify the factors of inclusive and equitable quality education, described in the current publication; determine the levels of inclusive and equitable education developed among rural school teachers in the province of Guayas; and discuss the relationship between the dimensions of inclusive, equitable and quality education. --- Development The right to education for all has been enshrined in international law since the adoption of the Universal Declaration of Human Rights in 1948, and the total number of out-of-school children has gradually declined . However, children and youth with disabilities continue to be T Attention to Diversity with Love and Respect. Inclusive, Equitable and Quality Education disproportionately excluded from any form of education and those who do attend school generally receive a poorer quality education and fewer years. Accurate data on the number of children with disabilities who are out of school and the quality of their education are not available, but United Nations agencies agree that at least one-third of out-ofschool children have a disability. Organizations of persons with disabilities came together to bring their concerns to the attention of the world's governments in the 1990s . The outcome sought by governments is a commitment to an inclusive education system at all levels, which means transforming education systems to be inclusive while providing the individual support services necessary for students to succeed. Social agencies hope to eventually achieve true inclusive education for all students; most of the member organizations were formed by parents whose children and youth were denied admission to regular schools, and many started the first special schools for children and youth with intellectual disabilities . The experience with special schools confirmed the learning potential of students with intellectual disabilities, but also led them to live apart from their communities. The inclusion of students with intellectual disabilities in regular classes, with support, led to better academic outcomes for students with and without disabilities and prepared them all to be members of inclusive communities. In addition, members recognized that there would never be enough resources to operate two separate systems at the same time, a special education system and a regular education system . The study by Klimentyeva et al., indicates that despite an overall increase in preschool enrollment, large equity gaps persist. Casanova and Rodriguez had already highlighted in their study that the Latin American community suffers from a legacy of discrimination that has contributed to widespread poverty, unemployment, homelessness and lack of access to education. School attendance among Latino children is sporadic at best. 10 years later, Cuevas et al., report that only 15 percent of children and 10 percent of the poorest children are enrolled in preschool. In primary and secondary education, while overall attendance is high at 98 percent in primary and 92 percent in secondary, for Roma children these figures drop to 69 percent and 23 percent, respectively. The rate is 27% for Latino boys and 18% for Roma girls for secondary education. No attempt has been made to enable the teaching of the Roma language or to ensure that the content of the curricula addresses the needs of children belonging to national minorities. The literature review of Castro indicates that access to education also remains limited for children with disabilities. Although students in some parts benefit from expert mobile teams, there are no programs designed to assist parents and other learners, as well as professional development programs for teachers. Rather, the education of children with special needs still takes place primarily in specialized institutions. This haphazard approach means that programs vary widely in the learning outcomes they allow children with special needs to achieve. In addition, many children with physical disabilities are unable to attend school or are restricted due to physical barriers that impede their access to school buildings. Castro adds that to ensure the progressive realization of the right of all children to inclusive learning, UNICEF will focus on advocacy, capacity building and technical assistance in the following areas:  Promote inclusive early learning through the continued expansion of preschool education programs.  Improving inclusive education both at the policy level and further institutionalizing good practices through support for model schools that include children with disabilities in regular schools and in the mainstream curriculum. In addition, UNICEF has supported the development of training modules for primary school teachers to support the inclusion of children with disabilities in the mainstream education process.  Promote peace building and social cohesion, and the strengthening of reconciliation at different levels, with a focus on adolescents and young people. The study entitled building management of inclusive education in primary education by Sulasmi Akrim had as the main objective of this research is to contribute to the implementation of the objectives of the fourth phase of the Inclusion and Social Protection Initiative and to support the implementation of the United Nations Convention on the Rights of Persons with Developmental Disabilities by raising awareness and developing a positive attitude. towards the inclusion of children TECHNO Review, 15, 2023, pp. 13-20 with developmental disabilities. The research findings may be useful to various governmental and public institutions, as well as to different civic and private organizations whose mission includes improving the situation of children and adults with developmental disabilities. The findings of this research lead to the conclusion that citizens are generally aware of the neglect, marginalization and lack of respect for the fundamental rights of children with developmental disabilities and the lack of help and social support for this group. However, many citizens have demonstrated certain stereotypical preconceptions and prejudices towards children with developmental disabilities, highlighting the need for activities to increase public awareness of the issue. Traditionally, the institutionalization of an ailing or sick person for others to care for is generally considered unacceptable and shameful. This attitude is rooted in the dominant concept of family, where parents are obliged to care for their children, who will then care for their parents in the future. Therefore, suggesting institutionalization of children with special needs would be considered a socially unacceptable response. Consequently, answers to questions on this topic may not accurately reflect the actual situation in society . The research findings of Molina indicates the need for additional efforts to inform citizens about children with developmental disabilities: who they are, their characteristics and abilities, emphasizing their similarities with other children. The social distance of these children should be reduced, increasing their inclusion both in education and in society in general. Latino society should do more to provide equal opportunities and rights to these children, no different from others, including the right to a dignified life and the right to an adequate education that allows them to develop to the maximum of their abilities. The hypotheses that were posed were: The teacher attends to children with situations of diversity; The teacher considers that his/her teaching is currently done with love; The teacher's behavior is respectful to the children; The teacher believes that he/she practices inclusive activities with his/her children; The teacher is equitable with his students; The quality he offers in his classes is the most optimal; therefore, a correlational study was carried out, through the postulates of the publications that have already been done in other realities in the world. This indicates that the results are reasoned in theory and justified by their methodological application. Children with disabilities are at much greater risk of being abused, bullied, coerced and patronized and it may be difficult for them to report this to anyone. There may be collusion by families and teachers in this mistreatment. It is important to be vigilant in research practice to keep children safe and, at the same time, provide opportunities for them to participate. Since protection should always be prioritized, it is the responsibility of adults to adapt research methods and processes to ensure safe and creative ways to support their participation. The tendency to overprotect children with disabilities results in exclusion from activities in which other children are invited to participate and contribute. Barriers to inclusion exist in society, not in the child. We need to be very careful not to impose roles and identities on children with disabilities based on assumptions we hold. Teachers researching children with disabilities should be encouraged to express their views freely and, as researchers, this means asking them what support they need, rather than assuming that it is already known. This helps to verify the participants' preferred level of understanding/communication format and adapt research approaches accordingly. Based on the evidence review in this paper, more research involving children with disabilities is urgently needed. Including them ensures that policy and practice are informed by the perspectives and concerns of children with disabilities themselves. When we have this, we can support a broader agenda for greater inclusion in society of children with disabilities. However, it is important to think more critically about the ethical dimensions of research, including methodological implications, to promote such inclusion. --- Methodology The research was developed under a non-experimental design, since the current positions of the studied dimensions were taken, within the correlational level, to know how they impact inclusive, equitable and quality education among themselves. The correlational design, of the quantitative approach, had to resort to the survey, which was made with nine questions, to professors who are currently studying education in a university in the city of Guayaquil. The sample that was developed was 384, thanks to the Attention to Diversity with Love and Respect. Inclusive, Equitable and Quality Education fact that the population that is working on inclusion is effectively unknown, and the sample calculation formula gave that number. The initial education teachers were randomly selected and contributed with their answers to a questionnaire with closed-ended questions on a Likert scale. For the reliability of the results, Cronbach's alpha was calculated and the result was 0.925, which is considered very high. In order to recognize the correct correlation method, a Kolmogorov-Smirnov Test was performed for one sample, in which it was found that the asymptotic significance was less than alpha in all cases . Therefore, it is logical to consider that the null hypotheses be rejected, but it is also understood that the correlation, should be done by the Rho Spearman test. --- Results After tabulating the results, the information was exported to the SPSS statistic, and after performing the reliability analysis, the cross table was used. For the quantitative phase, in this table the values strongly disagree and disagree were added to recognize the disapproval of the respondents; also the values Agree and Strongly agree were added to recognize the approval of the teachers investigated, in the descriptive analysis . The questions that were asked were validated with experts and are part of a broader investigation in the same sense of inclusion. What was found was that attention to diversity is very important for teachers, in 60.2%, but 34.9% disapprove of it, which although it is less than the previous one, it is considered that it should be reviewed because that percentage is the one that has been found among the theoretical reviews and is the one that is affecting the children, it is something similar to what happens with the teacher's love for the children, since their favorable response was 54.9%. On the other hand, the acceptance and denial of respect is almost similar, another variable that should be attended to by society, because children should have more respect. Fifty-seven percent of the teachers accept to practice inclusion with their students, 60.4% consider being equitable, which although they are high values, they are not consistent with a fully inclusive or equitable culture, and from the other data found it can be argued that this is happening similarly in all variables. The complete version of the descriptive study is in Annex 1. The correlations found in the study are generally very high, the analysis begins with the least intense and that exceed the value of 0.70, considered as highly correlated. Equity has an impact of 72.9% with the love that teachers give to their students, which is entirely understandable because an inclusive teacher is one who gives love in education. Inclusive education, has a 79% impact on teaching love, which is commented in the same way as the previous correlation and finally in this range is respect at 79.5% with love. Educational equity has a correlation of 81.6% with teacher respect, which is what every teacher must strive for, and sacrifice, to improve the quality of education for children. Inclusive education has a 91% impact on teachers' respect, which is also observed as a great social usefulness of teachers, and finally in the category of more than 0.80 is equity with an impact of 81% with inclusive education. Figure 1, completely rules out all the null hypotheses of the study, therefore, the minimum conditions in all variables are verified, but as observed in the descriptive study, the percentages that fail to fit this profile are the ones that call the attention of society, so that they can be adjusted, clarified and improved, in terms of inclusion and educational quality. --- Attention to --- Conclusions We can begin by concluding that, in accordance with the first objective of identifying the factors of inclusive and equitable quality education described in the current publication, it was found that the problems faced by families of vulnerable children should be clearly identified and families should receive adequate support. Citizens should be encouraged to get involved in defending the rights of these children. Plan an awareness campaign, local television stations should be considered the best channel of communication. The second conclusion, linked to the second objective of determining the levels of inclusive education and equity developed among rural school teachers in the province of Guayas, it was concluded that 62% of the teachers feel that they attend to children with situations of diversity, 60.4% are equitable with the students and 64.3% consider that their classes are of very high quality. The last conclusion was based on discussing the relationship between the dimensions of inclusive, equitable and quality education and it was found in the conclusion that educational equity has a correlation of 81.6% with teacher respect, this is what every teacher should strive for, and sacrifice, to improve the educational quality of children. Inclusive education has a 91% impact on teachers' respect, which is also observed as a great social usefulness of teachers and finally in the category of more than 0.80 is equity with an impact of 81% with inclusive education.

Educational inclusion, equity and quality of education should be considered from the perspective of diversity, love and respect for students. The teacher is the architect of this variable education, which is why this research was conducted. The main objective is to understand the degree of relationship between the dimensions of inclusive, equitable and quality education. A quantitative correlational study was carried out under a nonexperimental design, in which the perception of 384 teachers in the province of Guayas, all of them in initial and basic education, was collected through a survey. Cronbach's alpha was 92.5%.

Resumo O artigo mostra como a genética do câncer, em Cuba e no Brasil, tornou-se matéria de pesquisa, despertando maior interesse da saúde pública. Foram usadas pesquisas etnográficas realizadas em clínicas de genética comunitária e serviços de genética do câncer para averiguar como o conhecimento e as tecnologias associadas à nova área da saúde são convertidos e empregados por pesquisadores, profissionais da saúde, pacientes e familiares nesses dois contextos. Destaca, comparativamente, as semelhanças e diferenças na maneira pela qual a genética do câncer se posiciona em relação às prioridades em pesquisas transnacionais, na história e na política contemporânea da saúde pública e a vulnerabilidade incorporada ao câncer que reconfigura o escopo e o significado da genômica como a medicina "personalizada". Palavras-chave: genética do câncer; biomedicalização; Cuba; Brasil. care" in many nations in the "global north" is now unevenly filtering into other international medical practices. Emerging social science research is demonstrating how these interventions within Europe and North America, and increasingly beyond, are diversely situated across different landscapes of private or public healthcare, enlisting and producing various patient subjectivities, as well as practitioner communities, that challenge the notion of a uniform reproduction of genomics as personalised medicine . While cancer genetics constitutes a visible "absent presence" in some national contexts , in others an expanding emphasis on "prevention" fuels an effort to address the "unmet needs" of those not currently able to offer or access genetics testing services . Yet, alongside a moral discourse concerning issues of access or even "rights" to new, so-called "preventive" healthcare services in "underserved populations" there are also scientific questions about the meaning of genetic risk beyond high risk groups in Europe and North America. It has been argued that these questions can only be validated through the expansion and provision of genetics research and testing to include diverse populations . 2 In this way the involvement of what Rapp describes as "exquisitely stratified research populations" are essential to a transnational genomic healthcare enterprise, where they serve as both "potential global [scientific and medical] resources and [underserved] market beneficiaries". This is particularly evident in the context of BRCA genetics. This has been a flourishing area of transnational scientific research for some time, focused mainly on an effort to identify common founder mutations and or biomarkers that might significantly contribute to cancer risk in different national and regional contexts . High throughput "next generation" testing technologies that enable thousands of mutations to be identified quickly are central to this task, at least where such technologies are immediately available. Nevertheless, the complex epigenetic context of personal or so-called stratified medicine means the immediate healthcare application of post-genomic fields of inquiry such as pharmacogenomics are somewhat elusive -if still powerful -promissory dimensions of this expanding international area of cancer-related genetics research . The transnational research agendas which cancer genetics, and especially breast cancer genetics, unites and reproduces are therefore multiple. They are tied to practices of standardisation and localisation, while being informed by uncertainties concerning the meaning and significance of genetic risk information as well as the current limits of and on-going hopes for personalised medicine. Taking this transnational area of research and healthcare as a context to be both empirically examined and as a starting point for comparison, this article examines the socio-historical specificity through which cancer genetics is emerging as a focus for research and healthcare in two contrasting regions of Latin America: Cuba and Brazil. The article first examines how cancer genetics has developed in these two regions, tied to specific national histories and the contemporary organisation of public health services before examining how cancer genetics is pursued and practiced in each of these settings. There is a particular focus on how the lived reality of inequitable access to resources and basic healthcare needs are negotiated and ameliorated by those who work in these fields in Cuba and Brazil. The second half of this article reflects more directly on the kind of embodied patient subjectivities co-produced in the context of cancer genetics. It examines how the meaning of genetic risk for cancer patients and their families, in both Cuba and Brazil, is diffracted through particular ideas of embodied vulnerability that displace and configure genes in relation to the social environments and the lived experience of interpersonal relations. This raises further questions about how to understand the dynamic consequences of biomedicalisation and the meaning and significance of cancer genetics as "personalised medicine" in these two regions. The data presented here is based on ethnographic and qualitative research that developed differently in Cuba and Brazil. Research in Cuba was undertaken alongside genetics professionals in three different regions of the country, outside of the capital city Havana, as part of a collaborative project examining perceptions of risk for breast cancer. The fieldwork consisted of visiting families in their homes with Cuban medical geneticists as part of the routine collection of family history information and the use of a semi-structured questionnaire exploring perceptions of risk related to breast cancer. The questionnaire was completed by over two hundred Cuban women, all of whom were over 18. The work undertaken in Brazil included participant observation in cancer genetics clinics over a period of 18 months in three different urban centres in the south of Brazil , interviews with patients and family members visiting cancer genetics clinics , and interviews with practitioners and scientists working alongside cancer genetics specialists . Participating patients were visiting cancer genetics clinics and were either undergoing risk assessment, had received it, or were more frequently awaiting the result of a genetic test. While the majority were women, a number of men were also recruited in the study. All were over 18. --- Cancer genetics in Cuba and Brazil --- Cuba While medical genetics began to be incorporated into the Cuban healthcare programme in the 1980s, a rapid expansion of clinical genetics services has since taken place across the entire country, increasing the number of clinical specialists from 15 to around 184 . The founding of the National Genetics Network in 2003 led to the incorporation of community genetics services in all municipalities, primarily focused on pre-natal and neonatal screening and testing programmes. With the stated aim of developing "a national programme for the diagnosis, management and prevention of genetic diseases" there are now currently over 1,600 persons employed nationally in this field as genetic specialists, technicians, counsellors, and nurses as part of a regionally based national network of genetics centres and clinics. The working structure of Cuban community genetics therefore directly mirrors the system of local clinics established as part of the family medicine initiative in the 1980s, building on a nearly 40 year program of infant and maternal health. This has been central to the goal of the Cuban "socialist" project of realising the principle of "health for all" through a focus on primary healthcare pursued through close integration of hospitals, polyclinics and community-based services . It has also included the training of thousands of health professionals with many living and working in local community-based clinics. While now subject to the increasingly dynamic political goals of the Cuban government , it is nevertheless a framework that has provided the model for the expansion of community genetics clinics in Cuba. The main day-to-day focus of community genetics professionals relates to pre-natal and neo-natal screening and interventions. However, the collection of family history information for cardiovascular disease, breast cancer and diabetes has begun to form part of efforts to examine the role of genetic factors in common, adult-onset conditions, with reports of over 43,000 families on national disease registries . The utility of such data is significantly hampered by the limits of healthcare resources for interventions such as predictive genetics testing for common adult-onset conditions. At the time of my research, in 2007-2008, there was no widespread clinical or research genetics testing programme for commonly identified genes such as those associated with an increased risk of breast cancer, BRCA 1 and 2. Nevertheless, interest in conditions such as cancer within the context of community genetics is increasing. This reflects a changing national disease profile in terms of cancer incidence, with lung and breast cancers constituting a significant and growing public health problem . There is also a notable, long-standing area of scientific research in Cuba focused on cancer immunology and the development of cancer vaccines within the scientific pole in western Havana . While difficult to clearly define, there is also a growing association between community genetics in Cuba and the country's high profile biotechnology industry. This must be seen as part of what Reid-Henry refers to as the "local experimental milieu" of Cuban biotechnology, facilitated by a close relationship between state funded laboratory research and healthcare services. He suggests that an "alternative discourse of ethics and possibility" is pursued at this intersection, aimed at improving collective public health, strengthening revolutionary values and participating in international markets. The collation of family history data as part of the expanded work of community genetic health practitioners is therefore not only connected to a long history of public health tied to community health interventions, but also constitutes an invaluable resource for a growing "socialist" biotechnology sector. As Reid-Henry's work demonstrates, it is a sector uniquely focused on developing products that can not only provide cheaper solutions for the national and international healthcare needs of "underserved" populations, but also does not preclude participation in international research . While not extensive, there do appear to be a number of international collaborations in the context of Cuban Community Genetics, including collaborative research studies to characterise BRCA mutations in Cuba together with researchers in Canada and the United States . --- Brazil The development of specialist cancer genetics clinics and services in Brazil has emerged over the course of the last ten years in the wealthier and relatively more economically developed southern part of the country, where there are extremely high recorded rates of História, Ciências, Saúde -Manguinhos, Rio de Janeiro 100 História, Ciências, Saúde -Manguinhos, Rio de Janeiro breast and prostate cancer equivalent to the prevalence in the US population . During the time of my study, private predictive genetic testing for mutations on the two well-known BRCA genes was available for those able to pay the $2,000-$3,000 cost in the growing number of private clinics. More recently, in 2012, genetic testing was made available via some private health insurance providers, which are generally used by approximately 25% of the Brazilian population. However, the hub of emerging cancer genetic research and healthcare services is located within mostly public or mixed private/public health hospitals associated with universities and research institutes in the urban centres of the southern region of the country, including São Paulo, Rio de Janeiro and Porto Alegre. Nevertheless, because cancer genetic services are not currently offered by the Unified Health System for the majority of those identified as being at genetic risk for cancer, eligibility for any form of genetic testing is directly linked to national and transnational research collaborations between individual scientists and their research teams, most notably in France, the United States and Canada. As a result, cancer genetic services in Brazil operate in an arena where there is a close and fluid relationship between clinical services and research objectives. 3 That is, fledgling clinical services concerned with identifying, addressing, and ameliorating the risks for individual patients and their families are closely tied to research objectives linked to understanding how Brazilian cancer genetics can be mobilised in conjunction with transnational research agendas. At the time of my study, two primary areas of investigation were prominent. This included studies on the two BRCA genes associated with an increased risk of breast cancer, and genetic markers linked to rarer cancer syndromes associated with a range of different cancers, including breast cancer. While there are clearly differences in how cancer genetics is emerging in Cuba and Brazil, the fact that, in both cases, they are influenced by national and transnational research goals and/or resources is significant. Yet within both contexts cancer genetics is not easily or readily subsumed by a promissory vision of personalised medicine. The next section examines how, instead, it is mobilised within and in relation to a public health system that reflects the activism of researchers and practitioners. This configuration of cancer genetics must also be understood within the context of the very different histories of social medicine in Cuba and Brazil. --- Between the promise of personalised medicine, public health and the contemporary legacies of social medicine The fact that genetic testing for commonly identified inherited-susceptibility genes such as the BRCA genes was not routinely undertaken in Cuba reveals how certain key dimensions of transnational cancer genetic research constitute what has been described as an "absent space" . Nevertheless, the collection of family history information within the context of regionally focused community genetics activities also highlights continuities with long standing public health goals that facilitate and inform the localization of cancer genetics within Cuba. The routinisation and normalisation of collecting family history information in the local practice of Cuban healthcare to some extent enables the pursuit of community genetics for common complex adult-onset disorders such as cancer. As I've argued elsewhere, the work of collating family histories and managing the delicate inter-familial relations that are part of genetic interventions are central to tacit community caregiving that serves to reinforce rather than question the role of cancer genetics within the socialist goals of community genetics ). For the Cuban genetics health professionals that I met, attentiveness to the family in the context of community genetics practices re-enforced the socialist ideals of "revolutionary medicine" as collectively orientated, often in explicit contrast to what were seen as capitalist individualism and the prominent discourse of personal medicine associated with genomic healthcare. However, in the vocational narratives of these health professionals it was not only the humanitarian concerns of Cuban healthcare that were focused upon, but also a notion of community genetics as a modern cutting edge domain of medicine. Many of the predominantly female genetics health professionals noted how this field was in fact leading the way in addressing the challenges of informed consent and confidentiality that were particularly stark, they suggested, in Cuba given the on-going paternalism in the organisation and practice of state managed healthcare. As I've suggested elsewhere, this points to the heterogeneous ways in which the pursuit of cancer genetics in Cuba is being incorporated into the values and practices of state socialism . The focus on genetic risk information related to family history within the practice of community genetics could, however, also reveal shortcomings in basic healthcare in Cuba, such as the lack of mammogram machines in some regions of the country. At the same time, for these practitioners, the discomfort generated by this situation did not seem to lead to questioning the broader goals of pursuing cancer genetics as community healthcare. Rather, it sometimes served to reinforce the value and worth of socialist goals and a commitment to ensuring that issues of equity and access were positioned at the forefront of community genetics. In this context a promissory discourse of personal medicine was conspicuously absent, subsumed by a concern and desire to make cancer genetics part of the broader goals of socialist medicine. As one genetic practitioner involved in the collection of family history data related to cancer risk pointedly said, "we are not just trying to help one woman with breast cancer here, because science in Cuba is for everyone." The promise of personal medicine associated with cancer genomics was more easily incorporated in Brazil. Nonetheless, it is important to note how awareness and acknowledgement of inequities in access to healthcare in Brazil also inform how those who work in cancer genetics traversed the gap between the promise of genomics and the daily practice of cancer genetics. The constitutional right to health has been in place in Brazil for over 30 years, yet public healthcare provision is uneven and complex . Approximately 75% of the population only accesses healthcare through the public healthcare system , while the rest of the population makes use of variably affordable health insurance or health plans. However, while there are differences in the quality of provision both within and across public and private healthcare, the public system is often perceived by both middleclass patients with health plans, and those who use it, as inadequate; a situation described in terms of "excluding universalism" . This affects how and why those working at the forefront of Brazilian cancer genetic services seek to constitute cancer genetics as public health and also why patients seek out care in the context of cancer genetic research. For many healthcare practitioners the logic of cancer genetics seemed obvious in terms of a cost-benefit ratio of identifying those at risk compared to the costs of treating those with breast cancer. Moreover, despite the fact that Brazilian cancer genetics is practiced at the junction between the research activities of individual researchers or their teams involved in international collaborations and precarious or limited public health services, it is mobilised, by those who work in this domain, as part of a neglected "preventive" approach to health. A discourse of prevention has gained increasing prominence in recent years in Brazil in relation to breast cancer with new national guidelines for mammography screening implemented among a range of other interventions . In a similar way, while many of those I met who worked in cancer genetics were sceptical about the imminent realisation of "personalised" medicine in Brazil, cancer genetics was very directly related to the goal of preventive public health. As a result, there was widespread acknowledgement and a degree of exasperation among these health professionals that cancer genetics was only currently possible through globalised clinical and transnational research collaborations. This meant having to work in a context where screening interventions for those identified to be at genetic risk had to be negotiated with either health insurers or individual hospitals. Venting her frustrations, one doctor pointed out how "I can't keep doing research to offer testing for another ten years... with care being something peripheral that is just tacked on the side." 4 The commitment to cancer genetics as part of public health in Brazil led some practitioners to pursue what were described as "road trips" to rural parts of the country beyond the urban centres in which they worked . It was a practice that strongly reflected the contradictory contexts into which Brazilian cancer genetics is embedded and is practiced. Here the efforts to fulfil and expand the parameters of transnational research collaborations are conjoined with those to identify individuals and families affected by hereditary cancer syndromes as part of a preventive approach to public health. On one such road trip in which I participated, cancer genetics was explicitly presented as a chance to both "participate in research" and "prevent cancer in the future." The message was conveyed with an awareness that these families had somewhat limited choices -given that none had private health insurance and lived in a rural part of the state of São Paulo -and thus did not have easy access to healthcare services. While no practitioner explicitly stated to the families that being involved in research was a means to secure basic healthcare, it was implicit that this was an indirect benefit of participation. It is tempting to see such activities simply in terms of recruitment for research, but this does not do justice to the wider affective and moral framing of Brazilian cancer genetics by medical professionals as part of a preventive approach to public health which -like their Cuban counterparts -they actively pursue. While cancer genetics constitutes something of an "absent space" in Cuban community genetics as compared with Brazil, we can see in both cases how developments in this field of medicine are linked in direct and indirect ways to national and international research collaborations alongside efforts to intervene in public health. These intersections are very 103 much sustained by the actions and investments of health practitioners. These orientations to genomics as public health must also be understood in relation to the historical legacy and contemporary articulation of "socialist" medicine in both of these settings. These are most clearly evident in Cuba, with its long history of state socialist medicine, as outlined earlier. Yet this dynamic is also relevant in the Brazilian context. The influence of the Brazilian Hygiene Movement in the early twentieth century, which advocated widespread public health interventions as a means of addressing Brazil's social problems and the prominence and role of collective health from the late 1970s to the present , provide an important context for discussing why preventive public health is one of the vectors through which cancer genetics is being pursued in Brazil. The findings presented here suggest that the socio-historic specificity and dynamic contemporary articulation of social medicine in Cuba and Brazil shape the biomedicalised trajectories through which emerging fields of healthcare such as cancer genetics flow and in which they are unevenly constituted. The next section examines the experience of patients participating in fledgling cancer genetics services in Cuba and Brazil in order to consider the contrasting pathways through which they and their families enter this novel domain of healthcare. This provides a further means of reflecting on the complexities and dynamics of biomedicalisation. It examines how -in both contexts -genetic risk is perceived and acted upon as part of a wider understanding of embodied vulnerability to cancer. That is where the actions of genes on the development of disease are only made meaningful in relation to wider environments and lived social relations. This has consequences not only for how we might understand the molecularisation of bodies, but also the promise of personalised medicine and the wider translation of emerging epigenetic understanding of genetic risk. --- Patient-citizens and molecularised bodies The social context and pathways through which patients arrive in the domain of clinical cancer genetics in Brazil and Cuba is somewhat different, with the institutional configuration of these services affecting how patients perceive and engage with cancer genetic interventions. In Cuba, family history investigations are perceived as a routine part of "family medicine," where medical attention to and knowing about the health of family members is normal. In this sense, cancer genetics as part of the practice of community genetics is linked to longstanding efforts by the Cuban state to monitor and intervene in the population's health at a local and national level. As one female participant said, commenting on why she believed hereditary factors were important in relation to breast cancer, "because in consultations they [the doctors] are always asking if you have relatives who are sick." Nevertheless, while this routine attention to family history facilitates cancer genetics in Cuba, it does not necessarily entail a molecularised understanding of heredity and cancer risk, as I explore below. For the cohort of Brazilian patients that I met, cancer genetics was anything but normalised. Rather, it was more often than not seen as something valuable, a field of healthcare that was "leading the way in finding a cure for cancer" as one patient put it and sometimes also a means to access healthcare services inadequately provided through the public health system. Most of the patients I encountered during my fieldwork arrived in cancer genetics clinics following the treatment of a family member for cancer, having been referred by another health professional working in the same hospital. A large number were waiting for genetics test results linked to research protocols and many, as SUS patients, also received a significant amount of other basic screening and health monitoring services. As a result, simply being within the parameters of the hospital's care through participation in a study was often perceived as prevention. As one patient put it "everything is a blessing when you arrive at this hospital" referring to the relief she felt at being part of a cancer genetics research programme. For her this meant that she was able to access basic screening services, such as mammograms, that she would otherwise have to wait months for in a hospital that -to her mind -was substandard compared with the research hospital where she was part of cancer genetics studies. The way that rights, responsibilities and patient-citizens are constituted through being part of research protocols -particularly those linked to clinical trials -has been explored in other contexts within and outside Brazil ). In a context where clinical care and research are so closely entwined and where there are obvious healthcare disparities, exploitation and the reproduction of "bio-available" populations remains an ever-present possibility . Nonetheless given the way that participation in cancer genetics is routinised in public health in Cuba or can be an indirect means of accessing basic healthcare services in Brazil, resisting "uniform and unilateral diagnosis" seems necessary in order to account for the diverse ways that patients enter and participate in emerging fields of medicine such as cancer genetics. Despite the differences outlined above there appeared to be something of a convergence in the way Cuban and Brazilian research participants and patients understood, constituted and engaged with information on genetic risk for cancer. In Cuba, the notion of genes -or more specifically a genetic mutation -was not widely understood, nor considered necessarily relevant or sufficient to be the cause of cancer. By contrast, family history and hereditary factors were identified by a number of Cuban women who responded to open ended questions about risk factors for breast cancer. This must in part be read alongside an emphasis on family health linked to long-standing community management of health through local clinics and family medicine. However, more striking in the Cuban context was the relative dearth of attention to factors such as "stress" that were within the locus of the individual's control. Environmental factors, ozone depletion, and wars were all frequently identified as explanations for what were believed to be the causes of breast cancer. These responses reflected a general sense that health risks associated with cancer were not necessarily linked to individual behaviour or actions, but rather the wider context of daily life in Cuba. 5 I explore this further by reflecting on some frequent responses to questions about risk factors for breast cancer, specifically a physical or emotional blow, and also discussion of diet and nutrition. 6 In Cuba, rather than the idea that an excess of certain foods representing a risk factor for cancer, a deficit in the diet was most often associated with the disease, particularly a lack of vitamins, protein and milk. As one Cuban woman I met said, "It could be some food that you don't eat that causes your cancer, some fruit or vegetable that we don't have the habit of eating, perhaps because there aren't any." Soya beans were, however, identified as a food that might cause cancer if consumed in excess. While this idea is paradoxically in contrast to popular scientific understanding that consumption of soya in certain Asian diets may protect against cancer, in Cuba this must also be understood in relation to a long history of the use of soya beans by the state as a replacement for a shortage of dairy products. As one participant put it: "Since the 'Special Period' they have incorporated soya beans [into the diet] and there is a large number of people with cancer." 7 Another commonly held belief regarding risk for cancer among the Cuban women I met related to the notion of "a blow" or un golpe. More than half of the respondents who participated in the research project thought that a blow was a factor in the development of cancer. This term can refer to a physical or an emotional blow caused by, for example, a bereavement in the family. It was, however, most often the former meaning that seemed evident, particularly when ideas of female gender were incorporated into this understanding. Not only was the breast considered a "delicate area for women" but also there was also a belief that excessive physical force could be implicated in the development of a cancer. The gendered ways in which the notion of "a blow" was presented seemed therefore to reflect in part new anxieties about the changing role of women in Cuban society, including the dramatic shift to their inclusion in the work force during special period shortages . These examples draw attention to how factors beyond the control of the individual were most often articulated as risks for or the causes of cancer in Cuba. They reveal a specific articulation of physical vulnerability that must in part be understood in the context of changing socio-political dynamics related to the governance and provision of food in Cuba and women's changing social and cultural roles. Recent social science studies in Cuba, in urban centres such as Havana, have highlighted the extent to which the governance of health in a context of shortage is increasingly dependent on the entrepreneurial skills of individuals and families in seeking out and exchanging medical resources and products . Equally, changes in the organisation of healthcare provision have led to far fewer family physicians in Cuba, with the gradual appearance of health promotion drives that are beginning to emphasise the role of individual self management. Yet it is not so clear how these dynamics are playing out in rural and urban locales outside of Havana. As Elise Andaya's work demonstrates, any new emphasis on being "self-disciplined" in Cuba must be read in the context of a long and sustained history of state socialism. This includes examining how the historical emphasis on the health of the collective leaves traces in the contemporary lived embodiment of individual risk. This means, in Reid-Henry's terms, being attentive to how, a "socialist epistemology of the body" resonates in novel areas of healthcare such as cancer genetics, where personalised medicine is neither necessarily immediately valued or possible. For many of those individuals that I met while visiting cancer genetics clinics in Brazil, genetic mutations were more widely acknowledged as a risk factor for the disease. At the same time, they were nevertheless rarely believed to be the sole or sufficient cause of cancer. As in Cuba, genes were almost always necessarily seen as interacting with other factors associated with social, environmental and, in particular, interpersonal relations. História, Ciências, Saúde -Manguinhos, Rio de Janeiro The effect of strong emotions on the body was one of the risk factors most frequently discussed by those I met attending cancer genetic clinics in Brazil. For some, emotions were something that could enter and act upon or materially produce the body. As one patient I met succinctly put it; "these emotional questions affect our thoughts, they produce chemical reactions, and molecular and neurological changes inside our bodies. They transform into something concrete. Our thoughts are not just our thoughts, they happen inside our body as well." In my research with Brazilian patients, bodies were in fact seemingly mattered as a consequence of the conjoined effects of emotions and diet, the cyclical trauma of familial cancer, or problematic relations in the family. They were seen as having an agentive action in the body, being both a risk for and a cause of cancer that interacted with genetic or hereditary factors. I give some account here of these ideas of bodily vulnerability by reflecting in more detail on the experience and narratives of a few of the patients that I met. Ana Paula was in her early 40s. She lived in the southern town of Porto Alegre and worked in a shoe factory in one of the urban suburbs of the city. As well as having had breast cancer herself, a very large number of her family members had had cancer, including her father, a number of sisters and most recently her teenage daughter was being treated for a rare bone cancer. The family were under the care of the cancer genetics team at the hospital and were awaiting the results of blood tests to confirm if a genetic mutation associated with a rare cancer syndrome had been identified in the family. During our first meeting Ana Paula recounted in detail these traumatic cancer-related experiences she had experienced while growing up. Since she and her family had been treated for the disease in the local public health hospital, she had always heard that it "could be genetic" or, as she put it, "it could that different bolts of lightning fell in the same place at the same time." In fact she told me that she always thought that it was "emotional", elaborating she said I am always hearing interviews on the television with doctors about where breast cancer comes from? It comes from continuous hurt, from anger. During the time that I was undergoing treatment I heard this, that it comes from genetic inheritance so it could be emotional factors... and in my family too I'm beginning to think that it's this... to be certain that it's the two together because our head co-ordinates our body. I think that feelings are part of our daily existence and you don't know but one day you say something bad and you hurt someone or you are hurt. I think that it's this because my husband left me when I was unwell and then I started to hold my sadness, my hurt and then the breast cancer developed. I did all my treatment alone and finally my cancer was sleeping. But I was deceived by someone else I was living with and the cancer returned in another place. Similar ideas about the role of emotions were also evident in the way that others I met visiting cancer genetics clinics talked about how emotional problem marked the body. This was evident in the way that Marta -who was in her late 50's -talked about the importance of her catholic faith in dealing with the breast cancer she had had for 8 years and the role of emotions and their effect on the body and self. There is a psychologist currently on TV on the 'better life' channel, which is a catholic network. Anyway, she wrote a book called Good and Bad Anger and she talks about this Asking Marta more directly how emotions might cause cancer she went on to talk about the how they could affect the digestion of food and pass through the body via blood. Let's say you are anxious for some reason and you are going to eat, you are really not going to eat well it's not going to be good for you, you'll be eating irritated, with anger or with hate, a grudge or resentment or whatever… you might be eating in a rush, angry with whoever said something bad to you, but you start eating. Nobody knows what's happening inside you, but that feeling is talking to you, so I think at some moment that feeling passes to our blood, and moves inside of us. In all of these examples bodies were mattered through intersubjective emotions interacting with a seemingly porous notion of the body, in ways that contextualise, displace or redefine the meaning of genetic risk. These ideas of bodily vulnerability related to cancer in Brazil fit well-recognised understandings of the body -including the sick body -as subject to and produced through lived environments, outside influences, the suppression of anger and the emotional vicissitudes of selves in relation to others . More recent research examining new health technologies such as IVF, cosmetic surgery and hormone treatments in Brazil and other regions of Latin America have further illuminated how notions of biological plasticity and bodily porosity inform the response to and the uptake of novel health technologies . In her examination of contraception and menstrual suppression in Bahia, a state in the northeast of Brazil, Sanabria illustrates how humoural-inflected notions of blood are an expression of the body not as fixed or defined by rigid boundaries, but as contingent on and subject to the influence of external factors; an expression of "plasticity," which she argues is also tied to the history of the hygienist movement in Brazil at the turn of the twentieth century . Collectively, this work suggests then that the notion of bodily contingency has a significance in the Latin American region that also informs how genetic risk is translated and acted upon. In this sense, the expressions of embodied vulnerability articulated by patients and research participants in Cuba and Brazil cannot be understood outside the intersection of a range of specific socio-historic dynamics and cultural practices. This includes institutional cultures of state or socialist public health, contemporary histories of socialist medicine, the governance of food, anxieties about the changing role of women, contemporary religious practices and a prominent media informed discourse of self improvement. While these dynamics are clearly somewhat different in Cuba and Brazil, there are some notable similarities in the way in which notions of genetic risk are displaced and reframed. While studies in Europe and North America have also shown that genetic technologies and knowledge may not lead directly to the molecularisation of bodies , the findings presented here raise questions about the assumed pathways and consequences of biomedicalisation linked to the expansion of novel fields of healthcare such as genomics. While notions of bodily porosity or plasticity do seem to be more prominent in certain regions of South America , the relevance of a seemingly "Lamarckian" understanding of disease risk may also potentially be much more widespread, yet still perhaps somewhat under-examined not only in this region but also beyond. In Brazil and Cuba, genetic risk information does nevertheless appear to be layered into a set of meanings relating to bodies that were perhaps never molecularised in the first place. 8 In an era in which the master narrative of "the gene" is being replaced by the contingency and complexity of epigenetics and gene-environment interactions, it will be vital to monitor to what extent "popular" notions of risk for common diseases such as cancer intersect with and are themselves reframed by new understandings of genes, risk, bodies and environments in Latin America and elsewhere. --- Final considerations Analysing the contrasting contexts of Cuba and Brazil, this paper has examined the diverse trajectories in which one field of genomic medicine -cancer genetics -is being translated in Latin America. It has plotted the parallels and differences in the way that cancer genetics has emerged in each of these regions, linked to public health and histories of community genetics as well as in relation to in national and transnational research priorities and funding. It has outlined the diverse ways that health practitioners in these two contexts negotiate the gap between the promise of genetic medicine as personal medicine and public health in a context of scarce medical resources and entrenched health inequities. In each case the activism of practitioners in pursuing somewhat diverse ends has been at the centre of how and why cancer genetics is emerging in Cuba and Brazil. The pathways through which patients and research participants come to clinical cancer genetics are in part informed by the healthcare institutions that operate in each domain, while the ways in which embodied risk for cancer manifests points to convergences in how the relevance of genes are contextualised and in some cases actively displaced. The need to examine a "multiple politics of life" is now being directly taken up in comparative research of novel health technologies outside Europe, Canada, and the United States. The way in which diverse forms of public health governance are central to the mobilisation of medical research and health technologies is becoming apparent, particularly in those contexts where there are "very different historical and political rationalities" and "more collectivist mentalities" . Latin America offers a specific lens through which to examine these developments, given that the pursuit of social medicine has a long history and on-going contemporary relevance in the region as a whole . At the same time, the legacies of different histories of social medicine inform cancer genetics in Brazil and Cuba at the meeting point with transnational research priorities and funding agendas, that also shape the pursuit and practice of this emerging area of medical practice and research. The second half of this paper has examined how patients' engagement with genetic risk information is shaped by culturally 109 109 v.23, n.1, jan.-mar. 2016, p.95-111 109 significant meanings of embodied vulnerability to cancer that, in different ways, emphasise the role of exogenous factors and the seeming porosity of bodies. Yet, in a similar way, these articulations do not exist outside of novel injunctions relating to health responsibilities, often gendered in specific ways, that also form part of the transnational area of cancer genetics and which are so central to the promise of personalised medicine associated with this novel field of healthcare. In summary there is an on-going need to understand how novel fields of healthcare such as cancer genetics are uniquely informed by socio-historic specificity in dynamic response to broader global social processes. The translation of emergent areas of research and medical intervention, such as cancer genetics in Cuba and Brazil, makes visible the complex and frequently divergent pathways of biomedicalisation and the vital need to attend to how these are partially reproduced and heterogeneously localised. --- NOTES 1 Community genetics is often described in terms of an effort to bring clinical genetics to underdeveloped and developing countries and address the well being of the individual within society, see Raz . História, Ciências, Saúde -Manguinhos, Rio de Janeiro 110 História, Ciências, Saúde -Manguinhos, Rio de Janeiro Quarterly, v.27, n.4, p.531-549. 2013b. GIBBON, Sahra. Family medicine, "La Herencia" and breast cancer; understanding the continuities of predictive genetics in Cuba. Social Science and Medicine, v.72, n.11, p.1784-1792. 2011. GIBBON, Sahra. Genomics as public health? Community genetics and the challenge of personalised medicine in Cuba. Anthropology and Medicine, v.16, n.2, p.131-146. 2009. GIBBON, Sahra et al. Familial Cancer, v.7, n.3, p.275-279. 2008. ROMERO, Teresa. Changing the paradigm of cancer control in Cuba. Medicc Review, v.11, n.3, p.5-7. 2009 Quarterly, v.27, n.4, p.471-488. 2013. TERUEL, Beatriz Macheco. Cuba's national medical genetics program. Medicc Review, v.11, n.1, p.11-13. 2009. THORSTEINSDÓTTIR, Halla et al. Cuba: innovation through synergy. Nature Biotechnology, v.22, suppl., p.19-24. 2004. WHITMARSH, Ian. The ascetic subject of compliance: the turn to chronic disease in global health. In: Biehl, João; Petryna, Adriana . When people come first: critical studies in global health. Princeton: Princeton University Press. p. 302-325. 2013. uuuUUU

Translating genomics: cancer genetics, public health and the making of the (de)molecularised body in Cuba and Brazil Convertendo a genômica: genética do câncer, saúde pública e a formação do corpo (des)molecularizado em Cuba e no Brasil Recebido para publicação em novembro de 2014. Aprovado para publicação em junho de 2015.

Background Lack of physical activity and sedentary behaviours have been identified as risk factors for obesity, type 2 diabetes, cardiometabolic diseases, and depression [1][2][3][4]. Regular PA and reducing sitting time are associated with psychological benefits and better social development among adolescents. Also, physically active adolescents are more likely to avoid tobacco, alcohol, and drugs and therefore achieve better academic performance at school [5].An active lifestyle is not only necessary for physical and mental health in adolescence but also undoubtedly carried over into adulthood. Despite common knowledge on active lifestyle, a rapid decline in PA and an increase in sitting time have been observed in Chinese adolescents [6]. Over 50% of primary and secondary students in China spent less than one hour per day engaged in PA [7] and over 40% students spent more than two hours a day in screen-based activity [6]. Several studies have correlated PA of adolescents with their family support such as parents encouraging their children to be more physical active, watching their children engaging in PA, and involving themselves in their children's activity [8][9][10][11][12][13]. These supports have been suggested to affect children's behaviours differently [14]. However, instead of identifying their effects separately, previous studies mostly treated them as a whole. Therefore, we aimed to describe the levels of moderate to vigorous intensity physical activity and sedentary behaviours in a sample of junior high school students in Hangzhou of China; examine the associations between different types of family support and MVPA and sedentary behaviours among Chinese adolescents. The data of this article were part of the data collected in a multinational collaboration programme of the Community Interventions for Health [15]. The primary hypotheses in this analysis are that family support is one of the most important motivational factors for children to move more and sit less; children are motivated differently by different levels of support from verbally encouraging, watching, to involving. --- Methods --- Subjects and sampling methods Participants were students in Grade 7 and 9 from all junior high schools in Xiacheng, Gongshu, and Xihu districts of Hangzhou City. Two independent crosssectional surveys were conducted in 2009 and 2011 in the same schools. Both of surveys applied same sample size requirement and sampling design. The sample size estimation was based on the primary aim of CIH programme, that was, intervention group exposed to interventions regarding tobacco use, food choices, and PA had a 6% greater change in the prevalence of each of the three risk factors [15]. The final sample size was arrived at using knowledge of prevalence of the three risk factors in cities of China. The largest sample size across all three risk factors was selected as the necessary sample size. In addition, 20% was added to allow for non-response. The sample size for each of survey was 4800, with an equal size of 2400 subjects in grade 7 or 9. Stratified cluster sampling was used in both of the surveys. The list of classrooms was stratified by district, school, and grade. A simple random sample of classrooms were selected and all students in the selected classroom were eligible for the survey. Sampling fraction, that is, the ratio of sample size to population size, was applied to each strata to determine the number of classrooms and at least two classrooms in each strata were chosen. A total of 9271 students were investigated and the response rate was 94.8%. The study was approved by the Institutional Review Board at Peking University Health Science Centre . Informed consent ensuring privacy and confidentiality was obtained from both parent and student. --- Measures and variables Each student completed the CIH Youth Survey questionnaire addressing knowledge of, attitudes to, and behaviours in relation to unhealthy diet, physical inactivity, and tobacco use [15]. CIH Youth Survey was designed based on the Health Behaviour in School-aged Children [16]. While formal reliability and validity survey were not conducted in this population, this questionnaire has been shown to have good reliability and validity in other Chinese children [17]. Structured questionnaires were self-administered to the students. Physical examinations including height and weight were measured by trained field staff. The definitions of variables analysed in this article and their grouping methods are described as follows. --- Assessment of PA and sedentary behaviours The MVPA was assessed using the following two questions: How many days per week on average do you participate in MVPA ; and How many minutes per day on average do you participate in MVPA in the aforementioned days. According to global recommendations on PA, children and youth aged 5-17 should accumulate at least 60 minutes of MVPA daily [18]. Therefore, the frequency and duration of MVPA were dichotomised. In addition, the total level of MVPA per week was dichotomized as follows: meeting the recommendation: engaging in at least 60 minutes of MVPA daily; or not meeting the recommendation: less than 7 days of MVPA per week or less than 60 minutes of MVPA per day. Sedentary behaviours were assessed using two questions as follows: on school days, how much time per day on average do you spend in sedentary behaviours for leisure, including using a computer for pleasure, watching television/videos, playing electronic games, sitting and chatting, or other sitting activities ; and on weekends, how much time per day on average do you spend in sedentary behaviours for leisure. According to the global guidelines on youth sedentary behaviour, youth aged 12-17 should limit recreational screen time to no more than 2 hours per day [19]. Thus, the amount of sedentary time on school days and weekends was dichotomized as follows: meeting the recommendation: no more than 2 hours of leisuretime sedentary behaviours per day; or not meeting the recommendation: more than 2 hours of leisure-time sedentary behaviours per day. --- Assessment of family support Family support was assessed by the following three questions: how many days per week on average do your families encourage you to engage in PA; how many days per week on average do your families watch you engaging in PA; and how many days per week on average do your families involve themselves in your activities, that is, having PA together with you. Response options included: never or less than weekly; 1-2 days/week; 3-4 days/week; 5-6 days/week; and 7 days/week. The last two options were combined in the analyses. --- Assessment of other covariates Sex, grade , mother's level of education , health belief, and their perception of the neighbourhood environment, were reported by all students. BMI was calculated as weight in kilograms divided by the square of height in meters. Health belief was assessed by the following question: "To stay healthy, at least how much time per day should children like your age engage in PA" . We grouped health belief into two categories: less than 60 minutes or do not know/not sure; or at least 60 minutes. The perceptions of the neighbourhood environment were assessed by the following three questions: it is unsafe to walk or jog because of the busy traffic and no walking trails around my neighbourhood; parks or gyms are easily accessible around my neighbourhood; and it was safe to walk or jog around my neighbourhood during the day. Response options for these three questions were as follows: disagree strongly, disagree slightly, not sure, agree slightly, or agree strongly. We grouped the perceptions of the neighbourhood environments into three categories in the analyses. --- Statistical analyses Chi-square tests were used to compare the counts of categorical responses between two or more independent groups. Nonparametric tests of K-independent samples were performed on the equality of the medians for the continuous variables without normal distribution. Statistical significance was defined as P < 0.05. The statistical analyses were performed with Stata 12.0 . Three-level nested models were specified using the xtmelogit commands to address the dependency between the individuals in the cluster sampling data. Four dependent variables were tested: participating in seven days of MVPA per week ; at least 60 minutes of MVPA on days in which students reported having MVPA ; meeting the recommended level of at least 60 minutes of MVPA daily; and meeting the recommendation of no more than 2 hours of leisure-time sedentary behaviours per day . Multivariate models were fitted with different levels of adjustment. Model 1 included each type of family support separately. Model 2 included three types of family support simultaneously. Model 3 additionally included sex, grade, BMI, and health belief. Model 4 additionally included survey time , mother's level of education, and the perceptions of the neighbourhood environment. The tests for linear trend across familial support categories were performed by assigning the midpoint values of each frequency category and treating the variable as continuous in a separate model. We also examined associations between family support and MVPA and sedentary time across gender. Test for interaction were performed by means of likelihood ratio tests to compare multivariate models with and without the interaction terms between family support and gender. --- Results We excluded students who were unable to engage in PA due to disability, temporary or permanent disease and with missing in any variables . The final analyses included 3811 boys and 3475 girls. As compared with the excluded students, students kept in the analyses were more in Grade 7 and with more highly-educated mother. There was no statistically significant difference in sex and BMI. Overall, 31.2% of students participated in seven days of MVPA per week and 14.6% had at least 60 minutes of MVPA on days in which students reported having MVPA. Only 9.0% of students met the recommended level of PA for childrenthat is, at least 60 minutes of MVPA daily. 63.9% of students met the recommendation of no more than 2 hours of leisure-time sedentary behaviours per day, with a higher proportion on school days than on weekends. Higher proportions of active students, those who participated in seven days of MVPA per week, had at least 60 minutes of MVPA on days in which they reported having MVPA, met the recommended level of PA and met the recommendation of no more than 2 hours of leisure-time sedentary behaviours per day, received family support almost every day . The characteristics of the study participants according to familial support were presented in the Table 1. Multivariate-adjusted analyses showed that neither frequent verbal encouragement nor frequent watching were associated with higher level of MVPA . Students who reported familial support 1-2 days per week had lower frequency and duration of PA instead as compared with those who reported familial support less than once per week. However, frequent verbal encouragement and watching were associated less leisure-time sedentary behaviours. As compared with students who reported familial support less than once per week, the multivariate-adjusted odds ratios for verbal encouragement were 1.13 for 1-2 days per week, 1.03 for 3-4 days per week, and 1.29 for 5-7 days per week; the respective ORs for watching were 1.20 , 1.23 , and 1.19 . More details were shown in Additional file 1: Table S1 -S4. Multivariate-adjusted analyses showed that the involvement of family member in the children's activity in most of the days of the week was associated with both greater frequency and longer duration of MVPA and less leisure-time sedentary behaviours. The multivariateadjusted ORs among students who reported familial support 5-7 days per week, as compared with students who reported less than once per week, were 1.25 for at least 60 minutes of MVPA on days in which students reported having MVPA, 1.50 for engaging in seven days of MVPA per week, 1.67 for meeting the recommended level of at least 60 minutes of MVPA daily, and 1.48 for no more than 2 hours of leisure-time sedentary behaviours per day . More details were shown in Additional file 1: Table S1-S4. --- Discussions In this large sample size study of Chinese urban adolescents, we found that less than 10.0% of students met the recommended level of PA for childrenthat is, at least 60 minutes of MVPA daily. This study is also one of the few studies to explore the relationship between the family support and children's PA and sedentary behaviours in Chinese urban adolescents. We found that family members involving themselves in the children's activity had the most significant influence on children's behaviours as compared with other types of supports such as verbally encouraging and observing children's activity. Involvement of family members in the children's activity almost every day was associated with higher level of MVPA and less leisure-time sedentary behaviours. It has been suggested that strong family support increases children's self-efficacy to overcome barriers to being physically active [8]. Self-efficacy refers to a person's belief in his or her ability to succeed in a particular situation [20]. Children who have strong selfefficacy believe that they can successfully overcome obstacles to physical fitness and skills. Persuasive information, as in verbal encouragement from parents, can raise self-efficacy, but its effects can be transitory if subsequent performance turns out differently [21]. In contrast with verbally expressed expectation, the effects of parental expectations on children's behaviours seem greatest when a high level of parent involvement exists [21]. The familial involvement in PA may boost children's efficacy to improve MVPA, help children develop skills necessary to be physically active, and provide logistical support [8,22]. Sedentary behaviours have emerged as an important health risk factor independent of MVPA [23,24]. Although verbally encouraging and watching were not associated with a higher level of MVPA in this study, they were observed to be associated with less leisure-time sedentary behaviours. One possible explanation is that some bouts of light activity have replaced sedentary time under the familiar pressure on children. Several previous studies have reported that sedentary time was inversely correlated with light-intensity PA such as walking and playing [25,26]. While increasing MVPA should still be a public health priority, breaks in sedentary time and light-intensity PA may also provide a beneficial adjunct to the current 60 min/day of MVPA recommendation [27,28]. In the current study, we hypothesised that familial support such as verbal encouragement, watching, and involvement is one of the most important motivations for children to get outdoors in the leisure time at home. The more frequent the family members are concerned about their children's physical activity in whatever way, the more physically active children are. We used the weekly frequency as a measurement for the strength of familial support. Familial support could also manifest itself in other ways. For example, parents watch the exercise sessions which they book for their children or involve themselves in planned school events. In this situation, the proportion of parents attending scheduled events would be a better measurement for the strength of familial support. This study acknowledges a few limitations that should be noted. First, given the cross-sectional nature of this study, it is not possible for us to determine the temporal relationship between familial support and children's behaviours. Two-way causal relationship might exist. Stronger family support is associated with more physically active children, while for physically active children who like and specialize in sports, parents may save the efforts to push their children into playing sports. In the current study, we could not exclude the possibility that some physically active children consequently received less familial support which might have attenuated the positive influence of familiar support on children's behaviours toward null. Second, participants were more likely from educated households, possibly resulting in a potential bias [13]. However, the lack of family involvement in children's PA who were from educated household highlights the importance of increasing family involvement to promote children's MVPA and to reduce sedentary time. Third, residual confounding by other unmeasured or unknown factors such as children's preference for PA and parental level of PA remains possible although we have carefully adjusted for some potential factors for children's PA. Forth, the levels of MVPA, sedentary time, and family support were self-reported; therefore, some measurement error is inevitable. However, relevant survey instruments have been shown to have good reliability and validity in diverse populations [17,[29][30][31]. --- Conclusion In summary, in this large sample size study of Chinese urban adolescents, we found that family members involving themselves in the children's activity had the most significant influence on children's behaviours as compared with other types of supports such as verbally encouraging and observing children's activity. Any type of familial support including verbally encouraging, watching, and involving had effects on reducing leisure-time sedentary behaviours. --- Additional file Additional file 1: Table S1. The association between family influence and the duration of MVPA per day . OR for the duration of MVPA of child by family influence. Multivariate models were adjusted for the following factors: model 1 included certain family influence ; model 2 included all family influence ; model 3 additionally included sex, grade, BMI and health belief of daily PA time; model 4 additionally included survey time, mother's level of education and opinion regarding neighbourhood environment. Based on model 4, three-level nested models were specified with equal random effects variance and covariance being zero, and the coefficient is the SD of random intercept. Family influence was recoded as the continuous variable: less than weekly: 0; 1-2 days/week: 1.5; 3-4 days/week: 3.5; 5-7 days/week: 6. P-values for trend represent family influence group comparison. Table S2. The association between family influence and the frequency of MVPA . OR for the frequency of MVPA of child by family influence. Table S3. The association between family influence and total MVPA per week . OR for the total level of MVPA of child by family influence. Table S4. The association between family influence and sedentary time . OR for sedentary time of child by family influence. --- Competing interests The authors declare that they have no competing interests. Authors' contributions XW drafted the manuscript. JL developed the sampling design and reviewed the manuscript. XW and LJ undertook the statistical analysis. QML and YJR organized the field work. LML developed whole research designs, contributed to the interpretation of the results and reviewed the manuscript. All authors read and approved the final manuscript.

Background: Family influence plays an important role in a child's physical activity (PA). This study aimed to describe the level of moderate to vigorous intensity physical activity (MVPA) and sedentary behaviours among Chinese junior high school students and examine the associations between different types of family influence and MVPA or sedentary behaviours. Methods: Participants of two independent cross-sectional surveys, conducted in 2009 and 2011, were students in Grade 7 and 9 from all junior high schools in Hangzhou, China. The daily duration and frequency of MVPA, amount of sedentary time and frequency of family support were self-reported. Multi-level mixed-effects logistic regression was used to examine the associations between different types or levels of family influence and MVPA or sedentary behaviours. Results: A total of 7286 students were analysed finally. Overall, only 9.0% of the students participated in MVPA at least 60 minutes/day; 63.9% spent no more than 2 hours/day in sedentary behaviours. Frequent verbal encouragement and watching were associated with less leisure-time sedentary behaviours. The multivariate-adjusted odds ratios (ORs) for verbal encouragement and watching were 1.29 (95% CI, 1.08 to 1.55) and 1.19 (95% CI, 0.97 to 1.45) for 5-7 days per week. The involvement of family in the children's activity in most days of the week was associated with both higher level of MVPA and less leisure-time sedentary behaviours. The respective ORs among students who reported familial support 5-7 days per week, were 1.50 (95% CI, 1.21 to 1.86) for engaging in seven days of MVPA per week, 1.67 (95% CI, 1.19 to 2.32) for at least 60 minutes of MVPA daily, and 1.48 (95% CI, 1.19 to 1.84) for no more than 2 hours of leisure-time sedentary behaviours daily. Conclusions: This study found that less than 10.0% of urban Chinese adolescents engaged in MVPA at least 60 minutes/ day. Family involving themselves in the children's activity exerted the most significant influence on children's behaviours as compared with verbally encouraging and observing children's activity. Any type of familial support including verbally encouraging, watching, and involving had effects on reducing leisure-time sedentary behaviours.

Introduction Teaching Assistants1 constitute a major workforce in the United Kingdom, with their number increasing since 2010/11, from 221,500 to 281,100 . One reason for such an increase is that they are viewed as key figures in the inclusion of disabled students in mainstream schools . Nevertheless, the extent to which ta s are facilitating the inclusion of disabled students has been debatable, with the literature focusing on ta s' training, their deployment, and the types of support they offer . Disabled children's insights into ta s' support also reveal their mixed feelings towards it, with such support being experienced occasionally as enabling and other times as disabling . This paper explores the experiences of fourteen young people with dwarfism,2 who were supported by a ta in their secondary education in the United Kingdom. Drawing on their first-hand accounts, this paper aims to illustrate how support both re-inscribes cultural beliefs about disability as a problem to be 'managed' and/or 'fixed' and institutional power over disabled young people . To illustrate this, I draw on Foucault's box of tools and weave his discussion of surveillance, panopticism, and resistance into the participants' stories. Allan notes how this box of tools is particularly useful in questioning dominant discourses e.g. biomedical, deficit, individualistic, of disability, and how these are materialised in schools. Meekosha and Shuttleworh also argue that what makes Foucault's ideas so useful to cds3 is that they perform a radical de-familiarisation of modern institutions and practices as caring and benevolent and reveal technologies and procedures that classify, normalise, manage and control anomalous4 body-subjects. --- Teaching Assistants and Inclusive Education ta s have been discussed as key figures in the inclusion of disabled students in mainstream schools . As Webster and de Boer propose, 'In many schools and classrooms across the globe, inclusion has become ostensibly contingent on the creation and utilisation of a relatively new type of educator: the ta' . ta s have been described as the 'change agents' for inclusion and the 'solution to inclusion' due to the positive impact reported they have on disabled students in various areas , such as their academic performance and social integration . Nevertheless, ta s have also been viewed as hindering the inclusion of disabled students, shifting the attention away from the structural inequalities and root issues disabled students deal with , and acting as 'human resources to mind the disabled student' . Whitburn notes that 'support personnel are deployed in a way that perpetuates the special education tradition in inclusive education.' As Mortier et al. propose, 'these types of 'solutions' , enacted when deeper changes seem harder to achieve, risk bringing too much closure and lead to recreating forms of exclusion' . Such exclusionary practices are considered, for instance, in Holt et al.'s research with disabled students, where ta s moved the disabled child out of the classroom and in a separate space to run an intervention or when ta s were sitting next to the disabled child at a separate desk in the back of the classroom. Disabled students' insights into ta s' roles also demonstrate how the latter sometimes hinder rather than facilitate the inclusion of the former . Although Giancreco and Doyle note that disabled students' insights are 'notably absent from research about teacher assistant supports' in comparison to other stakeholders' insights' , such as teachers and ta s, there have been studies looking into how disabled students discuss the support they receive from ta s. Such research has involved disabled students with different impairments5 and took place in different countries.6 Some recurring themes in the studies exploring disabled students' insights are: -The stigma of support: disabled young people feel stigmatised by the presence of a ta next to them, as they believe that it stresses their 'difference' from their peers. -The threat to independency: disabled students report they want to do or to learn how to do things by themselves rather than having ta s helping them with everything. -The lack of input by disabled young people themselves regarding the support offered to them: it is quite common among disabled young people to be left out of the decision-making process concerning the support provided to them. Furthermore, some researchers explored how disabled students perceive their ta s, introducing different typologies, and the implications that each role have on disabled students. For instance, Broer et al. distinguish between four roles of ta s -mother, friend, protector and primary teacher, and Skar and Tamm discuss ta s as 'the replaceable assistant' , the assistant as mother/father, the professional assistant, the assistant as a friend, and the ideal assistant. Regarding the implications of each role, it is noted, for instance, how the 'mother' and 'friend' perception of a ta could have a negative effect on the relationship between disabled students and their peers . These perceptions also relate to the different deployment of ta s. For instance, there is the one-on-one model, with ta s working on a one-to-one basis with disabled students, which can result in the micro-exclusion of the students due to working separately to their peers . The second model is the class support model, during which ta s work with different students based on the class teacher's instructions . The literature on ta s and inclusion, however, has often focused on specific themes, such as their role, the types of support they offer, their benefits and drawbacks, their involvement with the inclusion of disabled students, their deployment, and their training, which Giangreco describes as 'status quo thinking' . This paper takes a difference stance towards ta s' support, that is, how such support comes at the expense of disabled young people finding themselves under incessant, adult surveillance, with the hidden agenda of normalisation. --- Methodology The research explored the secondary schooling experiences of 19 young people with dwarfism in the United Kingdom, adopting a narrative, qualitative approach . In particular, participants were provided with a range of options of how they preferred to share their stories, including oral storytelling , visual storytelling and digital storytelling . The presence and support from a ta was a theme that featured in their accounts. Out of the twenty young people, fourteen had a ta and it is their stories that are analysed in this paper. The inclusion of teenage participants was important, as they could reflect on how they experienced One-to-one assistant Online One-to-one Interview the support they were getting while they were still in school, whereas older participants were able to provide retrospective narratives of what their experiences with their ta s were like from an adult perspective. It is noteworthy that the teenage participants reported they had classroom assistants compared to the majority of the adult participants who had one-to-one ta s, indicating a shift in ta s' deployment. The differences between having a classroom assistant and a one-to-one ta are considered in the analysis of the stories. Recruitment was mediated by a range of gatekeepers, such as administrators of various Facebook groups and charities, and parents of participants who were below the age of sixteen. The most effective recruitment method was my attendance to annual, in person meetings of one association for people with dwarfism. Unlike the online advertisement of my research on platforms such as Facebook , which resulted in few participants expressing interest in participating, the personal interaction with participants in a safe space led to the recruitment of the majority of participants. This is understandable, especially considering the cyberbullying and mockery people with dwarfism experience online by non-disabled people . Snowballing was another effective method of recruitment . Ethical approval was granted by the University I did my doctoral research at. The ethical guidelines of the Economic and Social Research Council and British Educational Research Association were followed. All adult participants provided informed consent after reading the relevant information sheet, and all teenagers consented to their participation after gaining their parents' informed consent. Teenage participants were also reminded that, ultimately, it should be their decision and not their parents' if they wanted to take part in the research . Furthermore, a chaperone -the mother-was always present during the interviews with teenage participants. It was the mothers of the teenage participants that had consented for their children's participation in the research and, because this was our first in person or online encounter, they felt it was right to remain present during the interviews. Abbott also reflects on how parents of disabled young people were present in the interviews he had conducted and the implications of their presence e.g. what disabled young people chose to share or how parents' perspectives complimented the young people's ones. Additionally, the right to withdraw at any time was emphasised to all participants and anonymity was compulsory for the teenage participants and optional for the adult ones. Participants were also informed that I would strive for secure confidentiality, although this might be compromised due to the fact that this is a specific impairment group where many people know each other due to the common spaces they share, such as hospitals and associations . Interviews were transcribed verbatim and analysed through a narrative thematic approach . This method was chosen since it allows for the identification of themes within and across the stories . The analysis included the following stages: -I started with reading and re-reading the stories multiple times to achieve what is referred as 'familiarisation' with the data . Such familiarisation had already started during the transcribing process. -The familiarisation process was followed by manual colour-coding, that is searching codes within the stories and across the stories , which were highlighted with different colours . -Colour-coding was succeeded by the identification of themes and sub-themes. For instance, using the term 'teaching assistant' as a code, I went through the participants' stories and highlighted it whenever I came across it. Then, I read the full sentences within which this code had been encountered and started identifying common patterns in relation to when this code was used. --- Findings and Discussion The participants' stories are discussed next, illustrating the surveillance they endured under the guise of support from their ta s, and their attempts to resist it. A Foucauldian analysis of these stories take place by weaving Foucault's discussion of surveillance and resistance into them. Nevertheless, these stories also provide nuanced understandings of surveillance, which go beyond Foucault's panoptic model of surveillance. The findings have been divided into three sections based on the 'type' of surveillance: a) stories of panoptic surveillance b) stories of panaural surveillance and c) stories of synoptic surveillance. Each section starts with an explanation of the key concept that guide the analysis of the participants' stories, which are presented and discussed next. --- Stories of Panoptic Surveillance In "Discipline and Punish: the Birth of a Prison", Foucault describes schools as sites of surveillance, where a range of technologies are used to produce docile subjects. One of those technologies is panopticism. Panopticism consists of the internalisation of the sense that one is relentlessly watched , so that they behave in a specific manner . As Foucault argues, 'It is the fact of constantly being seen, of being always able to be seen, that maintains the disciplined individual in his subjection' . The idea of panopticism derives from the Panopticon , a prison designed in such a way so that inmates always feel that they are being watched. The purpose of the Panopticon is for their inmates to internalise the disciplinary gaze, even if surveillance is discontinuous, so that they self-govern their conduct accordingly. Rather than considering the Panopticon literally, Foucault suggests that it 'must be understood as a generalizable model of functioning; a way of defining power relations in terms of the everyday life of men.' The young people shared many stories of how they found themselves under the surveillance of their ta s in the panoptic classroom . Such surveillance was in line with two of the institutional roles of ta s: firstly, to facilitate the teacher with behaviour management, especially managing 'disruptive' behaviour and, secondly, to keep students 'on task' . The following two stories are illustrative of the first form of surveillance: Rania: And then, yes, say if you are breaking a rule, like my school, they absolutely hate chewing gum, so say if I was like being sneaky, had some chewing gum in the lesson, 'Rania, chewing gum' . It's like: I don't need you to tell me [..]You're not saying that to anyone else, which is quite cliché. You're just saying that to me, cause you're here to watch me, when actually that's not what you're doing, that's not what you're here to do. This it like to do, anyway, that was very annoying in that sense. Both Rania and Patricia reflected on how they were expected to behave in certain ways when ta s were around. Rania was aware of the surveillance she was under -you're here to watch me'-, but she felt conflicted by it -'that's not what you're here to do'. Moreover, Rania indicated that she already knew the behaviour norms -'I don't need you to tell me'-, but she felt she was being singled out, which she described as 'cliché' , showing that it was something expected for the disabled student to be singled out. The presence of the ta, as it was an one-to-one ta, meant that Patricia could not engage in any sort of behaviour that would be understood as 'naughty' or 'rebellious' . Patricia was also conscious of the limits of how naughty or rebellious she could be, as indicated by her use of the expression 'a little bit' . This 'a little bit' highlights also Patricia's struggle between resistance to ta s' power and conformity to the rules . According to Slater , the 'non-normativity' of young people e.g. being rebellious is 'only permissible to young people fitting neatly into other culturally privileged positions' . The non-normativity of those who do not fit into these positions, including disabled young people, leave them in much more precarious positions . This was evident in the two stories above: Patricia needed to self-govern her 'non-normativity' in terms of who she should be and who she could not be, and Rania was immediately singled out and corrected when she attempted to break the rules. In both stories, participants were conscious that they were being watched and that they had to self-govern their conduct accordingly, that is to turn themselves into docile subjects. Their observation by their ta s was hierarchical, in terms of being coercive, since the participants could not do much to cease it. ta s also paid close attention to the young people to ensure that they were always on task: Rania: If I was like, as a student, as you do, if you were sat next to somebody you were getting on with and had a chat with him. And then, they will always say: 'Oh, what are you doing? Oh, ok. Oh, are you alright with that question? Are you alright with this? Are you alright with that?' I am like: 'Fine' . 'Oh, come on, just get on' , like 'Let's get on with your work then.' 17 years old, pair interview, classroom assistant Lynn: Yeah, it is annoying. Like I was talking inside about the same. Oh, you need to pick four teaspoons of sugar and she goes "Shush, Lynn, Lynn is just talking." [..] She goes like: "Do your work". Christine: I don't talk too much if I am with my best friend. We giggle too much, we don't talk too much, we just giggle. And we do our work. And then, they just tell us off, even though we were doing our work. 13 years old, one-to-one interview, classroom assistant The ta s exercised their panoptic gaze and intervened when they thought that the young people were not focused on their work. What is common in these three stories is the persistence of ta s, who, even after being reassured by the young people that they were on task, insisted on the same course of action, reflecting the cultural mistrust towards young people, which is exacerbated towards those positioned as 'at risk' . Similar feelings of irritation from disabled young people who were under ta s' surveillance have been reported in other studies . Moreover, as noted by Rubie-Davies et al. , ta s seemed to be more concentrated on task completion rather than learning, which was also evident in the above stories. The stories of surveillance shared so far illustrate how ta s exercise panoptic surveillance on the young people in the classroom as part of their institutional role and with the purpose of producing the docile, self-governed subject. Nevertheless, surveillance extended to other spaces, such as playgrounds and dining rooms . Holt considers how disabled young people find themselves under adult surveillance in different spaces in the panoptic school. 'Outdoors' surveillance was also brought up in the stories of some young people: Patricia: Eh, yeah, she used to go for a coffee and then she kind of used to come and almost kind of find me sometimes. But again, she'd be quite good at kind of keeping her distance, so she just kind of stand almost in the background and just kind of make sure like, if I kind of say to her: 'Oh, can you come here for a minute?' And then she'd come and say like: 'What was it that you need' or whatever. So, em, so yeah, she was quite good at that really. 30 years old, one-to-one interview, one-to-one assistant Lynn: Like at class, when I leave five minutes early, cause I wait for my friend and someone else and , teaching assistant goes: 'Oh, I will wait out here for you.' […]She goes like 'No, I'll just wait.' And then my friend comes out and she walks with us. Rania: And then like someone who like walks around with you after. So, like every so often when you go to your lessons: 'Oh, where are you next?' 'I am over in the Science block. 'Alright, I will walk with you.' 17 years old, pair interview, classroom assistant The young people could not escape their ta s' surveillance, who did not leave them out of their sight8 at any moment, indicating the hierarchical nature of surveillance. In Patricia's story, this surveillance from a discreet distance made her feel quite safe, in terms of having immediate access to her ta in case she needed anything. However, it also indicated that panoptic surveillance could occur from a distance. On the other hand, Lynn and Rania did not appreciate being followed everywhere. Their stories show the institutional power ascribed to ta s permits them to exercise their panoptic gaze, as they did not have to ask the young people if they were alright with being followed everywhere. --- Stories of Panaural Surveillance Besides Foucault, various other scholars have adopted the 'Panopticon' as a metaphor to consider how surveillance occurs in schools and how everyoneteachers, ta s, students -get entangled in it . However, the dominance of the panopticon in examining surveillance has been understood as oppressive by various Surveillance Studies scholars , who have proposed moving towards post-panoptic understandings of surveillance . An important critique of panopticism that is pertinent to this research is its ocularcentrism, that is the emphasis on vision and surveillance's reduction to the gaze/being watched . Gallagher proposes that "dominant ocularcentric understandings of surveillance could be enlarged to encompass what might be termed 'panauralism9'". Panauralism refers to aural surveillance exercised through the sense of hearing by being overheard . The following stories illustrate how panauralism manifested in secondary schools. The presence and proximity of the ta, as discussed by the young people, affected the content of the conversation they could have with their friends: Louise: But if you have a ta, I don't know about you, but if you have a ta people are like, it's a bit weird , cause they'll be always with you and you just want to be with your friends. just to have a normal conversation. 16 years old, focus group, classroom assistant John: If I am talking to my friend, I won't be able to like, it's different, yeah. But like, when I am sat next to a teacher with me, even if it's in a classroom coming and checking on me, it's like, I'd rather not have them there actually, cause I want to talk, like, yeah, I want to talk to my friend about something and it's been more private. I don't want teachers just there. 16 years old, focus group, classroom assistant Louise, John and Mary reflected on how the ta's presence impacted on what they could discuss with their friends. All three young people mentioned how they would prefer their ta s not to be next to or around them all the time, especially when they were with their friends, indicating the toll that panauralism could have on their friendships. While John did not provide any examples of what he considered as a 'private' topic, Melrose and Patricia shared the following stories regarding potential topics that they would not like their ta s to be there for when discussed: Melrose: And obviously, when you are in high school, you talk about different things. Like, you know, you start your period, puberty; you don't want an older woman there while you're talking about things like that. So yeah, that was, that was, that was difficult. Patricia: When you kind of like, I don't know, if you like boys or whatever, and you wanna say: 'I quite like him' to your friend and, or your friend wants to say it to you or anything like that, yeah. Anything private that you kind of want to discuss almost you kind of don't want them to listen, it feels like another person kind of there. I mean sometimes that was quite nice, cause she gave quite helpful advice, but then, other times it was just like I want this privacy kind of thing. Melrose and Patricia expressed their hesitation to discuss more private topics with their peers in front of their ta s, such as menstruation and intimate relationships. They described this aural surveillance as 'difficult' and as a form of intrusion to their privacy respectively. Both Melrose and Patricia had oneto-one ta s, which intensified such surveillance. However, they had different perceptions regarding the age of the ta s and the impact it can have. For Melrose, the fact that her ta was an 'older woman10' exacerbated her hesitancy to talk about topics such as menstruation. On the other hand, Patricia recognised certain positive aspects of having an adult ta, as she could also provide her with helpful tips. What all of the above stories demonstrate is that the young people are conscious of this panaural surveillance and they have to self-govern/self-censor themselves in terms of what they could share with their friends. As Skar and Tamm claim, disabled young people experience their relationship with their ta s as 'an invasion of privacy' and as asymmetrical, in terms of them sharing everything about themselves, without them knowing anything about their ta s. --- Stories of Synoptic Surveillance As Foucault asserts, 'where there is power, there is resistance, and yet, or rather consequently, this resistance is never in a position of exteriority in relation to power.' Against the danger of portraying the young people as passive recipients of surveillance , the following stories explore the ways and the means that the young people used to resist such surveillance and the potential outcomes of such resistance. Prior to presenting the stories though, I consider how Foucault understood resistance, which I suggest aligns with a Critical Disability Studies consideration of resistance , and the necessary shift from panoptic to synoptic surveillance. Foucault argues that [We find] a plurality of resistances, each of them a special case: resistances that are possible, necessary, improbable; others that are spontaneous, savage, solitary, concerted, rampant, or violent; still others that are quick to compromise, interested, or sacrificial; by definition they can only exist in the strategic field of power relations. But this does not mean that they are only a reaction or a rebound, forming with respect to the basic domination an underside that is in the end always passive, doomed to perpetual defeat. This openness to what can be understood as resistance is in line with Gabels' and Peters' engagement with resistance theory, which 'allows disability studies to acknowledge the importance of all forms of resistance by disabled people' . This is particularly important in the case of disabled children and young people, who are often constructed as 'passive' and 'immature' subjects . However, my aim here is not to think of resistance as the 'panacea' to the ableist ideologies perpetuated in ta s' practices , but to use resistance as a concept to shed light on the existing power relations at schools and to indicate young people's knowledge of such power structures and the ways they negotiate them. Moreover, a turn to resistance means a shift from the panopticon to the synopticon. The notion of the synoptic, as discussed by Mathiesen , is a reversal of the panopticon: while the panopticon assumes that the few see the many e.g., the guardians watch the prisoners in the Panopticon, the synopticon assumes that the many see the few. Landhal applies the idea of the synopticon in the context of a classroom, where the teachers are also visible and exposed to the students' gaze. This reversal of the gaze, Landhal argues, is what opens space for resistance. The following stories illustrate how the young people exercised their synoptic gaze or how they re-turned the gaze to their ta s, and the outcomes of such resistance. Rania and Lynn were among the most outspoken young people regarding resistance and, hence, many of the stories following below are theirs. This resistant attitude could be partly attributed to the background of their families, who are active members of the organisations that I have worked with. Moreover, these two were interviewed together, so their stories were complimenting each other. As argued earlier, the purpose of surveillance is the production of docile, self-governed bodies , which is in line with the institutional role of ta s . Both Rania and Lynn shared stories whereby they acted as 'indocile' subjects against their ta s' surveillance: Rania: Whenever I was just like working and say if I was like, as a student, as you do, if you were sat next to somebody you were getting on with and had a chat with him and then, they will always say, and then they will always 'Oh, what are you doing? Oh, ok. Oh, are you alright with that question? Are you alright with this, are you alright with that?' I am Lynn: Yeah, it is annoying. Like I was talking inside about the same: 'Oh, you need to pick four teaspoons of sugar' and she goes: "Shush, Lynn, Lynn is just talking" and I am like: 'Yeah, she's talking to that group, not my group or to anyone.' She goes like "Do your work". I am like "I was, but you stopped me". . And then, like two minutes later, "Lynn, pack up, we're going". I am like "Seriously? And I was having fun." She makes me stop lessons, like miss the good things , cause she talks to me. Like I only did two minutes of that and I am supposed to do like five. 'You made me miss three minutes of that, so that's not nice or fun'. Instead of conforming with the ta s' orders, Rania and Lynn spoke back to them and resisted their governance, re-turning the gaze to their ta s. Rania's resistance built up gradually, from 'fine' to 'please leave me alone' , and Lynn called out her ta for not adhering to the instructions they were given: 'you made me miss three minutes' . However, in the case of disabled young people, resistance can lead to more surveillance and/or punishment, as they are constructed as the unruly subjects who need to be disciplined . For instance, Lynn was removed from the class -what could be seen as punishment from a Foucauldian perspective -for speaking out. Whilst Lynn's ta drew on the power assigned to her as a staff member to punish her, some young people also used the existent hierarchical power relations in schools, which distinguished between the class teacher and ta s, as a means of resistance: Rania: And like it's like you've got a second teacher just on your case. And it's like, if teachers say that I am chatting with my friend, we are chatting about French, she or we chat about languages, she's French. I am asking her about how do you learn a second language when you are in a German class. And I am like, you know, I am trying to get on with my German, just chatting about languages. And it's like: 'Stop with that, get on with your work.' And I am like, 'You're not, you aren't the teacher, and you're not here to help me with my learning. You are just here to help me, to support me if I got a problem with my back or with my legs or anything like that. 17 years old, pair interview, classroom assistant, added emphasis In this story, Rania first demonstrated her awareness and irritation for being under the ta's constant surveillance, who she described as a 'second teacher' . In her attempt to resist to the governance of her conduct by her ta, she made it clear to them that their responsibilities differed from the ones of a class teacher, with the latter being responsible for her learning and the former for physical support. As various studies have shown, young people are cognisant of the different roles between ta s and class teachers and their status in schools . Moreover, in line with the Foucauldian analysis, this story indicates the circular structure of power, meaning that power is not one's possession, but all individuals are caught up in hierarchical power relations . By exercising her synoptic gaze, Rania showed that ta s enact surveillance whilst also finding themselves under surveillance . Foucault also proposes that the Panopticon is structured in such as a way that those who exercise surveillance are also exposed to surveillance. Furthermore, the resistance in the above stories was against the subjectification of the young people based on fixed notions about their impairment . Such resistance manifested in Rania's story: Rania: And then, often I think, because you're small, there's always with lots of people, and this is gonna sound quite generalizing, but to be fair, it is with older people, which in my case my ta, the ta s that were shared were, they were generally older, near retiring age. There's always this like patronizing nature, cause they see like you're small. They 'll always see like, oh, like, like you got a younger mind, when actually with that definitely isn't the case with me. I am 17, at the age of 11 I passed a test to get to a grammar school. I was in the top 5% of my area, I am very capable. I am like, I am intelligent and I am fair and I am capable like looking after myself and doing my work, which a lot of them didn't like see, and it was annoying. So, when you got like a ta to help you with your bags or make sure you are ok with your back, I don't want them to try to help me with my work, I am very fine with my work, thanks. Rania's story illustrates her resistance towards the ta s' surveillance that see her and hear her in specific ways, such as 'small' and with a 'younger mind' . Not only does Rania resist the gaze that aims to discipline her , but she also resists the gaze that sees her as small . The constitutive gaze is not a neutral gaze, but a gaze which draws on cultural understandings of disability and smallness that are degrading and infantilising. Against her subjectification as small and with a younger mind, Rania framed her resistance drawing on the ableist register, meaning that her claims of capability built on discourses of able-mindness that are valued in schools, such as performing well in exams and achieving high scores that rank you among the top students. This indicates the struggle between resistance and conformity, meaning that the resistance against the subjectification as 'small' and its cultural implications e.g. of little importance, relied on the level of conformity with schools' ableist standards and expectations respectively, such as performing high in tests and being an independent learner . However, such resistance proved to be futile, since ta s "didn't like see and it was annoying", indicating the power of the constitutive gaze that refuses to see the subject otherwise. Finally, Rania acknowledged that most often it was older ta s that held such patronising attitudes towards her, indicating the ageist , dis/ableist power relations in place. --- Conclusions Rather than taking for granted that ta s' support always serves benevolent purposes and enables the inclusion of disabled students in mainstream schools, this paper focused on how such support also serves other 'hidden' agendas, such as the surveillance and normalisation of disabled students. Furthermore, ta s support re-inscribed cultural beliefs about disability e.g. as a problem to be managed and/or fixed, legitimating the institutional power over disabled young people. Specifically, the stories presented and discussed earlier have indicated how disabled young people find themselves under their ta s' constant surveillance, with the aim of the production of the docile, self-governed subject. Such surveillance takes place not only in the panoptic class, but extends to other spaces, such as the playground. Moreover, these stories have illustrated the multi-sensory nature of surveillance, that is, how surveillance was exercised through panopticism and panauralism . This has been one of the original contributions of this paper, with participants' stories illustrating that other modes of surveillance -beyond panopticism-are also exercised in schools, and their effects should be considered. Furthermore, the stories have demonstrated that disabled young people are not passive towards their surveillance, but they develop a range of resisting practices to the governance of their bodies and their conduct, such as the synoptic gaze. Their stories highlighted the relationality of surveillance, that is, those who find themselves under surveillance -on this occasion, disabled young people-also position those who surveil them -on this occasion, ta sunder surveillance. However, such relational surveillance is not devoid of institutional power relations. For instance, the disabled young people's resistance through the exercise of the synoptic gaze is either dismissed or it can even result in their punishment as a means of discipline due to the institutional power ascribed to ta s. This is another original contribution of this paper, which calls for more attention to be paid on how surveillance is exercised relationally and is always embedded in power relations. Moreover, it has been shown that ta s' panoptic gaze is a constitutive/subjectifying gaze that sees disabled young people in specific ways e.g. as disabled and as small, within a culture with degrading concepts of disability and smallness. Even when this gaze is defied, it still manages to interpellate subjects into fixed positions. Before concluding, I need to make an important clarification. Similar to Giangreco , 'I am not calling for the elimination of, or even necessarily the reduction of, ta s in schools' . The aim of this paper is not to hold specific individual ta s accountable for 'bad' practice or to undervalue ta s' labour and its importance. ta s' support can bring about important benefits and can be enabling . Moreover, the economic cuts on disabled children and young people have already widened the inequalities experienced between disabled and non-disabled children and young people in education. What I am arguing, however, is that that the institutional support offered by ta s to disabled young people, in its current form and in the context of a neoliberal ableist education , aims at their surveillance and normalisation. Nevertheless, if support is to be meaningful and enabling, disabled young people should be involved in the decisions of what support they need and how this could be provided, rather than specific types of support being imposed on them. Two criteria that McLaughlin proposes in relation to support are: ' it should not reduce the agency of disabled people to have a vital say in what they receive; support is more likely to be caring in its responsiveness when those receiving and providing it are recognised as --- Disclosure statement No potential conflict of interest was reported by the author.

Teaching Assistants (ta s) are described as key figures in the inclusion of disabled young people in mainstream schools. However, disabled young people's insights into ta s' support remain limited. This paper contributes to such insights by exploring the experiences of young people with dwarfism in secondary education with their ta s in the United Kingdom, drawing on their qualitative, first-hand accounts. I engage in a Foucauldian analysis of ta s' support, where I argue that, under the guise of support, disabled young people sustain ta s' panoptic surveillance, aiming at their normalisation. Nevertheless, the stories reveal modes of surveillance that go beyond panopticism, such as panauralism. Moreover, the stories demonstrate how young people resist to their surveillance by re-turning the gaze to their ta s, exercising the synoptic gaze. The paper concludes by considering the contributions of those stories, including the need to consider surveillance in relational terms, its multi-sensory nature, and its constitutive power.

Introduction Public open spaces are key built environment elements within neighborhoods intended to encourage various physical activities, provide a number of significant benefits, and serve various important functions that improve the quality of life in cities [1][2][3]. In urban and landscape planning studies, the quantity and quality of open spaces in a community have been eliciting increasing attention. Open spaces can promote residents' outdoor activities, which in turn help reduce stress and provide opportunities that promote relaxation [4][5][6][7]. Many dynamic factors and their complex interactions affect the influence of open spaces on human health in urban areas [8]. Based on studies on outdoor activities, many countries have formulated a number of policies and guidelines that guide the planning and design of public open spaces with the aim of promoting the use of open spaces by residents. Landscape design of public open spaces, which usually cover large areas with complex components and functions, have become a challenging task in the field of urban landscape design [9]. A growing body of literature has examined the association between the different aspects of open spaces and physical activity. Kaczynski and Henderson's literature review indicated that most studies have found that proximity to parks and recreational settings is generally associated with increased physical activity [10]. The spatial configuration of parks, their number, and their accessibility determine their access potential for residential populations [11,12]. The accessibility of open spaces is usually assumed to be the most important factor that influences their use [13,14]. A distance of 300 m to 400 m from a user to an open space is considered an important threshold. When the distance is greater than 400 m, the use frequency decreases rapidly [15]. Although the number of studies that focus on accessibility is increasing, several inconsistencies have been found in their results. For example, Lachowycz and Jones's literature review indicated that only 40% of related studies have confirmed significant associations between access-related measures of local green spaces and physical activity; the majority found weak or no associations [16]. Giles-Corti et al.'s study on the western Australian city of Perth revealed that among open spaces with similar scales, 70% of the respondents opt to visit open spaces that they find most attractive rather than the nearest ones [17]. Schipperijn et al. concluded that distance to green spaces is not a limiting factor for the majority of the Danish population [18]. Nielsen and Hansen also pointed out that only 3% of their respondents consider distance a barrier to the use of public open spaces in Denmark [19]. The definition of accessibility needs clarification and deep consideration. The definition of accessibility is based on a gravity model, which is conceived as a measure of the desire and ability of people to overcome distance or travel time to access a facility or activity [17]. Distance from the origin to an open space is still commonly measured with the straight line distance; however, street-network distance is argued as a better representation of the true relevant spatial distance [11]. A study also suggested the use of pedestrian networks and space syntax to understand the measures of proximity to public open spaces [9]. The representations of open spaces differ in different studies; sometimes, open space entrances are utilized as the destination point [13], whereas commonly, the geographic centroids of public open spaces are used [20]. These different research methodologies and results on accessibility confuse researchers, landscape architects, and policymakers. Thus, creating specific references for landscape design is difficult. The other factors that influence the use of open spaces are also varied and complicated. The attributes of open spaces per se provide cues about how they are used and by whom. Previous studies have examined the associations between the spatial attributes of open spaces and physical activity or walking. Holman et al. found that the use of parks is influenced by aesthetic features, presence of amenities, and park size [21]. However, Janet and Rachel found that the actual or perceived dimensions of open space size did not directly influence user preferences; instead, adherence to the venue design itself and the venue's opportunities for activities and landscape appear to be more important [22]. Schipperijn et al. indicated that generally, no association exists between outdoor physical activity and the size of, distance to, and number of features in the nearest urban green space; however, they found positive associations between physical activity and walking/cycling routes, wooded areas, water features, lights, pleasant views, bike rack, and parking [13]. Giles-Corti et al. used a score of open spaces called "attractiveness" to examine the association of open spaces with walking and found that people with good access to attractive open space are 50% more likely to achieve high levels of walking [17]. Goličnik and Ward Thompson reported that having a minimum supplementary space configured to allow for appropriate buffer zones is an important aspect to be considered in public space design and decision making [23]. These research discoveries indicate that physical activity in an urban open space might be stimulated by providing attractive facilities and landscape features. Additionally, the urban open space provision might be brought back to the open space itself. Aside from the attributes of open spaces and the distance between users and open spaces, the attributes of users are also important factors that affect the use of open spaces. Studies have found significant variations in the utilization of nearby open spaces by different groups. Among women and men and among younger and older adults, the relationships between activity and living near parks and parklands show significant differences [20]. Age, gender, marital status, and area of residence are listed as important factors that affect patterns in the use of nearby public open spaces [24]. In another study, gender, marital status, age, educational status, and income level together with the reflections of these features were proven to affect user profile at city parks [25]. Several personal factors, including having young children, old age, and poor health, have been proven to be negatively associated with open space use [26]. Preferences for and perceptions of open spaces are influenced not only by users' social characteristics but also by their recreation activities and specific interests and demands [27]. Existing evidence suggests that users generally prefer proximate, large, and attractive open spaces. Nevertheless, the influence of open space accessibility and scale remains disputable. However, if these disputes could be disregarded and focus is placed on the open space itself, the characteristics that affect the use of open spaces may become clear. What features of open spaces attract users to come and stay in open spaces? Do visitors actually prefer the various facilities inside the space or the attractive environment itself? What can landscape architects do to increase the use of open spaces? These questions still need to be addressed. Existing studies have mostly focused on developed cities in Western countries. The rapid urbanization of China in the past 30 years has led to the construction of many public open spaces in high-density communities, but many of them are not user-friendly and thus poorly used [28]. Although Western researchers have provided abundant articles on the use of open spaces, the city structures and characteristics of Western cities are different from those in China. Considering the enormous cultural differences between Eastern and Western societies, research results and recommendations for Western cities are not automatically valid for cities in China [29,30] The current study regarded the community open space as the specific study subject. On the one hand, community open spaces are usually located inside a residential community, not far from users' origins . Considering that several studies pointed out that distance to a green space is not a limiting factor for most residents [18,19], we focused on other important influencing factors, such as the attractiveness of a space. On the other hand, community open spaces are often small in scale [18,31]. Controlling the influence of the size of the space is possible to some extent. Therefore, we focused on the characteristics of an open space itself. To date, the factors that affect the use patterns of community open spaces are still unclear. Thus the goals of the current study are twofold: to analyze the specific characteristics of community open spaces that are associated with the use of open spaces in a typical community in China and to explore landscape design strategies through which community open space utilization can be strengthened. --- Data Acquisition This empirical study used the Shenzhen Overseas Chinese Town as a case study. We systematically investigated the large communities in the Nanshan District of Shenzhen, which is a pioneering area in China, to look for a mature community with multiple open spaces, good accessibility, and environmental diversity characteristics as a case for our research. The OCT community is located in Nanshan District, Shenzhen City . It is a typical, high-density, mixed community. The entire community is separated from other communities by urban streets in the north, east, and south and by a high wall with an urban village in the west. High-class residential groups are located in the northwest and the southeast, middle-class and low-class residential groups are located in the south and the southwest, and collective dormitory groups exist in the middle. The open space in this community is composed of several community parks, a hilly country park, several lakes, several pedestrian-only commercial zones, and other types of greenery. The distribution of green space in this community is relatively reasonable, and its landscape is ranked as first class in Shenzhen City. The access of residents to open spaces is very high, given that 94.0% of all the residents live within 300 m from an open space [32]. Therefore, this community with high-quality, high-accessibility open spaces is an appropriate case to study the characteristics that influence the use of open spaces. Before the specific data acquisition process, interviews were conducted with staff members of the OCT administration, the security department, and security guards to obtain a general overview of the community, the management and maintenance of the open spaces, and the public security situation. Through the interviews, we found that the open spaces in the OCT community were planned and developed with high-level standards and are maintained by the OCT Enterprise itself. Therefore, it has high utilization and good public security. The OCT community was developed and constructed in 1985. It was built and is administrated by a state-owned enterprise, the OCT Group, and has excellent internal facilities and a carefully planned environment [33]. Over the years, the OCT community has become a modernized seashore community that integrates tourism, housing, commercial offices, creative cultural industries, and other functions. According to the staff of the OCT administration, the community's residential population is 37,700, and the total employed population is approximately 21,800. The community is divided into two parts by Shennan Avenue. The southern part is mainly composed of tourist venues, including Window of the World Theme Park, Folk Culture Village, Splendid China Park, and OCT Happy Bay Entertainment Plaza. The northern part is mainly composed of residential and commercial venues. The northern part of the community, with a total land area of 306.2 ha, was selected as study area in this study. To establish a theoretical model, three instances of data collection were implemented. First, all the open spaces were identified and then divided into independent units. Second, an environmental scan was conducted to obtain the environmental characteristics of each open space unit. Finally, a systematic four-day observation was carried out to determine the number of users and their activity engagements in each open space unit. Before the specific data acquisition process, interviews were conducted with staff members of the OCT administration, the security department, and security guards to obtain a general overview of the community, the management and maintenance of the open spaces, and the public security situation. Through the interviews, we found that the open spaces in the OCT community were planned and developed with high-level standards and are maintained by the OCT Enterprise itself. Therefore, it has high utilization and good public security. The OCT community was developed and constructed in 1985. It was built and is administrated by a state-owned enterprise, the OCT Group, and has excellent internal facilities and a carefully planned environment [33]. Over the years, the OCT community has become a modernized seashore community that integrates tourism, housing, commercial offices, creative cultural industries, and other functions. According to the staff of the OCT administration, the community's residential population is 37,700, and the total employed population is approximately 21,800. The community is divided into two parts by Shennan Avenue. The southern part is mainly composed of tourist venues, including Window of the World Theme Park, Folk Culture Village, Splendid China Park, and OCT Happy Bay Entertainment Plaza. The northern part is mainly composed of residential and commercial venues. The northern part of the community, with a total land area of 306.2 ha, was selected as study area in this study. To establish a theoretical model, three instances of data collection were implemented. First, all the open spaces were identified and then divided into independent units. Second, an environmental scan was conducted to obtain the environmental characteristics of each open space unit. Finally, a systematic four-day observation was carried out to determine the number of users and their activity engagements in each open space unit. --- Open Space Recognition and Unit Division In 1906, The Open Space Act of Britain provided the first clear definition of open space as follows: "open space" pertains to "any land, whether enclosed or not, on which there are no buildings or of which not more than one-twentieth part is covered with buildings, and the whole or the remainder of which is laid out as a garden or is used for purposes of recreation, or lies waste and unoccupied" [34]. Open space has recently been defined as "managed space, typically green and available and open to all, even if temporally controlled" [35]. In this study, open spaces refer to areas left open for public use in urban communities, whether green or not, such as a plaza, park, playground, or courtyard. The open spaces located in high-class residential communities and villas are not included in this research because they are not open to the general public. In addition, streets, water surfaces, inaccessible hilly areas, and courtyards inside building blocks are excluded. Thus, in this study, open spaces mainly include community parks, outdoor playgrounds, public-access courtyards, water fronts, and small squares around commercial and public buildings. All open spaces in the OCT community were surveyed, particularly those where visitors can enter and stay to a certain extent. To explore the environmental elements that affect the use of open spaces and control the influence of open space size, large-scale open spaces were divided into space units based on spatial configuration and environmental characteristics. This process was also conducted to maintain the basic conformity or continuity of landscape elements inside the units and simultaneously achieve a large difference or obvious spatial separation between adjacent open space units. Moreover, to guarantee the size comparability of the basic space units, we ensured that the area of each unit is not less than 100 m 2 and not greater than 20,000 m 2 . The original largest open space in the community is Yanhanshan Hilly Country Park with long wandering walking paths and large woodlands; it was difficult to divide into units. Thus, several user-gathering squares and rest areas were recognized as space units. According to these principles, all open spaces in the research area were divided into 112 space units with an average area of approximately 2130 m 2 . The composition and functions of each unit differed. Among the 112 space units, 68 were individual land units and 44 belonged to a continuous network of open space or large public parks. Their accessibility had no significant differences [32]; most units were located inside residential or commercial areas, whereas others were located among residential areas. However, the use of the units differed significantly. Several units were always crowded, whereas many others were seldom visited. --- Environmental Scanning Full investigation and documentation of the environmental characteristics of all the space units were conducted. Existing evidence suggests that the characteristics that influence the use of urban public open spaces include the configuration of a place , facilities , and aesthetic features [9,13,36]. Referring to previous studies, the audit content in the current study mainly included three aspects with 17 characteristic variables. For spatial configuration, the total site area , accessible lawn area , woodland area , footpath length , and hard pavement were measured. The site facilities comprised eight types, namely, fitness facilities , commercial facility sites , total number of regular seats , auxiliary seats , rain and sun-shading devices , bicycle parking facilities , trash cans , and lighting facilities . The landscape setting surveys included the number of water landscapes , biological habitat types , sculptures , and motor vehicle parking volume . Among these variables, S1-S5, F1-F3, and L1-L4 are continuous variables, and F4-F8 are dummy variables. --- Observations of Public Open Space Users Site investigation of the use of all community open space units was conducted in four days in November 2014 . During the site investigation, the temperatures fell between 20 ˝C and 30 ˝C, and the weather was cloudy or sunny, which is suitable for outdoor activities. A total of 17 trained researchers observed the visitors in the 112 open space units on two weekdays and on two weekends . The observation time in each day was from 9 am to 10 pm, and documentation was performed every hour. Each observer was assigned to a residential group, a commercial area, or a park to systematically observe and record the number of visitors to these space units. Each space or space unit was subjected to a 6 min continuous visual scan every hour. Thus, we recorded a sufficient number of visitors, including those who stayed for a long time and those who only stayed for a short period or merely passed by. All visitors to a particular space observed within the 6 min visual scan were recorded as point data on a detailed map of the site. A total of 35,090 headcounts participating in various outdoor activities were recorded. The average activity population number in each open space unit was 313. During the day , 27,854 headcounts were observed, and during the evening , 7235 headcounts were observed. In total, 14,854 headcounts were observed during weekdays, and 20,236 headcounts were observed during weekends. The number of headcounts utilized in the subsequent text analysis comprises the summary of headcounts observed during the four-day site investigation. The data confirmed that the open spaces had the most number of visitors from 3 pm to 5 pm; the smallest number was recorded during lunch time and dinner time. In the other time periods within a day, the stream of visitors was relatively stable. After 9 pm, a rapid decrease in the number of visitors was observed . a day, the stream of visitors was relatively stable. After 9 pm, a rapid decrease in the number of visitors was observed . --- Model Building and Results Analysis --- Regression Models In the environment scanning process, 17 variables representing the characteristics of each space unit were scanned. These variables are potential factors that could affect the attractiveness of open spaces and physical activities. However, the initial data analysis showed that several variables should be excluded from the regression model. If the total site area is introduced to, it would result in a serious multicollinearity problem with several other variables ; hence, the model could fail. In addition, the information of S1 could be represented by a summary of S2, S3, and S4. Similarly, the variable hard pavement contains many trails and has a serious multicollinearity problem with other variables ; therefore, it was excluded from the model. For auxiliary seats , which is a dummy variable, the survey indicated that most of the space units provide a certain number of auxiliary seats, such as stairs, lotus pond edges, stones, stone pillars, and sculptures. Thus, the value of F4 in most space units was the same, so we did not introduce F4. For rain and sun-shading devices and bicycle parking facilities , only several space units provide such facilities. We observed that almost nobody uses them, so we excluded F5 and F6. For trash cans and lighting facilities , the survey indicated that almost all open spaces provide a sufficient number of these two types of facilities, so they were also excluded. After the initial selection of influencing factors, seven variables of the observed space units were excluded. The remaining 10 variables were introduced to the regression model, as shown in Equation : P i =C+αS i +βF i +γL i +ε i where P i represents the total activity in open space unit i; C is a constant term; εi is the disturbance; and S i , F i , and L i represent the spatial composition, facilities, and environmental elements of the open space units, respectively. The basic information of the 10 introduced independent variables and the dependent variable is shown in Table 2. The model illustrates the influencing factors of the total activity of open space unit i. Three regressions were completed, namely, total, weekdays, and weekends, because the activity patterns during weekdays and weekends are somewhat different. First, single-factor models were established to fit the three variable groups . Second, a multi-factor model was established to estimate the comprehensive influence of these variables . Third, weekday and weekend multi-factor models were established to test the different influences of the variables during weekdays and weekends. The regression results are shown in Table 3. To --- Model Building and Results Analysis --- Regression Models In the environment scanning process, 17 variables representing the characteristics of each space unit were scanned. These variables are potential factors that could affect the attractiveness of open spaces and physical activities. However, the initial data analysis showed that several variables should be excluded from the regression model. If the total site area is introduced to, it would result in a serious multicollinearity problem with several other variables ; hence, the model could fail. In addition, the information of S1 could be represented by a summary of S2, S3, and S4. Similarly, the variable hard pavement contains many trails and has a serious multicollinearity problem with other variables ; therefore, it was excluded from the model. For auxiliary seats , which is a dummy variable, the survey indicated that most of the space units provide a certain number of auxiliary seats, such as stairs, lotus pond edges, stones, stone pillars, and sculptures. Thus, the value of F4 in most space units was the same, so we did not introduce F4. For rain and sun-shading devices and bicycle parking facilities , only several space units provide such facilities. We observed that almost nobody uses them, so we excluded F5 and F6. For trash cans and lighting facilities , the survey indicated that almost all open spaces provide a sufficient number of these two types of facilities, so they were also excluded. After the initial selection of influencing factors, seven variables of the observed space units were excluded. The remaining 10 variables were introduced to the regression model, as shown in Equation : P i " C `αS i `βF i `γL i `εi where P i represents the total activity in open space unit i; C is a constant term; ε i is the disturbance; and S i , F i , and L i represent the spatial composition, facilities, and environmental elements of the open space units, respectively. The basic information of the 10 introduced independent variables and the dependent variable is shown in Table 2. The model illustrates the influencing factors of the total activity of open space unit i. Three regressions were completed, namely, total, weekdays, and weekends, because the activity patterns during weekdays and weekends are somewhat different. First, single-factor models were established to fit the three variable groups . Second, a multi-factor model was established to estimate the comprehensive influence of these variables . Third, weekday and weekend multi-factor models were established to test the different influences of the variables during weekdays and weekends. The regression results are shown in Table 3. To investigate the influence of existing facilities and environmental elements on the use of open spaces, we established a dummy variable model , in which all the variables of facilities and environmental elements are transformed into dummy variables. If a certain type of facility or environmental element exists in one open space unit, then the value of this variable is 1; otherwise, the value is 0. The regression results are presented in Table 4. --- Results In the regression models, the correlation among the independent variables was less than 0.5, and VIF was less than 2. No evident collinearity problems were observed [37,38]. The fitting result of the single-factor model revealed that two variables of each type significantly affect the use frequency of open spaces . The symbols and significance of the variables of the estimated coefficient were consistent with those in the multi-factor model. This consistency further verifies the stability of the models. Meanwhile, the attractiveness-promotion effect of the reasonable configuration of spatial resources was significantly higher than that of site facilities and environmental elements. The goodness-of-fit value of the former was 0.218, and the separate explanatory powers of the latter were less than 0.1. The goodness-of-fit values of the multi-factor model and the dummy variable model were 0.388 and 0.389, respectively . --- Influence of Spatial Configuration Space is the carrier of public activities. Among the different types of spaces, paved footpath , which had a high utilization rate, was the activity space of residents and the connection channel of different spaces. The regression results indicated that when the length of footpath increases by 10 m, the number of visitors in this area is expected to increase by seven. The influences of weekdays and weekends slightly differed . The accessible lawn area , which is important in attracting visitors, provided an open view and great accessibility. When the lawn area increases by 100 m 2 , the number of visitors in this area is expected to increase by nine and four during weekdays and weekends, respectively. The density distribution of total visitor flow during weekdays and weekends showed that accessible lawn and hard pavement are characterized by a large flow density . These spaces are also considered the main public areas for visitors to play and stay in. Contrary to general understanding, our findings indicate that in the community, the woodland is not significantly associated with the number of users. The model revealed its weak exclusion effect on open space users. The investigation showed that woodlands occupied much space, and most woodland open spaces lack a carefully landscaped design. In several observed space units, woodlands occupied over 80% of the total area, leaving little active space for users. Dense woods also cause safety problems at night. Several crimes have been reported in these densely covered woodlands according to our interviews. In the Ecological Square, the overall visitor flow was large, but the dense woodland areas were inaccessible . The units with dense vegetation were equipped with several walkways only; the lack of activity and places to stay leads to a low visitor flow and low activity density. --- Influence of Facilities Site facilities significantly influenced the improvement of site attractiveness and the use of the open spaces. Various commercial service facilities were found to be important in attracting visitors to the open spaces. The regression model showed that the addition of one commercial facility is expected to attract two and eight additional visitors to play in an area on weekdays and weekends, respectively. This trend indicates that commercial consumption occurs more remarkably on --- Influence of Facilities Site facilities significantly influenced the improvement of site attractiveness and the use of the open spaces. Various commercial service facilities were found to be important in attracting visitors to the open spaces. The regression model showed that the addition of one commercial facility is expected to attract two and eight additional visitors to play in an area on weekdays and weekends, --- Influence of Facilities Site facilities significantly influenced the improvement of site attractiveness and the use of the open spaces. Various commercial service facilities were found to be important in attracting visitors to the open spaces. The regression model showed that the addition of one commercial facility is expected to attract two and eight additional visitors to play in an area on weekdays and weekends, respectively. This trend indicates that commercial consumption occurs more remarkably on weekends than on weekdays. Likewise, visitors stay longer in commercial facilities on weekends than on weekdays. For example, our field investigation showed that the total activity density during weekdays and weekends surrounding the commercial facilities of the OCT Wal-Mart was five times higher than that of the surrounding areas. As the distance from Wal-Mart increases, the activity density decreases sharply. The number of benches in an area also influenced the decision of visitors to stay. Each additional bench is expected to attract five additional visitors to play in that area. This effect was almost similar during weekdays and weekends. Considering that weekends had more visitors than weekdays, we estimate that benches are probably intensely used and that additional seats should be provided. The field observations during the weekend showed that even in the space units with the most number of benches, all the benches are almost always occupied. Several older users have to sit on auxiliary seats because all benches are full. The strong guiding effect of benches on attracting visitors corresponds to their scarcity. If additional benches are added to the right places, then visitors would be encouraged to stay, which would enhance the attractiveness of the space for community residents, urge residents to participate in leisure, exchange, and other community activities, and strengthen community cohesion [39]. In contrast to the general concepts regarding the role of fitness facilities in urban open spaces, our actual results showed that in community open spaces, the availability of fitness facilities does not certainly attract more visitors. Not all facilities are fully used; regardless of the number of established fitness facilities in different space units, the number of users is almost the same. Increasing the number of fitness facilities does not attract more visitors to the space. In model 3, whether the supply of fitness facilities in the observed space unit significantly influences visitor flow is discussed. during weekdays and weekends surrounding the commercial facilities of the OCT Wal-Mart was five times higher than that of the surrounding areas. As the distance from Wal-Mart increases, the activity density decreases sharply. The number of benches in an area also influenced the decision of visitors to stay. Each additional bench is expected to attract five additional visitors to play in that area. This effect was almost similar during weekdays and weekends. Considering that weekends had more visitors than weekdays, we estimate that benches are probably intensely used and that additional seats should be provided. The field observations during the weekend showed that even in the space units with the most number of benches, all the benches are almost always occupied. Several older users have to sit on auxiliary seats because all benches are full. The strong guiding effect of benches on attracting visitors corresponds to their scarcity. If additional benches are added to the right places, then visitors would be encouraged to stay, which would enhance the attractiveness of the space for community residents, urge residents to participate in leisure, exchange, and other community activities, and strengthen community cohesion [39]. In contrast to the general concepts regarding the role of fitness facilities in urban open spaces, our actual results showed that in community open spaces, the availability of fitness facilities does not certainly attract more visitors. Not all facilities are fully used; regardless of the number of established fitness facilities in different space units, the number of users is almost the same. Increasing the number of fitness facilities does not attract more visitors to the space. In model 3, whether the supply of fitness facilities in the observed space unit significantly influences visitor flow is discussed. --- Influence of Ground Environment Elements The number of water landscapes remarkably affected the number of visitors. Each increase in water landscapes is expected to attract 73 and 88 visitors to stay on weekdays and weekends, respectively. This trend reflects the strong preference of visitors for water landscapes. This preference is also a driving force for real estate businesses to construct community water landscapes. Water --- Influence of Ground Environment Elements The number of water landscapes remarkably affected the number of visitors. Each increase in water landscapes is expected to attract 73 and 88 visitors to stay on weekdays and weekends, respectively. This trend reflects the strong preference of visitors for water landscapes. This preference is also a driving force for real estate businesses to construct community water landscapes. Water landscapes in OCT mainly include lakes, pools, fountains, and streams. The observations revealed a large visitor flow in almost all the surrounding water landscapes. The diverse activities include fishing, swimming, photographing, and sightseeing. Contrary to general understanding, our observation and regression indicate that there is no significant relationship between the number of sculptures and the activity density. The sculptures were set as accessories of the place, usually occupied the limited activity space, and were not integrated into users' activities. Tourists may come to have a glance at the sculptures or take photographs, but they leave immediately afterward. There are very few local residents opt to stay near the sculptures. No evident increase in visitor flow was observed around the sculptures. Therefore, placing more sculptures in the community open space will not attract more users. Animal species is usually a factor that attracts users, but the regression model revealed great exclusion between animals and crowd activities. In this observation, the term "animals" refers to wild animals instead of pets; these animals mainly include birds and mammals that live in quiet habitats. The number of activities in an open space also presented exclusion to the number of parking spaces . On the one hand, parking occupied the outdoor activity fields; on the other hand, vehicles were not attractive to visitors. The regression result was consistent with our expected finding, but it was not significant. --- Comparison of Regression Models Comparison of the dummy-variable model , the multifactor model , and the single-factor model revealed that the effects of most variables remained stable; this finding indicates model stability. Several differences were also found in the models. First, in the comparison of models 2 and 3, although the influence of fitness facilities was still not significant, the t value increased from 0.16 to 1.1 in model 3. Therefore, the installation of fitness facilities is highly essential, but the number of facilities inside a space unit is unimportant. Thus, the spatial layout of fitness facilities should be relatively dispersed instead of concentrated in community open spaces. Second, the significance of commercial facilities improved significantly in model 3. Therefore, the number of commercial facilities is relevant, but the presence or absence of commercial facilities is more important than the number of commercial facilities. As such, the layout of commercial facilities should be a combination of concentration and dispersion. This finding implies that large commercial facilities and numerous dispersed small commercial locations in the community are beneficial. Third, the presence or absence of benches showed insignificant effects , but numerous benches were more attractive to visitors than a few benches. Hence, benches or seats have a large demand. Increasing this kind of facility would attract more visitors, whereas individual facilities provide a limited influence. The layout of seats should be a combination of concentration and dispersion, similar to that of commercial facilities. In particular, service object, entrance distribution, and leisure path should be integrated into the pre-phase analysis to predict the spatial distribution of visitor flow and activities and determine the areas that should be equipped with seats. Finally, the exclusion of visitors from a parking lot was evident; this exclusion was slightly related to the scale and number of parking spaces. Having a parking space in an open space is not advisable; relevant measures should be implemented to prohibit vehicles from entering open spaces. If this situation cannot be avoided, then the best way is to place the parking lot at a relatively concentrated place, such as important individual entrances, to reduce its negative effects on the quality and attractiveness of open spaces. --- Discussion --- Discussion on the Influencing Factors of Community Open Spaces Some community open spaces are always crowded, whereas others have a few visitors. Traditional research on this topic in relation to accessibility has shown remarkable progress. The present study confirms that the landscape characteristics of open spaces are also important influencing factors. As concluded by Handy and Niemeier's study on accessibility, the quality and use of open spaces should be given special attention [40]. We found significant correlations between the use of green spaces and the characteristics of open spaces in our study. Scholars have been presenting questions regarding the attractiveness of open spaces since the last century. For instance, do visitors always prefer large-scale open spaces? Holman et al. argued that proximity and accessibility are important, but they alone do not explain patterns of use; many other factors, including environmental aesthetics, animals, maintenance, space scale, and footpath length, influence the use of open spaces [21]. Talbot and Kaplan pointed out that real and perceived scales are not directly related to users' preferences; conversely, site adhesion, opportunities, and landscape are essential [22]. Ye et al. indicated that in terms of daily leisure activities in urban open spaces, the quality of a "micro environment " is more beneficial than that of the overall environment [41]. Various peripheral factors, including the attributes of users, social psychology environment, surrounding facilities, and supply-demand relationship, may also affect the use of open spaces [17]. The current study further confirms that accessible lawn area, footpath length, benches, commercial facilities, and other factors are important characteristics that affect the use of open spaces. Many factors influence the use of open spaces, and these mechanisms remain unclear. Selecting among different open spaces with different characteristics is important because different open spaces complement one another [26]. When asked about the factors that they liked about the open spaces [42] or factors that influence their use of these spaces for physical activity [21] through questionnaires, the respondents mentioned trees, water features, bird life, and activity space. This scenario is consistent with our correlation analysis, except for trees and bird life, and the possible reasons are discussed in Section 3.2. Kaczynski et al. found that park facilities, including paved trails, water area, and playground, are more important than park amenities, such as drinking fountain, picnic area, and restroom for physical activity [43]. Our study also found a significant relationship between the use of open spaces and walking path, water landscapes, and accessible lawn; no significant relationship was found between the use of open spaces and open space amenities, such as excess decorations, heavy woodlands, sculptures, and fitness facilities. Schipperijn et al. found positive associations between physical activity and size, walking or cycling routes, wooded areas, water features, lights, pleasant views, bike rack, and parking lot of urban green space [13]. These results are mostly consistent with those of our study, except for some inconsistencies. Considering that our study focused on community open spaces that are small in scale in high-density residential areas in China, the characteristics that affect use patterns differ in open spaces of different types with distinctive characteristics. Numerous characteristics of open spaces may influence their use. Thus, in urban and landscape design, the numerous features of open spaces should be considered. Many Western countries have formulated a number of specific policies and guidelines that oversee open space planning and landscape design. However, in Eastern societies such as China, detailed policies and guidelines on the supply of open spaces are still absent, with only a few simple standards. For example, according to the national code for the classification of urban land use and planning standards, the planned urban green space per capita should not be less than 10.0 m 2 , of which the green land area per capita should not be less than 8.0 m 2 . In this code, the supply standard for required facilities and landscape amenities is absent [44]. Thus, many open spaces are characterized by a single type, a monotonous function, scarce facilities, and a low utilization rate. Meanwhile, the construction of city parks remarkably considers the quantity requirement of greening coverage. For example, the Shenzhen Urban Planning Standards and Guidelines [45] recommend that the greening rate of a park should not be less than 65%, and the width of the green belt in the community park should not be less than 8 m. Therefore, the construction of parks remarkably considers greening; these parks unilaterally emphasize landscaping, gardening, and ornamental features with lush trees and rich vegetation. As a result, the activity space for users is limited or absent. This study suggests that the most attractive factors of open spaces to users are accessible lawn area, footpath length, and site facilities, not woodland, flowers, plants, or greening coverage. These attractive factors are directly related to the demand of users. In terms of the supply of open spaces, the Hong Kong Planning Standards and Guidelines provide the per capita indicator of open spaces and recommend a passive open space-to-active open space ratio of not less than 3:2 [46]. This recommendation ensures the demand of sufficient land for all kinds of dynamic activities and serves as a helpful reference. --- Implications for Landscape Architecture Well-designed open spaces are an important community asset because they serve as a venue for outdoor physical activities and social interaction. In open spaces in high-density residential communities, attracting more users is possible by providing more accessible mown lawns, developing more well-maintained footpaths, providing more seats and commercial facilities, and creating water landscapes. Encouraging the increased use of public spaces would have a positive effect on attracting users to public spaces, thereby increasing the security of public open spaces [47]. Redesigning existing spaces is also important for the promotion of public open spaces in communities [17]. Redesigning open spaces by adding accessible fields for users, including accessible lawns, hard-covered fields, and footpaths for multiple users-walkers, joggers, cyclists, and passive recreational users -is highly recommended. In this manner, an important community resource is maximized. Open spaces are of different types. For example, in the UK, open spaces are classified into nine categories, namely, parks and ornamental gardens, natural and semi-natural green spaces, green corridors, outdoor sports facilities, amenity green spaces, provision for children and young people, allotments, cemeteries, and civic spaces [48]. The provision and distribution of different categories depend on the function of the spaces. The primary function of community open spaces is to provide predominantly passive recreation and sitting areas and children's playgrounds to serve local residents close to home; these spaces are passive in nature and positive in providing a venue for resting [46]. The provision of community open spaces is based on the number of residents. Thus, community open spaces are usually much smaller than urban and regional open spaces. Small urban parks should be designed with natural components, shielded from disturbing surroundings, and furnished with some seating to promote opportunities for restorative experiences and to function as social meeting places [49]. This study also showed that open space units providing facilities, activity spaces, and amenity landscape features are more welcoming to visitors. Therefore, the landscape design of open spaces in high-density residential communities should be concordant with the function of such spaces, that is, providing active spaces for local residents. Excessive ornamental vegetation, sculptures, and dense woodlands are not recommended for community open spaces. In some community parks in China, the greening coverage occupies more than 65% of the public space [50], leaving little actual open space for visitors. Providing sufficient regular and various kinds of auxiliary seats is also important [51]. The total length density of seats in community open spaces in Hong Kong is about 10 m/100 m 2 , but in the OCT community, the total length density of seats is less than 3 m/100 m 2 . Fitness facilities are also essential for communities, including fitness facilities for the elderly, children's entertainment facilities, table tennis tables, badminton fields, and the like. --- Conclusions and Future Studies This study confirms that large accessible lawns, well-maintained footpaths, seats, commercial facilities, and water landscapes are important characteristics that could increase the use of community open spaces. It argues that increasing the green vegetation and the number of sculptures or landscape accessories in community open spaces exerts limited effects on residents' outdoor activities. Too much woodland and excessively large green coverage do not significantly influence the residents' outdoor activities in community open spaces. Sculptures or sketches serve as decorations and do not encourage citizens to stay. Thus, to increase the use of community open spaces, landscape designers should focus on creating user-oriented spaces with facilities that encourage active use instead of spending excessive money on ornamental vegetation and accessories. This study focused on the characteristics and utilization of community open spaces; the characteristics of users were not recorded. The characteristics of users, including age, gender, and group, may also affect the use of community open spaces. In addition, the characteristics of an open space that attract one population group may also affect its use by other groups. Given the limitation of the observation method, we could only record the headcounts of visitors who appeared in a certain open space. We could not distinguish the headcounts of those who stayed for a long period and the headcounts of those who only stayed for several minutes or were merely passing by. These differences may also be associated with the landscape characteristics of an open space. Studies could be conducted after systematic questionnaires have been distributed and investigations have been conducted. With this recommendation in mind, future studies can explore other types of public open spaces after comparing the characteristics of different open spaces and utilization across different groups. Comparative studies could also be conducted after performing additional case studies. --- Author Contributions: Yiyong Chen and Xiaohuan Xie conceived the research, implemented the field investigation, and wrote the paper; Tao Liu analyzed the data; Barbara Goličnik Marušić gave professional suggestions to the paper and polished the manuscript. ---

A well-designed open space that encourages outdoor activity and social communication is a community asset that could potentially contribute to the health of local residents and social harmony of the community. Numerous factors may influence the use of each single space and may result in a variety of visitors. Compared with previous studies that focused on accessibility, this study highlights the relationship between the utilization and characteristics of community open spaces in China. The Overseas Chinese Town community in Shenzhen is regarded as an example. The association between the number of visitors and space characteristics is examined with multivariate regression models. Results show that large areas with accessible lawns, well-maintained footpaths, seats, commercial facilities, and water landscapes are important characteristics that could increase the use of community open spaces. However, adding green vegetation, sculptures, and landscape accessories in open spaces has limited effects on increasing the outdoor activities of residents. Thus, to increase the use of community open spaces, landscape designers should focus more on creating user-oriented spaces with facilities that encourage active use than on improving ornamental vegetation and accessories.

water. The research involved participant observation of different areas in both neighbourhoods, extensive and repeated interviews with a cross-section of people from both neighbourhoods, and interviews with municipal staff, non-governmental organizations, and community groups working on sanitation and water. The production of local conceptions and practices of entitlement in the two neighbourhoods plays out in quite different ways. It is closely related to a sense of moral economy in relation to the everyday state, private sanitation providers, and the exclusion of particular groups. Our argument is that sites of entitlement emerge in large part through a series of contextually specific moral economies and by linking a wide variety of everyday practices across different spaces. These particular issues are those that matter most in the informal settlements where we conducted the research. Our larger point is, however, precisely to underline this spatial contingency as a more active, vital basis for thinking about universalist claims to the right to sanitation and water, rather than as an afterthought to those claims. --- II. COMPOSING EVERYDAY ENTITLEMENTS We begin by asking a seemingly straightforward question: how are people's sanitation and water entitlements produced? To consider this question, we must first clarify what we mean by entitlement. The language of entitlement is widely used in two ways: to invoke universal normative rights, or in relation to specific rights attached to certain groups. It is important to distinguish here between "rights" and "entitlement". While they are interrelated, we take entitlements to be what people claim as a product of the rights available to them. Importantly, while rights generally take the form of legally binding statements, entitlements are produced through social relations and based on people's experience and perceptions. Amartya Sen's influential writings are useful here. For Sen, a person's entitlements are the totality of thingsresources, commodities, infrastructures, services, and so onthat she can command based on all the rights applicable to her. Sen argued that while it is usual to characterize rights as a relation between particular actors, such as people and the state, entitlement emerges by virtue of provision of rights. If Sen tends to conceptualize the basis of these claimsthat is, the source of their legitimacy in narrowly legalistic terms, some of his work on gender and poverty does attempt to account for the role of social conventions that govern the basis of claims beyond courts. A number of scholars have sought to expand the concept of entitlement to include the things an individual can command through claims with a basis in other sources of legitimacy. Charles Gore, for instance, has argued that entitlements are shaped not only by legal rules but also by socially shared rules. Naila Kabeer has called for a reformulation of Sen's concept of entitlement to consider the different basis of claims on resources that prevail in a society, and that are embedded within social relations and practices rather than legally binding contracts. She argued that entitlements are generated through rules, norms and practices characterizing different institutional arenassuch as market-based exchange, state provision, community and kinshipthat determine who gets what and on what terms. For Gore, these socially shared rules are best understood through the concept of "moral economy". There is a long history to this idea, of course, and its use in the social sciences owes a great deal to James Scott's 1976 book The Moral Economy of the Peasant, which emphasized the role of shared traditional methods and models put to work by rural communities in Southeast Asia to cope with change. Broadly speaking, moral economies are social regulations related to but not determined by legal rules, which help set the conditions through which entitlements are conceived, claimed and contested. Moreover, they vary within and between different groups and spaces, and over time. As Gore has argued, although the concept of a moral economy refers to socially shared rules, based, for instance, on moral obligations, patron-client networks and so on, senses of entitlement are not necessarily shared by all and need to be seen as an outcome of struggles and active processes of negotiation. We conceptualize entitlement as a claim that relates a person/persons to services/infrastructures in a way that seeks to fulfil their access to what they need and value, and which is produced through diverse practices, social relations, and moral economies. While we value the emphasis by scholars such as Scott, Gore and Kabeer on nonlegalistic claims to entitlement through moral economies, we do not differentiate between socially shared rules articulated through a moral economy and legal rules, since the latter are often an outcome of a particular moral economy and can also feed into moral economies. Rather, we emphasize the importance of unpacking the moral economies, social relations and practices through which individuals or groups both shape sites of entitlement and are positioned as deserving or undeserving of entitlement. As studies on informality, urban politics and citizenship reveal, it is through individual and collective practices of negotiating urban space and resources that marginalized groups attempt to expand their claims. While active collective mobilization is important in these accounts, this body of literature suggests that entitlements are produced through different modalities of urban struggle and people's everyday experiences and practices. We consider it vital that research on entitlements focus on the production of entitlements from the perspective of residents themselvesin this case in relation to informal settlementsand from the perspective of their moral economies. Quite apart from the centrality of moral economies to people's sense of entitlement, there are two other important reasons for focusing on sites of entitlement. First, the entitlement of residents is too often understood through a state-centric lens. This leads to an overly legalistic, formal view of entitlements. We are not arguing that the state is anything but critical to how residents understand entitlement, nor are we suggesting that the state exists "above" the daily life of the neighbourhoods. The state, in the form of local officials, ward councils, specific projects, or particular politicians, is present in specific ways in the daily life of the neighbourhoods and in shaping a sense of entitlement. Instead, we argue that rather than focusing on the state's treatment of particular informal settlements as a basis for understanding entitlement, it is more fruitful to examine the currents of everyday life and why and how the state features in people's understandings of entitlement. As research on the state has shown, this approach typically reveals a highly variegated sense of how the state shapes local perceptions and practices. Ethnographic studies reveal the state not as an entity "up there", above society, but as constituted by a multiplicity of actors, sites and practices that permeate and are reshaped through society and everyday life. Residents of informal settlements encounter and experience the state through the practices of elected representatives, municipal officials, the police and others in a variety of modalities, whether violent, oppressive, beneficial, welfarist or governmental. These relations influence and are influenced by the moral economies of these different actors. While the state-centric focus on entitlement has been enlarged in recent decades to account for the increasing importance of frameworks of institutional pluralism, and especially the state's role in facilitating service delivery through civil society and the private sector, the everyday experience and perceptions of communities around entitlement remains largely neglected. As non-governmental organizations , community-based organizations and the private sector have increasingly become service providers or partners in service provisionincluding in sanitation in Mumbai the shaping of entitlement is more and more understood as a relational process in which the state sets vital parameters. We would not of course deny the vital role these actors play in shaping perceptions of entitlement, but too often the claims and views of established organizationsespecially states, NGOs and social movementsare taken as representative of people's entitlements. As a result, social relations of family, kin, community and locality are often removed from view. Rather than look to mediating organizations as representatives of people's sense of entitlement, a sites of entitlement approach aims to empirically examine the different social relationsand civil society groups and states are usually and inevitably part of thisthat compose conceptions and practices of entitlement. Second, a sites of entitlement approach focuses on how residents compose a sense of entitlement not just across space, but over time. As we shall see, this reveals senses of entitlement in variable amounts of flux and ambivalence. It also reveals how these understandings and the moral economies that help shape them become contested, undermined, or altered through new relations, including changing political relations. Of course, not all practices to improve access to sanitation and water are about entitlements per se. Residents in Rafinagar and Khotwadi do not always have a clearly articulated sense of entitlement that informs their practices. Often, their practices are about fulfilling their needs in conditions of profound uncertainty and inequality. However, when established practices to fulfil needs are under threat, residents express a sense of injustice and, alongside this, a sense of their legitimate claims. --- III. MORAL ECONOMIES, LEGALITY AND THE STATE What, then, is the role of the state in producing sites of entitlements? We begin in Rafinagar, with a key issue that shapes but, as we shall see, does not determine, the nature of understandings of local entitlement: legality and the state. Salimbhai is an elderly shopkeeper who has owned a furniture workshop in Rafinagar since 1985. He explained that demolitions had been common before he came to Rafinagar. Along with some other residents and shopkeepers, he established a committee and succeeded in obtaining the support of a local politician -Javed Khanto prevent further demolitions. Khan was a Member of Legislative Assembly for the constituency between 1985 and 1995. Salimbhai explained that the committee had tried to get Rafinagar declared a notified slum with Khan's support but had been unsuccessful. But Khan had supported the committee in other ways and Rafinagar got its first water connections, streetlights and toilet blocks as a result in the late 1980s, more than a decade after its emergence. Salimbhai argued that if Khan's reputation had not been damaged by involvement in the 1992 Hindu-Muslim riots, he would have been impossible to defeat in the area because of his good work. An informal settlement must be notified as a slum under the Maharashtra Slum Areas Act, 1971, for it to legally qualify for the provision of basic amenities such as water, toilets, streetlights and drainage. Rafinagar's experience shows, however, that despite no legal recognition, the relations with Javed Khan initiated the cultivation of relations with local elected representatives and a kind of extra-legal recognition. Residents were thus able to slowly carve out claims through negotiations in political societya disparate realm largely outside of formal citizenship claims in which community groups make moral claims, supplementing electoral relations, on particular politicians and officials. However, political society has brought mixed results, such as they are, to Rafinagar. Negotiations have led to some successes in gaining infrastructure and services in Rafinagar Part 1, the older and more established area of the settlement, but have been highly uneven. For example, water cannot be legally provided to residents who arrived in the neighbourhood after 1 January 1995, meaning that these residents are doubly excluded through both the legality of the space and the time of inhabitation . Residents estimate that only 1,500 out of a total of 4,000 households in Rafinagar can prove pre-1995 residency. Later we will consider the unsuccessful attempts by residents in Part 2, the area settled after 1996, to negotiate entitlements through political society. The distribution of entitlements has also been significantly shaped by each family's personal circumstances and its networks and negotiations with elected representatives. Consider the experience of Shobha, one of Khotwadi's poorer residents. Even though Khotwadi is a notified informal settlement and Shobha and her neighbours are pre-1995 residents, obtaining a shared water connection as per the policy of the Municipal Corporation of Greater Mumbai has been a fraught process. Encounters with unresponsive municipal officials often force residents, even when they are legally entitled to a shared water connection, to approach their representatives for what can also be a long, frustrating process. Even after Shobha and other households had each paid the municipal corporator Rs. 1,200 for a shared water connection, the infrastructure had yet to materialize. Shobha said she was now worried that if the municipal corporator changed she would have to begin the whole process again, adding: "The government should do something about this. I can't spend much money on this. We have already paid Rs. 1,200 each. How much more will we have to pay who knows? If [the money] is adequate then he will give us a connection. It might work for two years, four years…but a sarkari [government] tap is a sarkari tap, right?". Shobha's narrative of her ongoing attempt to get a water connection reveals that even when legally entitled to certain infrastructure and services, residents are often forced to negotiate in political society to realize these entitlements in practice, an experience that can be a source of immense anxiety and uncertainty. Shobha argued that the state should take responsibility towards residents like herself and provide sarkari taps, i.e. unmetered municipal standposts, in her neighbourhood. Khotwadi has often fared better in water provision than other localities in part because it is predominantly Marathi. Therefore it is favoured by the dominant political party in the city, the Shiv Sena, a nativist and historically anti-Muslim party that has run the BMC since 1995. However, provision has been uneven: the Shiv Sena municipal corporator extended the network of unmetered standposts in some of Khotwadi's predominantly Marathi neighbourhoods by taking a one-time fee from some residents, even as elsewhere in the same locality such standposts were removed by the BMC and replaced by water connections that entail monthly bills. What is striking about Shobha's account is that she expects the state to provide despite these spatiotemporal variations. Here, a moral economy of state provision remains, uncertain, frustrating and stuttering though it may be. Such an expectation of the state is less common in Rafinagar, where for some a sense of entitlement from an often hostile state is an alien notion. Indeed, even through the medium of political society, the opportunities for provision from the state in Rafinagar have dwindled rather than expanded over the years. Many older residents argue that water problems have increased and intensified since Javed Khan's time. Neither elected representatives nor municipal officials have taken serious steps to resolve this water crisis. Instead, the BMC, with the police, carried out raids on water connections in Rafinagar, Shivajinagar and other nearby areas in 2009 in the name of cracking down on "water theft" to deal with Mumbai's "water shortage". Both legal and illegal connections were cut during the raids without adequate alternative provision, destabilizing provision previously gained, and Rafinagar was thrown into an even deeper water crisis than before. This is not a context where many feel that they can reliably make claims on state provision. Nonetheless, many residents have a strong moral economy of state provision of water, not so much for unmetered standposts as Shobha has in Khotwadi, but for water connections entailing monthly bills. The history of state provision and denial thus influences the moral economies of residents as well. The Shiv Sena and the Maharashtra Navnirman Sena have often run campaigns against north Indians and Muslims in the city. This chauvinistic ethnic and religious identity politics has influenced the workings of municipal government and shaped the hostile attitudes of municipal officials towards areas like Rafinagar and Shivajinagar whose residents are predominantly Muslim and sometimes north Indian. Moreover, municipal officials often see these areas as harbouring illegal Bangladeshi immigrants, despite the lack of clear evidence. The claims of residents, in particular certain ethnic and religious groups amongst them, to urban space and resources are increasingly delegitimized. This is particularly visible at moments such as that of "water shortage". Many residents attempted to prove the legality of their connections that had been cut in the raids and thus obtain BMC permission to re-join their taps to the water mains; some succeeded, others did not. Residents also repeatedly directed collective appeals and protests at their elected representatives. This led the BMC to install two temporary water storage tanks near the entrance to Rafinagar. These were to be filled by BMC water tankers free of cost to residents. However, tankers came irregularly, sometimes once a day, sometimes four times. Residents had to frequently approach their elected representatives to ask them to pressure the BMC to send more water tankers. Residents who lived further from the entrance pressured elected representatives to install more temporary tanks near their homes, which eventually led to the installation of five tanks over six months. Sites of entitlement are constantly unfolding across spatial and temporal configurations, and are continually negotiated through political society. Similar stories emerge around sanitation. Most of Rafinagar's residents explained that Bullet Patil, the well-known former municipal corporator, did not respond to sanitation complaints unless they were prepared to repeatedly plead for help at his office. Many residents had given up approaching him altogether and instead pooled money or their own labour to clean their drains or construct makeshift toilets. Given this, it is not surprising that some residents did not even consider the municipal cleaning of drains as a service they were entitled to. Some residents had a clearly defined spatial imaginary of entitlement here, differentiating between the smaller drains in their lanes that they felt responsible for and the larger drains on the main streets that they felt the BMC was responsible for. In contrast, in Khotwadi the cleaning of all drains has long been a relatively regular municipal service, and residents readily described it as an entitlement they expect from the state. However, it is not the case that residents in Rafinagar do not believe in state provision. For example, three attempts had been made over the past few years to demolish a public toilet block in Rafinagar and replace it with a private blockevery time, residents came together to successfully protest and drive away the demolition crew. Residents cited their concern about having to pay more for a private block; some also explained that the three other private blocks in Rafinagar remained closed at night from 12-5 am, and conversion of this block would also mean no night provision. One resident explained that one private block was locked in the middle of the day if there was inadequate water from tankers. Residents do not necessarily feel a sense of entitlement to state-provided toiletssome said they did, others had less of an expectationbut they have a keen sense that state provision is preferable. In other words, while there is a strong moral economy of public sanitation in Rafinagar, as a result of bitter and often violent experience of demolition this does not necessarily translate into a clear sense of entitlement from the state. In sum, the role of the state in shaping sites of entitlement varies by place and issue . If we delve deeper into the everyday practices and perceptions of residents in the neighbourhoods, we see that sites of entitlement are shaped by changing conditions and shifting moral economies of service and infrastructure that extend beyond the state. --- IV. SHAPING ENTITLEMENT BEYOND THE STATE The division between public and private is often blurred. Two of Rafinagar Part 1's three private toilet blocks were made possibledespite its "non-notified" statusthrough the support of elected representatives to the private organizations that built and operate them. Some residents who were earlier part of these private organizations were also proactive. Like most private services, these exceed but are not straightforwardly "separate" from the state. Nonetheless, they are first and foremost private blocks, and residents do not expect them to operate as public blocks. This does not mean that residents do not have compelling notions about what constitutes fair access to privatized sanitation. The sense of entitlement to private blocks is distinct, shaped through individual and collective negotiations around specific concerns such as user charges and water provision. Consider one of the private toilet blocks in Rafinagar Part 1 where the operator attempted to increase the user charge from Rs. 1 to Rs. 2 per use some years ago. Naina, arguing that the toilet operator had increased the charge to increase his income, explained the reaction of resident users: The public created a scene. They went and sat down [to defecate] anywhere, in the maidan [open ground], the garden, on the road, near the clinic…So that he [the toilet block caretaker] will also not be able to sit there, he will also get the stink, no? Tomorrow he can even say it is Rs. 3. Should we drink water from his hands?... Meaning if he keeps increasing the money, should the public keep giving? And at that time if anyone came and said why are you sitting [to defecate] outside, then people just said go and tell him, he has made it Rs. 2. Then the area became dirtier…We told others. This way people stopped going there. Naina explained that ultimately the toilet block operator had to decrease the charge to Rs. 1 again. Her narrative suggests that many residents felt that the increase was unaffordable and unjust, and that their individual actions linked up to register a protest against the increase. More recently, another private toilet block in Rafinagar increased its user charges to Rs. 2. Mumtaz explained what she thought about this: "Rs. 1 is enough to keep it clean. Why do they need to take Rs. 2?...Maybe because there is a water problem…But women complained so they take Rs. 1 from us…Women will not part with more money for the toilet. She will think of what she can get for her children instead." These are not transformative protests, of course, but they are part of a wider set of small acts through which a moral economy of sanitation is defended and a particular sense of entitlementnot to free private toilet blocks, but to reasonable chargesis maintained. Mumtaz's sense of being entitled to access the toilet block for Rs. 1 and her rejection of the operator's attempt to charge Rs. 2 emerged then from the competing priorities that had to be negotiated in everyday life. This led to negotiations with the caretaker, who relented and allowed the women to use the block for Rs. 1. Some refused to pay the cost of using the toilet, or paid less. This was a gendered practice. Many men were observed entering the block after giving the caretaker Rs. 1 or even half a rupee. At times, the caretaker tried to stop them from entering, but often he seemed resigned, muttering once that men just paid what they wanted to. He clearly did not think this was okay but allowed it because he felt helpless to do anything about it. He said the toilet block was difficult to maintain because "here there are more taporis", a term for the figure of the rowdy male who is difficult to control. There was, then, a distinct gendered dimension to the caretaker's attitudes, and a distinct sense of entitlement across different groups of women and men. Other differentials and expectations occur: when women bring their young children to the block with them, the caretaker does not always charge for the child. In the caretaker's moral economy, women's claims to using the block for less than Rs. 2 were seen as more legitimate than men's, but men were more likely to ignore the caretaker's moral economy. Sites of entitlement are, then, relationally formed through interactions among men and women, caretakers, and the nature of the toilet block . Moreover, sites of entitlement change over time as circumstances alter. Take the example of a toilet block built under the Slum Sanitation Programme and run through a community-based organization in Khotwadi. Despite it being community managed, most users saw it as a private toilet. Before our research began, the CBO had closed off all piped water connections to prevent "water wastage" caused, according to the CBO, by women leaving the water tap running too long or not switching it off. Women users were then required to bring their own water from home or buy water from the caretaker: Rs. 1 for half a bucket of water and Rs. 2 for a full bucket. Shobha did not mind that the water had been closed off, and sheas well as other women users -echoed the CBO's narrative of "water wastage". But the caretaker also refused to provide water to clean the toilet if it was "dirty" when Shobha arrived to use it, and she resented using her own water to clean up after others. She added, further, that this response was class-based: "If there is a known person, a well-off person, they give water to them, but not to us. They see who is it and then give accordingly." In this example, Shobha's moral economy of water is shaped by both a local discourse of "women wasting water" and her questioning of the CBO's access practices. She saw it to be the CBO's responsibility to at least provide her with water at no cost if the toilet had been left dirty by a previous user, and she believed that all users were entitled to equal treatment. She raised these issues with the caretaker, to little avail, but not with the CBO itself: "Who would listen to one person?" she asked. In other words, while a moral economy of service provision helps Shobha to shape a sense of entitlement to a certain standard of provision, she had little faith that this could be realized in practice. Partly as a result, from time to time conflicts erupt between residents around who is entitled and who is not, and under what circumstances. --- V. NEGOTIATING EXCLUSIONARY ENTITLEMENTS In a context of inadequate sanitation facilities, conflicts sometimes erupted amongst residents over the use of toilet blocks in both Rafinagar and Khotwadi. Large numbers of users and the lack of water provision at the blocks often led to unclean toilet conditions, with residents often viewing particular groups as responsible for making the toilets dirty. In Khotwadi in particular, this led to discourses and practices through which residents sought to establish their entitlements by delegitimizing the claims of others, especially daytime garment workers from Khotwadi's many textile workshops. Many public toilets in Khotwadi have been appropriated by particular groups of residents who regulate access to them through lock-and-key arrangements. Residents articulated the legitimacy of these appropriations by arguing that their own well-being was threatened. Lakshmi explained: "Illnesses spread through the toilet. Jaundice, diarrhoea and other diseases spread like this. The workers of the garment workshops make the toilets dirty. This is why we have put locks in. Women garment workers throw cloth into the toilet. This is why we do not allow them [to use the toilet]." In this and statements like it, residents demonstrated a moral economy of access that defends "their" resource, and underpins a sense of entitlement that links cleanliness to social identity. In some blocks, appropriations were sometimes buttressed by historical territorial claims. A number of residents explained that the toilets in their chawls had been built 40-50 years earlier by the informal landlords who built the chawl and rented out the units. Many toilet blocks in Khotwadi had thus been originally built for the tenants. However, the blocks had been rebuilt 25-30 years before by the BMC to convert them from dabba sandaas to pour flush toilets. As municipal funds had been used, it was no longer clear whether these historical territorial claims could be sustained. Prabha, for instance, was aware of this history, but considered it the "right" of residents of her chawl and some residents from an adjoining chawl to lock the toilet cubicles to prevent garment workers from using them. After all, they contributed money for the cleaning of the block and also for repairs and maintenance, so why should it not be appropriated more formally? She explained that residents in fact did not pressure the municipality to clean it, for then the toilet would be understood as public, which would create difficulties for the chawl residents: "If the BMC cleans it then everyone can go, so we don't force the BMC. If BMC cleans it then everyone will have a right . If we do it ourselves then it will remain private and personal." Here, then, historical ambivalence and poor municipal performance combine in the production of a particular moral economy that justifies semi-private control over a private-turned-public toilet, and produces a fragmented sense of entitlement that asserts the right of particular groups over other less powerful groups. But this was not always the end of the story. Locks on these appropriated blocks were often broken, whether by garment workers or others, and in some cases residents relented by leaving open one cubicle for public use while the others were kept locked. Even then, locks were sometimes broken, and in one case residents had been forced to abandon the locks altogether. This does not mean that the claims of garment workers to the toilets were recognized by residents. As one garment worker explained, women residents from the area often shouted at them, blaming them for the unclean toilets. She added that one garment worker had even been slapped by a woman and was now scared of even going to the toilet. Although this did not stop her and other garment workers from using the block, verbal abuse and violence were sometimes used to attempt to establish territorial rights to the toilet blocks in the face of inadequate sanitation and the daily experiences this created for different groups. For some, the issue of entitlement was not just one of controlling access to blocks but of charging at all. Shobha again: "We should not be charged. We are from here only, right? You can take money from outsiders, from the garment workers. Not from us. Why should you take from us? It should be a sarkari [public] toilet. [The CBO running the block] is also earning so much… from the garment workers. From one person they take Rs. 2 for a bucket of water. It was Rs. 1, now they take Rs. 2…They take Rs. 10-15 for bathing. So they must be earning so much." Sites of entitlement, then, are produced not just through moral economies rooted in notions of the public and the private, but through ideas about who should and should not gain access to these provisions, and about what constitutes acceptable forms of formal and informal regulation. --- VI. CONCLUSIONS Sites of entitlement are not fixed. They are ongoing processes that are reshaped through changing interactions between residents, between residents and states, between residents and civil society and private actors, and through the experience of living in often uncertain, precarious neighbourhoods. Sites of entitlement emerge not through one-way causal relations triggered by particular events, legal conditions or moral economies, but as contingent assemblages in which these different moments and processes shape one another in fluid and sometimes provisional ways. While we entirely support universalist claims to sanitation and water, our research leads us to argue that the understanding and production of entitlements in practice demand that research, policy and practice pay far more heed to spatial and temporal heterogeneity. Understandings of entitlement, like the provision of services and infrastructure themselves, are not just translated into local contexts; they are actively made and remade through spatial and temporal processes entangling multiple sites and actors. The consequence for global debates on sanitation and water policy and practice is in our view significant: any effort to understand the extent and nature of the challenge and the process of working through solutions must engage with the heterogeneous nature of sites of entitlement not just between cities, but within them too. We are not suggesting that "quick fixes" or engineering-based approaches are no longer relevant, but our evidence shows that the success of any intervention depends on a strong understanding of people's lives, expectations and preferences, andcruciallythe ways in which they vary within and between spaces. To be clear: we do not think it sufficient to continue with a gesture to how "local contexts matter": instead, we argue that the shift in focus from abstract entitlement to sites of entitlement holds the seeds of genuine success in understanding and addressing sanitation inadequacies. We need invocations of legal rights, state provision and community mobilization, but we need equally to recognize that such invocations often do not take us far in understanding the problems or working to solutions. Working in and with sites of entitlement at the neighbourhood level and across multiple actors is in our view the most effective route to lasting sanitation and water improvement, even though it is slow and particularist. We end by highlighting two challenges for future research on sites of entitlement. First, the relation among legality, rights, moral economies, entitlement and everyday life is fundamental to both people's experiences and perceptions of water and sanitation and the possibilities of developing solutions. Working through solutions that assist better sanitation and water in Rafinagar must by necessity be distinct from those in Khotwadi. In Rafinagar, illegality is central, as is a more widespread politics of religion, and as is the politics of state hostility, manifested for instance in the profound challenges with water. In Khotwadi, the state is far more supportive and central to service delivery, including in relation to water, and the context for intervention is as a result radically different. These relations can and will change of course: a key part of the research challenge is to understand how the problems and solutions shift over time as conditions change. Second, our research has demonstrated the need for better comparative understandings of basic urban services in informal settlements, not only between cities but within them. The fact that central aspects of the sanitation and water problems in Rafinagar and Khotwadi vary so significantly, including the production, logics, actors and form of the problems and potential solutions, reflects the value of comparative research. Comparative research at the neighbourhood level is particularly valuable here because neighbourhoods generally allow for a manageable unit of analysis between the multiple complexities of everyday life and the city more generally. Research, policy and practice tend to assume that spatial differences will be more pronounced between cities, and especially between countries, than within cities. The extent of the differences between Rafinagar and Khotwadi leads us to insert a note of caution here. Spatial diversity may be as strong within cities as it is between them, and we may find that particular neighbourhoods across different cities in different countries provide challenges in sanitation and water that have more in common with one another than with other neighbourhoods in the same city. --- BIOGRAPHY Colin McFarlane is an urban geographer whose work focuses on the experience and politics of informal neighbourhoods. This has involved research into the relations among informality, infrastructure and knowledge in urban India and elsewhere. A key part of this has been a focus on the experience and politics of sanitation in informal settlements in Mumbai, including the everyday cultures and contested politics of sanitation and water in two informal settlements. His current work examines the politicisation of informal neighbourhoods in comparative perspective, including African and South Asian cities. He is author of Learning the City: Knowledge and Translocal Assemblage , a book focused on the intersections among urban inequality, materiality, resistance and learning, as well as related edited collections, including: Infrastructural Lives: Urban Infrastructure in Context , Urban Navigations: Politics, Space and the City in South Asia , and Urban Informalities: Reflections on the Formal and Informal . Address: Department of Geography, Durham University, DH13LE, UK; e-mail: [email protected]

This paper develops a conception of "sites of entitlement" as a basis for better understanding how infrastructure and services are perceived and experienced in informal settlements. While legal and policy frameworks are often viewed as the source of entitlements to infrastructure and services, the complexity of provision, access and negotiation in informal settlements demands a conception of entitlement that exceeds those domains. Based on ethnographic research on sanitation and water in informal settlements in Mumbai, we focus on the ways in which people's everyday experiences, interactions and practices constitute sites of entitlement. These sites are unevenly produced, contested, often in flux and ambivalent, sometimes made through collective struggle and at other times through quiet individual practice, and always constituted by social relations. Sites of entitlement emerge in close relation to moral economies, and are characterized by often profound and for research, policy and practicechallenging levels of spatial and temporal variation. We argue that sites of entitlement are vital for thinking through the possibilities of realizing the universal right to sanitation and water. KEYWORDS informal settlement / moral economy / Mumbai / sanitation / sites of entitlementIncreasing concerns over the consequences of inadequate urban sanitation and water for poverty, health, livelihoods, and education have spurred global declarations on the human right to sanitation and water throughout the past decade. (1) However, the social and spatial heterogeneity of urban poverty is often missing in global policy debates. (2) Sanitation interventions usually focus on informal settlements that are officially recognized, and there is also a generally weak understanding of the micropolitical realities of urban services in informal settlements over time and space. While research, policy and practice debates routinely invoke the human right to sanitation and water, there is comparatively little grasp of how local understandings of entitlement are produced and contested, or how that varies across social groups and spaces. There is a pressing need for greater attention to the geographies of entitlement in approaches to water and sanitation. In this paper, we offer a focus on what we are calling "sites of entitlement" as a basis to carry this approach forward. We make our case based on ethnographic fieldwork conducted in informal settlements in Mumbai. Entitlements are usually considered at the level of state, legal and policy frameworks, and in relation to how these interact with global agreements and discourses such as the Millennium Development Goals. While the complexity of provision, access and negotiation around sanitation and water in informal settlements is often gestured to in these domains, it is also routinely demoted to secondary importanceto mere context to the abstractions of broader principles and agreements. Our research suggests a different view is required. The geographies of sanitation and water provision and access across and within urban informal settlements are profoundly diverse and contingent on various social, economic, political, institutional, technological and historical processes. We see this within one city, let alone between cities and countries. Research over the past decade has widened our understanding of these urban geographies and processes in significant ways. (3) This diversity and contingency should leave us in no doubt that the urban poor have strikingly inadequate formal rights. As Partha Chatterjee has so influentially shown, the fact that so many of the urban poor live in, at best, an ambiguous relationship with dominant frameworks of private property and citizenship rights leads to an often ad hoc, piecemeal provision of basic services and infrastructure across the global South. (4) Even when certain rights are granted, they are often not realized because of the uneven power relations that shape the everyday life of the city, and the poor very often turn to diverse informal arrangements, social networks and negotiations to meet their sanitation and water needs. The argument we make is more than a gesture to spatial diversity. We believe this spatial diversity demands that we refute any easy separation of abstract principlesuch as the right to sanitation and waterand messy everyday contexts and struggles. More fundamentally, we position entitlement alongside this spatial heterogeneity, and on the same conceptual plane. Rather than seeing context as the "micro" and principles of entitlement as the "macro", we contend that it is impossible to think about one without the other. Universal rights to sanitation and water can only emerge through a focus on the everyday experiences, claims, negotiations and struggles that continually take place, sometimes through overt individual or collective action and sometimes through quiet processes of subversion, in informal settlements. This is what we mean by "sites of entitlement": the always coexisting principles, laws and norms with social and spatial differences in the everyday production of claims. Sites of entitlement are characterized by changes over time, often require ongoing negotiation, and are frequently characterized by uncertainty and ambivalence. Sites of entitlement are the lived geographies of abstract claims to universalist rights, fundamental to how rights to sanitation and water are delivered or might be delivered in the future. We propose that "moral economies" play a significant role in producing sites of entitlement. We define moral economies as both collectively understood informal regulations around expected behaviour and an individually held sense of what is expected that may or may not coincide with that shared collective view. While moral economies are collectively shaped in the lives and local/translocal relations that people have, individuals often construct their own preferences and contestations in relation to those collective views, sometimes in oppositional ways, and this dialectic shapes a sense of what individuals feel they might reasonably expect from others. Moral economies provide a legitimate, though not necessarily legal, basis for claims and affect how entitlements are conceived, claimed, contested, and ultimately realized or denied, along with other factors such as local power relations, state action or regulation, and so on. Sites of entitlement are relationally formed between different groups, spaces, actorsfrom states to individual politicians to civil society groups or private actorsand moral economies. Moral economies do not play a one-way causal role here. In practice, moral economies and sites of entitlement are constructed fluidly. Moral economies can delimit a provisional sense of who is and is not entitled to provisions and under what conditions, while a sense of entitlementwhether locally based on cultural historical practices or legally basedcan play a part in generating the operative moral economy. The relation between the two emerges more as a contingent assemblage than a linear causal process. Sites of entitlement reveal a complex and contingent relation between moral economies and entitlement. For example, while some people, and in relation to some services, might feel a strong sense of moral economy around state delivery, experience may indicate that such provision is so unlikely that to think of it as an entitlement borders on absurd. Moreover, this relation between moral economy and entitlement may vary from group to group or neighbourhood to neighbourhood. Moral economies are important to shaping entitlement, but not in a linear way. We argue that our focus on the relations between everyday life, moral economies, and sites of entitlement lays out an important and poorly understood research and policy agenda. From this we can develop a more nuanced understanding of the challenges and, through this, a more socially equitable provision of basic services. The paper is based on a nine-month ethnographic study in two informal settlements in Mumbai. Rafinagar is a poor unauthorized or "non-notified" settlement in the north-east with an older, more established Part 1 and a newer, poorer and still expanding Part 2. Khotwadi is an authorized, well-established settlement in the north-west. ( 5) Khotwadi, with a population of approximately 2,000 households, has 24 toilet blocks and a total of 180 seats, whereas Rafinagar, with approximately 4,000 households, has six toilet blocks and 76 seats. Rafinagar, then, has twice the population and half the toilet seats. There is no legal requirement for the state to provide services and infrastructure to non-notified Rafinagar, but all informal neighbourhoodsnotified and non-notifiedcan access services through specific projects. One such project is the Slum Sanitation Programme (initially supported by the World Bank in the mid-1990s, which provided toilet blocks to Rafinagar (6) ). These are unevenly distributed: Rafinagar Part 2 has only one toilet block, provided by the Maharashtra Housing and Area Development Authority (MHADA) in 2011, although also in use are temporary hanging latrinesprecarious self-built structures usually over open drains or watercourses. While most residents in Khotwadi have a degree of secure water access through unmetered municipal standposts, metered group connections and wells, most of Rafinagar's residents face profound difficulties and incur high expenditures for water and/or time and effort in collecting

BACKGROUND Studies consistently find that depressed Latinos are less likely than depressed non-Latino Whites to receive adequate treatment, even when controlling for socio-demographic differences. [1][2][3] Understanding cultural perspectives on depression and depression care may be a critical step in reducing depression care disparities. An emerging literature has focused on the way Latinos identify what causes depression, [4][5][6][7] how they manage physical and emotional symptoms of depression, 8 and how they understand depression care treatment outcomes. [9][10][11] There is a well-recognized, strong association between depression and IPV. [12][13][14][15][16][17] Several studies have confirmed the relationship between IPV and depression in Latina women. 12,18,19 IPV also may negatively affect mental healthcare-seeking behaviors, with lower use of mental health services in depressed women with a history of IPV than in depressed women without an abuse history. 20,21 A history of IPV may potentially compound depression care disparities for Latina women. Little is known about how Latina women conceptualize the relationship between violence and depression. This type of information can help us understand how to provide culturally appropriate depression care for the vulnerable and understudied population of Latina women who have experienced IPV. Our objective was to understand Latina women's beliefs, attitudes, and recommendations regarding depression and depression care, with a special focus on the impact of gender, ethnicity, violence, and other social stressors. --- METHODS --- Community-Based Participatory Research We used a community-based participatory research approach throughout the project. We formed an academiccommunity partnership consisting of academic researchers, healthcare providers, domestic violence advocates, IPV survivors, community members, and community leaders from both the Latina and African-American communities. The principal investigator had originally intended to work with community members to help increase the cultural relevance of a health-system-based depression care intervention, so she and the executive directors from two community agencies applied for an additional grant to support the CBPR aspects of the project. However, once the project began, community partners felt that the intervention would not be successful within the health system, so the entire project moved to the community, including the focus groups and the intervention. Community partners and the PI acted as equal partners in each phase of the project. The entire team met at least monthly and worked collaboratively to design protocols and materials, collect data, analyze results, and use findings to design the intervention. The African-American and Latino portions of the study were designed together, but were then implemented and analyzed separately. This manuscript addresses findings from the focus groups with Latina women. Findings from focus groups with African-American women are presented elsewhere. 22 --- Translation and Interpretation Issues Joint meetings with the Latino and African-American portions of the team were conducted using simultaneous translation. Latino team meetings were conducted in Spanish. Whenever possible, we used survey instruments that had previously been translated and validated in Spanish. The remaining data collection instruments, as well as all recruitment and consent materials, were translated from English into Spanish and then back-translated by two community members to ensure accuracy and cultural relevancy. Focus groups were conducted and transcribed in Spanish and translated into English. Some members of the team analyzed transcripts in English while others analyzed them in Spanish. Meetings to decide on themes and reconcile differences were conducted in Spanish. --- Recruitment and Eligibility Community partners recruited participants primarily by word of mouth. They also distributed flyers about the study at social service agencies. The flyers and recruitment scripts did not mention violence or depression so as to not bias the sample. Potential participants were screened for eligibility either by the community partners or a Latina research assistant. Focus group participants were offered $50. To be eligible, participants had to be female, at least 18 years of age, consider themselves to be Latina, Hispanic, or of Latin American descent, speak Spanish, score 15 or higher on the depression scale of the Patient Health Questionnaire 23 and have a lifetime history of IPV. Physical or sexual IPV was assessed using 2 items . Emotional abuse was measured by a score of 20 or greater on the Women's Experiences of Battering scale 24 , modified to ask about lifetime experiences. Emotional abuse was only assessed systematically if participants denied physical or sexual IPV. Eligible participants were invited to a focus group by the community partners or research assistant. Potential participants were warned that their attendance would signify that they had depressive symptoms and a history of IPV to other participants attending the focus group. --- Data Collection Focus groups were held in private community settings. Informed consent was obtained prior to the start of the focus group discussion. The consent form, jointly created by academic and community partners, stated that we will ask all participants to keep what is stated in the focus groups confidential, but that we could not guarantee that other participants would maintain their confidentiality. It also explained that excerpts of what they said would be used in publications. Participants first completed a short demographic questionnaire. Then focus groups began with a discussion of "ground rules" that stressed the importance of respecting other's confidentiality and reminded participants that they could step out at any time if needed. On a few instances, individual participants did choose to take a break when they became emotional. They received immediate counseling from the PI or from community partners who were experienced promotoras/domestic violence advocates. All felt well enough to voluntarily re-join the focus group session. Focus groups used a structured interview guide created collaboratively by the academic and community partners. First, women were asked several questions about their experiences and beliefs about overall health, mental health, depression, depression treatments, and violence, as well as the interrelationship between physical health, mental health, and violence. After being asked to share their recommendations for improving depression care, they were presented with the teams' ideas for possible depression interventions and asked to respond to them. At the end of the session, to increase validity, the facilitator summarized what she had heard the women say and asked them to respond. Finally, each woman was asked to state the single most important point she wanted to make. An expert in focus group research and the PI conducted an all-day training for the African-American and Latina community members of the team on how to facilitate focus groups. Training included role-play sessions where learners took turns moderating mock focus groups. Community partners had prior experience facilitating domestic violence support groups. Role-plays focused on the similarities and differences between the facilitator's role in a support group vs. a focus group setting. We also included scenarios for how to handle difficult situations. Two Latina community partners and one Latina research assistant facilitated the focus groups. The PI listened to the focus group discussion via a remote headset in a separate room and met with the facilitators during breaks. Interviews were recorded and transcribed. --- Data Analysis Both academic and community team members participated in the thematic analysis 25 using an inductive approach at a semantic level, with an essentialist paradigm . The PI and the lay facilitators met regularly to discuss preliminary themes. After all focus groups were completed, team members met to create an exhaustive list of preliminary codes. The PI and a research assistant formally coded all transcripts using Atlas-Ti software . Three community partners independently coded the transcripts by making notations onto the transcripts. The group then met to compare findings and refined what they felt were the key messages. The group collaboratively decided on a framework that collapsed codes into major themes and sub-themes. A different Latina research assistant, , re-checked the coding to assure that it fit with the final coding structure. The group met one last time to resolve any inconsistencies. --- RESULTS We screened a total of 78 women, 48 of whom were depressed. Of those, 45 women had a history of physical and/or sexual IPV and the remaining three women had a history of emotional battering alone. We invited the 45 eligible women to the focus groups;31 women attended one of five focus groups. All focus group participants had a lifetime history of physical or sexual IPV. Table 1 describes their demographic characteristics. --- Common Themes around Depression We identified a number of common themes regarding women's beliefs about depression. Community and academic partners used these themes to create a model describing how participants understood the relationship between life stressors, depression, and health. Violence and Other Life Stressors Affect Health and Well-Being. Participants discussed many life stressors they felt effected their health and well-being. Such stressors included exposure to childhood sexual abuse, experiences of emotional, physical and sexual abuse from intimate partners, their partners' use of alcohol or drugs, difficult relationships with family members, poverty, and stressors related to their immigration status. "Every time I was raped-because I was raped more than once-my uncle raped me from when I was 8 years old until I was 11 years old -every time he was raping me he told me I was 'meat to feed the dogs'-and everything, every single thingfrom headaches-everything comes back to that. I mean, everything, in my case, everything is related." The stressors themselves were often inter-related-for example, many women talked about childhood sexual abuse leading to a lack of interest or comfort with sex, which then led to marital conflict or inability to sustain intimate relationships. "I was sexually abused… and it is horrible. It doesn't go away. It is the worst; and that's why you can't have relations with your partner sometimes; because you feel-and you remember-you were hurt. You feel revulsion having sex with your husband." "Holding Things Inside" Causes Depression and Physical Illness. Participants often felt what was most harmful to their health and well-being was the need to keep negative experiences hidden. "When you shake a bottle-a soda, for example-it has gas, so shaking that bottle will make it explode. It is the same. When you are holding up all the things inside-in your heart-it damages people-like me personally. There are many things that I, since I was little, I had been through…. No one knows what it is inside of my heart; I am the only one that knows." --- They Viewed Depression as a Direct Result of Holding Things Inside "There are a lot of things, things like violence, or a lot of things that, maybe I can't finish saying them, no, but each person has a lot of things, physically, emotionally, in your children, in your family, friendships, well, of everything that happens in the world. To me, if you don't have an exit, it goes into your blood and becomes a depression." Inability to Talk. Many women felt a strong need to talk about all the things they had been holding inside. "Every person-every person is and knows what is on their heart-every person knows-and it is our duty to take it out so we don't have it inside anymore." However, they often lamented about not being able to do so. Sometimes, they had practical reasons for not speaking. However, most often they simply felt they "couldn't talk". "I can't talk. At times, it isn't because I don't want to talk about it, but because it is so difficult to get the words out of my mouth. I feel like there is something in my, something here that blocks my throat. I can't get the words out to explain what I want to say." Depression Leads to Physical Illness. They also felt that there was a direct connection between depression and physical illness: "Yes, because you know that violence brings depression and depression makes you sick, and sicknesses brings more sickness." "The mind controls your body, your eyes, everything you do. If your mind is-like people say-if depression attacked it, you don't have the control of doing things. It starts with your heartthe circulation of blood." Desire to be Heard. Women felt the key to healing was being able to talk about their depression and the experiences they were holding inside. --- Common Themes Around Depression Care "Talking will cure you. Cure is when you talk and get it all out." Unfortunately, they often did not feel comfortable discussing mental health issues or life stressors with --- Depression Care Desire to be Heard Though they were in the minority, a few women did feel like they had succeeded in finding their voices: "I had to do things that were so ugly… I felt like no one would believe me or maybe I felt like I would never be able to talk about it, but this thing that was so bad … It is something that I thought I would never talk about, that would never be revealed and that no one would ever know what I had lived through, but no. There came a time when I figured out what I had to say." Desire for and Appreciation of Health Information. Women were very interested in obtaining factual health information. They often felt that a depression diagnosis was a great relief: "I didn't know what depression was…. But when they started giving me the magazines… the symptoms that were described there-that's what I was feeling." In contrast to our findings with non-Hispanic White women, 26 who were concerned that providers would attribute physical symptoms to their mental health or abuse histories, women appeared to be grateful when providers told them that physical symptoms were caused by depression, often seeing it as a source of new information. "I, thank God, went to urgent care because my depression stopped this part of my hands from working so that I wasn't able to hold onto a gallon of milk. And at night my fingers would hurt and my hands would swell. And so I went to urgent care, but the nurse told me, "Look, … it could be that the depression is affecting your hands." And they referred me < to counseling>, thank God." Desire for Depression-Care Program that Addressed Real-Life Needs. Latinas in our sample indicated that an ideal treatment program would not only address their mental health concerns, but would ensure confidentiality, provide childcare, help them address the violence in their lives, and give them accurate information and practical skills. Negative Attitudes About Antidepressants. Though a few women related positive experiences with antidepressants, most had relatively negative attitudes toward antidepressants, primarily due to experiences with side effects. "They gave me medication and instead of feeling better, I felt worse. The medication inflamed this whole part and my heart felt like it was accelerated badly and my veins were irregular here in my feet and in my hands so that's how I was with the medication and then later I told the doctor, 'You know what? Instead of feeling good, I feel worse'." They also were concerned that antidepressants were addictive: "I took antidepressants, but no, like for a month only. I didn't want to be addicted to those pills, true. Because I saw relatives that, if they didn't take them, they were like, they could not handle it, and they were bad with their children and everybody around them; so I didn't want to be addicted to it." Negative Experiences With Healthcare Attributed to Lack of Health Insurance or Class Issues. Women often discussed experiences of discrimination in healthcare. Though there were mixed opinions as to whether or not the discrimination related to their race or ethnicity, there was general concensus that the problem was due to a lack of good health insurance. For example, after hearing of an experience where a participant felt her family was not treated as well as a White family, the facilitator asked: "So you feel that was because of your color?" The participant replied: "Not because of the color-because of the insurance. Because you can see that the parents of that kid had money." Some participants related insurance status back to ethnicity: "Well, for example, we, Latinas, we don't have the same [health insurance] benefits that Americans do, or even other races. Some of them have more benefits than we do. I feel like Latinos, Hispanics, are the ones with less benefit after all. Latina women were less interested in the race or ethnicity of health care providers than were African-American women 22 , as long as providers spoke Spanish fluently and understood their culture. "We really don't have fear of racism or what have you, but we want them to understand what we want to say." --- DISCUSSION The concept of "keeping things inside" was key to Latina IPV survivors' understanding of the cause of depression and physical health problems. Ultimately, they wanted to find their voice-a voice that had been suppressed due to a multitude of social factors. They desired programs that would help them address their experiences in a confidential, culturally-appropriate manner and would provide them with accurate information and practical skills. Participants in our study saw a strong link between their depressive symptoms and their experiences of childhood sexual abuse, IPV, and other life stressors. Several other qualitative studies of Latinos have noted that Latinos perceive depression as being caused by an accumulation of interpersonal and social stressors. 4,5 Our study adds to this literature by further examining participants' understanding of the nature of that link. Participants seemed to feel that depression was not simply caused by having had these experiences, but more specifically by not being able to talk about them. Just as they had not felt comfortable talking about such experiences with family or friends, they also avoided talking about them with healthcare providers. Our sample, unlike the samples in other qualitative studies on depression in Latinos, was limited to women who had experienced IPV. Additionally, a majority of women spontaneously disclosed a history of childhood sexual abuse. It is possible that abuse survivors place a greater emphasis than non-abuse survivors on the importance of talking about life stressors because they, in fact, have greater barriers to discussing life stressors, be it due to the classic isolation that abuse survivors experience, social taboos against discussing abuse, or the dangers associated with disclosing abuse. 27 However, given the high lifetime prevalence of abuse amongst women with depression, it is likely that other samples also included many abuse survivors. Participants in our study knew that everyone in the room had experienced abuse and thus may have felt more comfortable talking about what they normally cannot discuss. Women in our study felt that depression could directly lead to physical illness and expressed appreciation toward providers who made them aware that their somatic symptoms could be caused by depression. This finding is in direct contrast to what we have previously reported from a focus group study of non-Hispanic White women with depressive symptoms and histories of IPV. 26 Participants in that study were often angered by suggestions that their physical symptoms were "caused" by mental health issues or abuse, and they worried that disclosing information to healthcare providers about abuse or mental health would lead providers to treat them "like hypochondriacs". Participants in the current study had an opposite response to providers attributing physical symptoms to depression. Studies have noted a relationship between level of acculturation and the identification, perception, and role of somatization in Latinos in the US. 28,29 Some have argued that somatization is more culturally accepted in Hispanic culture than it is in the US, and serves as an important modality for eliciting support. 29 It is possible that our Latina participantswho had low levels of acculturation-did not have as strong negative perceptions of somatization as non-Hispanic IPV survivors and were thus more open to healthcare providers attributing physical symptoms to depression or mental health. Among our participants, there was a strong desire to know what was wrong, to find ways to get better, and to demystify the many physical and emotional symptoms that made their already complicated lives less manageable. Participants' motivation to understand and receive a diagnosis partly supports Nadeem's findings that depressed immigrant Latinas were more open to treatment than depressed non-Hispanic White women. 30 Although they cited stigma-related concerns when seeking mental health treatment, it was not as pronounced as in other minority women such as African, African American, or Caribbean Blacks. In our study, Latina women seemed less focused on concerns about racial discrimination than were African-American participants. 22 Whether or not they made the connection between ethnicity and insurance status, they most often attributed negative experiences to lack of health insurance and wished for greater access to care. Only 15% of participants currently had a primary care provider, which highlights some of their difficulty accessing care and may contribute to their not having felt comfortable discussing abuse. Their openness to depression treatment, however, did not indicate receptiveness to antidepressant medications. Multiple other studies have noted an apprehension toward antidepressants in minority populations, including Latinos and African-Americans. 4,5,31,32 Our study adds information as to how women conceptualize their aversion toward antidepressants. For example, in the African-American portion of the study, 22 participants also had negative attitudes toward antidepressants, but they framed their objections in terms of the need for selfreliance and a distrust of the White providers prescribing the medications. Latina participants voiced less overt distrust of the healthcare system, but often justified their hesitation toward antidepressants by citing adverse effects they had experienced or heard about from others. Our study has a number of limitations. It describes a small, non-random sample of women in a single city. As with most qualitative studies, we were most interested in including informants who could best provide rich, in-depth information about the topic of interest. Results may not be generalizable to other populations, especially those with a greater degree of acculturation, without histories of violence or abuse, or with better access to healthcare services. Moreover, we did not separate focus groups by participants' county or origin. Some studies have noted significant variation in IPV rates amongst different Hispanic sub-groups. 33,34 A majority of women in our study were born in Mexico. Our results may be less applicable to women from other countries. Despite these limitations, our study has some important implications for individual providers and depression care programs. Healthcare providers may use our findings in a number of ways. First, they should recognize that it may be very challenging for some of their Latina patients to discuss stressful life experiences. Still, they should actively inquire about life stressors, including experiences of violence and abuse, and encourage patients to talk about their experiences. Reinforcing the need to "not keep things inside" may be an appropriate way to encourage mental health counseling for Latina women, especially those that might hint at having had difficult life experiences. Providers should try to name depression as a clinical diagnosis and offer practical information to patients about depression and its treatment. Providers should still prescribe antidepressants when appropriate, but should pay close attention, both proactively and after starting treatment, to patients' concerns about side effects and addiction, as such concerns may be a significant barrier to adherence. Rather than assuming that patients will be offended by attributing physical symptoms to mental health, providers should recognize that some Latina patients may be appreciative of information explaining the relationship between life stressors, mental health, and physical symptoms. Our study also has implications for community or healthcare organizations interested in creating depression care programs for Latina women. Such programs should make sure to actively address life stressors such as childhood abuse or intimate partner violence, offering women a safe way to find their voice. Though race or ethnicity of staff seemed to be less of a factor for our Latina participants than for women in the African-American portion of our study, programs should make sure that staff speak fluent Spanish and are culturally competent. Programs would likely benefit by highlighting policies around confidentiality and offering practical supports such as childcare or skills training. Our academic-community partnership used results from this study to create a community-based depression care intervention for Latina women with a history of intimate partner violence. Future studies are needed to test the effectiveness of such depression-care programs for Latina women. --- Conflicts of Interest: None disclosed. Corresponding Author: Christina Nicolaidis, MD, MPH; Division of General Internal Medicine, Oregon Health Science University, L475, 3181 SW Sam Jackson Park Road, Portland, OR 97239, USA .

BACKGROUND: Latinos are less likely than non-Hispanic whites to be adequately treated for depression. Intimate partner violence (IPV) is strongly associated with depression. Less is known about how Latina IPV survivors understand depression. OBJECTIVE: To understand Latina women's beliefs, attitudes, and recommendations regarding depression and depression care, with a special focus on the impact of gender, ethnicity, violence, and social stressors. DESIGN: Focus group study. PARTICIPANTS: Spanish-speaking Latina women with a lifetime history of IPV and moderate to severe depressive symptoms. APPROACH: We used a community-based participatory research (CBPR) approach to conduct a thematic analysis using an inductive approach. RESULTS: Thirty-one women participated in five focus groups. Women felt depression is caused by "keeping things inside". They also felt that keeping things inside could lead to physical illness or an inability to function. Their inability to talk was fueled by issues such as stigma, fear, isolation, cultural norms, or simply "not having the words". They felt that the key to treating depression was finding a way to talk about the things that they had kept inside. They greatly valued information about depression and appreciated learning from providers that their physical symptoms were caused by depression. They wanted confidential depression care programs that not only helped them deal with their depression, but also addressed the violence in their lives, gave them practical skills, and attended to practical issues such as childcare. They had negative attitudes toward antidepressants, primarily due to experiences with side effects. Negative experiences with the health care system were primarily attributed to lack of good healthcare insurance. CONCLUSIONS: The concept of "keeping things inside" was key to participants' understanding of the cause of depression and other health problems. Clinicians and depression care programs can potentially use such information to provide culturally-appropriate depression care to Latina women.

Introduction Depression and suicidal ideation are about two to three times higher in women than men. [1][2][3][4][5][6][7] Theories explaining this higher risk focus on biological issues, despite the evident impact of social variables and inequities. Although scientists have reported this gap for many decades, empirical evidence explaining this difference is scarce. [8][9][10] Violence is a global public health problem and women are at a higher risk of victimization, especially from family members or acquaintances. [11][12][13][14][15] According to the United Nations' Sustainable Development Goals Report 2019, from 2005 to 2017, 18% of women between 15 and 49 years of age had experienced physical and/or sexual violence by an intimate partner in the previous 12 months. The majority of victims of intimate partner/family-related homicides in 2017 were also women. 16 Violence may lead to death in the most extreme cases, and when it does not, it could have important consequences for the victim's mental health and quality of life, independently of the type of violence . 17 Many studies have demonstrated the association between violence and mental illness, and that certain kinds violence, such as that by intimate partners or acquaintances, are more prevalent in women. Other studies have shown that violence might play different roles in depressive disorder in men and women in adulthood. 12,13 Gender inequality and exposure to severe adversity, especially violence, might explain the higher prevalence of certain mental disorders in women. 10,12,13 Nevertheless, little is known about the role of violence in the causal pathway of mental disorders in both sexes, and none have explored it using a counterfactual framework. Therefore, the aim of this study was to investigate the mediating effect of violence in the association between biological sex and depression or suicidal ideation in adult respondents to a nationwide Brazilian survey. --- Methods This study used data from the 2013 National Health Survey , a cross-sectional population-based survey. The National Health Survey covered the entire territory of Brazil based on tracts used in the 2010 census. Areas with special characteristics or small populations were excluded. The sampling process was carried out in three stages: census tracts were the primary sampling units, households were the second-stage units and residents aged 18 years or over were the third-stage units. The original study selected 69,954 households and involved 13.9% losses and refusals. Through a simple randomization process, one individual per household was selected and invited to participate in the face-to-face survey. Trained interviewers applied the survey using personal digital assistants. Written informed consent was obtained from all participants. Based on the probability of sample participation, weighting was used to ensure national, regional, and state representativeness. Further details on the methodology, sampling and weighting factors can be found in the National Health Survey technical report. 18 The outcomes considered in the present study were major depressive episode , determined through the Patient Health Questionnaire , which identifies cases according to DSM-IV criteria, and suicidal ideation, which was determined with the final question of the PHQ-9: ''In the last two weeks, how many days did you think about hurting yourself in any way or think that it would be better to be dead?'' Possible answers to the question were: None, Less than one week, One week or more, and Almost every day. If the respondent's answer was anything other than ''none'', he/she was considered positive for suicidal ideation. The PHQ-9 has been validated for adults in the general population of Brazil. 19 We investigated whether respondents were victims of family violence with the question: ''In the last 12 months, did you experience any violence or aggression from an acquaintance ?'' The variable biological sex was our main exposure variable. Sociodemographic data, collected through the general questionnaire, were used as covariables: age in complete years ; self-reported skin color ; education ; and wealth index quintiles . For WIQ classification, the respondents were asked about residential characteristics and household belongings, such as: house or apartment, number of bedrooms, having a domestic worker, and the number and type of home appliances . The five quintiles ranged from the poorest to the richest . Data analysis was performed in Stata version 14.0 , and the clustering effect was considered using the ''survey'' command. We calculated total and stratified prevalence of the variables and used chi-square tests for heterogeneity. Odds ratios and 95% confidence intervals were estimated with logistic regression to estimate the association between sex, MDE, and suicidal ideation. The analysis was stratified according to violence, and a multiplicative interaction term between both sexes and violence was included in the regression models to determine if there was a modifier effect. We also calculated the relative excess risk due to interaction, a measure similar to additive interaction that uses risk ratios or odds ratios rather than risks. 20 To estimate the potential mediating effect of violence, we used the G-formula, a parametric computation that uses Monte Carlo simulations. The G-formula offers greater flexibility than traditional mediation methods, including modelling non-linearities and allowing for mediator-outcome confounding and exposure-mediator interaction. We calculated the total effect, the natural direct effect , the natural indirect effect , and the controlled direct effect . 21 The NDE represents the effect of exposure on the outcomes that is not explained by the mediator , while the NIE estimates the effect that is explained by it. The sum of the NDE and NIE represents the total effect, and the quotient of the NIE divided by the total effect represents the percentage of the effect explained by the mediator. On the other hand, the CDE represents the direct effect of exposure on the outcome if the mediator is controlled . In our case we investigated what would happen if no one was exposed to violence, regardless of gender. We created and tested two different models, one for MDE and one for suicidal ideation, adjusting for age, skin color, education and WIQ . Education was included in the model as an intermediate confounder. --- Ethics statement All procedures in this study complied with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration. All procedures involving human participants were approved by the Brazilian Ministry of Health's National Commission of Ethics in Research for Human Beings . --- Results Of the 60,202 interviewed individuals, 52.9% were female, 21.6% were between 25 and 34 years of age, most selfreported their skin color as white , completed secondary school or had incomplete higher education , and belonged to the poorest quintile . Overall, 4.1% presented MDE and 3.8% presented suicidal ideation and 2.5% had suffered family violence in the last 12 months . Older men and women with less education and a lower WIQ had a higher prevalence of MDE and suicidal ideation. Violence victimization was more frequent in poor young black women with low education . Table 2 shows that women's odds of MDE and suicidal ideation were 2.36 and 2.02 times higher than men, respectively. When the analysis was stratified according to violence victimization, female victims had a higher risk of MDE and suicidal ideation . However, confidence intervals from multiplicative interaction terms included the reference, suggesting that these differences could be due to chance . When the additive interactions were calculated using the OR from our regression models, we found that the relative excess risk due to interaction for depression was 5.97 and that of suicidal ideation was 6.20 . Since both were 4 1, this implies a positive additive interaction between sex and violence in both outcomes. G-computation estimates of the NIE, NDE and CDE for suicidal ideation and MDE were similar. Figure 1 shows that, in the mediation analysis, the NIE corresponded to 10.6% of the total effect of sex on MDE and 8.0% of the effect on suicidal ideation. --- Discussion We found that women are at higher risk of MDE and suicidal ideation than men, and that this association is mediated by suffering violence from a family member or acquaintance in the last 12 months. The NIE should be interpreted as the proportion of the association between sex and MDE and between sex and suicidal ideation that is mediated by violence. In other words, approximately 1/10 of the higher risk that women have of these mental conditions is due to their increased risk of violence victimization. Since there was no multiplicative interaction between exposure and the mediator, it was expected that the NDE would be equal to the CDE. Thus, we could also interpret the same result in terms of the mediator. If no violence had occurred in this population, 10% of the association between sex, MDE and suicidal ideation would have been prevented. However, we should point out that there was a positive additive interaction, which suggests that female violence victims would benefit more from interventions to reduce the risk of depression and suicidal ideation. Our finding that there is a higher prevalence of MDE and suicidal ideation in women is similar to other results in the literature. [1][2][3][4]7 Although biological theories of depression involve hormonal function and the performance of neurotransmitters and neuromodulators, 22 social adversity, including maltreatment, gender roles, and violence, is higher in women, and its psychological impact can be very significant for both depressive disorders and suicidal ideation. 7,9,10,23 Even when social impairment is discussed in the literature as a possible explanation for the mental health gap between men and women, most of the discussion about likely associations has focused on the fact that sex differences in depression and negative mood may be associated with biopsychosocial factors such as hormonal differences, family environment, childhood stressors and social and cultural factors. 4,5,7,9,10,23 One of the most discussed hypotheses is that women's hormone levels change more throughout life due to menarche, pregnancy, menopause and contraceptive use, and this could contribute significantly to the research gap. These Figure 1 The mediating effect of violence in the association of being a woman and: major depressive episode, suicidal ideation. The main exposure was being a woman, with men as the reference group. As a mediator, violence was categorized as victimization vs. not having a major depressive episode vs. not having suicidal ideation. The natural direct and indirect effects were calculated with G-computation. Confounders included age, skin color, wealth index, and education . The percentages are based on the natural direct and indirect effects and represent the proportion of the association explained by each pathway. MDE = major depressive episode; NIE = natural indirect effect; NDE = natural direct effect. changes could be directly related to mood and negative/ positive affect, and could interact with gendered intra-and interpersonal factors, such as stress. In addition, women might have a blunted cortisol response. 10 Nevertheless, some research has proposed that this gap might be related to greater exposure to severe adversity, especially violence, and to structural gender inequality at macroand microlevels, despite the present lack of empirical evidence. 5,9,10 In a context where suicidality is more likely a consequence of depression, these theories might be relevant, although some of the burden of suicidality is not necessarily due to depression or depressive-like symptoms. Therefore, other important predictors and mediators should be explored. 24 However, in general, the reasons women are more affected than men by depression and/or suicidality remain unresolved in the literature, and biological factors do not seem to fully explain it. Violence against women can take many forms, such as physical, sexual and emotional/psychological, and these forms can coexist. 14,25,26 Several studies have shown female violence victims are at an increased risk of mental disorders. [25][26][27][28][29] Furthermore, women who have experienced repeated domestic violence and abuse, other forms of violent victimization, or direct expressions of rage or aggression have a strong tendency to suppress their emotions. 5 This could be related to other behaviors, such as substance use and other psychological consequences of violence, including suicidal ideation. 26,27,29,30 When combined with the submissive nature of some women, Data presented as odds ratio and 95% confidence interval. p-values for interaction term in a logistic regression: * p = 0.183; w p = 0.050. continuous exposure to stressful events, such as violence, could lead to extreme vulnerability, fear and feelings of inability or powerlessness, which could cause or contribute to mental health problems. 5,26,27,29 Sociocultural factors, such as the stress associated with traditional female roles, can affect women's mental health. 5,7,8,10 In this study we found important associations between sociodemographic characteristics and depression and suicidal ideation in women, a relationship that has been demonstrated in other studies. 1,4,31 For the men in our sample, no associations were found between skin color and either outcome, between education and violence victimization, or between WIQ and MDE. In Brazil, skin color, education and WIQ are markers of social vulnerability, which can be associated with mental health conditions. 32,33 The fact that important social determinants were associated with outcomes for Brazilian women but not for men could indicate that these determinants are among the sociocultural markers that affect women's mental health. We also found that women are more likely than men to suffer violence by a family member or acquaintance, which has also been reported in previous studies. 11,12,15 However, these findings also indicate the importance of investigating these mental health conditions and their greater occurrence among women from a social determinants of health perspective. It is important to consider some limitations of our study. First, since we did not have access to all the necessary information on losses and refusals, we could not assess their impact on our measures of association. However, the losses were relatively small, and they do not seem related to exposure, since the prevalence of females and depression/suicide were similar to that reported in other studies. 2 The PHQ-9 uses an algorithm to assess depression that reproduces DSM-IV diagnostic criteria. Nevertheless, as with any screening test, classification errors could have occurred, since a clinical interview was not conducted for diagnostic confirmation. Similarly, suicidal ideation and violence were self-reported and evaluated with a single question. Thus, underreporting could have also been an issue in this study. However, we considered that if there was a classification error for any of these cases it would be non-differential, thus the measures of association reported in this study could be underestimated. Such underestimation could influence our findings, leading to underestimation of the mediating effect of violence, for example. In addition, although longitudinal measures would be preferable for assessing the mediating effect of violence, we find it more plausible to think that depression and suicidal ideation are a consequence of violence, at least in this case, and not the other way around, and there is evidence of such a longitudinal association in the literature. 12,13 In addition, respondents reported the occurrence of violence in the last 12 months, but depression and suicidal ideation in the last two weeks. We only studied violence from family members or acquaintances, which might have limited our capacity to determine which kind of violence would be more important in the explored mediators. However, the available data was inappropriate for answering this specific question, since the type of violence was self-reported and the occurrence of violence was likely underestimated. Thus, the answer to this question should be explored in other studies with higher specificity. On the other hand, we used a representative sample of the entire Brazilian population that included more than 60,000 individuals. We also assessed mediation with a novel method, using a counterfactual framework that avoided overadjustment and extracted measures that could be considered confounders. Since violence and mental health outcomes could both be consequences of sociodemographic characteristics, to correctly calculate NDE and NIE, we had to adjust our models for those variables, and given that adjusting for sociodemographic characteristics in a regular regression model would have been inadequate , they had to be treated as post-confounders. Thus, estimating from G-computation is more adequate for evaluating the mediating effect of violence while adjusting for sociodemographic variables. 21 Finally, our study has shown evidence that violence is a mediator in the causal pathway between sex, and depression and suicidal ideation. Avoiding any kind of violence is already considered an important objective worldwide, and it should be pursued independently of any association with mental health. However, we find it relevant to highlight the importance of psychosocial issues, especially violence from family members or acquaintances, in the etiology of mental health conditions, since they partially explain why women are more affected than men. Our results indicate that one likely reason women have a higher burden of depression and suicidality is the violence they suffer, i.e., vulnerability to these mental health problems is not as much an inherent condition as a consequence of external psychosocial factors. Therefore, it is important to consider these issues in the planning and implementation of public mental health policies, as well as in more specific clinical approaches, especially those regarding women's mental health. --- Disclosure The authors report no conflicts of interest.

Objective: We aimed to investigate the mediating effect of violence by a family member or acquaintance on biological sex, depression and suicidal ideation. Methods: We used data from the 2013 Brazilian National Health Survey, a cross-sectional nationwide survey. Major depressive episode and suicidal ideation were evaluated with the Patient Health Questionnaire. Violence victimization and other sociodemographic variables were self-reported. We used logistic regression to estimate the OR, 95%CI and G-computation to calculate the natural direct and indirect effects. Results: A total of 60,202 individuals were evaluated. Women had higher prevalences of major depressive episode (OR = 2.36; 95%CI 2.03-2.74), suicidal ideation (OR = 2.02; 95%CI 1.73-2.36) and violence victimization (OR = 1.73; 95%CI 1.45-2.06). The mediation analysis showed that 10.6% of the association between sex and major depressive episode and 8.0% of the association between sex and suicidal ideation is explained by violence. Conclusions: Women in Brazil have an increased risk of major depressive episode and suicidal ideation, and this association is mediated, in part, by the fact that they suffer more violence from family members or acquaintances.

Background About 17 percent of all married women in the world would prefer to avoid a pregnancy but are not using any form of family planning [1]. Over the past two decades, reproductive health and family planning programs, seeking to reduce this unmet need for contraception, have increasingly included components to address women's empowerment [2]. Yet virtually all the evidence comes from cross-sectional studies rather than interventions or longitudinal research. Women's empowerment has been defined as the ability of the woman to formulate strategic choices and control resources and decisions that affect important life outcomes [3]. Women's decision-making power, measured through the Demographic and Health Surveys , can be analyzed using a participation model which considers any form of women's active role in decision-making versus all the decisions being made by the husband. Two-dozen DHS country reports demonstrate that contraceptive prevalence depends on the number of household decisions in which the woman participates [4][5][6][7][8][9]. Recent meta-analyses in 31 to 46 countries using DHS data confirmed the link between women's participation in decision-making and their use of maternal health services [10] or family planning [11]. Age, education, and work for pay are the characteristics that most consistently determine women's participation in household decision-making [12]. Yet, these variables are not factors that reproductive health and family planning programs can address directly through their services. Education represents a long-term investment, and affecting women's employment would require the application of vast material resources. Specialized microcredit programs for women have produced mixed results [13][14][15][16][17][18]. Since the associations between directly measured women's decision-making power and contraceptive use do not simply reflect the magnitude of the proxy variables of age, education and work [11,19], it is theoretically possible to improve contraceptive use through initiatives that address women's empowerment by other means. Do organizing services to reduce barriers for women, training providers to be aware of gender-based power dynamics, and involving men in counseling and services empower women? Furthermore, does increasing the availability of family planning information and services benefit women with lower levels of decision-making power to the same extent as it benefits women with higher levels of decision-making power? Generally, it is assumed that more empowered women would have the advantage in attaining access to family planning. Could a program's efforts to widen access equalize access for women, regardless of their empowerment levels? Finally, is there an interaction between women's decision-making power and literacy? Are women who have high levels of decision-making power but who are illiterate similarly affected by a family planning intervention as literate women who have lower levels of decision-making power? This paper applies the participation model of women's decision-making power in the analysis of data from a study of a family planning intervention in Jharkhand state in India to test if the intervention had any effect on women's empowerment. Our intervention in Jharkhand, which expanded the availability of family planning information and methods and stressed the need for couple decision-making and communication, significantly reduced women's unmet need for contraception in the intervention area compared to the control area [20]. The earlier analysis of the intervention data, however, did not consider women's empowerment variables. This paper seeks to determine whether literacy moderated the effects on women's decision-making power of the family planning intervention in Jharkhand and whether literacy and women's decision-making power together moderated the effects of the intervention on met need for contraception. Met need is used because it more directly measures the ability of the woman to overcome barriers to access and thus is a more appropriate dependent variable for studies designed to test the effects of program efforts to increase access; contraceptive use confounds access to contraception with fertility desires and risk of pregnancy. --- Methods The institutional review board of Georgetown University at Washington, DC approved all aspects of treatment of human subjects of the study design. --- Study setting Unmet need for contraception was declining in India and it reached 13% in 2005-2006. Jharkhand, a state in Northeastern India, has been described as more patriarchal than Southern states in India [21]. In Jharkhand, according to the latest National Family Health Survey , conducted in 2005-2006 [22], only 37 percent of married women of reproductive age are literate and 60 percent are not exposed to any media. The NFHS-3 found that 35.7 percent of married women of reproductive age used contraception, including 28.2 percent in rural areas of the state. Female sterilization dominates the method mix, with 23.4 percent of Jharkhand's married women of reproductive age using that method, followed by the pill , the condom , and scant use of other modern methods. Less than 5 percent of women use a traditional method. The method mix in rural areas mirrors that in the state. The public health system in these areas consists of subcenters reporting to a primary health center. In each subcenter, one auxiliary nurse midwife provides services to an average of five villages. --- Intervention The family planning intervention that was designed to reduce women's unmet need for contraception [11] was undertaken in three rural blocks in Jharkhand in 2004-2007. It was comprised of strengthening family planning generally and introducing a new family planning method, and was carried out in the Ormanjhi block, encompassing 89 villages and about 76,000 inhabitants. Burmu, encompassing 101 villages and about 77,000 inhabitants, served as a control site for the study. A third block that also received the intervention, Kanke, was excluded from the analysis in this paper, given its more urban structure than the other two blocks. Krishi Gram Vikas Kendra , a nongovernmental organization that works in Jharkhand and manages six subcenters in Ormanjhi, coordinated the intervention. To strengthen family planning in the experimental villages, an NGO specializing in street theatre and puppet shows was hired to provide information about contraceptive methods, couple communication and decision making related to family planning and women's reproductive rights. Public and private providers posted signs announcing that they offered various family planning methods. Wall paintings in public areas also informed village residents about the availability of a range of contraceptive methods, and providers conducted health fairs in villages. Providers were trained to offer family planning information and services which took into consideration gender power dynamics. No individual contraceptive method was stressed in the messages and information. The second component of the intervention introduced a relatively new method, the Standard Days Method® , to public health centers run by the Ministry of Health and sub-centers run by NGOs. SDM is appropriate for women with cycles that usually range 26-32 days. It identifies days 8 to 19 of the menstrual cycle as the fertile window, i.e., the days when pregnancy is most likely. To prevent pregnancy, the couple avoids unprotected intercourse during the 12-day fertile window, which is identified by using a visual aid representing the menstrual cycle, a color-coded string of beads called Cyclebeads®. SDM efficacy rates, established in a clinical trial, are comparable to those of male condoms, the failure rate being less than 4 per 100 women years of correct use. Adoption of this method requires agreement by the couple rather than the wife alone. Providers were trained in SDM counseling; service delivery points were supplied with Cyclebeads®, the visual tool that supports correct use of SDM, and with simple leaflets presenting the method as a new family planning option in the context of informed choice. Anganwadi workers and village animators were also trained to offer information on family planning including SDM, and to provide SDM. The community health workers in particular were encouraged to reach out to men and couples. The intervention was not designed to address women's empowerment broadly, but its emphasis on empowering women to know about and use family planning, and its promotion of couples' joint decision-making concerning family planning use, were relevant to women's empowerment in the intervention site. Furthermore, given that SDM is a couple method, and that it was being introduced as a new method of family planning in the intervention area, special care was taken to ensure that men were reached through the IEC [information, education and communication] efforts. In the control villages in Burmu, women had access to family planning through regular service channels; however, no special information was provided about availability of services nor was SDM offered in those villages. --- Data Married women ages 15-49 who lived in these blocks had an equal opportunity to participate in the study. A research firm in New Dehli conducted the baseline and endline community surveys. The former was conducted in both blocks in late 2004-early 2005, three months before the intervention started in Ormanjhi, and the latter was conducted after the intervention had been in place for close to three years . The pretest and posttest surveys were independent of each other. To minimize the number of randomly selected households lost due to addresses not found, the final sampling frame was obtained after thorough physical inspection of addresses. All married women of reproductive age were eligible to be interviewed within each selected household. Up to three repeat visits were undertaken early in the morning or late in the evening, to minimize the number of respondents not found at home. Women who had no need for family planning, because they desired more children soon or were not at risk of pregnancy, were excluded from the analyses for this study, which focused directly on women with a need for contraception. Also, given the very small number of followers of religions other than Hinduism and Islam, only Muslim women were selected for comparison with Hindu women, who were the majority in the sample. --- Variables Women's Decision-making Power. In this study, women's decision-making power was measured though normative beliefs concerning whether women should be involved in various household decisions. The questionnaire included a 5-item question on household decision-making, consistent with questions in the DHS: "In a couple, who do you think should have the greater say in each of the following decisions: the husband, the wife, or both equally?" The decisions were: making large household purchases; making small daily household purchases; deciding what to do with the money she earns for her work; deciding when to visit family, friends, or relatives; and deciding how many children to have and when to have them. The few cases choosing the response options "Don't know/depends" were excluded from analysis. Implementing the participation model in the measurement of normative beliefs, 1 point was assigned to Wife and Both equally, and 0 points were assigned to Husband only. Need for Contraception. To determine eligibility for inclusion in the analysis, a score of 1 was assigned to: pregnant women who had wanted their last child later or had not wanted more children at all, nonpregnant women who were using any family planning method, and nonpregnant women who were not using family planning despite their not wanting a child in the next two years and being at risk of pregnancy . A score of 0 was assigned if a pregnant woman said that the child was intended, or a nonpregnant woman was not using family planning because she wanted to have children in the next two years or was not at risk of pregnancy . Met Need. A score of 1 was assigned to the woman in need of contraception if she said that she was using any modern family planning method and a 0 if she was pregnant, was using a traditional method, or was using no method. Age. The woman's age was calculated considering her birth date and the date of the interview. Two questions were asked: "In which month and year were you born?" and "What age did you reach in your last birthday?" Inconsistencies were corrected where possible. Children. Women were asked how many living children they had. Literacy. Women were asked, "Have you ever gone to school?" and, "What was the highest year of studies you attained?" Women who responded "Primary instruction" or less were given a card with a sentence and asked to read it. To translate educational attainment into a single score, the two variables were combined and produced the following scale: 0 = unable to read, whether the woman had formal education or not and 1 = reads part or all the sentence and/or has primary, secondary, or higher education . Work. The woman was asked whether she worked at the time of the interview and whether she had worked in the past 12 months. Her responses were coded 0 = did not work and 1 = worked and/or is working now, regardless of whether she received payment or not, and the type of payment. Religion. Women were asked whether they were Hindu, Muslim, Christian, Sikh, Jain, or Buddhist. Muslims received a score of 0, Hindus a score of 1, and the others a missing value. Listening to Radio. Women were asked if they listened to radio and frequency of radio listening: never , less than once a week , at least once a week , or almost every day . Watching Television. Women were asked if they watched television and frequency of television watching: never , less than once a week , at least once a week , or almost every day . Visited Health Center. Women who had visited a health center in the past 12 months were given a score of 1. Those who did not received a 0. Received Visit. Similar scores were assigned to women who in the past 12 months had received the visit of anyone who talked about family planning. --- Analyses Chi-square and the t-test for independent samples were employed to analyze simple pre-post data, and the following generalized linear model evaluated the effects of the intervention on women's decision-making power: λ ¼ β 0 þ β i δπ ð Þ þ β j ξ k h i þ εð1Þ where λ is the decision-making power measure and β and ε are the regression coefficients and the error term. The main target of the analysis was the δ x π interactive effect on decision-making power, where δ is a fixed-effects treatment factor defined as 1 = Burmu pretest, 2 = Ormanjhi pretest, 3 = Burmu posttest, and 4 = Ormanjhi posttest and π is a fixed-effects literacy factor with two levels . The other component of Equation 1, ξ k , is a set of main-effect covariates that includes age, religion, children, radio, TV, and work. A similar model, but with the focus on the 3-way treatment x literacy x decisionmaking power interaction and having met need for contraception as the dependent variable was employed to assess the moderating roles of women's literacy and decision-making power in the intervention process: γ ¼ β 0 þ β i δπλ ð Þ þ β j ξ k h i þ εð2Þ Results Table 1 presents descriptive statistics which reveal a moderately high level of met need in rural Jharkhand as well as an increase from pretest to posttest in the intervention area, Ormanjhi, but not in Burmu, the control area. Regarding the covariates, the results show that literacy, watching TV, proportion of Muslims, and family planning visits received, all increased from pretest to posttest in both groups and thus these changes can be regarded as parts of maturation processes in rural Jharkhand. The proportion of women working increased in the control but not in the intervention site; since the difference between blocks at the pretest was significant , this dynamic may be understood as the women from Burmu catching up with their neighbors. The fact that the proportion of Muslims was greater in Burmu at the pretest may be relevant in this respect, considering their general employment level . Visits to health centers significantly decreased in Ormanjhi but not in Burmu. The results shown in Figure 1, which refer to the combined intervention and control samples, suggest that a practical, egalitarian model of domestic decision-making prevails in rural Jharkhand: while making decisions about small purchases appears to be within women's autonomous decision-making purview, the other four are ones in which joint decisions with the husband are expected. Table 2 and Figure 2 present results from generalized linear models which had women's decision-making power scores as dependent variables. Neither decision-making power to make large purchases nor to make small purchases was affected by the intervention. Regarding the role of literacy as a moderating variable, the pairwise comparisons indicate that in the intervention area, women's power to decide on having children increased among illiterate women but not among literate women. The intervention was associated with increased decision-making power for women regarding use of wife's earnings and making visitsfor both illiterate and literate women. Regarding the other covariates, younger women held normative beliefs suggestive of greater power in decisions on having children; number of living children positively influenced decision-making power for small purchases and disposal of earnings; women who listened to radio never or infrequently tended to have more decision-making power regarding making visits; women who worked were more likely to say that they should participate in decisions on large purchases for the home and regarding disposal of earnings; and those who were visited by a community health worker to talk about family planning considered themselves more empowered to make decisions on visiting family and friends or having children. Nonresponse was greater for the question on use of wife's earning than for the other questions, because women who did not work for pay tended to choose "no response" in answer to this question. To avoid a loss of cases, a sum of women's decision-making power scores was computed without the question on disposition of women's earnings. With a total number of four items dichotomously scored as 0 or 1, the summed score could range from 0 through 4. The internal-consistency reliability of the sum of the four items' scores was not fully satisfactory . That is, women who were powerful in one area were not necessarily powerful in a different area. That the decision-making power scale αs were weaker than reported in an analysis at the national level for India can be explained by the fact that the current analysis referred to normative beliefs and used the participation model of decision-making while the study at the national level entailed self-reported behavior using a control model of decision-making which defines a hierarchy of women's self-perceived decision-making power that ranges from lack of participation to joint decisions with the husband to women's autonomous decision-making [11]. For the analysis of the function of women's decision-making power in the intervention to increase met need for contraception, the sum was dichotomized at the median, yielding a group of women with relatively less decision-making power and a group with relatively more decision-making power . Table 3 and Figure 3 present results from the generalized linear model with met need as the dependent variable. A maturation effect is suggested by the adjusted means: the eight subgroups presented increased met need levels from pretest to posttest. However, the rates of change differed. It is evident that the intervention did not improve met need among women with relatively less decision-making power who were illiterate, but the change for women with relatively more power who were illiterate was greater in the intervention area Ormanjhi, than in the control area Burmu. Similarly, the steepness of the curves for literate women with relatively less decision-making power was virtually the same at the two sites, while it was steeper in the intervention area than the control area for literate women with relatively more decision-making power. Thus, the findings strongly suggest that, regardless of Table 2 Odds ratios from generalized linear models predicting women's decision-making power from treatment x literacy interaction and covariates, per decision-making item *p < .05, **p < .01, ***p < .001. whether the woman was literate or illiterate, she was more likely to have her need for family planning met at the posttest in the intervention than the control site if she had relatively more decision-making power than if she had relatively less decision-making power. As for differences observed between sites at the baseline, they had opposite signs depending on whether the women were literate or illiterate. It can be seen in Figure 3 that Burmu presented slightly greater levels of met need than Ormanjhi among illiterate women at the pretest, while the opposite is clearly visible among literate women. Concerning the covariates, religion had the strongest effect on met need. Hindus were nearly five times as likely as Muslims to have a met need. Older women who had more children and were frequent TV viewers were more likely to have a met need than younger women who had fewer children and were less frequent TV viewers. --- Discussion That women of the intervention site were more empowered in terms of decision-making over the three year period of the study than those of the control site is well established by the research results: the differences in normative beliefs about wife's participation in decision-making between women in Burmu and Ormanjhi at the baseline were all non-significant, while significant pretest-posttest changes only occurred in the intervention site, Ormanjhi. These findings suggest that the intervention had an effect on women's power in decision-making and that alternative interpretations related to selection bias, local history, maturation, regression towards the mean, attrition, and other threats to the internal validity of quasi-experimental designs [23] can be reasonably ruled out. It is generally accepted that a nonequivalent control group quasi-experiment with pretest and posttest approximates answers from randomized experiments [24], especially if it tackles the main limitations of the research design: selection bias and attrition [25]. In this study, attrition was avoided through independent sampling of respondents at pretest and posttest. Selection bias was ameliorated by having a comparison group living close to the intervention group. Burmu and Ormanjhi are contiguous blocks within the district of Ranchi in the state of Jharkhand that share the same ecological, socioeconomic and cultural configurations. Their comparability was made evident by the finding of similar pretest-posttest changes in literacy, watching TV, proportion of Muslims, and visits received, although the observed increase in women working in Burmu but not in Ormanjhi and the differences in the proportion of Muslims and women working at the baseline suggest that some historical processes may have differentiated the groups. The observed changes, however, did not involve all the decision areas and sub-groups of women. Both literate and illiterate women of the intervention site increased their levels of decision-making power with respect to disposition of wife's earnings and visits to relatives and friends but not regarding large or small purchases. A Figure 2 Adjusted power means for literate and illiterate women at pretest and posttest, per site and decision area, and significant changes . possible explanation for this finding is that the nature of the intervention affected dimensions of women's empowerment differently. It is likely that the messages used in the intervention were more relevant to women's decisionmaking about their own earnings and about freedom of movement, namely to visit family and friends, than about specific material possessions, namely making purchases. A revealing finding of the study in this respect is that decision-making power regarding making visits and having children was significantly greater among women who had received family planning visits than those who had not. The finding that literate women, whose normative beliefs entailing earnings and visits were changed by the intervention, did not seem to have their beliefs changed about their decisions on having children, can be explained by a ceiling effect. Whereas normative beliefs regarding fertility decision-making had considerable room for change among illiterate women in Ormanjhi, this was not the case for literate women. Regarding the specific causes of the observed changes in women's decision-making power, it seems unlikely that their interaction with providers at service delivery points in the intervention group was the only cause of the observed empowerment. The women who visited a health center during the period of the study did not present greater decision-making power levels than those who did not. But the decision-making power levels of those who received a visit of someone who talked about family planning did increase, which suggests a direct influence of the intervention's outreach efforts. Normative beliefs concerning visits to family and friends and having children were likely changed by the visits received at home. Outreach workers were trained to emphasize the convenience of traveling to a health center or sub-center and to discuss decision-making about fertility with women and men, which would explain why decisionmaking power changes were observed in these decision areas and not those related to purchases. It is not possible to determine whether the influence of the intervention was limited to the 27% of women in the intervention site who were visited during the study period. It is also possible that the women who were visited in turn influenced some of their neighbors. If this was the case, the stable unit treatment value assumption of the Rubin model, which establishes that the treatment obtained by one individual has no effect on the set of potential outcomes of any other individual, was violated [26,27]. Nonetheless, while it is very likely that some women in Ormanjhi influenced by the intervention in turn influenced others, the likelihood that women in Ormanjhi affected the outcomes of women in Burmu is much less likely. As for the street theatre and puppet shows, these may have also contributed to the observed empowerment. Since outreach actions are frequently implemented by family planning programs, an important practical question refers to the extent to which such actions by themselves, without community-level media , are able to produce the effective empowerment seen in this analysis. Despite these questions, the research showed that women's empowerment in several household decision-making areas can be enhanced by family planning interventions. Whether the empowerment contributed to meeting the needs of the women for contraception is more difficult to establish because all the subgroups of women at both control and intervention sites underwent positive changes regarding met need for contraception, which suggests an underlying shift in Jharkhand society. Compelling evidence suggesting a causal link with empowerment is the observation that the woman was more likely to exhibit an enhanced met need at the posttest in the intervention than the control site if she was literate and had relatively more power in decisionmaking, than if she was literate and had relatively less power in decision-making. The intervention might have enhanced met need by moving some literate women from the relatively less decision-making power to the relatively more decision-making power category. A typical ethical limitation of the study was the assignment of one of blocks to the control condition, which implied that Burmu women lacked the opportunity to receive the enriched information and services received by Ormanjhi women. The state of Jharkhand, however, was expected to scale-up the intervention. --- Conclusions Theoretical, methodological, and practical lessons can be drawn from this study. First, whereas most research has emphasized the cultural and social determinants of women's empowerment, the analysis in this paper demonstrates that women's empowerment not only is determined by such macro social forces but can be influenced by family planning programs that include attention to gender dynamics. While the intervention in Jharkhand did not have stated women's empowerment objectives, it did include attention to gender dynamics, and its materials and messages incorporated gender transformative messages related to family planning use. The specific paths through which programs can empower women and affect contraceptive use that were discovered in this study suggest that outreach efforts by themselves are not enough to empower women and increase contraceptive use. Visits received by women also increased in Burmu. The Burmu outreach workers, however, were not encouraged to stress the gender issues that were addressed in Ormanjhi. It is of course important to note that these findings pertain to Northern India in areas of a state which already had a moderately high level of met need for contraception when the intervention started. Whether similar results are found in different contexts, including in Sub-Saharan Africa where contraceptive use generally remains low, and the links between empowerment and family planning may be more tenuous [28], only can be established by replications of the study outside India. Still, the findings of this analysis support the hypothesis that family planning programs that address gender are contributing to the empowerment of women. --- Competing interests The authors declare that they have no competing interests. Authors' contributions FRL designed the study, participated in the construction of data-collection instruments, conducted the data analyses, and drafted the manuscript. RL participated in the study design and coordination, participated in the construction of data-collection instruments, supervised the fieldwork and helped to draft the manuscript. IS participated in the study design and coordination, participated in the construction of data-collection instruments, and helped to draf the manuscript. RS participated in the construction of data-collection instruments and conducted the intervention and datacollection phases of the study. VJ conceived of the study, participated in the study design and helped to draft the manuscript. All the authors read and approved the manuscript. --- Figure 3 Adjusted met need means for women with relatively more power and women with relatively less power at pretest and posttest, per site and literacy level, and significance of changes .

Background: Virtually all the evidence on the relationship between women's empowerment and use of contraception comes from cross-sectional studies that have emphasized macrosocial factors. This analysis tested whether literate and illiterate women are empowered by an intervention designed to provide information addressing technical and gender concerns and expand contraceptive choice, and evaluated the effects of women's decision-making power on contraceptive behavior. Methods: The data came from a three-year quasi-experiment conducted in two comparable, yet not equivalent, rural blocks in Jharkhand, India. At the intervention block, a new contraceptive method was introduced at Ministry of Health health centers, providers were trained to offer family planning information and services which took into consideration gender power dynamics, and promotional messages and information about contraception were disseminated community-wide. Married women ages 15-49 who lived in the intervention and control blocks were sampled and interviewed before and after the intervention by a professional research firm. Data analyses included generalized linear models with interactions and covariate control. Results: Women's normative beliefs concerning wives' power in decisions regarding money earned and visits to relatives and friends vis-à-vis their husbands' power were increased by the intervention; similar was the case among illiterate, but not literate, women regarding decisions related to childbearing. Concerning met need for contraception, the change for women with relatively more power who were illiterate was greater in the intervention than in the control area.The findings suggest that women were empowered by outreach visits that addressed gender dynamics and that their empowerment contributed to their met need for contraception. Generalizations to other settings, however, may be limited by cultural differences.

However, increasing consent is a daunting task. It has taken 20 years of concerted effort to increase family consent to solid organ donation from 40% to 60%. 3,5 The challenge for tissue donation is greater because it is poorly understood by the public. A survey of families who donated tissues indicated only one-half distinguished tissue donation from organ donation. 6 Moreover, the public is generally unaware of the details of tissue donation, such as the preparation and distribution process, which can involve for-profit companies. 4 Previous studies examining tissue donation have been limited by scope, sample size, and analytic method. Most have small sample sizes and exclusively focus on families who consented to donation. [6][7][8][9] Furthermore, only one study used a multivariate model to identify predictors of family consent. 7 This study is the first to examine a large, national sample of potential tissue donors. The goals of this study were to describe the process and content of the requests for tissue donation and to identify the factors associated with family consent. --- METHODS Tissue Bank Sample Sixteen tissue banks representing a spectrum of tissue procurement organizations across the United States were sampled. In addition, two organizations procured tissues nationally, whereas the remaining 14 procured tissues from within proscribed geographic areas. Specifically, five organizations were based in the Southwest, four in the Southeast, three in the Midwest, three in the Northeast, and one in the Northwest. Finally, 6 hired and trained their own requesters, whereas 10 contracted at least a percentage of their requests to services that provide full-or part-time telephone coverage for the purpose of contacting and requesting tissue donation from the families of tissue-eligible patients. Data collection at each tissue bank involved the identification of families from whom tissue donation was requested; self-administered surveys completed by tissue bank requesters immediately after discussing donation with the families; and audiotaped telephone interviews with the family decision makers who participated in the donation decisions. Data were collected from February 2003 through 2006. --- Tissue Bank Requester Sample Eligible requesters were defined as those tissue bank staff who discussed donation with family members. Each case was associated with at least one staff requester; 13 cases included more than one requester. Requesters completed a self-administered survey at the completion of each request for tissue donation. A total of 226 tissue requesters participated in the study, and 99.1% of cases had a completed tissue requester survey. Because of the high volume of tissue donation requests made each month, tissue banks were randomized to specific data collection days. During these days, requesters completed a brief, selfadministered survey after each request. --- Family Decision Maker Sample FDM were identified using the tissue banks' records. Two months after the death of adult patients and 3 months after the death of pediatric patients, a letter explaining the purpose and methods of the study were sent to a random sample of families from whom requests were made. Past research examining family consent to organ donation has found that this time period allows families to grieve without compromising recall of the events surrounding the patient's death. 10 Ten days after the letters were mailed, families were contacted through telephone and invited to participate in the study. We interviewed 62.5% of individuals contacted about tissue donation , of whom 1,008 donated tissues and 463 did not. Because the interviews relied on FDMs' memory of the donation request, only those reporting good recollection of the donation conversation were included in these analyses; 53 cases in which respondents reported poor recall were excluded for a final sample of 1,418 cases. The family interview consisted of two parts. The first contained a series of structured and semistructured questions that systematically collected data on the quality of care the patient received at the hospital, interactions with tissue bank requestors, the donation decision, and the reasons for that decision. The second section measured family members' attitudes about and knowledge of tissue donation and transplantation. The interview was a modified version of a well-validated interview used extensively with families that were asked to donate solid organs. 10 All interviews were audio-taped and transcribed verbatim; interviews lasted 45 minutes to 60 minutes. The study was approved by the appropriate institutional review board, and verbal informed consent was obtained from all participants. --- Outcomes The primary outcome measure was the decision whether to donate tissue . Consent data obtained from FDMs during the interviews were verified through tissue bank reports. Basic demographic information, including sex, ethnicity, and age, were collected for all subjects. In addition data such as income, education, and marital status were obtained for FDMs. Factors thought to influence family consent were assessed during the interviews and are described later. For all scales, higher scores indicate higher levels of the measured variable. Family members rated the quality of care the patient received while in the hospital, the quality of communications with the health care providers, and health care provider's care and concern using three 5-point Likert Scales. In addition, FDMs indicated who first approached them about the option of tissue donation and their level of comfort with that person. FDMs also reported their initial response to the donation request and level of surprise at being asked to donate. Both variables were measured on a 7-point Likert Scale. Decision makers provided the rationale behind their donation decision. Similarly, tissue requesters reported their perceptions of families' initial response to the request. Congruence between requesters' and family members' reports of FDMs' initial response was also ascertained through a paired comparison yielding a variable scored as either concordant or discordant. The relationship between this variable and consent was then assessed to determine whether requesters' ability to accurately gauge FDMs' reactions to the donation request influenced the donation decision. Two measures were used to examine FDMs' understanding of donation: the sum of correct responses to seven tissue donation knowledge questions; and correctly identifying the tissue requester's organizational affiliation. A 14-item, 5-point Likert Scale of attitudes toward donation and transplantation used and validated in several studies of organ donation was modified for tissue donation and administered to FDMs. 10 Scores are summative and range from 14 to 70. Family respondents also recalled the donation-related topics discussed with the tissue requester, the type of tissue requested, and the total time spent in discussion with the requester. FDMs' level of satisfaction with their decision was obtained using the Decisional Regret Scale, a standard 5-item, 5-point Likert Scale. 11 To contextualize this measure, an openended question allowed respondents to indicate what, if anything, they would have changed about their decision. By using three 7-point Likert questions, requesters rated their comfort speaking with the family and answering the family's questions as well as their overall satisfaction with the process. Berlo's Source Credibility Scale, a 5-item, 7-point Likert Scale, assessed FDMs' perceptions of requesters' training, experience, qualifications, skill, and knowledge. Scores on the scale were summed and ranged from 5 to 35. 12 Family respondents also indicated whether the tissue requester asked if they had any questions and, if so, how satisfied they were with the requesters' answers. Satisfaction was rated using a 7-point Likert Scale of agreement. Two 5-point Likert Scales assessed respondents' perception of requesters' display of care and concern and the quality of the requester's communications during the donation conversation. Families also completed a 14-item, 7-point Relational Communication Scale. 13 Item responses were based on level of agreement and were summed to range from 14 to 98. --- Statistical Analysis Associations between the independent variables and family consent were performed using the Fisher exact test or the χ 2 test, for variables with two categories or three or more categories, respectively. The Mann-Whitney U test was used to examine ordinal-level variables; the Student's t test was used for interval-or ratio-level variables. Because the sample was skewed toward FDMs who consented to tissue donation, a weighted analysis was conducted. No significant differences were found between the analysis weighted for sampling bias and the unweighted analysis; the results of the unweighted analysis are presented here. We performed a log-linear regression to describe the relationship between family consent to tissue donation and factors describing the consent process. A theoretical model of donation decision making was used to determine the constituents of each factor. 10 First, the study's independent variables were grouped to create six factors, each representing a different conceptual domain . Second, variables that were significantly associated with donation within each of the domains using bivariate techniques were retained. Third, separate logistic regression analyses were performed on each of the domains using consent as the outcome. This process created six variables representing the estimated probability of consent for each conceptual domain. A cut point of 0.5 was chosen to transform each of these variables into a dichotomous variable. Finally, a log-linear regression was conducted to analyze the interrelationships between the six dichotomous variables and consent. The analyses were performed using SPSS 16.0 for Microsoft Windows and SAS . --- RESULTS --- Sociodemographics and Consent Patients' mean age was 56.1 years ; 83.3% were white and 68.6% were men. Younger, white patients were more likely to become donors than older, nonwhite patients. FDMs were predominantly white and women with a mean age of 51.3 years . Although most decision makers were the spouse of the deceased, others were adult children , parents , siblings , or other relatives, significant others, or legal guardians . Fifty-one percent of family members reported their religious affiliation as Protestant and 22.8% as Catholic. Nearly three quarters of the families had yearly incomes > $30,000 and 14.0 years of education. A minority of family members held health-related occupations . Although only 66.5% of families had signed a donor card, nearly all stated a willingness to donate their own tissue. FDMs who donated were more likely to be white and Catholic , to have an annual income > $59,000 , and more years of education , compared with FDMs who refused donation. Donor families, compared with nondonors, were also more likely to have signed a donor card . Donor families cited implicit or explicit knowledge that the patient wanted to be a donor , families' desire to help others , and families' general inclination toward donation as reasons for their final decision. Of the families who donated, 150 donated with conditions-donating some, but not all tissues requested. Reasons for families' decision to donate conditionally included concern for potential disfigurement or mutilation of the patient's body , assessment of the patient's eligibility , an emotional attachment to a certain body part , family compromise , and the implicit or explicit knowledge of the patient's wishes . The most common reason for refusing donation was patient and family fatigue . Other reasons why nondonors refused included implicit or explicit knowledge that the patient did not want to donate , concern over potential disfigurement , a belief that the patient was not eligible to donate , and feeling uninformed about tissue donation . Overall, FDMs were generally satisfied with their donation decisions , although families refusing donation exhibited greater regret than did donating families . Tissue requesters were predominately women and white . The average age of a requester was 34.2 years and most had at least a college education . On average, requesters had 1.5 years of job experience ; half of the requesters held degrees in a health-related field. Requesters' sociodemographics were not significantly associated with consent. --- Families' Initial Reactions to the Request and Consent Thirty-three percent of families were surprised to be asked about tissue donation. Consistent with research examining families asked to donate solid organs, 11 family members who refused donation reported higher levels of surprise than consenting family members . Although many families were surprised at the request, the majority of family members reported being in favor of donation when first asked; 24.3% of families were initially unsure about donation and 15.5% were unfavorable. Tissue requesters correctly judged families' initial reactions to the donation request 64.4% of the time. Family consent was more likely when the tissue requester correctly assessed the FDM's reaction to the request than when the assessment was discordant . --- Attitudes, Knowledge, and Consent Donor families reported significantly more favorable attitudes toward donation than did nondonor families . Conversely, nondonors were more likely to feel "squeamish" about the idea of tissue donation , to agree that donor families should share in any monies made from donated tissue , and to agree that families should have some degree of control over how donated tissue is used than donors. Few families were in favor of using donated tissue for cosmetic purposes or involving for-profit companies in the processing and distribution of donated tissue . On average, respondents' knowledge about tissue donation was moderately high, with a mean of 5.4 correct answers out of 7 . However, donating families displayed more knowledge of tissue donation than families declining donation . Nondonors commonly and incorrectly thought that tissue donors could not have open casket funerals and that a signed donor card or a license marked "donor" did not apply to tissues or corneas . In addition, although more than half of all FDMs were able to identify the organizational affiliation of the tissue requester by name or type of organization, significantly more donors than nondonors did so correctly . --- Request Process Factors and Consent Past research in organ donation has indicated that family members' perceptions of the care and concern exhibited by the patient's health care team affect the decision to donate. 10 The use of the telephone in requests for tissue donation is standard practice. In this sample, 61.0% of the requests for donation were initiated by telephone, and all requests were followed up and completed by telephone. Requests initiated by telephone were significantly more likely to result in refusals than when families were given advance notice in person about the request by hospital personnel . Families who consented to donation were most likely to have their first substantive conversation about tissue donation with a tissue requester ; nondonor families were most likely to have the issue first discussed with clergy, social workers, coroners, funeral directors, or other unspecified hospital personnel . A similar discussion pattern was observed in requests for organ donation. 10 Consenting families also reported greater comfort with requesters than did families who refused donation . In addition, families who knew of the patient's desire to become a donor or who thought the patient would have wanted to donate were more likely to grant consent than FDMs who did not . --- Quality of Communication and Consent Discussions with families about donation ranged from 1 minute to 240 minutes, with a median length of 20 minutes . Consenting families spent considerably more time discussing donation than did families who refused consent . Family members consenting to donation discussed significantly more topics with requesters than did family members who refused . For donors, conversations about donation most often included discussions of the patients' medical history and the use of donated tissue to save lives and treat disease . The least commonly discussed topics for donors included the involvement of for-profit organizations, families' religious or spiritual beliefs, and the tissue requesters' point of view on donation. For nondonors, the most frequently discussed topics were the time to make the decision and the patient's donation wishes, whereas the processing and storage of donated tissue and the involvement of for-profit companies were discussed least. Almost three quarters of respondents recalled being asked whether they had any questions concerning donation. Families consenting to donation were more likely to be solicited for questions than were families who refused . Donor families also rated the communication with the requester of higher quality than nondonors . Donors thought that the requester was more caring and concerned , more credible , and possessed stronger communication skills than did nondonors. --- Multivariable Analysis of Determinants of Consent Six key aspects of the tissue request process known to affect consent were examined to determine how interrelations among the factors affected consent in this population. Results revealed that all six factors were directly related to consent . The strongest association was found between the tissue donation-related topics discussed with FDMs. Families who discussed key issues, such as costs associated with donation, the time available to make the donation decision, the impact of donation on funeral arrangements, and the ability to have an open casket funeral, were 17 times more likely to donate . The quality of communication was also associated with consent, as were donation-related attitudes and knowledge, the role of the individual who raised the issue of donation, and patient and family sociodemographics. Significant associations among the individual factors were also found. --- DISCUSSION The results of this study reveal that family consent to tissue donation is affected by many of the same factors that influence consent to organ donation. For example, the number of donation-related topics discussed with the FDM and the family member's degree of surprise at the request predict family consent. In addition, the role of the person who first approached the family about donation , the amount of time spent discussing donation, and the FDM's attitudes toward and knowledge of tissue donation were significantly associated with families' donation decision. These findings are mirrored in work on organ donation by Siminoff et al. and others. 10,[14][15][16] There are several significant differences between organ and tissue donation. First, tissue donation can take place many hours after the patient dies. This has encouraged the use of telephone requesting. Indeed, much attention has been paid to the use of the telephone in requests for tissue donation. 4,6,[17][18][19] We found that FDMs were more likely to consent when approached first in person rather than by telephone; this is consistent with past research examining the use of the telephone in requests for corneal donation. [17][18][19] However, the large number of individuals eligible for tissue donation coupled with the substantial numbers of eligible deaths occurring outside hospitals, make in-person requests often impracticable. Furthermore, Wilson et al. 6 found that >90% of families were satisfied with the telephone request process; they and others find that telephone requests can be acceptable for requesting tissue donation. 4,6,[17][18][19] Tissues and organs are also differentiated because tissues are frequently processed and go through several levels of handling before transplantation. As such, some tissue come to resemble a commercial product much more than we might like, given that these are tissue donated freely by deceased individuals. Moreover, for-profit companies may ultimately sell some of the donated tissue, a fact only 12% of families knew. Nearly a quarter of family members interviewed were also unaware that a signed donor card was, in effect, an agreement to donate tissues and corneas as well as organs. Even fewer family members understood that these documents were legally binding and the respondents often confused tissues and organs. Data from this study and others show that families frequently know little about tissue donation, confuse tissue and organ donation, and are surprised to receive these requests. It is notable that the element of surprise significantly decreases consent rates. 4,6,9 Clearly, the American public still lacks basic knowledge of the differences between organ and tissue donation. The study also demonstrates that information that might be considered critical to informed consent-the uses, procurement, and distribution of tissue-go largely undisclosed. Indeed, both the Department of Health and Human Services' Office of the Inspector General 20 and the National Kidney Foundations' National Donor Family Council 21 consider these topics critical to families' decision-making process. The Office of the Inspector General 20 also recommends that, at minimum, families should be provided with, among other things, a copy of the consent form. Nonetheless, only 28.8% of consenting families in this study were asked if they would like a copy of the consent form and, of those responding affirmatively, only 32.7% reported they received one. Given these findings, we propose the following suggestions for improving consent to tissue donation in the United States. First, a general education campaign, targeted primarily to minority communities, is needed to inform the American populace about the nature and benefits of tissue donation. A simple solution would be to distribute pamphlets about tissue donation to all families and patients in hospitals. Our data indicate that preparing families to receive a telephone request can add to the likelihood of a successful request. However, research promoting family consent to organ donation has shown that requesters spending more time with families are more likely to obtain consent. 16 Although requesting tissue donation by telephone may be cost effective, consent rates may improve if more requests for tissue were made in person, like those for organs. Tissue banks should also routinely ensure that all families are provided information on the processing and distribution of donated tissue and of the involvement of for-profit companies. Because families stated these topics would be influential to their choices about donation, they should be included in discussions about donation to provide families with the information needed to make informed donation decisions. Moreover, tissue banks must be vigilant in distributing informed consent documentation to all families who consent to donation. Informed consent is a standard element of medical care and should not be overlooked in this context. Although this is the largest, most comprehensive examination of the process of requesting tissue donation in the United States, it has limitations. Few minority respondents were of Hispanic or Asian decent; most were African American . However, this study's demographic profile is similar to that of the US population. 22 Future research should use stratified sampling techniques to provide a more in-depth analysis of Asian and Hispanic attitudes and behaviors regarding tissue donation. In addition, because the interviews were held 2 months to 4 months after the patient's death, there is a chance that some FDMs' failed to accurately recall these events. Recall error is a common threat to validity in retrospective studies but was minimized in this study by careful cognitive interviewing techniques and the vividness of the events. Finally, although the sample was large for a study of this type, there was a significantly higher participation rate for families consenting to donation than for those who refused . Nondonors' participation in similar studies examining organ and tissue donation ranges from a low of 21% to a high of 60%. 4,23,24 $watermark-text $watermark-text $watermark-text --- Supplementary Material Refer to Web version on PubMed Central for supplementary material.

Background-Family consent to tissue donation currently falls well below that for solid organ donation. Moreover, research suggests that Americans have limited understanding of tissue donation, an activity largely overshadowed by the more publicly visible organ donation. This research sought to identify determinants of families' consent to tissue donation.-Data collection included a brief, self-administered survey completed by tissue bank staff and telephone interviews with family for a sample of tissue donor-eligible deaths reported to a national sample of 16 United States Tissue banks from 2003 to 2006. Family members (N = 1,418) and tissue bank staff (n = 226) involved in the tissue donation decision participated in the research. Results-The families of 1,015 (71.6%) tissue donor-eligible patients consented to tissue donation; 403 (26.8%) families refused. Results revealed a multitude of associations between study variables and the donation decision. A subsequent multivariable analysis identified determinants of family consent to tissue donation, including the discussion of key donation-related issues during the request (adjusted odds ratio [AOR], 17.22; 95% confidence interval [CI], 11.61-25.54), the quality of communication during the request (AOR, 12.39; CI, 7.76 -20.02), families' tissue donation attitudes and knowledge (AOR, 10.01; CI, 6.47-15.50), families' initial reactions to the request (AOR, 7.86; CI, 5.13-12.05), families' advance notice of the request (AOR, 3.95; CI, 2.41-6.46), and patient and family sociodemographic characteristics (AOR, 3.32; CI, 2.01-5.48). Conclusions-Family consent to tissue donation is affected by many of the same variables influencing consent to solid organ donation. Recommendations for practice are provided.

There is a large unmet need for family interventions for homeless adolescents. Family interventions with homeless adolescents are, however, a relatively unexplored area even though their family relationships are often characterized by conflict, inappropriate problem-solving and poor communication . While research demonstrates that newly homeless adolescents will return home , those who have more troubled relationships with their parents have a more difficult time remaining at home . Getting homeless adolescents and their families to participate in family interventions, crucial for intervention effectiveness, is a known challenge , especially for at-risk adolescents that often have low completion rates . This may be due to barriers to participating in terms of access, relevance, and time commitment . One strategy often used for increasing access for these challenged families is to provide family interventions during key life transitions for adolescents when families naturally partake in their children's activities and, thus, might be receptive to participating in an intervention that could help with the transition . For homeless adolescents, a key life transition time may be when they first leave home, are newly homeless, and may want to reconnect via a family intervention. The success of this strategy, however, has been limited for homeless adolescents . The importance and challenges of participation in interventions have been highlighted as key area for further research in Type 2 translation research to implement and scale up evidencebased interventions by Spoth et al. . Identifying the "key factors" and "best strategies for enhancing participation" are critical to the Implementation Phase of their Translation Science to Population Impact Framework . Mauricio, Gonzales and Sandler have applied the TSci Impact Framework specifically to family interventions to highlight areas for further research. These included the need to have consistent definitions of participation and to define participation in terms of both behavioral and attitudinal/cognitive dimensions. This work provides an overarching frame for our examination of participation in a family intervention for homeless adolescents. Retention, as indicated by attendance over time, is one aspect of the behavioral dimension while treatment satisfaction is one attitudinal/cognitive dimension of engagement . Like other studies of intervention implementation, retention, the behavioral dimension, was operationalized as the completion of a substantial portion of the intervention . Engagement, the attitudinal/cognitive dimension, was operationalized as satisfaction with the therapeutic alliance between the families and the intervention facilitators-it is treatment satisfaction . A personal bond with a treatment provider positively impacts family engagement in therapeutic interventions . Staudt proposed a framework for thinking about family engagement that is closely aligned with the conceptual framework of Mauricio et al. in that it takes into account the attitudinal component of engagement, in addition to the behavior of the facilitator. The attitudinal component of engagement is related to five dimensions: intervention relevance and acceptability, daily stresses, external barriers to the intervention, cognitions about the intervention, and the therapeutic alliance. We focus on one of these dimensions, the therapeutic alliance, because this dimension is an indicator of a personal bond between the facilitator and family . Many studies assert that better retention is associated with demographics such as higher levels of income and educational attainment, family characteristics such as having more family stress, and intervention characteristics such as trust ; however, findings are mixed and evidence is still emerging relevant to culturally humble ways of engaging participants . Engagement has been associated with demographics such as higher levels of educational attainment and family characteristics such as parental involvement . Interventions having project goals that are consistent with the goals of the parent/adolescent are also linked to better retention . While previous research has emphasized the importance of addressing how to best engage and retain participants in interventions to reach more families and have more widespread intervention uptake, more research defining and examining contextual factors related to participation is needed . Specifically, given high levels of current mental and behavioral problems among homeless adolescents -identified as potential barriers to participation -further understanding how these issues as well as their readiness to change might relate to program participation would be informative. Furthermore, research on how to increase participation in family interventions has often only examined one dimension of participation , and many studies have focused on the behavioral dimension . In this manuscript, we examine the association of demographic, family, and intervention characteristics with two dimensions of participation, retention and treatment satisfaction, in a sample of participants in STRIVE , a family intervention for homeless adolescents and their parents or guardians that has been found to be efficacious in reducing delinquent behaviors, substance use, and sexual risk taking behaviors . We explore how individual characteristics of parents and adolescents and family characteristics relate to the retention and treatment satisfaction of homeless adolescents and their parents or guardians in STRIVE. Directional hypotheses were not made for how these individual and family characteristics would be associated with retention and treatment satisfaction. Our goal is to begin to identify how these factors relate to participation in this underserved population to gain a better understanding of how to better implement family interventions with populations that are not often provided evidence-based family interventions . --- Method --- Procedure From March 2006 to June 2009, newly homeless adolescents and their parents or guardians , who will be referred to as parents going forward, were recruited from homeless adolescent-serving, community-based organizations and from direct recruitment in Los Angeles and San Bernardino Counties to participate in a randomized controlled trial of STRIVE. STRIVE is a brief five session psychoeducational intervention administered to the adolescents and parent together to improve families' problem-solving, emotional regulation and conflict resolution skills. Eligibility criteria were that adolescents were 12 to 17 years old, away from home ) for at least two nights in the past six months, were not away from home for more than six months, and were at a transition point to reconnect with family . In addition, no current abuse or neglect, no active psychosis, or no current substance intoxication could be present. These additional criteria were screened for after informed consent was given during a baseline assessment. Written informed consent/assent to participate was obtained from both parents and adolescents in the study. The Institutional Review Board of the University of California at Los Angeles approved the protocol for this study. After the adolescent and parent assented/consented, they completed baseline, 3, 6 and 12 months follow-up assessments. The face-to-face computerized assessments were conducted by a highly trained and ethnically/racially diverse assessment team. Audio computer-assisted self-interviewing was used for sensitive measures . Eligible families were randomly assigned to either an intervention or control group . Further details on the study design and intervention may be found in Milburn et al. . This manuscript focuses on the baseline responses as they relate to attendance at intervention sessions that were scheduled to take place between the baseline and the 3 months follow-up assessment. Therefore, analyses are conducted on baseline data from the 68 adolescents and parents who were randomized to the intervention condition. --- Sample Parents.-Of the 68 parents in the intervention arm of the study, 88% were female , and the average age was 41 years . 57% reported their race/ethnicity as Hispanic , 21% as Black/African American , and 21% as White/European American . Among the Hispanic parents, 41% were Mexican American . 63% reported being born in the United States , and 60% reported their primary language to be English, followed by Spanish . Parents who reported not being born in the US reported living in the US for 21 years on average . Parents received a mean of 13 years of education . 51% of parents reported previous year total income of less than $25,000, and 63% were currently employed. The majority of parents were either the mother , father , or grandparent of the adolescent in the study. 78% of parents reported their child to currently be living at home, and 4% reported in a shelter. Adolescents.-Of the 68 adolescents in the intervention arm, 78% were female , and the average age was 15 years . 62% of adolescents reported their race/ ethnicity as Hispanic, 16% as Black/African American, and 10% as White/European American. 91% of adolescents were born in the US, and 87% were primarily Englishspeaking. 68% of adolescents had consumed alcohol in their lifetime, 56% had used marijuana, and 26.5% had used any illicit substances other than marijuana . Rates were lower for use in the past three months: alcohol , marijuana , and hard drugs . Analyses of substance use focus on alcohol, marijuana, and hard drugs, due to low reported numbers of use of individual substances other than marijuana. Table 1 summarizes both parent and adolescent demographics. --- Measures Measures reported by both parents and adolescents, by parents only, and by adolescents only are summarized below. Further details on measures are found in supplementary Table S1 , including the number of items, Cronbach's alphas, reporter, and scale ranges. --- Parent and adolescent report. Demographics.: Demographic characteristics included age, gender, race/ethnicity, and primary language. Parent demographic characteristics also included Mexican versus non-Mexican Latino ethnicity, marital status, household income in the previous year, amount of money received in the previous month from all sources, highest grade/year of education completed, being born in the US, number of years living in the US, being currently employed, parent's relationship to the child , and whether the adolescent in the study was living with a biological parent. Household income was categorized into $5,000 increments and treated as a Likert scale, e.g., "Less than $5,000" , "$5,000-9,999" , "$35,000-39,999" , and "$40,000 -And Over" . Adolescent demographic characteristics also included sexual orientation. --- Mental health.: The Brief Symptom Inventory assessed parent and adolescent mental health symptoms during the previous week. A global summary measure of distress was calculated along with nine sub-scales for symptoms of depression, anxiety, somatization, interpersonal sensitivity, obsessive-compulsive, hostility, phobic anxiety, paranoid ideation, and psychoticism. In addition, a binary caseness variable was calculated that provided a clinical cutoff level based on the global summary measure and the nine subscales as outlined in the BSI scoring manual. --- Family functioning.: The Family Functioning Scale measured parents' and adolescents' perception of their family's functioning across seven different constructs: cohesion, expressiveness, disengagement, democratic family style, laissez-faire family style, authoritarian family style, and conflict. We adapted the original response scale labels "Strongly agree" to "Strongly disagree" to read as "Very true for my family" to "Very untrue for my family". --- Readiness for change.: The University of Rhode Island Change Assessment assessed parents' and adolescents' readiness for change through four subscales: pre-contemplation, contemplation, action, and maintenance. Readiness was computed by summing the last three subscales scores and subtracting the pre-contemplation subscale score. Both the full readiness scale and subscales were examined to assess whether specific stage of readiness might be related to participation. --- Substance use.: The Short Michigan Alcohol Screening Test ascertained alcohol use and abuse among parents. Problematic alcohol use was identified in one of the following three ways. Participants indicating lifetime alcohol abstinence were not administered the SMAST and classified as not engaging in problematic alcohol use. Participants who gave a "no" response to the first SMAST item, "Do you feel that you are a normal drinker?", were classified as engaging in problematic alcohol use. Participants who indicated being a normal drinker were further queried on 12 yes-no indicator items assessing feeling and actions related to drinking. Participants with a total score of 3 or more were also classified as problematic alcohol users as suggested by Selzer et al. . The adolescent version of the Michigan Alcohol Screening Test was administered to adolescents. The last two items were modified and queried individuals on whether they had ever been "arrested or gotten a ticket when you have been drinking " and arrested for "drunk driving". Follow-up questions on the number of arrests were not administered. A yes-no indicator variable was created to identify borderline problematic alcohol use. Participants who indicated lifetime alcohol abstinence were not administered the MAST and were classified as not engaging in problematic alcohol use. Individual MAST items were weighted and summed based on scoring guidelines. Participants with a total score of five or more were identified as problematic alcohol users based on guidelines suggested by Selzer . The Drug Abuse Screening Test ascertained drug use and abuse among parents through items assessing feelings and actions related to substance use. Following diagnostic validity statistics and recommendations presented in Gavin, Ross, and Skinner , the presence of a substance use disorder was identified by a score of six or more. The adolescent DAST ascertained drug use and abuse among adolescents in a similar manner. The AIDS Risk Behavior Assessment assessed alcohol and drug use among adolescents over their lifetime and the past three months, frequency of use in the past three months, and method of use. As a supplement to the DAST, the ARBA explicitly asks about the following substances: alcohol, marijuana, cocaine or crack, amphetamines, ice , heroin, a mixture of heroin and cocaine, a mixture of heroin and speed, non-prescription methadone, and other opiates. As in Milburn et al. , we created yes-no composite indexes of "hard drug use" for lifetime and past-three-months use of any substance other than marijuana. Comfort discussing sexual and reproductive health.: Teaming African American Parents with Survival Skills measured the degree to which a parent was comfortable discussing risk behaviors for sexually transmitted infections, including HIV/AIDS, with their child. Three summary sub-scales covered: ever talking with one's child about sex; talking with one's child about sex currently or in the past three months; and talking with one's child in general about topics that include sexuality, reproduction, alcohol and drug use in the past three months. The TAAPSS also included 11 yes-no questions for adolescents on whether their guardian discussed sexual behavior, alcohol, and drug use with them in the past three months. Items were summed. Retention and engagement.: Retention, defined as completion of the intervention sessions by parents and their adolescent children, was ascertained by session attendance. Most parents and their adolescents attended all of the five possible sessions. Remaining parents and their adolescents attended zero to one session , two sessions , three , or four sessions . Hence, we created a binary outcome for attendance, classified as having attended all five sessions versus fewer than five sessions. Engagement, defined as satisfaction with the therapeutic alliance, was assessed by the Working Alliance Inventory, Short Form . A higher WAI summary score indicates a better working relationship and trust between facilitator and participant. Scores were reported for facilitators and participants . Correlation between facilitator and participant scores was low for parents and moderately high for adolescents . On average, parent-reported scores were 3 points higher than facilitator reported scores ; adolescent and facilitator-reported scores did not differ significantly. Parent report.-The Conflict Tactics Scales, Form A ascertained the frequency of parent conflict with the adolescent over the past three months or currently across three scales: reasoning, verbal aggression, and physical violence. The Parker Parental Bonding Instrument assessed parent perception of the degree of bonding between adolescent and parent across two subscales, care and protection. The Adult Attachment Scale ascertained adult attachment styles and closeness of relations with partners and other individuals across three subscales: difficulty in depending on others, and anxiety and discomfort in closeness to others. Higher scores represent less secure attachment. --- Adolescent report. Emotional regulation.: The Difficulties in Emotion Regulation Scale measured components of emotional regulation among adolescents and was analyzed as a summary measure of the first 14 items from the original 36-item scale. --- Trauma.: The adolescent version of the UCLA Post Traumatic Stress Disorder Reaction Index for DSM-IV assessed adolescents' exposure to traumatic events , symptoms of PTSD , and DSM-IV PTSD diagnostic criteria. Criteria B, C, and D represent reexperiencing, avoidance, and increased arousal symptoms, respectively. Items are scored from 0 to 4 . Based on the UCLA PTSD-RI manual, responses of 2 or higher indicate the presence of symptoms for individual items. Partial PTSD is considered likely if Criterion A is met along with any two of criteria B, C, or D. Full PTSD requires all Criteria to be met. --- Sexual behavior.: The ARBA also assessed sexual behavior among adolescents, including: lifetime participation in anal or vaginal sex, number of anal or vaginal sex acts during the past three months, and condom usage during sex in the past three months. We analyzed the average of Likert-scale scores for condom usage during anal and vaginal sex. --- Data Analysis Primary analyses compared parents and adolescents by retention items , and engagement items on demographics, mental health factors, perceptions of family functioning, attachment/bonding, parenting practices, substance use, adolescent sexual practices/ behaviors, and readiness to change behaviors. T-tests were conducted to compare Likertscaled measures that were appropriately treated as continuous variables by session attendance groups; the Satterthwaite adjustment was applied if diagnostic tests indicated unequal variances across groups. Since the WAI scores were continuous, we used a different analytic approach to examine potential correlates. Categorical measures, which mainly consisted of demographic characteristics, were treated as covariates of WAI scores in linear regression models. We examined Pearson correlation coefficients between measures that were appropriately treated as continuous measures and WAI scores. We report t-statistics or Chi-square statistics, degrees of freedom , and p-values. Each of the three parental retention items-session attendance, facilitator and participant WAI scores-were compared across 49 measures. Similarly, each of the three adolescent retention items were compared across 41 measures. In addition to the standard .05 alpha level, statistical significance is discussed in terms of the Holm-Bonferroni adjustment to give context to the large number of comparisons and Type I error rates. For each retention item, comparisons were ranked by their p-values. The alpha level for the smallest p-value was calculated as .05 / , the alpha level for the next smallest p-value was calculated as .05 / , and so forth, where n is the number of comparisons. For parents and adolescents, the smallest p-values needed to be less than alpha levels of approximately .0010 and .0012, respectively. We report mean differences between session attendance groups in terms of standardized effect sizes where SD is the pooled standard deviation between groups, ES = MD/SD. Following Cohen's rules of thumb, ES of .1, .3, and .5 are considered to be small, medium, and large respectively . --- Results --- Retention and Engagement Outcome Analyses Parents.-Table 2 shows all of the parent demographic characteristics and remaining measures that differed significantly by session attendance. Parents attending all five sessions versus fewer sessions reported a higher household income , less conflict-related reasoning on the Conflict Tactics Scale , and higher anxiety on the Adult Attachment Scale, on average . Next we report relationships between WAI scores and parent measures, with significant findings summarized in Table 3. Higher facilitator-reported WAI scores were associated with a higher likelihood of reporting White/European American race/ethnicity versus other racial/ ethnic groups , less education , a lower Adult Attachment closeness score , and fewer mental health symptoms on the BSI global distress index and on the somatic , phobia , and psychoticism sub-scales. Higher participant-reported WAI scores were associated with a higher likelihood to be female , and Mexican versus non-Mexican Hispanic ethnicity; participant-reported WAI scores did not significantly differ between Non-Hispanic and Hispanic ethnicity. Furthermore, higher participant-reported WAI scores were associated with receiving less money last month , and increased levels of hostility on the BSI . No other significant differences were found for parent demographics, mental health, sexual behavior, and substance use in comparisons with session attendance and WAI scores. Based on the Holm-Bonferroni adjustment, none of the parent results were statistically significant. 4 shows all of the adolescent demographic characteristics and remaining measures that differed significantly by session attendance. Adolescents attending all five sessions versus fewer sessions reported more mental health symptoms on the BSI, both on the global distress index , and on the depression , anxiety , somatization ,obsessive-compulsive , phobic anxiety , and psychoticism subscales. A higher percentage of males attended all five sessions than females , though this difference was not statistically significant . --- Adolescents.-Table Next we report relationships between WAI scores and adolescent measures, with significant findings summarized in Table 5. Higher WAI scores were observed for females versus males and for URICA contemplation scores . Higher participant-reported WAI scores were also associated with higher URICA action and readiness scores , as well as lower precontemplation scores . Adolescents talking more about sex, alcohol, and drugs with their parents in the past three months according to the TAAPSS had higher facilitator-reported WAI scores . No other significant differences were found for adolescent demographics, mental health, sexual behavior, and substance use in comparisons with session attendance and WAI scores. After Holm-Bonferroni adjustment, statistical significance remained for mean BSI somatization score differences by session attendance, as well as gender, URICA contemplation and action score differences by participant-reported WAI scores. --- Discussion There are several important results from these analyses. Parents with more income and less perceived family conflict were more likely to complete the intervention. These findings are in keeping with several other studies , suggesting that families with less hardship and distress can be better retained in interventions. Relatively more interesting is the finding that the parents who completed the intervention also reported more anxiety symptoms, suggesting that some parents with greater need for intervention were actually better retained. With respect to adolescents, the retention results are quite surprising. Adolescents who completed the intervention, relative to those who did not, reported more depressive, anxiety, somatizing, obsessive-compulsive, phobic, and psychotic symptoms. This suggests that, as with the parents in this study, adolescents with greater distress and thus greater need were more apt to finish the intervention. This finding differs from previous research on family interventions that more often reports families who have more distress have lowered retention in interventions . But, research on youth serviced by the public sector , also reports a similar pattern of higher use of mental health services . There are a number of reasons why retention of at-risk participants, such as homeless adolescents, is challenging for prevention programs, including psychological distress, systematic and societal barriers . However, for the adolescents in this study, a distressed emotional state seemed to be a potential mechanism for retention. This suggests a number of areas for further research on how greater stress may be associated with completing the intervention, such as through the related increased awareness of the need for support, higher motivation, and/or experience of more immediate benefits. Compared to other family interventions, STRIVE may be more acutely culturally responsive to the needs of distressed adolescents. This may be an artifact of both the curriculum and training of the facilitators. STRIVE facilitators are trained in how to implement the intervention in a culturally-humble way while keeping health disparities, sociopolitical climate and culture in mind. The curriculum is manualized to maintain fidelity, but it is also highly encouraged that the facilitator individualizes the examples to the family's culture. Establishing this positive cultural response may have assisted with retention. This is consistent with emerging research on being culturally responsive in engaging racial/ethnic minorities . The process of engagement as assessed by the therapeutic alliance was sensitive to several individual-level differences for parents. From the standpoint of the facilitators, a greater alliance was perceived with White/European American parents and parents with fewer mental health symptoms. This is consistent with prior work that suggests engaging with racial/ethnic minority families and families who are in more distress is more challenging . Interestingly, however, parents who identified as racial/ethnic minorities, women, those with lower income and higher hostility symptoms reported an increased sense of alliance with the facilitators. Fully understanding, for example, why ethnic minority parents report greater alliance for themselves requires more research in areas such as access. Ethnic minority parents, whom often have limited access to interventions, may have had positive perceptions of the alliance because of the accessibility of the intervention. Thus, in the current study, being more distressed or being a racial/ethnic minority person was not a barrier to alignment. Such findings are uncommon in family interventions, challenging the notion that the neediest families are more difficult to engage in building a relationship with the facilitators. This result speaks to the important role the facilitator have of engaging families of "high need" and knowing that alliances can indeed be fostered between the facilitator and parents of atrisk youth. This study supports the notion that the parent buy-in to any intervention is essential in the retention and completion of the program. In a larger scale, this can assist with lowering the racial and ethnic disparities in mental health services . The findings around the therapeutic alliance and engagement with the adolescents were less conclusive. From the perspective of both the facilitators and the youth, female participants built more productive therapeutic alliances. Adolescents who reported greater comfort discussing sex with their parents were identified by facilitators as having a stronger therapeutic alliance. This last finding may suggest that youth who have an easier time discussing difficult topics may be more amenable to family interventions because they are open and "ready" to discuss difficult topics such as family conflict. The remainder of the findings were not significant at the conventional p < .05 level, and we hesitate to interpret them further. There are a few limitations to this study. First, these data come from an efficacy trial of a family intervention, not an efficacy trial of intervention implementation. As such, we have limited information on key engagement processes aside from the therapeutic alliance. Second, numerous statistical tests were conducted on data from a relatively small randomized control trial, making it difficult to adjust for Type I error without preventing us from reporting on relationships which exist in the population . Third, on a related point, we focus on bi-variate associations and not multivariate models because these analyses are both exploratory and have limited statistical power for such models. In addition, small Cronbach's alphas less than 0.70 were found for a number of scales, indicating less than adequate reliability between scale items; especially low alpha values are noted for the Conflicts Tactics Reasoning scale, the discomfort-in-closeness subscale, and a number of adolescent-assessed components of Family Functioning. Last, the use of self-report measures was a limitation of this study. Self-report measures rely on the participant's report of their symptoms, behaviors and experiences. Honesty, introspective ability and interpretation ability can influence participants' responses. The implication is that self-report measures may not be assessing the underlying constructs they were intended to capture in this sample, and the results should be interpreted with caution. Despite these limitations, we believe these data are informative for several reasons: relatively few studies have looked at retention and engagement in the context of prevention programs , and fewer still in HIV prevention programs ; no studies have looked at retention and engagement in prevention programs with homeless youth ; we found some very informative counter-intuitive findings with respect to family distress which suggest that for these families who are almost all distressed, some modest increased levels of additional distress may be a motivation for retention and engagement in treatment. The implications of these findings are significant in understanding the complexities of engagement for at-risk youth and their families. It is also important to note that both participant and provider perceptions are important in development of a strong therapeutic alliance. Given providers' perception that their alliances were stronger with White/European American families and those with fewer mental health symptoms, these findings warrant further exploration of the development of cultural humility, awareness training, and improving mental health literacy for facilitators. Because working with families who may have more distress, economically and health wise, is perceived to be more challenging, preparing facilitators to work with these families is paramount. The fact that families who were experiencing more distress had better therapeutic alliance than their counterparts suggests that there may be additional need for the development of psychoeducational interventions for families in crisis. More work needs to be done, but we can say that brief psychoeducational family interventions can be engaging and efficacious for at-risk adolescents and their families, but for these to be effective, we may need to determine who among this population is the best target for these types of family interventions. This study contributes to understanding how to engage and retain underserved populations in family interventions within the Implementation Phase of the TSci Impact Framework . These findings can help in the development of better strategies for improving participation in family-based interventions that can improve the health outcomes of young people over their lifetimes. --- Supplementary Material Refer to Web version on PubMed Central for supplementary material. ---

Behavioral family interventions are an effective way to intervene to prevent negative developmental outcomes for adolescents. Participation in family interventions encompasses behavioral and cognitive/attitudinal dimensions, among others, indicated by retention and engagement, respectively. Two dimensions of participation, retention and engagement, in a family intervention were examined in a sample of newly homeless adolescents and their parents or guardians. Correlates of participation included parents with more income and less perceived family conflict, and adolescents with higher endorsement of depressive, anxiety, somatizing, obsessive-compulsive, phobic, and psychotic symptoms on the Brief Symptom Inventory (BSI). Stronger therapeutic alliance was correlated with being more distressed (i.e. lower income, more hostility), being a female adolescent participant, and having greater comfort discussing sex with parents. Furthermore, parents and adolescents with greater distress and thus greater need were more apt to finish the intervention. The finding that families who were experiencing more distress had higher alliance scores suggests that there is an additional need for development of interventions for families in crisis. Both participant and provider perceptions are also important in development of a strong therapeutic alliance. This study's findings have implications for further exploration of the development of cultural humility and improving mental health literacy among facilitators of behavioral interventions.

INTRODUCTION Since the democratization of South Africa, great attention has been directed toward single motherhood and the challenges that mothers face in raising children alone [1 -14]. This is partly due to the rise in trends, both globally and in South Africa, regarding households run by single mothers [1,2,6,7]. One reason for such an increase in attention to single mothers has been the prevalence of the absence of fathers [7,8,15,16]. According to Posel and Dewey [17], South Africa has the second highest rate of father absence in Africa. Findings reported by Van den Berg [8] showed that more than 40% of households in South Africa are run by single mothers compared to only 3% of those run by single fathers. According to Statistics South Africa [18], Black Africans are among the most affected, with about 41.92% of reported absent fathers in households. Another study by Heartlines [19] reported that coresidence data from Statistics South Africa revealed that 36% of children live with their biological mothers in the same household, 34% live with both biological parents, and only 2% live with their biological fathers. The increase in studies on single mothers' experiences, predominantly Black single mothers, might also be contributing to the predominant negative perception and the demonisation of Black fathers, who are often described in the literature and the media as absent and uninvolved [20,21]. Therefore, understanding the experiences of single fathers in South Africa is crucial as it allows for a critical assessment of the extent to which reported experiences align with or contradict commonly held views about Black men both in South Africa and globally. Many studies have been conducted on father absence and its related consequences [7,8,16,18,22]. However, very little is still known about fathers' involvement in raising their children, especially as single fathers [2, 4, 23 -25]. For this article, single fatherhood only included biological fathers who were raising their children on their own due to the death of their partner, separation, or divorce [24,26]. However, we are aware that in traditional African societies, the status of 'father' is not limited to the biological process, but this may refer to other male figures, such as grandfathers, uncles, brothers, etc., who may fulfill a role of care, protection, and provision in relation to 'children' [8,25,27,28]. This study only focused on single fathers and their experiences of fatherhood to enrich the currently existing literature [14, 29 -32]. Furthermore, various benefits of fathers' involvement have been reported in the literature. These include improved behavioural, cognitive, and socioemotional development in children, increased child tolerance for stress, less depressive symptoms and higher academic performance, and improved postnatal mental health for the mother [19,25,33,34]. Other benefits include children's increased empathy, self-esteem, and social competence [7,23]. Involved fathers have also been reported to be healthier than those not involved in raising their children [23,35,36]. Other topics related to fatherhood reported in the literature include the interconnection of manhood and fatherhood, fatherhood practices, and parenting [33, 37 -41]. In many cultures, including in South Africa, becoming a father is still perceived to be a rite of passage into manhood [27,28], despite a reported increase in the adoption of progressive attitudes and beliefs on manhood post-apartheid [41]. Furthermore, fatherhood is still understood in terms of the role played by men, which often relates to traditional gender stereotypes and African parenthood practices [19,41]. For example, men and women are still expected to conform to traditional gender roles, with women's identity centred around their ability to give birth and be primary caregivers, even when they are employed. On the other hand, men's identity is still centred around their ability to provide financially [19]. Another body of research to consider in the context of our current study is parenting practices, especially in the African context. Studies on Baumrind and Maccoby and Martin's parenting styles have yielded mixed findings on children of different cultural and racial groups [38,42]. While some studies have argued that collectivistic societies, such as African societies, tend to emphasise the use of authoritarian parenting styles [38], recent studies have challenged these notions and argued otherwise [39]. In addition, studies that have argued that collectivistic societies often exhibit authoritarian parenting styles may often assume that the use of increased control is always negative regardless of the context. These studies often overlook the impact of western socialisation on parenting practices in non-western societies and the diversity of family structures within non-western societies, such as in South Africa. Therefore, while it is important to acknowledge the influence that living in a collectivistic society may have on a child's upbringing, it is also important to avoid polarising parenting styles based on "western" or "non-western" narratives. Moreover, such polarisation often tends to position western notions of parenting as "better" than non-western and may overlook positive aspects that have been reported about parenting styles perceived as non-western [39]. Therefore, this study serves to add to the literature on gender and establish whether and to what extent gender stereotypes and traditional notions of parenthood in African societies are reflected in participants' experiences. Moreover, current research on manhood and masculinity was also discussed in our study. These include research on hegemonic, hybrid, and Black masculinities. Hegemonic masculinity is a term coined by Cornell [43] and refers to an "ideal" form of masculinity. Contrary to Cornell's initial assertion, further research has revealed more than one form of hegemonic masculinity [44]. These are often informed by gender ideologies and involve the attribution of gender roles based on biology [45,46]. The performance of hegemonic forms of masculinity often reflects relationships of power between men and women and men and men across racial and socioeconomic classes [44]. An example of this is the idea that a husband is 'supposed' to be the breadwinner and dominant earner, whereas the wife is responsible for domestic labour and child care [46]. Hegemonic gender ideologies also acknowledge that gender meanings constantly change over time across social, cultural, and institutional contexts [44,46,47]. Moreover, these ideologies are informed by factors, such as race, age, class, and the cultural context, in which individuals find themselves. In instances when a man is unable to conform to the expectations of the hegemonic forms of masculinity within their context , they often compensate for this lack by engaging in gender roles traditionally ascribed to females , or in domestic violence and adultery [47]. However, due to limited research on single fathers, the current literature has not reported on how single fathers who assume both the provider and caregiver roles negotiate these roles. Furthermore, research on hybrid masculinities has documented how cisgender, straight, white, middle-class men often engage in non-hegemonic forms of masculinity without losing their masculinity status. Men who engage in "hybrid masculinities" incorporate both hegemonic and non-hegemonic masculinities into their experiences [47]. Hybrid hegemonic masculinities have also been used to discuss the intersection between gender, class, and sexual privileges [48]. These forms of masculinity are often characterised by increasing levels of equality with less hierarchy and include diverse populations of men across a wide range of contexts. Participants in our study can be considered to engage in some forms of hybrid masculinities by attempting to integrate both traditionally male and female gender roles in their experiences as single fathers [47]. Moreover, studies on Black masculinities have documented the role played by historical and structural factors, such as slavery in the United States, labour migration during apartheid in South Africa, and unemployment and low education in shaping Black men's experiences [43,49]. Debates on Black masculinities are often premised on the idea that there are versions of being a man specific to Black men and draw from hegemonic masculinities that focus on predominantly negative narratives. These debates are also often focused on a history of Black masculinities shaped by patriarchal and negative stereotypes of what it means to be a man, specifically a Black man. In addition, Black men's failure to live up to hegemonic masculinity expectations and perform traditional gender roles due to structural factors have contributed to perpetuating negative discourses and stereotypes of masculinity based on race. Such examples include Black men being demonised and perceived as hypersexual, violent, dangerous, irresponsible, absent, in crisis, a failure, and in need of improvement [20,21,43,49,50]. These negative discourses have led researchers and scholars to advocate for a "new Black man" [50,51]. Moreover, prevalent negative stereotypes and narratives about Black men have also obscured alternative positive stories and experiences of manhood and fatherhood. The current study hopes to address this problem by highlighting alternative discourses of Black masculinities and parenthood. Recent literature on Black masculinities has challenged the one-size-fits-all and predominantly negative approach reported in earlier studies about Black men. Social class has been acknowledged as crucial in differentiating positive and negative forms of masculinity among Black men [45,49,52]. The extent to which individuals have access to economic and educational opportunities, rather than race alone, seems to influence Black men's choices about which forms of masculinity to engage with. Moreover, Black working men with some form of education often integrate dominant norms with those valued within the Black culture [53]. Participants interviewed for our study were all working middle-class citizens who had some form of higher education. However, the extent to which these factors influenced their experiences is yet to be established through our research findings and discussion. Considering the current context of manhood, masculinity, fatherhood, and parenthood, and after reviewing the literature on these topics, we are convinced that investigating the experiences of Black single fathers is crucial. Understanding Black single fathers' experiences will provide deeper insights into parenting practices and enrich our views of the dynamics within relationships where fathers take the leading role as both providers and caregivers. This is especially so, considering that fathers have often been reported to take a secondary and supportive role in the childrearing process [28]. --- RESEARCH QUESTIONS Based on its aim, this study had four objectives to achieve: Identifying key themes and narratives based on participants' stories of their experiences as single fathers. Discussing how single fathers negotiate their role and identity as single fathers. Discussing the implications of the findings on parenting, with a specific focus on differences between raising boys and raising girls. Identifying differences in raising children of different developmental stages. These objectives translated into four questions which were all addressed through the research: 1) What are the experiences of single fathers in South Africa? 2) How do single fathers negotiate their role and identity as single fathers? 3) Are there differences between raising boys as opposed to raising girls? If there are, what is the nature of such differences? 4) Are there any differences in raising children of different developmental stages? If there are, what is the nature of such differences? --- METHODOLOGY --- Theoretical Frameworks This study was based on two paradigms: social constructivism and the critical approach. Whereas social constructivism aims to make sense of the subjective world of human experiences [54], a critical paradigm questions these experiences. A social constructivist approach understands social phenomena "through the eyes of the participant rather than the researcher" [55]. In addition to reporting on Black single fathers' experiences, our study also challenged notions of parenting by creating a space where single fathers can express themselves and become part of the broader conversation about fatherhood and parenting in South Africa. Furthermore, being single fathers sharing their experiences served to challenge common discourses about fathers in South Africa and promote alternative marginalised narratives. The social constructivist paradigm advocates for multiple realities that are constructed through social interaction and ascribed meanings within specific contexts [54,56,57]. In addition, it recognises the influence of social values and subjectivity on the research findings [58]. For our study, knowledge about the experiences of single fathers was considered to be a result of interactions between the first author and research participants during the narrative interview process. Each participant's perspective was equally valid and constituted a crucial part of reality that served to enrich our understanding. Moreover, the presence of the first author as a researcher, the types of follow-up questions asked during the interviews, and how participants answered these provided access to various realities about being a single father shared by the participants. This, according to social constructivism, refers to the interdependence between the researcher and the participants in co-creating reality [54]. This view was expanded on by the critical approach, which allowed us to perceive participants' realities as shaped by interacting cultural, political, ethnic, gender, and religious factors. In line with the critical paradigm, our discussion of Black single fatherhood attempted to remediate the political, social, and economic problems resulting in social oppression, conflict, struggle, and power structures within various and varying levels of the social-ecological spheres [54]. We also considered how constructions about fatherhood in relation to parenting had influenced discourses on the role of fathers in raising their children. Within a critical paradigm, the overemphasis on the narratives of single mothers was understood as contributing to silencing or marginalising the experiences of single fathers. Using a critical approach, the historical, social, and cultural contexts that shape participants' stories were considered, and discourses and issues of power embedded in participants' stories were identified [55]. From a social constructivist perspective, meanings derived from the content of the interviews were negotiated through the interactions that occurred prior to, during, and after them. This was informed by the notion that meaning is constructed through interaction between the researcher and the participants [54]. The interpretivist paradigm uses data collected through interviews, discourses, text messages, and reflective sessions [54]. Data was collected using qualitative methods and analysed using an inductive approach, which consisted of the first author, through thematic analysis, identifying patterns embedded in the collected data in order to generate themes and increase our understanding of single fathers' experiences as a social phenomenon [54]. The critical paradigm focuses on respecting norms within a cultural context while at the same time challenging notions that serve to legitimise certain bodies of knowledge and delegitimise others. In our study, notions that served to perceive parenthood as a role that is only fit for women, as reflected in research on single mothers, were challenged by allowing marginalised voices of single fathers to become part of the discourse. Therefore, by creating awareness of the experiences of single fathers, this study adopted a social justice perspective, one which challenges those notions of fatherhood and parenthood that often lead to the social oppression of the participant group. --- --- Materials Narrative interviews were conducted with seven participants who all identified as Black single fathers [59 -63]. In addition, a snowball sampling method was used where participants who agreed to participate in the study were also asked to invite other participants who met the criteria for selection [64 -66]. Questions used during the interview are included in Appendix B. For this study, two key forms of narrative analysis were used in combination: thematic and dialogic/performance analyses [67 -71]. While thematic analysis focuses on the identification and analysis of patterns or themes within the data, dialogic/performance analysis examines the way certain narratives are told, the context in which they are told, and the impact telling the narrative in a specific manner has on the identities of individuals involved [71 -74]. Moreover, dialogical/performance analysis also alludes to broader cultural and societal issues and how these relate to the narrative being analysed [71]. These two forms of analysis were in line with the aim and objectives of this study. Furthermore, Bamberg's little story approach was incorporated into the analysis of themes identified in the interviews. Incorporating Bamberg's little story approach into thematic and dialogical/performance analysis was beneficial as, unlike the above-mentioned forms of analysis that contribute to the construction of a big narrative story, the little story approach takes into consideration the impact that small stories have in influencing individuals when constructing a sense of self or identity [75,76]. Bamberg's approach also served to demonstrate the interrelationship between continuities and discontinuities in participants' narratives over time, pointed to the self-vis-à-vis others and uncovered how participants portrayed themselves as both agents and undergoers in their stories [75,77]. Finally, Bamberg's approach was crucial in closely evaluating the inconsistencies, ambiguities, contradictions, moments of trouble and tension, and participants' constant navigation and process of refining various versions of selfhood in interpersonal and interactional contexts [75,77,78]. --- Procedure Ethical clearance was obtained from the School of Human and Community Development Ethics Committee and the Registrar of the Faculty of Humanity at the University of Witwatersrand. Participants were recruited using online adverts that were posted on the researcher's Facebook account and then invited for face-to-face narrative interviews. Due to challenges related to the COVID-19 pandemic, alternative means of data collection were put in place [79 -82]. These included conducting individual interviews via Microsoft Teams or Zoom. Out of the seven interviews conducted, four were virtual , and two were face-to-face. All key ethical considerations were considered, including signing consent forms for recording and maintaining privacy and confidentiality using pseudonyms [83 -85]. --- RESULTS AND DISCUSSION Two key themes emerged from the thematic analysis of the data collected: 1) Parenting as a single father and 2) Life beyond single fatherhood. These themes reflect the big picture of the experiences of singles fathers related to Bamberg's theory. A chronological approach was adopted in identifying the themes, in line with both a social constructivist framework and a narrative approach to data analysis [54,56,58,60]. The section below discusses these themes and their related subthemes. --- Theme 1: Parenting as a Single Father Part of the participants' experiences as single fathers involved managing various tensions between dual parenthood and single parenthood, navigating what it means to parent boys versus parenting girls, and reflecting on the various support systems available to them or lack thereof. Some participants had to be there for their children as a way to compensate for the absence of their own biological fathers, as participants stated: "I remember I had issues, and my grandmother said, irrespective of whether he is in your picture or not, he is your father! We all brag about our parents. I cannot brag about my dad as such because I do not know him! Like here is my dad. But I do not know him! So, me with that in mind, I got a kid, I got responsibilities, people that are looking up to me, my kid, and my own siblings back home. It made me stronger. And I had no option but to follow that route." "I was never raised by a man. You know was been raised by a single mom, we never had a father figure. " By being there for their children, P3 and P6 hoped to break the cycle of fatherlessness that they experienced growing up. In addition, participants also viewed themselves not just as victims of their past experiences but as key players in influencing changes to the broader narrative about single fatherhood [75]. This also reflected efforts to challenge negative stereotypes about absent Black men and fathers that are part of the dominant discourse in South Africa [7,8,16,18,22]. P7, on the other hand, grew up with his grandmother, away from his family, something he reported has motivated him to be a present father. Moreover, P3 wanted to set a new example for his siblings and those who looked up to him. Through his commitment to being involved in his own children's life, P3 was also engaged in the process of redefining what Black masculinity means and reaffirmed that Black parents are just as capable of being involved in their children's life, therefore changing the narrative, as stated by participants: "I grew up without my father. My father passed away when I was 12. But there is one thing I have learned from my dad. My dad was very passionate about family. So, the family was very important. And that, that has… It was well glued in my mind ." "But for me, it started from a young age when I wanted to have a family of my own. I did not want to regard myself as a sponsor, whereby I would just send money home. It was important for me to be with my family ." Many fathers in this study grew up without fathers. They wanted to be good fathers and break the cycle of absent fatherhood in South Africa. These accounts provided a new understanding that goes beyond material provision and includes being emotionally available. Heartlines [19] reported in a study on perceptions of fatherhood that while financial provision was not the only measure of fathers' levels of involvement, it was the primary factor. Ratele et al. [39] also alluded to the provider role in their research on discourses of fatherhood and found that the concept of "being there" for one's children also involved providing financially for them. Therefore, although single fathers assumed other female roles in raising their children, they still subscribed to the traditional male roles as measures of their involvement. This also seems to confirm findings from other studies on fathers' involvement [6,18,28] that many fathers see their role mainly as a financial provider rather than being emotionally available for their children. Participants' emphasis on being financial providers also reflects their complicity in reinforcing hegemonic ideologies of masculinity that position men as sole providers [44,53,86]. This is especially due to participants' ex-partners' reluctance and unwillingness to contribute financially. Participants in this study acknowledged that fathers need to be available emotionally, but all emphasised the role of provision as the key marker of being a good father, further reflecting their commitment to traditional and patriarchal gender-role stereotypes about men and fathers [45,47,52,87]. --- Parenting Boys versus Parenting Girls Participants also reflected on their experiences of parenting boys versus parenting girls. Different participants reflected different views related to the parenting style that needed to be adopted between boys and girls. One of the participants shared his views as: "…The boy, he does not talk. You know boys. They do not have the same emotions as girls. So, they do not talk, or they do not express their feelings like girls…Now recently, I have been spending more time with him, but still, he is not opening up ." In commenting, "they do not have the same emotions as girls", P7 seems to allude to more traditional notions of how boys and girls should be treated. In addition, in the above passage, P7 points to a gender stereotype that boys do not display emotions while at the same time revealing a struggle to connect emotionally with his son. By referring to the difference in emotional openness, P7 also supports the view that boys should be treated differently from girls. Research has shown that teenagers' relationships with their fathers appear to be more important to their development and achievement in school [88 -90]. Therefore, P7's son, being in his teen years, may benefit from a closer relationship with his father, as his father suggested in the extract above to be "spending more time" with his son. This need to treat boys differently is also reflected in P1's struggle to answer the question on the difference between raising a boy and a girl : "… That is a dicing question… I do not know, but based on my personality, I think I will approach the same situation, the same way it is happening now. I think with boys, you have to be extra careful and very cautious when you are dealing with them…" By using the expression "with boys, you have to be extra careful and very cautious", P1 may be referring to the need to use more control when raising boys compared to girls. This may also reflect an assumption that girls are easier to control than boys. Hence, stricter measures may be required for boys. This view also seems to align with research on parenting styles, which suggests that African parents tend to use authoritarian parenting styles [38]. However, further investigation is required to ascertain this hypothesis as there is little research focusing specifically on parenting styles within African households. The available research on this topic seems to contradict the above view [39]. Regarding age, participants shared mixed views on how they treated their children. While some reported being stricter with their younger children, others used more authority, especially during their children's puberty. Moreover, P5 reported being stricter and a disciplinarian when his children were younger to compensate for his ex-wife's permissiveness. "We would have different approaches, my wife was too soft with them, and I needed to be hard in making sure that they studied ." In being stricter with his children, P5 reported wanting to strike a balance in parenting, which is often believed to be beneficial to a child's development. However, he also reported adjusting his parenting style as his children became adolescents. This need to use force can also be understood as an attempt to establish a relationship of power between a child and a parent that has been characteristic of parenting within patriarchal societies in which men often feel entitled to having control over women and children [44,86]. "My son, at 20 years old, he is now spending more time with me. And without any prompting, without any guidance from anyone, he just started saying, I do not want to go to mum, I am fine. I want to stay with you… And we are no longer sticking to those bound rigidity that we used to when we started because now there is a lot of maturity ." This further reflects ways in which P5 has justified his exercise of power over his son and used age as a justification for his son's freedom and as a rite of passage into adulthood. In doing so, P5 may have further perpetuated discourses that children need to be controlled by adults just because they are children, therefore reflecting ageism [91]. Moreover, other participants were permissive with their younger children, as stated by one of the participants: "When she wants anything, Whether I have money or not, I will drop everything, make sure that she gets what she wants ." "I was never saying no to my children. I could maybe say no to my wife but never say no to my children ." Participants who were permissive with their younger children also reported challenges in adjusting to their children's increasing and developmentally appropriate need for autonomy as they became adolescents [5]. They stated that: "I remember when she was starting to be rebellious. I almost got physically…I will shout and everything ." " …a person that was giving me a lot of grief and rebellious inside the house, and she will lay into sleep this time of the day like she will put a big, huge blanket over, while it is 30 degrees outside… I forced her. She went back to school, she did, and she passed matric ." 'Then the boy being a boy, trying to be a man, you know when boys like 13-14… he thinks now he is a man, he can make things move. They did not get along with the stepmother. So now there were a lot of conundrums between the two ." The extracts above further reflect the idea that men, and in this case, parents, should strive to control their children, even if that means using power to exert their authority. Participants, by virtue of being parents, felt entitled to impose their will on their children, perhaps as a way of enacting discourses about discipline and raising children that they themselves had been socialised into. Research on intergenerational parenting styles has shown that past parenting styles have an influence from one generation to another [21]. These views also support research on parent-adolescent relationship dynamics. The relationship between a parent and their child has been reported to shift during the child's adolescence, with parents often realigning and negotiating the dynamics and offering increased autonomy to the child [92]. When the experiences of participants on how they related to their children when they were young versus when they were adolescents were compared, differences were identified. Most participants spoke about being more permissive with their children when they were younger. These differences may be related to both the fathers and the children's personalities and temperament, among other factors that have been identified to influence parenting and the relationship between parents and their children [5]. However, reflecting on their experiences, participants expressed a sense of inadequacy in raising adolescent daughters compared to adolescent sons, especially regarding their daughters' menstruation. While there was an emphasis on menstruation as a rite of passage into girls' adulthood, little was mentioned about boys' process to manhood [93 -95]. Furthermore, fathers framed their girls' menstruation as a problem to deal with and expressed anxiety in facing this "problem," as they stated: "You know, the only challenge was when my daughter came out of puberty, when she had her menstrual period. That was my biggest problem… And so, I had to now go to the shop and buy pads for her. And I get to the shop, and I now realize oops, now I do not even know what to buy, you know? " "And then sometimes, there come times when somebody asked you that, you know, daddy, I have got my periods. I need money from you. How do you, as a dad, handle it? Here is a young girl getting her first period. I had to go buy a pad ." Participants spoke of having to buy pads for their daughters as a shameful experience. This attitude toward menstruation may be reflective of broader narratives around the meaning of menstruation as something to be hidden, controlled, and treated in secret [95]. In addition, fathers' shame around their daughters' menstruation could be understood as a reaction to an act of transgression against both the hegemonic and traditional gender roles of men, particularly Black men [49]. In other words, dealing with menstruation seemed to have been perceived as an offence to participants' Black masculinity as it blurred the lines between what is traditionally perceived as a masculine versus a feminine task [95,96]. This is particularly so considering that dealing with menstruation is often considered to be the role of the mother, not the father. Hence, participants may have felt the need to "rectify" this offence by reaching out to women for assistance. These experiences also reflect binary ways in which gender roles are still defined within South African families [19]. Furthermore, treating menstruation as a secret may also indicate to the girls that the very experience that proves their womanhood is unclean and something to be treated in secret. The discomfort displayed in the quote above also reflects participants' treatment of menstruation as a "women's problem" and a taboo topic. This may also limit participants' understanding of this phenomenon and may serve to perpetuate problematic narratives on menstruation and how girls are perceived by boys and men in society. Boys, on the other hand, seemed not to experience any pressure when transitioning from boyhood to manhood. Women also seemed to reinforce these problematic narratives and discourses around menstruation and how men should handle it. This is reflected in the comment below made by P1: "I remember when she was now moving into the adolescence stage, menstruation kicked in and everything. My sister said to me, you know what? There are things that this child cannot talk to you. Yes… let me just stay with her for that period so that she can adjust. Because menstruation is something to other kids…I mean, you are a man. Then I said, okay, fine… You know… ." P1's sister seemed to endorse views that men are not "allowed" to engage with their girls on menstruation, a view reported to be common among women, reflecting traditional patriarchal ideologies about how gender roles should be performed [86,95,96]. Moreover, the comment above reflects a narrow understanding of roles that men and fathers should play in their children's lives and exposes women's complicity in perpetuating particular discourses about womanhood and manhood that dictate what men and women "should" and "should not" do [53]. Furthermore, it reflects the idea that just because women alone go through menstruation, men cannot fully understand this experience. As a result, menstruation may have been treated as disgusting, something to be feared and avoided by men and boys, an experience that only the "others" should be concerned with [95,96]. Moreover, not only were boys not engaged in issues surrounding their own menarche , but nothing was mentioned about boys being educated on the topic of menstruation. This further confirmed findings from other studies that have discovered that women or mothers, more than fathers, function as sexuality educators in the home [95]. Furthermore, boys were reported to use informal and unreliable sources, such as friends or the media, to gain information on menstruation. Therefore, by not engaging boys on topics around menstruation, in addition to treating girls' menstruation as a secret, participants may be communicating to boys and girls that menstruation is obscene, therefore indirectly promoting notions that use menstruation to stigmatize girls and women as moody and difficult to deal with during this period [95]. By holding such views, relationships of power and men's assumed superiority over women might also be promoted, where women are perceived as weak and emotionally incapable because of these unique experiences. --- Support Systems Participants acknowledged the role played by various support systems in helping them fulfill their function as single fathers. All participants reported receiving support. However, the nature, purpose, and degree of the support varied from one participant to another. Furthermore, participants mentioned receiving support at different stages in their journey and depending on the number and age of children they raised. One of the participants stated: "But fortunately, my mom came, and took my son away from me to go and look after him. And I continued working. I am telling you, if I start to complain, some people will be by my door, wanting to know what is happening. I am very much thankful, and I am very grateful to the families, my family, my mom, my uncles, and everyone in my family because they are always there; support, help, how do you? " As P3 reflected in the extract above, some participants had to temporarily let their children live with extended family members or temporarily live with family members to balance caregiving and the demands of work. This is a common practice in South Africa, highlighting the role that the family can play in raising a child [4,7,97,98]. It also supports the common narrative prevalent in collectivistic societies and mentioned by P4 that it takes a village to raise a child [7]. In other words, raising a child is not just a mother or father's responsibility but also involves family members. Moreover, participants mostly mentioned receiving support from their own family members, as opposed to those of their ex-partners' families, especially for divorced participants. This support what has been reported in the literature and further highlights the role that conflicts between couples can have on children [5,86,99]. However, participants whose wives died reported receiving support from both sides of the family. This may indicate that the nature of the cause for becoming a single parent influenced how support was granted. Participants did not only receive support from family members but also from community members, friends, and colleagues who provided emotional, caregiving, and in some cases, financial support during bereavement and the divorce settlement. P5 prioritised his children's needs and was proactive in establishing measures to prevent him and his ex-wife from using children in the divorce battles. Co-parenting is often more beneficial for children, and this seems to be evident in P5's divorce experience [5,99]. However, other divorced participants received no support and experienced resistance from both their partners and their partners' family members, as stated: "But since the child is with me, she is not interested in coming to see her. Yeah, so it is like I am on my own. It is like, okay then, you have got what you wanted; let's see if you will cope ." "My son was brought to me that you must look after ." "For years now, I think she still has a grudge against me. We just greet. There was a time we never even used to greet each other ." In the three cases above, participants' partners left the task of parenting to the fathers. In addition, for P2 and P6, the mothers used their absence as a punishment to the father for wanting to become the child's primary caregiver. These mothers may also have used their absence to test the fathers' fitness to be parents. "One of the court orders was that we have an arrangement that on weekends she will be with her, yeah, but they cannot go there . She can be with her on the weekend… like not there… wherever she goes, you must make sure that, that boy is not there . For some reason, she just stopped coming. So, I cannot go after her and try to force her… there is nothing --- I can do ." This sentiment was also expressed by P3 and P6. As a result, fathers had to move their children between their own homes and different family members' homes. This instability in the home may have an impact on the children's emotional and overall development [4,5]. Moreover, fathers who did not receive support from their ex-partners reported either financial difficulties or issues with maintenance. For example, P2 mentioned that his girlfriend, despite being ordered by the court to pay maintenance, defaulted on payment. P6, on the other hand, spoke of his ex-wife's lack of involvement: "So, there was only one income. I did not ask… Their mother never sent money to me… she used to say, okay then, you wanna play the father Christmas, and be the Good Samaritan? You know, carry on… I never picked up my phone and called her for one cent ." These participants seemed to have experienced financial difficulties due to their partners' lack of involvement. The situation was even more difficult to manage as participants' exes appeared surprised by their ex-partners' involvement and then used this to challenge them to be "real" men and provide. The pressure to prove to their partners that they could provide may have increased the difficulty of approaching their reluctant partners to ask for support. Therefore, the children were the most impacted [4,5]. Participants' willingness to accept financial responsibility could also be reflective of their adherence to gender norms that equate fatherhood with financial/material provision [42]. In taking on this financial responsibility without objecting, these fathers, with the complicity of their ex-partners, seem to have been affirming their role as breadwinners, hence reinforcing prevalent hegemonic gender ideologies that to be a "real man" in South Africa, one needs to provide financially for one's children [19,44,47,92]. Moreover, issues with support experienced by participants also highlighted inequalities within the maintenance system in South Africa, which seems to operate under the assumption that only women and single mothers need money, not men and single fathers. This issue may have been influenced by discourses that position men only as capable breadwinners and women as recipients of maintenance money, therefore overlooking the complexities and diversities within the modern family structures and the active role that men, not only women, play in raising children [6,100]. Participants may have felt marginalised as not only are they faced with the challenge of not having to receive financial support from their partners, but societal expectations for them to be breadwinners, and structural barriers that still hold narrow descriptions of roles that men and women "should" play in their children's lives. Therefore, there is a need for the justice system, in general, and the maintenance system to acknowledge fatherhood in an active everyday form and shift it from the provider role [100]. "Absent mothers" were either preoccupied with other relationships or deliberately withheld the necessary support from fathers because of their marital issues. However, P7, who became a single father after his wife died, also reported experiencing challenges with support, even though he quoted relationships within his family of origin as a barrier to accessing this support: "They make sure that they respect my boundaries. I take my own decisions, so they do not interfere with that… There is a lot of uncertainty because they do not know what is in my head. I do not know what is in their head, but they are my family, and I love them to bits. But it is hard for them to interfere in how I do things because I had my way or I have my own way of doing things ." P7's lack of support from family related to his reluctance to seek it and a belief that he can handle the situation on his own, despite the challenges he faced with balancing various roles. As a personal trainer, P7 may have adopted a more individualist approach in addressing his challenges, which may have influenced his reluctance to seek help from the family structures available. This approach to problem-solving could also reflect notions that a man is supposed to be strong and take responsibility, with responsibility in this instance referring to not asking for assistance from others [47]. Whereas other participants referenced family members and friends, he mentioned briefly attending therapy, hiring a helper, and then relying on his own internal locus of control. P7 was not the only one who hired a helper, as P1 also used a helper to compensate for duties that he could not fulfill. However, P7 was raised by his grandmother, away from his family members, and acknowledged receiving some support from her, albeit minimal, as he stated: "So, I would say the support that I got from her is just making sure that when the kids are at home, she is there, as an adult. Nothing more, nothing less than that ." P7 added that he would only visit his family during the holiday seasons. This early experience may have created some distance in his interpersonal relationships with family members and influenced his perception of the role that family plays in one's life. Nonetheless, P7's experiences add to the diversity of single fathers' experiences and provide insights into how the men who were interviewed used support differently, despite having gone through similar experiences. Furthermore, participants received support from their current wives and partners. However, like the help received from family members, the help received from partners was primarily with caregiving. This further reinforced traditional gender-role ideologies, which hold that women are meant to be caregivers in the family [44,53,86]. Moreover, P6, who experienced challenges with finances, mentioned receiving money from his boss: "Luckily, I worked for a very good boss at that time, that, sometimes I was like honest with my boss and said to my boss, you know, today there is nothing at home. And my boss would say, okay then I will give you a thousand rand. I will give you a thousand rand just to add up for the girls ." The extract above also indicates that participants may have felt more comfortable approaching alternative and conventional normative sources of help, such as colleagues from work, for maintenance money as opposed to the court. This further reflected more traditional gender-role ideologies [45,52]. Moreover, participants and their adolescent daughters received guidance and parenting advice from participants' ex-partners and other mother figures available in their daughters' life. In addition to the support from family members and acquaintances, participants used their religion and spirituality, their own self-efficacy, and physical activities, such as running, as coping mechanisms. Most participants quoted their belief in God as a crucial source of strength and hope in raising their children. Furthermore, participants referred to the influence of historical, social, economic, and cultural factors in how they assumed their role as single fathers. Intrapersonal conflicts experienced by participants were influenced by the tension between previously held beliefs about manhood and fatherhood and their current and, at times, contradicting realities. There was tension and discomfort among participants, particularly when assuming roles traditionally attributed to mothers. This further highlighted the interrelationship between personal and broader narratives around manhood and fatherhood and revealed a struggle to relinquish hegemonic masculine ideologies. Moreover, the process of constructing a coherent narrative around participants' experiences as single fathers also reflected moments of tensions, contradictions, and inconsistencies in the process of navigating and refining various aspects of their selfhood or their identity [75]. --- Theme 2: Life Beyond Single Fatherhood Participants' experiences also extended beyond their role as single parents and involved their relationship with their adult children, healing from the divorce and death of their wife, finding new meanings to life, and entering new relationships. Participants also reflected on mistakes made during their journey and shared lessons they had learned from their experiences. Participants' relationships with their children changed over time. While some fathers reported being closer to their children, others said that they experienced distancing from their children. P3 and P6 reported being closer to their sons now than when they were young. Both participants still share the same household with their children. This may have influenced the need for both parties to maintain closer bonds. P3, who now lives with his 28-year-old son, reported that conflict with his son mainly arises due to his need for independence and differences in views, an experience reported to be common among parents raising adolescents and young adult children [92]. "Maybe now that I think he is independent. Because I think he knows, it upsets me that he thinks that he can reason for things ." In the extract above, P3 also seemed to imply that a child will always remain a child regardless of their age and that a father's views are not to be challenged by virtue of being a father. This reflects patriarchal thinking and a belief in traditional notions of parenthood that confer to an adult man and a father continual power over their children [92]. Although participants did not provide details on points of conflicts between them and their children, some issues that were noted to contribute to conflicts between adult children and their parents include communication and interactional styles, habits and lifestyle choices, values, politics, religion, and ideology, work-habits and household standards and maintenance [88]. Moreover, issues around adjusting to children moving out of the house or the "empty nest syndrome" were also reported [4]. For example, P1, whose daughter is now studying at a university in Cape Town, reported feeling lonely and anxious about his daughter's independence. He had to initially maintain contact with her through constant phone calls. "She is now alone on her own. I am no more there… no daddy, she is on her own. That was challenging me emotionally. I would have to call her almost every day in the morning to check if she is fine ." However, P1 and others who had adult children learned to accept their children's independence and the shift in their role as primary caregivers. This shift in the relationship between fathers and their children also reflected a moment of transition in the family life cycle, which necessitated an adjustment [4]. Participants maintained their relationship with their children through regular face-to-face visits and from a distance. For example, P6, whose three daughters had all moved out of the house due to marriage and work commitments, mentioned talking to his children using Skype during the lockdown period. "We like to Skype each other and speak over the phone and still have that father and daughter relationship because they were not working far from me. So, Erick , I had jobs where they were not far from me… I got them jobs right next to me ." Finding jobs for his children in closer proximity to his workplace was another way for P6 to maintain his relationship with his daughters. While some participants were grateful and content with the state of their relationship with their children, others believed that it was still a work in progress. P1, P3, and P5 all reported experiencing a sense of joy and satisfaction in their relationship with their children: "It gives me great pleasure. It gives me great pleasure that there are people… okay, not necessarily here… out there, even when I walk out the street here, you know they are people that say you have got a big boy? We still walk together, make jokes, we laugh… people admire… ." "So, I want to believe that the decision I took to be there for them is now paying off. I do not have an alien relationship with my kids… They grew up understanding that, yes, this is how it is. But we still have these two people in our lives. So, yeah, it helped lay that foundation from the onset… I am very happy, being a single father. The fact that I am spending time with my kids, but also, the fact that they also have equal access to their mother, and she spends as much time with them. I am happy. I know it is not an ideal father-and-mother family relationship ." Whereas P3's sense of pride was from the feedback he receives in social settings, P5 was proud to have had a relationship with his children beyond their early formative years and despite the divorce between him and his wife. P5's unique experience with his children became his little story, further highlighting the benefits related to parents' joint custody [5]. Although the mere presence of a father in his children's life does not lead to effective parenting, children whose fathers are present and involved, have been reported to benefit from the involvement of both parental figures [5,25,33,99]. Benefits related to fathers' involvement include children improved behavioural, cognitive, and socioemotional development, improved tolerance for stress, reduced depressive symptoms, high academic performance, increased empathy, self-esteem, and social competence [7,33]. Fathers' involvement also benefits fathers. The advantages of being involved in a child's life include improved health, lower levels of depression, and high levels of satisfaction. Overall, all participants reported experiencing a sense of pride and satisfaction to have been present in their children's life. For some, children were a motivation to remain committed to their role as a father, especially during stressful situations. However, not all fathers were satisfied with their relationship with their children. For example, P7 reported that his relationship with his children was still a work in progress. In addition to challenges in healing from the impact of his wife's recent death, P7 also mentioned issues such as not living in the same city and household with his children because of work. P7 only visits his children during the holidays or at the end of the month and cannot be present with his children daily. He reported experiencing stress and worry about not sharing the same household with all his children. This dynamic may also reflect the impact that cohabitation with children may have in their life. --- Limitations and Future Directions Although this study is significant and contributed to the body of knowledge on fatherhood in South Africa, it also presents some limitations. A very small sample of participants was interviewed. This means that views shared by participants interviewed may not necessarily be representative of all Black single fathers in South Africa. In addition, this study only included Black fathers as part of the research sample, hence limiting the relevance of its findings to only one racial group, especially for a country as racially diverse as South Africa. Furthermore, only fathers were included as participants and not their children. Therefore, the findings of this study are representative of one group of individuals involved in the process of parenting: fathers. Meaningful insights that children of single fathers may have shared to enhance the quality of the study have not been included, further limiting the relevance of this study. Interviewing children raised by single fathers may provide more information to be compared with what has been shared by fathers in the study. Despite the above-mentioned limitations, this study has been valuable in creating awareness of issues faced by single fathers as a marginalised community and has contributed to the diversity of voices and insights into fatherhood and parenthood in South Africa. Based on the limitations of the current study identified above, future studies should include a larger sample of participants to ensure that the views and experiences of single fathers are more representative of the larger Black South African community. In addition, including the voices of fathers from different racial groups and the experiences of children from diverse racial groups raised by single fathers will be beneficial. Furthermore, research focusing on understanding single fatherhood and fatherhood within non-heteronormative relationships, as well as research investigating "Black fatherhood" as a specific topic within the South African context, is needed to deepen our understanding of fatherhood and parenthood. --- CONCLUSION This study was narrative research on the experiences of Black single fathers living in Johannesburg, who raise children on their own. The study was crucial as it allowed for the inclusion of the voices of single fathers that have been underrepresented in the literature and discussion on fatherhood and parenthood. Using social constructivism and critically situated epistemologies as theoretical frameworks [20,49,51], the study used narrative interviewing as a method for data collection, as well as thematic analysis and Bamberg's little story approach to analyse data. Three key objectives were addressed in the discussion: identifying key themes and narratives based on participants' stories of their experiences as single fathers, discussing how single fathers negotiate their role and identity as single fathers, discussing the implications of the findings on parenting, with a specific focus on differences between raising boys and raising girls, and finally identifying differences in raising children of different developmental stages. From this study, it can be concluded that, contrary to the dominant narrative about fathers in South Africa, especially Black fathers, not all fathers are absent. The study demonstrated that fathers are just as capable as mothers of integrating roles traditionally associated with both fatherhood and motherhood. These findings offer an alternative to common narratives that portray Black fathers as absent, irresponsible, and uninvolved in their children's upbringing. In addition, they cast a positive light on parenting and Black masculinities in South Africa by allowing Black fathers to rewrite their narratives and contribute meaningfully to the fatherhood discourse. These findings reflect participants' attempts to move away from traditional and patriarchal ideologies about masculinity and fatherhood that only position men as financial providers. By embracing their ability to be caregivers, participants embraced more hybrid forms of masculinity, which allowed them to integrate both hegemonic and non-hegemonic masculinities [53,86]. However, the study also demonstrated that both a father figure and a mother figure are crucial to a child's optimal development. In as much as children growing up with maternal figures may benefit from the presence of paternal figures in their lives, children raised by single fathers also stand to benefit from having mother figures in their lives. This view was further supported by participants through suggestions to other single fathers, such as avoiding divorce as much as possible and building a solid and stable home environment for children. Findings further pointed to the idea that children are not raised in isolation but may benefit from the help of various family and community members [7]. Moreover, the results of our study further highlighted the caregiver role that mothers and grandmothers still play in their children's and grandchildren's lives. A closer analysis of these results reveals that participants may still be committed to patriarchal discourses of parenthood that consider heterosexual relationships as the "ideal" type of relationship for raising children [45]. For example, this was reflected in most fathers' attempts to solicit support from mother figures, especially when raising adolescent girls and faced with their daughters' menstruation. Participants spoke of the involvement of female partners and did not consider alternative family structures and dynamics outside of heteronormative relationships. These results support findings from other studies establishing that, despite progressive gender and family policies and legislation adopted by the South African government, traditional and patriarchal ideologies about gender roles are still prevalent in South Africa [19,101]. Therefore, participants' experiences reflected both an attempt to challenge dominant narratives about being a parent and an enactment of traditional masculine beliefs and ideologies of "gender" [51,102]. Moreover, participants referred to the influence of historical, social, economic, and cultural factors in how they assumed their role as Black single fathers. Considering such factors allowed us to appreciate the complexities, difficulties, conflicts, and tensions embedded in participants' experiences of having to negotiate gender roles in contemporary South Africa, especially for a generation of individuals raised in a patriarchal society in which gender roles are expected to be performed according to traditional norms. Despite the fact that participants identified as Black, they did not specifically refer to the role of race in shaping their experiences as fathers. However, the role that race has played in shaping the realities of many South Africans was considered [86]. Participants' narratives included experiences shaped by how they understood their identities as Black men in South Africa. For example, participants attempted in their stories to distance themselves from negative stereotypes about being a Black man and a Black father in South Africa by rejecting the discourse that Black fathers are absent and uninvolved in their children's lives. By doing this, they were also engaged in a protest against hegemonic and negative stereotypes about South African men. The intrapersonal conflicts participants reported in their experiences may be understood, through the lens of Bamberg's theory, as influenced by the tension between previously held beliefs about manhood and fatherhood and their current and at times, contradictory realities [75,77]. There was tension and discomfort among participants, particularly when assuming roles traditionally attributed to mothers. This further highlighted participants' fear of transgressing hegemonic notions of being a man and a father and the interrelationship between personal and broader narratives around fatherhood and thereby, between processes of personal and social change [47,53]. In addition, participants' experiences also showed how they navigated among the intersectionality of gender, race, and social class. While they benefited from the privileges of being men in a patriarchal society, they also grappled with challenges related to being Black and being a marginalised group of single Black fathers. The process of constructing a coherent narrative around participants' experiences as single fathers also reflected moments of tension, contradiction, and inconsistency in the process of navigating and refining various aspects of their selfhood or identity [75]. Participants found that prioritising the needs of their children over marital conflicts was important for building relationships with the children. Other participants suggested prioritising building a solid relationship with children and healing from the loss of a loved one or from divorce before entering a new relationship. Overall, participants' experiences as single fathers were characterised by intrapersonal and interpersonal tensions that revolved around choosing between their needs and those of their children. There are a number of implications to this study's findings. Firstly, in light of the results presented above, there is a need to engage all men in the process of gender resocialisation that focuses on renegotiating gender roles. In addition, the results of our study have highlighted the importance of creating opportunities for fathers to be more involved in raising their children by engaging in tasks beyond their traditional role as providers. This is crucial as it will initiate the process of breaking down gender-role stereotypes and create equal relationship dynamics within families. The changes suggested above can and should be implemented both at an individual as well as a broader societal level. This study has inspired new optimism about fathers' involvement in their children's lives beyond traditional norms of masculinity. In addition, the study has meaningfully contributed to the project of social justice by giving a voice to a marginalised community and by challenging limiting and problematic notions about fatherhood and parenthood in South Africa. --- LIST OF ABBREVIATIONS --- HUMAN AND ANIMAL RIGHTS No animals were used in this study. All procedures performed in studies involving human participants were in accordance with the ethical standards of institutional and/or research committee and with the 1975 Declaration of Helsinki, as revised in 2013. --- CONSENT FOR PUBLICATION Informed consent was obtained from all the participants. --- AVAILABILITY OF DATA AND MATERIALS All the data and supportive information are provided within the article. --- STANDARDS OF REPORTING COREQ guidelines were followed. --- CONFLICT OF INTEREST The authors declare no conflict of interest, financial or otherwise. 2. How did you become a single father? --- Appendix 3. How has your experience of being a single father been like? . 4. How has your experience as a single father shaped your identity as a male? 5. How has your experience been raising a boy/boys or a girl/girls or both? applicable to you). 6. How has your experience been raising a younger child/younger children or an older child/older children, or both? applicable to you). 7. What support do you draw from as a single father?

This study aimed to bridge the gap in the literature on parenting by investigating the experiences of Black single fathers living in Johannesburg, who raise children on their own.Social constructivism and critical theory served as theoretical frameworks, while narrative interviews were used to collect data. The study involved seven Black male participants, who were selected using a snowball sampling method. Dialogical analysis and Bamberg's little story approach were used to analyse the data collected.Two themes emerged from the thematic analysis: parenting as a single father and life beyond single fatherhood. The study's objectives were integrated into the discussion of themes and subthemes that emerged.The study results revealed that, contrary to the common narrative, not all Black fathers are absent. In addition, it pointed to the fact that single fathers are also capable of being effective parents. However, it was discovered that children benefit the most from the active involvement of both parental figures. Implications for these findings and recommendations for future studies were also discussed.

Background Adolescence is a time of considerable sexual development, driven by biological and cognitive changes entering puberty, coupled with changes in social expectations and interactions [1]. For both biological and sociocultural reasons, adolescents are at increased risk of poor sexual health outcomes such as sexually transmitted infections and unintended pregnancy [2]. Furthermore, adolescence is a time of increased vulnerability to sexual coercion and violence due to lack of experience, social pressure, and the progressive development of self-identity and self-esteem [1]. Sex education is one of the core strategies for achieving improved adolescent sexual health [2,3]. Desiging effective sex education has been the subject of numerous studies and characteristics of succsessful school sex education programs have been extensively decribed [4,5]. Many of these studies have shown a positive effect on knowledge of safe sex practices, however knowledge alone may not always be enough to ensure safe sex behaviours. In their literature review, Marston and King [6] point to wider barriers that inhibit safe sex practices, beyond a lack of knowledge. These include poor communication, myths and stigma related to condom use and gender stereotypes inhibiting safe sex practice or partner communication [6]. Thus to be effective, sex education should focus on increasing knowledge, developing skills and challenging sociocultural barriers to safe sex. In addition, sex education programs need to consider the challenges and opportunities presented by the unique sociocultural and political settings where the programs are implemented. In Chile this includes taking into account issues surrounding legal rights , religion, indigenous worldviews, immigration, societal and cultural norms and gender roles. Much literature on sex education is focussed on formal education in schools, however studies from settings as diverse as the United States [7][8][9], Vanuatu [10], Nigeria [11] and Brazil [12] attest to adolescents receiving and actively seeking sexual health information from a range of formal and informal sources simultaneously. Furthermore, the explosion of media globalization with the Internet has opened up new forums for sexual health and sexuality information [13]. Independent of where this information comes from, it may be ambiguous and contradictory, clouded in secrecy and emotional overtones [14], and shaped by the attitudes, ethics and values of the adult world [15]. In the Chilean context, previous quantitative studies name schools [16] and parents [17] as the main sources of sexual health and sexuality information utilized by adolescents. From these studies it remains unclear what information is provided by these sources, how this information is communicated and how adolescents judge the trustworthiness of the information they receive. This current research aims to address this gap in the literature. This study had the main objective to explore sources of information and adolescent learning about sexual health and sexuality in Chile. This paper reports on four subobjectives exploring 1) the sources of information Chilean adolescents use to learn about sexual health and sexuality; 2) the content of this information; 3) the techniques used to communicate this information; and 4) the strategies Chilean adolescents adopt to judge the trustworthiness of the information. --- Study context Chile is a traditionally conservative country, reflected in strong links between religious and political power, conservative "pro-family" legislation and strongly prescribed gender roles [18]. In the years immediately following the return to democracy in 1990, Chile made great advances in civil and political rights, however moves to improve sexual and reproductive rights were slow and inconsistent [15,18]. From the mid 2000s onwards, considerable legal advances were made. In 2010, a law was passed ensuring the right to sex education, access to information and services for the prevention of adolescent pregnancy, and outlining the legal responsibility of health professionals to report suspected sexual abuse [19]. This final point is of key importance since a 2012 national survey found 7.3 % of school-aged children had experienced some form of sexual abuse, whilst 22.4 % of women surveyed had experienced sexual violence at some point in their lives [20]. In 2012 an anti-discrimination law was passed which included discrimination on the grounds of sexual orientation [21]. Although discrimination based on sexual orientation and/or gender identity still exists, there has been a reduction in reported cases of violence against homosexual, lesbian, bisexual or transgender individuals since the peak of 186 cases in 2011 [22]. Even though advances in sexual and reproductive rights have been made, major issues remain, such as the penalization of induced termination of pregnancy, including therapeutic abortion. Chile has also seen changes in epidemiological trends in adolescent sexual and reproductive health. Adolescent pregnancy rates peaked in 2008-2009, followed by a downward trend thereafter [23]. In 2012, 14.4 % of all live births were to adolescent mothers 10-19 years old, down from 16.6 % in 2008 [23]. Regarding STIs, the most commonly diagnosed infections in Chile are genital warts and syphilis [24]. The highest number of new notifications for STIs and human immunodeficiency virus are in the age group 20-39 years [24,25]. It is unknown how many of these adults may have been infected during adolescence, especially given that prevention of pregnancy, rather than prevention of STIs has been described as the primary motivation for using condoms by Chilean youth [16,25]. --- Methods --- Study design This study is based on four months of field research in Santiago, Chile in 2013. Data collection included two pilot interviews with university students, two focus group discussions with a total of 14 high school pupils 18-19 years old, two FGDs with 10 university students 18-19 years old, 20 semi-structured interviews with high school pupils 16-19 years old, and seven semi-structured interviews with key informants working with adolescents. The semi-structured interview method was chosen to collect data on individual experiences, perspectives and opinions, whilst the focus group discussion method was chosen to collect data on overarching themes, as well as group dynamics. The initial pilot interviews were conducted to practice the interview structure, test the interview questions and tailor the interview guide to the Chilean context. Similarly, the first two FGDs with high school students helped adjust the interview guide through discussion of sources of information previously identified from the literature and encouraging discussion of any new sources the adolescents identified. Furthermore, these discussions provided a valuable opportunity to observe how adolescents discussed the study topic in a group setting. For individual interviews, a total of 20 adolescents were recruited from three high schools in the municipalities of Independencia, Recoleta and Las Condes. These three high schools were chosen to capture a diversity of youth experiences as the schools differed in size, socioeconomic status, academic rigour, geographic location, religiosity and sex education curriculum. 1 One school was private and Catholic, whilst the other two were partially subsidised, secular schools. All schools differed in their way of approaching sex education, and even in the same school between different year levels there were significant differences. Upon completion and preliminary assessment of adolescent interviews, seven semi-structured interviews were held with key informants working with adolescents, including three school psychologists, three health professionals and one Catholic priest. Finally, upon completion of all individual interviews, a further two FGDs were held at a public university to stimulate fruitful discussions on the preliminary findings. All interviews and discussions were conducted/moderated in Spanish by the first author with the support of a local research assistant to resolve language misunderstandings and explain culturally specific terms. --- Ethical considerations Verbal and written consent/assent was collected from all participants after talking through the intentions of the study, structure of the discussion/interview and confidentiality. Written parental consent was required for participants under 18 years of age. There was no direct benefit to participation and the potential harm was assessed to be low, as the aim was to study adolescent sources of information and learning about sexual health and sexuality, and not actual sexual experience. There was no financial incentive for participation, however snacks and drinks were provided. This study was approved by the Board of Ethics at the Faculty of Medicine, University of Chile, Santiago and the Norwegian Social Science Data Service. --- Sampling and recruitment Conducting research within the constraints of the high school setting meant that sampling and recruitment needed to be both focused and flexible, juggling the busy schedules of the school psychologists aiding recruitment, the pupils and the interview team . In response to these challenges and to ensure a gender balanced sample with maximum variation, a range of sampling techniques were utilized. This included maximum variation sampling, purposeful sampling of adolescent parents and purposeful random sampling [26]. For FGDs, homogenous sampling was the primary approach used, however poor participant turn out on each day of the high school FGDs mean that opportunistic sampling was necessary [26]. The project was presented by the first author to entire classrooms of high school or university students or to a selection eligible pupils . 11 adolescents over 18 years of age volunteered and seven of these were interviewed. Those volunteers under 18 years of age were given an informed consent form to take home to their parents. In total 18 out of the 39 volunteers returned parental consent forms and 13 of these were interviewed. For sampling key informants, information from adolescent interviews and FGDs highlighted the importance of schoolteachers and health professionals in provision of information. The influence of religion on sexual health discourse in Chile was also discussed, therefore a representative of the Catholic Church with extensive experience working with adolescents was recruited. These participants were all contacted directly by the first author and invited to participate. --- Interviews The final sample included 10 females and 10 males, three of whom were adolescent parents. The interviews opened with a general request for the participant to introduce themselves, before being asked to describe in their own words what they understood by the terms sexual health, sex education and sexuality. This was in order to ensure intersubjectivity: that the researcher and participant had a common understanding of the core terms used throughout the interview. Participant responses to interview questions were understood in relation to their unique understanding of these terms and probes were used to integrate elements of the interviewer's understanding. The core of the interview included questions and probes on each source of information identified from the literature, pilot interviews and FGDs: family, school, friends, Internet, health professionals, television, films, advertisements, radio and religion. It was neither the aim nor was it possible to ask all questions in all interviews. Some adolescents required a more exploratory approach with few open questions, whilst other adolescents required a more descriptive approach with specific questions and extensive probing. For key informant interviews, separate interview guides were developed for each group after pre-analysis of adolescent interviews. These guides included exploratory questions as well as examples of anecdotes or relevant themes from adolescent interviews and FGDs brought up for discussion. Interviews were conducted in private offices in the schools and key informant interviews were held in the offices of each participant. All interviews were audio recorded and lasted 45-75 min. --- Focus group discussions Four gender separated FGDs were conducted with a total of 24 adolescents 18-19 years old. For the first two FGDs in the high school the discussion opened with an 'ice breaker' activity whereby all participants were asked to write anonymously a list of all the words they knew related to sex, sexual health and sexuality. After discussion of a selection of these words, open questions were posed about different sources of information and how they believed Chilean adolescents utilized these to learn about sexual health and sexuality. There was also room for probing on emergent themes. For the final two FGDs a new thematic guide was developed with an 'ice breaker' activity where the participants were presented with the lists of words created by participants in the first focus groups and asked to comment on these. Following on, the guide included a number of themes or anecdotes from adolescent interviews that were interesting, surprising or particularly noteworthy. Again, there was room for spontaneous probing on emerging themes. The FGDs were held in an empty school library and a university conference room. The discussions were audio recorded and lasted 50-80 min. --- Data analysis Data analysis started upon entering the field through debriefing, journaling, prompt verbatim transcription of interviews/discussions and eight formal pre-analysis discussions in the interview team. All interviews and discussions were transcribed verbatim in the field by the first author and research assistant. Afterwards a selection of transcripts were interchanged between the two and crosschecked for accuracy. During the pre-analysis sessions transcripts were reviewed and recurrent or emerging themes were recorded. Throughout the data collection process the interview guide was continuously adjusted and through pre-analysis discussions it was possible to determine a point of adequate saturation of data in relation to the research objectives. Upon completion of fieldwork, the formal data analysis drew on Taylor-Powell and Renner's steps in content analysis [27]. Initially, in order to form an understanding of the data set as a whole, all interviews were listened through and transcripts were reviewed. Thereafter, the data was manually coded "descriptively" with the creation of 36 descriptive codes by source of information stemming from the research objectives and interview guides . Following on, the data was coded analytically, and through a flexible process of code expansion and deletion, categories developed based on the content, rather than source of information . These categories were recorded in colour-coded tables with examples the data, translated by the first author from Spanish to English, and checked by a native Spanish speaker. In the final step the data was analysed within the context of relevant theories and previous empirical research, leading to the development of three major themes . --- Results The primary sources of information utilized by Chilean adolescents in this sample are family, school and friends. There were gender differences in the utilization of family as a source, with females more likely to talk with one of their parents than males. Many adolescents talked with friends or partners about topics related to sexual health and sexuality. Although these conversations were seen as important, there was often resistance to the idea of friends and partners being sources of information in the formal sense. There were visible gender differences in the use of secondary sources, with females using health care services as a source of information whilst males were more likely to utilize Internet. Alternative media sources such as television, films, advertisements and radio were only sporadically mentioned, therefore they will not be reported on further in this paper. Similarly religion, or specifically representatives of Catholic or Evangelical Churches were not seen as sources of information per se, however religion was discussed in relation to other sources such as schools. --- Most trusted sources of information The most highly trusted sources of information were teachers, parents and healthcare professionals. Reasons for this included the innate belief that these adults wished them well, the life experience that these adults had and the length of time the health professionals and teachers had spent studying for their careers. Triangulation of information between these three trusted sources increased trustworthiness. --- Biological teachings All the adolescents interviewed received information about the biological aspects of sexuality from school and/or their parents. This included the topics of pregnancy, sexually transmitted infections and contraception . In schools these topics were most often covered in biology classes, with an academic focus, through powerpoint presentations with limited pupil participation. If covered in orientation classes, sexuality was often taught alongside the topics of drugs and alcohol, with a focus on risk. Many participants described their feelings of fear and/or disgust at being shown pictures of infected genitals during these sex education classes. Even though they were subsequently informed to use condoms to prevent STIs, only one participant witnessed a practical demonstration of how to put a condom on a broom. One participant shared his opinion of the absence of practical teachings of how to use a condom: This is what is also missing in school […] In school they don't teach you how to put a condom on. So obviously, you could arrive to the act and it doesn't help you to know all the theory if you don't know the practice. Formal school education focused on heterosexual relations and reproduction, therefore the only opportunity adolescents had to learn about the broad spectrum of sexual acts or other types of unanswered questions was through anonymous question sessions. During these sessions, adolescents posed questions about a range of topics moving far beyond coitus, for example incorporating topics of oral/anal sex and masturbation. One participant in the Catholic school described an example from such an anonymous question session: We were so young that the questions were like… "what does semen taste like". Similarly, two key informants described holding anonymous question sessions and being asked about a range of topics including bestiality and their own sexual experiences. Parents were another source of biological information with 17 adolescents having received information from their parents about pregnancy, STIs and/or contraception. Most commonly these conversations were triggered by a specific event such as an upcoming school camp, upon request from the school, when starting a romantic relationship or in the case of many females, with their first menstruation. One participant described her experience: I started dating and then came the massive attack from my parents about adolescent pregnancy and things like that. […] it was like they assumed that I was no longer a little girl, I was now big and they needed to start informing me. Parents commonly provided non-specific messages to "protect yourself " without explicit information on how to do this. Conversations were either one off experiences of "the talk", or the start of ongoing communication. These triggered conversations were most often experienced as embarrassing for the adolescents, however those with subsequent ongoing communication described increasing trust and diminishing embarrassment over time. In total four participants stated that their parents had themselves been adolescent parents, therefore communication was triggered by a fear that they would repeat the same scenario as their parent. This is exemplified in comments made two by female participants: My mother had me at 18, so she always had a fear, the whole family had this fear that I would repeat my mother's experience. Ever since I was very young she made me scared about birth, about the pain and all these things so that I would not fall pregnant. Thus, similar to the school setting, fear was also a tool used by mothers in communicating to daughters about sexual health. Midwives and gynaecologists also provided reproductive and contraceptive information for females who attended health centres with their mothers after their first menstruation, or when starting a romantic relationship. Only two adolescent males had accompanied their partners to see a gynaecologist, however one health professional described a new trend in both public and private health sectors with more males accompanying their partners for contraceptive advice. --- Emotional and relational teachings Compared to the biological teachings, few adolescents received information about the more emotional and relational aspects of sexuality, falling under the Spanish umbrella term afectividad. In this study, adolescents and key informants described afectividad loosely as encompassing themes such as love, romance, attraction, desire, pleasure, affection, caring, self-esteem and communication. When probed about afectividad, two male participants associated this topic exclusively to religion. Unsurprisingly therefore, the Catholic priest interviewed discussed at length the importance of sexual relations being within the context of a committed, loving relationship. If topics related to afectividad were taught in schools, this was frequently during religion classes. One female participant described how, unlike secular schools, religious schools automatically assumed it was their role to teach adolescents about afectividad. Another female described how she had learnt about afectividad in religion classes, but that as she got older the focus of her school teaching changed: They started saying, I don't know, that "love is not anything magical, instead it is just biological reactions" […] like they tried to remove all of the magic. Thus a form of biological reductionism converted the magic of love to a purely biological process of chemical reactions, separate from emotions. This separation of the biological and emotional/relational aspects of sexuality in mainstream teaching was critiqued by a psychologist whom stressed the importance of integrating psychosocial elements of sexuality including love, when teaching sex education: Often I also tell them [the pupils] that one of the risks of sexuality at such a young age, is not just pregnancy and sexually transmitted infections, it is that you fall so deeply in love, that you fall into a depression. Key informants who taught sex education gave a number of reasons for why educators focus on biological sexuality, stating it was a reflection of narrow educational and public health agendas in Chile and that biology was easier to teach than afectividad. Furthermore, they stated that to teach about sexuality in depth, educators needed to acknowledge their own myths regarding their sexuality. Both adolescents and key informants described self-confidence and trust as prerequisites for communicating effectively about sexual health and sexuality. Love and affection were not spontaneously mentioned by any participant, however when probed, six participants had talked about these topics at home. From home these messages were either that love and sex were two separate concepts or that it was necessary to have love before engaging in sexual relations. One participant stated that parents were those responsible for teaching about values and love to their children, letting schools and health professionals take responsibility for teaching biological sexual health. He described what he had learnt from home: When I was young they talked a lot to me about sex being the most beautiful thing that exists in life and that it is a form of sharing love. This positive message of love and sex contrasts to the direct association of sex to risky behaviours such as consuming drugs and alcohol, a common element of school sex education programs. --- Sexual pleasure and masturbation None of the participants had learnt about the topic of sexual pleasure in formal sex education and all four key informants that actively worked with school sex education discussed reasons for this omission. The main reason was the belief that talking about pleasure counteracted the aims of sex education; namely to delay sexual debut and reduce adolescent pregnancies. One psychologist stated her opinion that teaching on the topic of pleasure was a challenge, but not impossible, if approached from the angle of emotions. One adolescent shared her opinion of teaching on pleasure: I think that if they cover the topic of pleasure, they have to cover it together with the diseases, always subtly, but each one in their place. […] one can't just cover the diseases to make them scared. The only element of sexual pleasure which was incorporated into formal sex education was male masturbation. Four males recounted how school psychologists had talked of masturbation as a natural part of male sexual development. One male recounted his experience: The professors tell you that " for you guys it is normal". Even when the topic of male masturbation was broached in the school setting, none of the participants recalled any messages related to female masturbation. One participant critiqued this: I remember very strongly in school the taboo topic of knowing oneself, […] It has always surprised me, like, "why can't women touch themselves?" This gendered taboo of masturbation was universally reflected in the adolescent interviews. --- Sexual diversity The topic of sexual orientation was also often absent in the formal education and one key informant explained why: One approaches what one sees, approaches sexuality from the point of view of prevention of adolescent pregnancy, which in its essence is what scares us adults. Thus again, the education and public health agendas limit the breadth of teaching on sexuality. This same health professional also said that sexual orientation was often omitted in schools since many people in Chile still viewed sexual diversity as a question of core social values. If discussed in schools, the topic of sexual diversity was commonly taught in one of two ways. Firstly, five students were taught about respect and not discriminating against someone on the basis of their sexual orientation and in one school there was a student debate about social and legal rights of homosexual people in Chile. All three school psychologists dismissed the idea of sexual orientation being a cause of bullying in their schools, however one participant described discrimination of a classmate based on suspected homosexual orientation and female FGD participants described a negative attitude towards sexual diversity in their school with peers pointing out students believed to be gay or lesbian. The second perspective was a biological explanation of sexual diversity. Four participants described learning in biology classes that homosexuality was the result of a hormonal imbalance. For example one female described how her teacher had explained why homosexuality occurred: Sometimes they had lots of male hormones or female hormones and that is why it happened. Sexual orientation was sporadically discussed in the home setting, often in reaction to issues surrounding sexual diversity in the media or through comments on the increased visibility of same-sex couples on the streets of Santiago. These discussions focused primarily on personal opinions supporting or rejecting sexual diversity, or the importance of not discriminating. Two health professionals and one school psychologist described referring adolescents suspected of being homosexual, lesbian or bisexual to clinical psychologists and/or psychiatrists for further follow-up. One health professional described her role supporting a homosexual, lesbian or bisexual young person: Explaining to him that it [homosexual/lesbian/ bisexual orientation] […] could be something transitory or something definite but he has to work on it, because when he is ready to reveal his sexual orientation, he faces a society that is still not prepared. On the one hand, these teachers and health professionals medicalized non-heterosexual sexual orientations as a conditions in need of specialist follow-up, whilst on the other hand, referral opened up a supportive space for counselling adolescents who may receive limited support in their home or school/peer settings. --- Abortion and emergency contraception The topic of abortion was rarely discussed in the school setting, the exception being the Catholic school where five participants described a student debate on decriminalization of therapeutic abortion. Compared to other participants, these adolescents talked about abortion with fluidity and reflectiveness. Emergency contraception was scarcely covered in schools and no participant recalled having received information on the biological functioning of the pill or how to access it. One adolescent and one psychologist stated their opinion that teaching about emergency contraception would encourage unprotected sex. If covered, emergency contraception was discussed from a moral standpoint, focusing on responsibility and blame for unsafe sex. One male participant described the message he received in religion classes: We need to care for our bodies and avoid having a sexual relation where the female could fall pregnant, instead of just coming and taking the pill, like taking the easy way out. Although a number of female adolescents had discussed contraception with their mothers, few had talked about emergency contraception. One female participant described the message her mother had given her and her sister: My mother says that, if one day someone rapes us, it would be the only time when she would accompany us to take it [emergency contraception]. Thus, the moral aspect of taking emergency contraception was discussed in both the school and home settings with a strong discourse of blame and responsibility for having unprotected sex. --- Sexual abstinence and defining sex Surprisingly, the topic of sexual abstinence was almost absent in adolescent discourse. Only one participant, an adolescent mother, recalled learning in school that "abstinence is the best contraception" . The idea that Chile was going through a generational change in relation to abstinence was common, with one female stating that abstinence was "no longer relevant for Chilean society" . A number of adolescents, particularly male FGD participants, laughed at the concept of abstinence. In response to this observation, one key informant shared her opinion that it was important that educators included the topic of sexual abstinence as some adolescents still saw it as a valid option. School teaching on sexual abstinence was criticised by male participants from the university FGD, who described a double standard whereby youth followed the teaching to abstain from vaginal sex, instead having unprotected oral or anal sex: For example, they say "let's abstain and do it orally", and in the end they still catch AIDS or some other thing. You still catch other things, […] like abstinence "no I can't", "ok, from behind then" . But that still won't do. This highlights the ineffectiveness of teaching on sexual abstinence without first understanding how young people define sex, virginity and abstinence. It also underlines the limitations of safe sex messages that focus primarily on prevention of pregnancy rather than prevention of STIs. Only one evangelical participant discussed abstinence at home, with female friends and partners. A further three participants with religious parents had talked about abstinence before marriage at home, however in all cases, abstinence was only ever referred to as an ideal, rather than a strict rule. --- Sexual violence, limits and negotiation When it comes to sexual violence, none of the adolescents in this study described being taught in schools about setting personal limits, issues of consent or partner violence. A school psychologist critiqued this by quoting a Chilean saying: I think that it is serious that it [violence in relationships] is not talked about. It is considered part of our culture. "The one who loves you, beats you". Two female participants recounted being warned at home about the risk of kidnapping and random sexual violence, however only one participant described how her mother told her to set limits with a partner to avoid verbal, physical or sexual abuse. No participant described having received guidance on knowing when one is ready for sex. Similarly, none of the participants reported having learnt skills for communicating with a partner about the timing of sex and negotiating condom use. --- Talking about sexuality in the informal peer network Participants also described conversations with friends and a partner about a variety of sexuality issues. Friends and partners encouraged varied levels of trust, some participants describing increased trust if the friendship had evolved over a long time. However, there was also scepticism about where these friends had found their information. Friends occasionally broached the topics of contraception or testing for STIs, usually with the purpose of giving safe sex advice. Notably, the three adolescent parents interviewed saw themselves as trustworthy sources of information as they had first hand experience of how tough it was to become parents so young. One of these mothers described her role: I always tell them, "please protect yourselves. It's wonderful to have a baby, but, at our age… because it's so early, you're still in school, entering university…" so I'm always like the mother of the group. It was common for participants to talk with same sex friends about various sexual acts. One female participant described a lunchtime conversation with a sexually experienced classmate: We asked her everything, how did she do it, why did she do it, how did it start, […] we had so many questions, since we haven't done it yet, we're virgins, so we wanted to know everything. Conversations with same sex peers could also take the form of social pressure to engage in sexual acts. One male described a joke made about a fellow classmate who had not had sex with his girlfriend after two years together, asking "when are you going to become a man?" . This example reinforces the gender stereotype that having sex defines "being a man". One psychologist described building selfesteem as fundamental part of sex education so that adolescents do not to submit to this kind of social pressure. Although communication between male and female friends about sex was uncommon, sexual health topics were often discussed with a partner. All participants said it was necessary to discuss contraception with a partner and ten participants recounted previous conversations with a partner. One male described how he felt talking to his partner about contraception: It was evident that the participant felt mature in this situation and confident in sharing responsibility for sexual health choices with his girlfriend. This stands in stark contrast to one adolescent mother's response when asked if she had talked with her partner about sexual health or sexuality before her pregnancy: This quote illustrates a traditional gendered responsibility for contraception which one health professional described as less and less common nowadays. She observed of a cultural shift with newer generations of young females expecting their partners to accompany them to seek contraceptive advice and demanding shared responsibility for contraception. I When probing adolescents on topics that they thought should be discussed with a partner, 16 participants described the importance of discussing contraception and protection from STIs. Four males and no females spontaneously included pleasure as a topic for discussion. --- Information on the Internet and attitudes towards pornography Internet was rarely referred to as a source of sexual health information in a formal sense. A number of participants had utilized Internet for school projects, however these participants were sceptical to the truthfulness of the information. All participants who used Internet described different triangulation techniques for assessing critically the trustworthiness of information, often prompted by a trusted adult advising them to be critical of certain websites. These techniques included crosschecking information on various websites to assess the congruency of the information, or double-checking the information found on the Internet with a trusted adult or friend. Alongside asking sexually experienced friends, searching on the Internet and specifically watching pornography, were described as ways of learning about different sexual acts. When asking about the acceptability of pornography, there was a universal gender difference in the responses. Males described viewing pornography as normal, although potentially only acceptable up to a certain age, whilst females regarded it as taboo. One participant named pornography as a causal factor in adolescent pregnancy: There's a misinformation. Because of this I think that sometimes youth, younger than me, have babies […] Because basically in a pornographic video you see people having sexual relations, right? They don't show you how to put a condom on. The participants in the two university FGDs discussed how pornography filled a gap in formal sex education. One female participant explained: I think that pornography talks about precisely the topics that aren't touched upon in sex education, which is like coitus, like the act, like how to be sensual, how to be desirable, like, the whole part about pleasure […] in school they talk to you about sex as something so scientific, like "ok, so there is penetration, ejaculation and a baby is formed bla bla bla bla". This more reflective and positive attitude towards pornography as a source of information on explicit sexual acts contrasted to the dismissive attitude of many younger, particularly female adolescents interviewed. --- Discussion Two major themes emerging from the findings will be discussed in this paper: the biological and heterosexual focus of sexual health information from trusted sources; and the lack of information on practical skills for ensuring safe and pleasurable sexual experiences. --- Biological reductionism and heteronormativity Adolescents in this study received considerable information on biological aspects of heterosexual sexuality and reproduction as it related to pregnancy and STIs. Although this is an important part of their sexual health knowledge, the potentially negative effects of this overriding focus on biological and heterosexual sexuality are worth exploring. When appraising the content of the biologically focussed sexual health information, it becomes clear that adolescents are expected to take rational decisions based on being cognitively aware of negative biological outcomes of unprotected sex. Focussing on biological sexuality without also discussing the emotional/relational aspects of sexuality effectively separates the mind from the body, and the individual from their social context. This artificial separation fails to take into account the emotional and physical impulses that drive sexual attraction, nor does it account for the social forces, gender inequalities and power relations that influence a person's ability to initiate positive health behaviours [28]. To underline this rational decision-making process, fear arousal was a tool utilized by both teachers and parents to warn adolescents about risks and consequences of unprotected sex. However, this information was rarely paired with practical explanations of how to access and use contraception, nor how to negotiate safe sex with a partner. Research has shown that utilization of fear arousal techniques may be counter productive, leading to risk denial [29], particularly exacerbated in adolescence by a belief in their own invulnerability or invincibility [1,3]. This rational perspective also artificially separates the individual from their social context, disregarding the influence of peers, partners and parents on sexual decisionmaking. In this study negative peer pressure was exemplified though jokes and comments about virginity as well as discrimination of supposedly homosexual peers. This peer pressure may negatively influence sexual health decision making, as found in a South African study where peer pressure undermined health promotion information encouraging safe sex and HIV prevention [30]. Conversely, peers may have a positive influence as seen in this study with the adolescent parents who described their role in promoting safe sex in their peers. When it comes to parents, a study from Cameroon found that perceived parental support for condom use had a strong association with actual condom use [28]. Therefore it seems important that educators promoting safe sex behaviours also include discussions about peer, partner and/or parental attitudes that may greatly influence sexual decision-making. Rationalization also fails to take into account the influence of conflicting messages from media, in particular pornography and erotic music. In this current study viewing pornography was described as normal for males and attitudes of towards pornography were gendered, consistent with previous research [31]. The sheer extent of adolescent exposure to sexually explicit material on the Internet makes a strong case for broaching the topic of pornography with adolescents to raise their "media literacy" in understanding how media is produced and how to be critical of behaviours promoted through the media [13,32]. Adolescents in this study were already critical consumers of information on the Internet in general, therefore there is every reason to believe that through critical debate about pornography and other media eroticisation, adolescents can be encouraged to reflect on the conflicting messages between rational risks of unprotected sex and media eroticisation. Another limitation of biological reductionism is the effect it has on adolescent understandings of sexual orientation. Although there were examples of adolescents whom had learnt about sexual diversity from the perspective of respect and non-discrimination, this approach was far from universal. With an overriding heteronormative, biological focus on prevention of pregnancy in the school setting, the topic of sexual orientation was commonly omitted or explained in biological terms as a hormonal disorder or "condition" making someone different. Additionally, medicalization in the form of referring suspected homosexual, lesbian or bisexual adolescents to psychologists and psychiatrists also supported an understanding of these orientations as conditions in need of medical support. This may further alienate youth with non-heterosexual orientations already at greater risk of poor health outcomes and bullying, whilst failing to explore with these adolescents the emotional and relation aspects of their developing sexuality [33,34]. Providing support for adolescents in the form of discussing same-sex attraction from a normality perspective focusing on feelings and attractions rather than biological differences may be important in preventing negative social and health outcomes [33,34]. This support could come from formal sources such as teachers or health professionals. Somewhat paradoxically, through the medicalization of sexual diversity, health professionals in this current study were given a point of contact to provide this emotional support for these youth. A heteronormative focus may also distort the understanding adolescents have of the wide spectrum of sexual behaviours, regardless of sexual orientation. Adolescents learn about this spectrum of acts primarily from informal sources through conversations with friends and partners, Internet and pornography. In schools, adolescents learnt about the spectrum of sexual acts and particularly noncoital sex through anonymous question sessions. This reflects findings from a study from the United States that analysed the content of anonymous questions written by 795 pupils 11-12 years old and found three times as many questions related to non-coital sex, compared to vaginal sex [35]. This trend indicates that adolescents see their teacher as an appropriate source of this information on diverse sexual acts, however the current taboo of the topic means they only ask in the context of anonymity. The heteronormative nature of safe sex and abstinence messages were also discussed in the male university focus group where participants observed that many adolescents view oral and anal sex as both physically safe and morally safe . The idea that oral and anal sex maintain virginity is not unique to the Chilean context, and may be influenced by conservative religious ideologies promoting abstinence and virginity [36]. As these participants discussed, implications of this may include low perception of risk related to anal and oral sex, which could lead to reduced condom use and increased risk of STIs [36]. A further serious implication may be an underreporting of sexual abuse not regarded by adolescents as either sex or abuse. Another limit to this biological focus was the lack of information on emotional and relation aspects of sexuality in the home and in schools. In the home setting discomfort in talking about afectividad was reflected in those parents who provided triggered, short, preventative messages linked to risk of pregnancy and contraception, without developing ideas of love, relationships, feelings, desire or sexual impulses. In schools the limited teaching of afectividad in mainstream teaching was linked to the narrow educational and public health agendas focussing on pregnancy and STIs. This is of particular importance in light of results from a Chilean study showing that the main motivations for becoming sexually active described by adolescents are precisely love, emotions and pleasure [17]. Therefore, in order to communicate effectively with adolescents about broad topics of sexuality beyond biology, both parents and teachers may need guidance in confronting their own sexuality, myths, cultural taboos and other barriers to communication on these topics [37]. An interesting emerging finding about school sex education was that themes such as love and what one participant called the "magic" of sexuality were often associated directly to religion classes. This raises the concern that non-religious adolescents may dismiss the value of learning about emotional and relational aspects of sexuality in school, if they only associate these teachings with religion. Similarly, by linking the idea of sexual abstinence only to religion, an important part of sex education, namely acknowledging the decision to wait to have sex until after marriage as a valid choice independent of religion, may be dismissed by adolescents and educators alike. In their study from the United States, Jones et al. [8] found that adolescents had a range of definitions of abstinence and that that abstinence messages were often seen as compatible with contraceptive messages. This differs from findings in this current study where abstinence and contraceptive messages were described as mutually exclusive. Therefore, exploring the meaning of abstinence in contemporary Chile presents an interesting area for future research. Finally, as seen in this study, the narrow public health and educational agendas focused on reducing the number of unplanned pregnancies and STIs meant that sexual pleasure was absent from formal sex education. The only exception was male masturbation, however the complete absence of the topic of female masturbation emphasises a gender disparity that denies female empowerment and entitlement to sexual pleasure [38]. Fine and McClelland [39] criticize the separation of risk and pleasure in formal school sex education, claiming that it distorts the way adolescents understand human relationships and desire. What both male and female adolescents in this current study showed through anonymous questioning in schools, conversations with friends and watching pornography, is that it is very often the topics of pleasure and desire that spark their curiosity. A focus on the acceptable biological, heteronormative teachings across formal and informal sources is not unique to the Chilean context. Results from quantitative and qualitative studies from countries as varied as Brazil [12], Norway [40], New Zealand [38], Vanuatu [10], the United Kingdom and Australia [37] describe similar situations, suggesting that the heteronormative, biological and risk focus is a tendency in a diversity of global contexts. --- The theory-practice gap A second emerging theme is the lack of focus on the practical aspects of ensuring optimal sexual health and sexuality. A number of adolescents criticized information from schools and parents for not adequately preparing them with tools to practice safe sex. To underscore the importance learning practical skills, an extensive systematic review of 83 school sex education programs claims that the most effective programs in delaying sexual debut and encouraging protected sex are comprehensive sex education programs that incorporate teaching skills related to partner communication and contraceptive use, as well building self-efficacy to be able to use these skills [4]. Information about contraception was often provided by schools, parents and/or health professionals. Condoms were the primary form of contraception discussed, however the actual mechanism on how to put a condom on was only shown to one adolescent in school. Casas and Ahumada [15] describe how practical teaching about condom use in schools touches on a sensitive nerve in the Chilean context, where there still exists a myth that teaching about contraception, and showing how condoms work will encourage early sexual activity. This myth has been dispelled by extensive evidence showing that comprehensive sex education is more effective than abstinence only sex education in delaying sexual initiation [4,5]. Another major contraceptive issue raised in this current study was the lack of morally neutral and medically correct information about the emergency contraceptive pill and how to access it. Adolescents had either not received information about the emergency contraceptive, or received vague information as to its efficacy, overshadowed by moralizing messages of guilt and responsibility. Although access is now guaranteed by law in Chile [19], conflicting and ambiguous information about the emergency contraceptive, may impede an adolescent's ability to make an informed decision about its use. Chile is not unique in this respect, as debates on the function and legal status of the emergency contraceptive are common throughout Latin America [41]. A core practical skill for promoting pleasurable and safe sex practices and preventing situations of coercion and miscommunication, is learning how to communicate effectively with a partner about sex [5]. In this study the adolescents overwhelmingly discussed the importance of communicating with a partner on topics of sexual health and sexuality, however actual experiences were varied. These differences may be explained in relation to differing levels of self-efficacy felt by adolescents in power relations with their partners. Breakwell [14] describes the link between the theory of self-efficacy and power, stating that effective communication helps set rules of engagement on when and where sex takes place. Conversely, infrequent partner communication, fear and/or perceived lack of ability to negotiate condom use has been shown to be directly associated with infrequent condom use [42]. This supports the need for an increased focus on communication skills necessary for effective negotiation of consensual, safe and pleasurable sexual experiences. The final skill relates to setting personal limits and learning to respect other's limits in order to prevent situations of violence and sexual coercion. Few participants in this study reported conversations with trusted adults about setting personal limits and partner violence, representing a gap in sex education. Globally, the majority of gender-based violence is perpetrated by men against women and most of the time this is by someone known to the victim, often their partner [43]. Interestingly, a Chilean study surveying university students found males experienced more dating violence than females, however females were more likely to be physically hurt when exposed to dating violence [44]. Few of these victims had talked with a friend about the violence and none had gone to the police, thus the authors advocate for open discussions with adolescents about partner violence, what it is, how the law can protect them and how to support friends that report violence [44]. Alcohol and drugs may be involved in many coercive sex situations, and although their influence should not be overlooked, focussing only on what potential victims can do to reduce risks may merely perpetuate victim blaming rape myths. Thus it is important to engage both males and females in discussions on issues of respect, consent, personal limits and violence, so they can be respected and empowered in their sexuality. It is a human right to have a sexual life free from coercion, discrimination and abuse, therefore sex education that includes building skills for both setting and respecting personal limits represents a great opportunity to help break the current patterns of sexual violence in Chile. --- Limitations This study has limitations. Firstly, the voluntary nature of participation means that the sample may be skewed towards adolescents with a strong interest in the topic, or adolescents with particularly open communication with their social network about sexuality issues. The need for parental consent may have bias the sample towards adolescents with open communication with their parents, however three participants whom reported no communication with their parents on sexual health issues still received consent to participate. Finally, there may have been recall bias since participants were asked about experiences they might have had many years earlier. Follow-up interviews and participant validation of the data may have reduced this recall bias, however this was not feasible within the constraints of sampling school pupils. --- Conclusion Chile has made considerable advances on the sexual and reproductive rights agenda in recent years. Findings in this study suggest that the traditional taboos restricting dialogue about topics such as sexual diversity and abortion seem to be broken and certain prescribed gender roles may be changing, whilst challenges to discussing topics like emergency contraception, sexual violence and sexual pleasure still remain. Therefore it seems an opportune moment to re-evaluate the way in which trusted adults engage with adolescents on these topics. Parents, teachers and health professionals were the most trusted sources of information in this setting, therefore it seems logical to focus interventions on these channels. In order to promote critical thinking, trusted adults are encouraged to build on the foundation of biological sexual health information currently provided and guide adolescents in reflection on broader topics of sexuality. These topics include sociocultural and media influences on sexual decision-making; sexual orientation, gender and the diversity of sexual practices; peer pressure, discrimination and violence; and emotional and relational aspects of sexuality. Furthermore, it is important to promote development of practical skills in relation to contraceptive use, partner communication and personal limits, as well as building self-esteem and self-efficacy necessary to use these skills. Moving away from fear-based, moralizing approaches towards more reflective and participatory dialogue with adolescents represents a positive step on the way to encouraging adolescent to be critical thinkers and empowered in the decisions they make regarding their sexuality. --- --- Abbreviations STI: Sexually transmitted infection; HIV: Human immunodeficiency virus; FGD: Focus group discussion. --- Competing interests The authors declare that they have no competing interests. --- --- Authors' information Not applicable.

Background: Sexual and reproductive rights include access to accurate and appropriate information in order to make informed decisions. In the current age of media globalization and Internet, adolescents are exposed to information about sexual health and sexuality from a myriad of sources. The objective of this study was to explore sources of information and adolescent learning about sexual health and sexuality in Santiago, Chile. Methods: Data collection included four focus group discussions with a total of 24 adolescents 18-19 years old, 20 semi-structured interviews with adolescents 16-19 years old, and seven interviews with key informants working with adolescents. Audio recordings were transcribed verbatim and analysed using content analysis. Results: The primary sources of sexual health and sexuality information were parents, teachers and friends, whilst secondary sources included health professionals for females and Internet for males. Information provided by the trusted sources of parents, teachers and health professionals tended to focus on biological aspects of sexuality, particularly pregnancy and sexually transmitted infections. Limited emphasis was placed on topics such as love, attraction, pleasure, relationships, abstinence and sexual violence. Information focused primarily on heterosexual relations and reproduction. Adolescents learnt about relationships and sexual acts through friends, partners and, for many males, pornography. Findings indicate a lack of available information on partner communication, setting personal limits, and contraception, including morally neutral and medically correct information about emergency contraception. Conclusions: This study highlights numerous gaps between adolescent information needs and information provided by parents, teachers and health professionals. The priority these trusted sources place on providing biological information overshadows learning about emotional and relational aspects of sexuality. This biological rationalization of adolescent sexual behaviour neglects the way gender inequality, peer-pressure, coercion, media eroticization and religion influence adolescent sexual decision-making. The heteronormativity of information excludes other sexual orientations and disregards the diverse spectrum of human sexual behaviours. Finally, the limited provision of practical information hinders development of skills necessary for ensuring safe, consensual and pleasurable sexual relations. Trusted adults are encouraged to engage adolescents in critical reflection on a broad range of sexuality topics, dispelling myths, and building knowledge and skills necessary to make informed decisions.

INTRODUCTION An internationally recognized method of advancing gender equality is female normalization . The process of gender mainstreaming was starting through reliable strategies to achieve gender equality . This is part of Human Right issues under Beijing Platform for Action . In an effort to remedy glaring cluelessness for Women, gender mainstreaming was introduced in year 1995. On the whole, gender mainstreaming encompasses all aspects of planning, implementing and monitoring any social, political or economic actions The concept was developed by United Nation through international platforms and gradually from there it came to be adopted as important of policy initiative in most of the developed and developing countries. Malaysia also was being one of the countries who are agree to implement the policy. Hence, process of Gender Mainstreaming is involving relationship at international and national strategies happening to achieve gender equality in Malaysia. GM that was introduced internationally and adapted in Malaysia will be discussed based on approach Political System Analysis and Policy Process that was discussed by Birkland 2001. The adoption of GM is based on global issues that have influenced the national issues and resolved through Malaysia's policies. Acceptance of global policies as the solution of national issues has shown that the process of implementation has resulted GM implementation practices in Malaysia. Policy formulation at the national level in Malaysia was implemented at national and local level to achieve gender equality. One widespread misconception is that changing organisational "internal" processes will also need changing operating "outside" processes. The former relates to adjustments that must be made inside organisations in order to embrace the principles and aims of mainstreaming gender and to adapt policies and practises to achieve these objectives . Political leaders will be compelled and able to include gendered standpoint parity in their programs by reforming the policy-making process. This approach will result in a fundamental shift, eradicating gender prejudices and reorienting policy to support the objective of women's rights . And therefore, GM is required for Malaysia to attain equality between the sexes, which is determined by the participation of women in centralized decision. --- BACK GROUND OF MALAYSIA Malaysia now has a greater population, up from 32.4 million in 2018 to 32.6 million in 2019. Males outweigh females in 2019 . As of 2018, there were 16.8 million more men than there were in 2018. Likewise, there were 15.8 million more women than there were women as of 2018. Since 2013, there are 107 males for every 100 females in the population as a whole . The Malaysian women's population is increasing as well as their achievement in education as compared to the men, they are still experiencing challenges to entered remain at higher level management. Malaysia has one of the lowest rates of female labour participation in Asean, per a survey by Moody's across the Asia-Pacific . Despite the fact there are more women than men enrolling in higher education institutions and that girls often perform better in school, just 39% of Malaysia's workforce is made up of women, according to a World Bank Study published in January in working-age of labour force the 80.4% of were men , but women it is only 55.2%. But, more than 50% of graduates are women. Within all major educational groups, women's average earning are 70.6% to 83.2% of men's average earnings. In 2017, 22.1% of managers in Malaysia were female. Overall, women participation in the public sector in Malaysia has increased. However, Women continue to be disproportionately represented in positions of higher leadership and decision-making . Besides, Zaharah mentioned that the promotion for women at work is slower than for men. It was influenced by many factors such as family and work life balance, discrimination, gender stereotype to women as a leader and, not comfortable to women's leader. In many nations around the world, there are still few women who are involved in making decisions in the government sector. Statistics showed a rising trend for women in positions of greater responsibility in some countries. For instance, according to U.S. Bureau of Labor Statistics in 2002, women occupied 46% of managerial and administrative posts in industrialised countries like the United States, up from 18% in 1972 . Although the percentage of women holding high managerial and administrative positions had increased, there is still very small percentage of women at the highest level of management in most men-dominated organizations. According to Bernama , the World Gender Gap Report, Malaysia ranked 70th overall out of 144 nations. The Malaysian Gender Gap Index , which showed the mean score for four sub-indices as 0.697 or 69.7%, was taken into account in order to determine the rank, according to Mohd Uzir Mahidin, chief mathematician at the Malaysian Department of Statistics . The rating increased from 0.692 in 2016. Meanwhile, in terms of women's representation in government, Malaysia was placed at 110, while behind Philippines, Vietnam, Cambodia, and Indonesia. The rating of political participation was based on the proportion of women in legislature and cabinet posts, as well as the length of time a nation has had a female head of state or administration. Due to this fact, Malaysia has put tremendous effort to rectify the situations through Malaysia Plan. According to the Malaysian 9th Plan, Malaysia sought to reach a goal of 30% female representation in decision-making groups inside an institution. As per Ng , the Ministries of Women, Family, and Community Engagement and the United Nations Development Program have done study on the attainment of 30% female participation in judgement in 2007. As a result of the plan, they have developed an action plan that aims to be completed by the year 2015. In both the public and private sectors, they discovered that women make up, on aggregate, 14% of judgement positions. As a result, MWFCD had intensified the efforts to increase the number of women in the decision making level to at least 30 percent. In 2010, the 30 percent mark was finally achieved in the overall Malaysian Public Sectors but it is a different situation in the private sectors where more efforts are needed to accomplish a similar achievement. The question then is: Is there an orderly process for women to involve in decision making groups or does it only exists in gender mainstreaming? Thus, this paper will analyse what the factors which influencing gender mainstreaming process in decision-making process at public sector of Malaysia. --- LITERATURE REVIEW Development of the country influences by reflection and action to the changing in the an environmental, economic, politic, institution and cultural attitudes . All the changes impacted to women status in development especially in the context of the changes in cultural attitudes. The changes of attitudes develop a new challenging in women's world especially in the context to survive in male dominated. So, women need supported by the government through policies to ensure the survival of women in challenging world. By means of international platform, United Nation make an action by Sustainability Development Goal . In the SDG, one of the targets is integrating gender equality in development of country. According to Peterson and Camegie , integrating gender equality was one the development objectives to support economic achievement, food security, poverty and, hardship alleviation of the featured rural community-based. Hence, an action that considering is so important to achieve gender equality is gender mainstreaming. The discussion of gender mainstreaming was starting since 1980s . However, it was drastically discussed in year 2005 when majority of academicians were debates the achievement of gender mainstreaming after Beijing Platform was announced in 1995. It part of research finding about the process and progress of gender mainstreaming at many countries that agreed to implement an action at national level . The fundamental tenets of women's empowerment are to strengthen the position of women and give them greater power to make decisions and deal with issues linked to their human rights, poverty, young women, violence against women, and other problems affecting women . According Mukhopadhyay in Changachirere , gender mainstreaming targets two broad vision. One of the targets is infusing gender analysis in development policy and programmes at all levels. Second target is to support initiatives that facilitate society participation in decision making. According Nur Syakiran Akmal and Ummu Attiyah in the context of decision making, it open involvement from grassroots to higher decision makers. Hence, social inclusion is a tactic to encourage policymaking and choice that takes into account the needs and desires of both men and women . There are already a lot of research that provide a thorough explanation of particular ways for gender mainstreaming in leadership. Gender mainstreaming was used by many organizations to improve the management in the context to achieve gender equality . While, success in implementation of gender mainstreaming has been achieved by some organizations, there does not shown conclusively proof that this approach returns preferred outcome to gender equality . Women continue to be under-represented in senior leadership roles and many confront a "labyrinth" of "gender-organization-system" challenges throughout their careers . This is because some of manager or top management level not willing to share new knowledge in the organization such as gender sensitive. The organization should have managers who can improve knowledge-sharing activities in their organizations, which will have important implications in terms of fostering an environment such as gender sensitive. So we can conclude that, even many action by countries that claim support gender mainstreaming in organization, but majority finding shown that this action not fulfill the target. According Anderson, etc , the groups that, in their investigation, did not include some of the people who were the focus of the strategy . Women's subject positions in organisational policies were those of needing assistance and having insufficient abilities and competencies. Organizational procedures and institutions were usually undetectable in the materials, in addition to men and the patriarchal standards. The idea of gender equality was mostly devoid of authority and struggle. Hence, Åsa Corneliusson suggest that strategies of how a collective organizing of "outside-inside" gender practitioners might push the internal work of implementing feminist perspectives forward. So, gender mainstreaming should be use as fundamental organizational change to foresight program development in the public sector . Males report more possibilities, increased degrees of inclusion, and less employment discrimination than do women. Throughout all career levels, there are different patterns of gender differences. Men and women experience the greatest gaps in opportunity and inclusiveness perspectives at the beginning of mid-career . This situation proves that gender mainstreaming is very important to implement in all organization especially in public sectors to achieve gender equality. Some knowledge sharing practices that influences nature of knowledge, working culture and, motivation to share to support this issue . --- METHODOLOGY The research utilizes qualitative conversations with sixteen informants, ten of whom are women and six of whom are males who are in managerial roles at two selected Public Sectors companies. Both formal and unstructured interviewing techniques were used to speak with the informants. Results will be discussed in order to describe how GM is actually used in the organizations. The preponderance of the personnel at Organization X are women, thus that is why it was selected as the example illustrates for this research. According to recent data, there are 279,545 female employees compared to 127,171 male employees. Yet, male employees controlled the positions in the organizations that made decisions. All names stated are not the real names of the informant. --- FINDINGS AND DISCUSSION Malaysia is moving in this direction as half of its work force consists of women and this motion also served as step towards achieving gender equality. As a member of United Nation, Malaysia constantly involves in international conferences especially the ones that are related to issues of women. Malaysia also responded positively to the agreements achieved in these conferences. For example, Malaysia has agreed to implement the policy of eliminating discrimination against women in the country as drawn in the Convention on the Elimination of All Forms of Discrimination against Women. Since the provisions were believed to be at odds with Malaysia's domestic legislation, the nation first showed some reluctance to abide by the first accession to the Conventions. Nevertheless, with the establishment of Supporting Women as a method to promote gender equality and the debates on contemporary challenges facing women at that time, the Malaysian delegates pledged to carry out the Beijing Framework for Actions and to evaluate the Beijing Declaration issues of reservation placed under CEDAW. The head of delegation said; It is a sign of our dedication to and willingness to adapt that we are gathering at this Fourth Global Conference. Our plan to bring about a profound shift will be put into practise through the Platforms for Actions. The Malaysian government is dedicated to ensuring that men and women have equal opportunities, privileges, and obligations In 2009, the Malaysian Government introduced the Second National Policy on Women as step to ensure the continuous development for women. The policy adopts GM as a strategy to achieve gender equality as suggested in the Beijing Declaration. The implementation of the policy portrays the seriousness of the Malaysian Government to promote gender equality and emphasize priority for women in the decision making groups. Uplifting the position of women especially in the field of management such as having more women in the decision making posts is a form of continuous action by Ministry of Women, Family and Community Development. Means that, Malaysia put the priority to support women empowerment through gender mainstreaming. Hence, two organizations were selected to explore gender mainstreaming process in public sectors in Malaysia. The organization namely X and Y. Therefore, this study tries to identify the gender mainstreaming process in the decision making from both organizations as to highlighted earlier. The informants involved in the interviews were selected from both genders to reflect the overall findings. The requirement for a location as a decision maker at top management level is grade of 54 and above . The findings showed that both organizations offer opportunity for women to hold senior positions in the decision making groups. This demonstrate, both organizations constitute both gender and did not practice gender bias in the decision making structure. Yet, majority men dominated all positions in the management and policy making levels in organization X, despite the majority of staff were female. It can be assumed that imbalance of opinions and values in decisions existed and worked heavily in favor of the men. The situation was self-confessed by the Deputy Secretary General of Organization X and she was not so sure of the factors that led to such situation. According to Hassan, women decision makers are mostly attended in every meeting. His agreed that women are more committed with their work. Even though, the decision making group consisted mostly of men, the meeting attendance showed more women were present than the men. However, Hassan stated that the meeting outcomes were never gender bias even though the meeting compositions were dominated by men or women were presented more than men. Meanwhile, the decision making structure indicated that women exceeded the 30 percent in the Organization Y. Referring on the table above, the organization's Deputy Secretary General post is also held by a Deputy Secretary woman and followed by other main administrative positions. Umi mentioned that since women hold majority of the senior positions in the organization Y, there has been a presumption that it is a feminine organization. On the other hand, she emphasized on the balance of gender in which men must also hold important positions among the women decision makers. The word "must" from her statement proved that there are some successful women who are unsupportive and have doubtful thoughts about women empowerment in organization. This situation usually happens when women are in comfort zone and they believed that women are content with what they have attained. According to Nora, the committee meetings among the decision makers, at the top of the decisionmaking pyramid, men still dominate. The majority of senior roles are held by women, though, thus she acknowledged that in some situations, topics raised can be gendered. She asserted, nevertheless, that gender concerns do not affect policy-related sessions. The findings above proved that women had participated as decision makers at the management level in both organizations and hence, both organizations have gender mainstreaming process. The success of the gender popularization process, however, in the context of concerns relating to gender equality is not just restricted to the representation of women in top management positions of decision-making, but also as a content of contribution to gender matters. The lady should be able to fulfil her responsibility as a spokesperson on all matters pertaining to women's rights. This also implies that the gender mainstreaming process is applicable in order to guarantee sound judgement, which will eventually lead to the realization of gender equality. Regarding the finding, since majority of informants in the interviews have no prior knowledge pertaining to the concept of GM, the assumption is: the process of GM was not implemented in their administrative routines. In contrast, the findings indicated otherwise; the practice of GM proses did occur among the decision makers but not intentionally. The process of GM can be identified through discussion regarding gender issues in their main services which involves gender stereotype illustrations in their resources for target group. The depictions were said to have instigated the culture of gender bias in daily routines. Rita said that she had been discussing the issue for the last ten years in meetings among the decision makers. Kiong is generally familiar with the idea of women's rights. In one of his managerial practises at Organization Y, he had inadvertently used the GM method. After working late into the night, he once made sure the parking lot was secure and safe for the personnel to board their cars. On the other side, Umi had stated that certain judgements ought to be made according to gender. For example, while sending staff members for training, she occasionally has to take into account gender viewpoint in addition to the staff members' ability and qualities. A mentor-mentee system is one of Organization Y's career support program for its staffs. Even though, the system is still at an early stage, the designers have created the system to the best of their capabilities. The system is part of tacit knowledge that manager should use to increase the effectives of the organization . One of the designers was Yati, who among others had outlined that the mentor and mentee must be of the same gender as the safety of the two parties can be greatly ensured. The findings also showed that indirect GM approaches occurred among decision makers in determining or conducting an action. However, only three persons in organization Y had taken actions based on the GM approach as compared to only one in organisation X. Following that, it is assumed that decision-makers did not employ any GM methods during discussions at their various firms based on their awareness of the GM idea. The findings, meanwhile, indicated that while women are involved in decision-making in both organizations, there has been little progress towards implementing a women's participation strategy. The issue of gender stereotype in resouces for their target group was also discussed in Organization X as mentioned by Aminah. As a result, all the resources are closely monitored by the organization as to avoid gender stereotype illustrations or examples. The critical shortage of male staffs in Organization X was also discussed among the decision makers. Inadvertently, this circumstance has raised questions about the effects of the male crew, which formerly dominated every notable sector in the world of human resources. In addition, during the meeting of decision-makers, sexual misconduct was also covered. The debate on this subject revealed that both organizations' meetings have discussed the GM approach. Sexual harassment problems have received a lot of attention from the two groups, and it was on the agenda for one of the meetings. Due to the fact that it affects their primary business, Organization X is extremely worried about this problem. But in the case of Organization Y, the Secretaries General's concerns as a woman are more likely to be the cause of the emphasis. The many scenarios demonstrated that a woman in the top position inside an organization will inadvertently use a GM strategy and be more understanding of gender politics. It means, nature of knowledge of information of the women at top management level such as Secretary General in Public Sector also can influence the cultural of gender sensitive. One of the key parts of the framework for supporting employees is the issue of childcare for staff. According to the research findings, decision-makers have discussed this issue. Yet, departmental conferences in each company brought up this issue. The childcare issue is viewed as an unmistakable support for female employees. As a result, this topic was considered in the higher tier conferences by both groups. Despite being stressed in each and every term of the Malaysian Plan, offering nurseries at employment is still only at a minimal, which suggests that this problem continues to be in its adolescence. This result revealed that GM approaches occur in the exercise of decision-making. The judgments, nevertheless, do not fully understand the concept's genuine meaning, hence the methods used were considered oblique. Resulting of the finding analysis, the research found four themes that can be used as elements that must be considered when implementing the process of GM in an organization. The next section will discuss these elements as suggestions in implementing process GM in organization. --- SUGGESTIONS OF GENDER MAINSTREAMING PROCESS IN PUBLIC SECTOR? The results have shown four main elements should have in the process of GM practice in an organization. The four components are I knowledge; understanding; empathy; and practise at the beginning of the GM process. These components are theme findings derived from analysis of source responses and the following part is the discussion of the four elements found in the analyses. --- a. Knowledge Based on the tables 4 and 5, With the exception of one informant from Organization Y, neither of the two organizations' whistleblowers who were questioned had ever heard of the GM notion. In studies on most countries that practice GM found that majority of the workers are incapable to relate the concept of GM to their work and organizations' objectives. As result, the practice of GM was not seriously considered . Consequently, these workers need more exposure on the issues in gender through specific trainings and programs. The interviews results have shown that most informants were unaware of the concept of GM and had asked for an elaboration at the beginning of the interview. There were some informants who did not respond to the GM concept related questions. In fact, when the same question on the subject was re-posed to the informants for the second time after three months of the first interviews, none had answered satisfactorily. Below are some examples of the answers given: […] mainstreaming means to implement in the main stream, meaningto have more women in the administrative field […] […] to my understanding, trying to give more attention to gender, meaningspecific attention to gender in the public sector. That is all I know […] Intan was the only informant who has knowledge about the concept of GM. Prior to her current position she had served with Organization Z which is the only organization to completely implementing GM in its overall operations based on current needs. There, she had experienced GM implementation through Gender Responsive Budget. However, the researcher concluded that majority of informants still lack the knowledge of GM concept. --- b. Comprehension Since the majority of informants lack the knowledge of GM, the researcher concluded that the informants also have no comprehension regarding its concept. In reality, if a person has knowledge of a certain subject then the person will also have an understanding of its concept. However, the case of Intan, she was exposed to the approach of GM during her tenure at Organization Z; therefore, she has the grasp of the concept. In other words, the researcher inferred that certain extend of knowledge of the GM is required in order to have a comprehension of its concept. --- c. Compassion Compassion of the concept of GM can only be thrived if the informants have an in-depth comprehension of GM. The researcher concluded that since almost all informants do not have significant comprehension regarding the concept of GM then compassion cannot be inferred to exist. On the other hand, Intan, as the only informant who had experienced and comprehended the concept of GM, had demonstrated the elements of compassion through her justification on gender's needs. --- d. Practice Based on the results above, the author's assumption is that actual GM approach was never practice in management routines. However, decision makers do refer to gender perspective needs in certain cases even though they claimed that their action were gender neutral. In fact, implicit GM approach did occur either in individual or collective decision making. --- CONCLUSION Overall, this paper focuses on the issue of mainstreaming gender in a company's decision-making processes. The importance of measuring the gender mainstreaming process in organization to know the level of support from public sectors in the context to support women policy's in their organization. It is also important to know the opportunity of women as decision makers in public sectors. The survey found that while there are many possibilities for women to engage in the decision-making process, there are very few of them in top managerial positions. The study's findings also demonstrated that both organizations have gender popularisation policies in place. The implementation and existence are, however, still quite minimal. This is a result of the president's lack of commitment to putting suitable policies and initiatives in place to promote gender parity. As a result, the promotion of GM is still lacking and most organizations do not have adequate information of it. Insufficient knowledge of GM affects the effectiveness of its implementation in organizations as the decision makers do not have the comprehension of the concept. Because of this, the administration should place a focus on the assertive marketing and delivery of knowledge regarding gender issues to companies in order to build a solid and effective gender balance throughout the country.

The purpose of this study is to explore the impacts of gender mainstreaming process in the public sector in Malaysia. It is an attempt to see the achievement of gender mainstreaming process in the organization for public sector. Therefore, this paper reviews on the Gender Mainstreaming process impact by looking at decision making among leaders in the gender issues. The research relies on indepth conversations with participants, both men and women, who work at high management levels in two particular government departments. The findings indicate that both organizations have women participating in the decision-making processes, but that multicultural education is still in its early stages. Results suggest that to increase support the decision making process, four main elements that

INTRODUCTION Early studies of elder mistreatment indicate that prevalence varies by gender, socioeconomic status, prior victimization, psychiatric history, low social support, and substance use . Demographic factors such as race and ethnicity also have been the foci of study. Considering race, results have been mixed, with some studies showing that Caucasians at increased risk relative to African Americans and Hispanics , and other studies placing African Americans at greatest vulnerability . However, this variance in race-based risk estimates across studies may be due to confounding effects of income, education, age, gender, geographic location, and respondent willingness to report the crime . Considering ethnicity, Laumann reported reduced risk of verbal or financial mistreatment in Latinos relative to African-Americans and Whites. Nevertheless, very little research has focused on ethnicity, and reasons for differential prevalence are not yet clear. Frequently, explanations of low or non-existent elder mistreatment in the Latino population cite family solidarity and its association with higher levels of social support as potentially protective . This seems logical when considering that Latino culture is deeply rooted in the concepts of familia and respeto , especially with reference to older adults. This is typically a positive aspect of family care of the elderly, but as Lowenthal states, may lead to the protection of intra-family perpetrators and a rejection of involvement of authorities. La vergüenza , which goes against the "family institution", also may lead to inhibited reporting. Thus, these often beneficial values do not assure that elder mistreatment is not occurring. Instead, while often protective, they may also contribute to underreporting when the event does occur relative to other ethnic/race groups . Some investigators posit that very specific experiences of Latinos and minority group members might actually increase the probability of abuse, such as the process of immigration with all its associated stressors . Of course, when considering ethnicity or race as a risk or protective factor, it is also important to consider other risk factors for elder mistreatment, both independently and insofar as they relate to ethnicity per se. For example, older women seem to be at increased risk relative to older men . Age also plays a role, with "younger old" reporting more verbal abuse and financial mistreatment than "older old" . Acierno et al. recently reported that low social support is a very strong risk factor for mistreatment, and may be more relevant than either gender or age in predicting these events. This finding was supported by other studies where authors noted that reduced participation in social activities may be a risk factor for abuse, particularly among men. Prior research has not focused on the associations of race and ethnicity, per se, on elder mistreatment insofar as these factors have been considered along with other demographic factors, but have not themselves been the target of higher level analyses or statistical control. Therefore, the present investigation sought to isolate the relevance of race and ethnicity to mistreatment potential by controlling for specific factors identified as highly important in previous studies and as central in the culture of Latinos and African Americans, including history of prior traumatic event exposure, low income, poor perceived health, and low social support . --- METHODS The present study used Random Digit Dialing methodology to derive a random, nationally representative sample of 5,777 communityresiding older adults. RDD was selected over in person interviews in light of successful epidemiological research on interpersonal violence in which rates of physical and sexual assault, including assault of older adults, was documented . Moreover, this telephone based method was contrasted directly to in-person interview in a sample of older adults in terms of reporting rates, comfort with the topic, and hearing or technical difficulties. No differences were found on either prevalence or feasibility variables. In fact, responses to open-ended questions at the end of the interview indicated that participants felt more comfortable talking about interpersonal violence over the telephone than directly to another person physically present in the room with them . In addition to demographic factors, participants were interviewed via telephone in English or Spanish about emotional, physical, sexual, and financial mistreatment as well as neglect occurring since they were age 60. --- Variables EMOTIONAL MISTREATMENT-Emotional mistreatment was defined as an affirmative answer to any one of the following four questions. After it was determined that such an event had occurred, the timing was defined. 1. "Now we want to ask you about some things that people in your life might do that make you feel bad, such as saying very mean things to you, or being rude to you. A lot of people say this happens to them, and we really need to find out how often it happens. Sometimes, we call these things emotional mistreatment. The person who might do these things could be a romantic partner, spouse, family member, friend, or someone who helps take care of you. Has anyone ever verbally attacked, scolded, or yelled at you so that you felt afraid for your safety, threatened or intimidated?" 2. "Has anyone ever made you feel humiliated or embarrassed by calling you names such as stupid, or telling you that you or your opinion was worthless?" 3. "Has anyone ever forcefully or repeatedly asked you to do something so much that you felt harassed or coerced into doing something against your will?" 4. "Has anyone close to you ever completely refused to talk to you or ignored you for days at a time, even when you wanted to talk to them?" PHYSICAL MISTREATMENT-Physical mistreatment was defined as an affirmative answer to any one of the following three questions. As was the case with emotional mistreatment, descriptive parameters of the event were collected after respondents indicated that such an event had occurred. 1. "Another type of stressful event that people sometimes experience is being physically hurt by another person. The person doing these things could be a romantic partner, spouse, family member, friend, or someone who helps take care of you. Has anyone ever hit you with their hand or object, slapped you, or threatened you with a weapon?" 2. "Has anyone ever tried to restrain you by holding you down, tying you up, or locking you in your room or house?" 3. "Has anyone ever physically hurt you so that you suffered some degree of injury, including cuts, bruises, or other marks?" SEXUAL MISTREATMENT-Sexual mistreatment was defined as an affirmative answer to any one of the following three questions. After it was determined that such an event had occurred, the timing was defined. --- 1 "I am going to ask you questions about unwanted sexual advances that you may have experienced over your lifetime. People do not always report such experiences to the police or discuss them with family or friends. The person making the unwanted advances isn't always a stranger, but can be a friend, romantic partner, or even a family member or someone you trust to help you or help take care of you. Such experiences can occur anytime in a person's life. Regardless of how long ago it happened or who made the advances, has anyone ever made you have sex or oral sex by using force or threatening to harm you or someone close to you?" 2a "Has anyone ever touched your breasts or pubic area or made you touch his penis by using force or threat of force?" 2b "Has anyone ever touched your pubic area or made you touch their pubic area by using force or threat of force?" 3a "Has anyone ever forced you to undress or expose your breasts or pubic area when you didn't want to?" 3b "Has anyone ever forced you to undress or expose your pubic area when you didn't want to?" ETHNICITY-Ethnicity was defined as an affirmative response to the question "Are you of Hispanic origin or decent." RACE-Race was first defined specifically in response to the question "In which of the following categories do you feel you belong: white; Black, Pacific Islander, American Indian or Alaskan native, Asian or something else?" However, due to small cell sizes for most race subtypes, categories were collapsed into White vs. Non-White. INCOME -Low income was defined as cases where the entire household income was less than $35,000 the previous year. HEALTH STATUS -Health status over the prior month was assessed using the general health question number 1 from the World Health Organization Short-Form 36 Health Questionnaire . Participants were asked to rate the following question: "In general, would you say your health is "Excellent, Very Good, Good, Fair, or Poor." These responses were dichotomized into Poor Health and Good Health . This assessment is consistent with previously validated single item measures of general subjective health, which have shown both good reliability and validity , and has been found related to morbidity and mortality . Poor health was considered as a possible risk factor due to the associated increase in caregiver burden resulting from increased medical visits, healthcare regimens, and caretaking responsibilities. --- EXPERIENCE OF PRIOR TRAUMATIC EVENTS - Participants were asked to report if they had been exposed to the following events and indicated fear that they would be killed or seriously injured during this exposure: natural disasters such as earthquake, hurricane, flood, or tornado; serious accident at work, in a car, or somewhere else; or being in any other situation where you thought you would be killed. Past research indicates that exposure to one form of trauma is associated with increased risk of interpersonal violence . SOCIAL SUPPORT -Perceived social support during the past month was assessed via a modified five-item version of the Medical Outcomes Study module for social support . Participants were asked about emotional ; instrumental ; and appraisal social support and responded to items using a fourpoint scale from "none of the time" to "all of the time" (sample range=0-20; M=15.9 [SD=4.8]). Low social support was operationalized as a score in the lower quartile of the sample ratings, and the comparison high social support was operationalized as a score in the upper quartile of sample ratings. Social support has been identified as a protective factor for mental and health outcomes in prior research , and was predicted to be a potentially modifiable protective factor for elder mistreatment insofar as social connections would lessen the likelihood that abuse would occur due to increased activities with others in the social network, leaving less time in the isolated company of a perpetrator. --- --- Statistical Plan Prevalence of each type of mistreatment was compared across race and ethnicity via univariate chi square analyses. Chi square analyses also were used to determine ethnicity and race based differences in contextual variables that might influence risk, including income, social support, perceived health status, and exposure to prior traumatic events. Contextual variables that were significantly different across ethnicity and race were included as covariates in multivariate logistic regression analyses for each mistreatment type to determine if control of these factors affected ethnicity and race based risk of mistreatment. Note that logistic regressions did not include gender and age in order to limit the number of non-modifiable variables in the logistic regression and maximize power. Thus, all logistic regressions were re-run with the addition of gender and age to verify that controlling for these variables would not change observed patterns of risk relationships. No risk relationships were altered by this addition, and logistic regression results reported below omitted these two variables. --- RESULTS The cooperation rate was 69% for the sample. 60.2% of the older adults were women and 39.8% were men. Table 1 presents the prevalence for each mistreatment type in terms of ethnicity and race. No significant differences were observed for any mistreatment type with respect to ethnicity. Considering race, Non-Whites were at twice the risk of Whites for physical mistreatment , but there were no observed differences for emotional or sexual mistreatment. Considering differences in terms of contextual variables presented in Table 2, greater proportions of Hispanics reported low income , poor social support , and poor health . There were no differences in terms of exposure to prior traumatic events. Similar patterns were observed with race, with a greater proportion of Non-Whites reporting low income , poor social support , and poor health . Thus, logistic regression analyses were conducted to examine risk of mistreatment in terms of ethnicity and race, controlling for the significantly different contextual variables of low income, social support, and poor health. Ethnicity based logistic regressions continued to reveal no predictive risk of being Hispanic for any form of elder mistreatment. However, low social support increased risk of all forms of mistreatment, even after effects of ethnicity and other factors were controlled. Similarly, poor health was associated with both emotional and physical mistreatment. Race based logistic regressions also indicated that Non-White status was not, in itself, associated with any increased prevalence of any form of mistreatment, including physical mistreatment once income, social support, and health status were controlled. As with ethnicity based multivariate analyses, low social support was independently predictive of all forms of mistreatment, and poor self-rated health was associated with increased emotional mistreatment. Thus, overall, no differences for ethnicity or race were observed for any mistreatment type in multivariate analyses. --- DISCUSSION There is significant interest in identifying and explaining race and ethnicity-based differences in the prevalence of interpersonal violence. However, studies on children and adolescents, as well as data presented here on older adults from the National Elder Mistreatment Study, indicate that these differences are less apparent than one might expect. In this study, race and ethnicity exerted no effect on risk of any form of mistreatment when other contextually related variables were controlled. Indeed, the only significant univariate finding, that of increased risk of physical mistreatment for Non-Whites, was eliminated when effects of income, perceived health, and social support were considered. Of particular interest was the finding that the most important predictive factors for all forms of mistreatment were those that are potentially modifiable, specifically social support and health. This finding provides some specific direction to those developing preventive and secondary interventions for elder mistreatment in that it may well be worthwhile to leverage existing agencies and organizations that could potentially facilitate social support or improve health. For example, church or college-based volunteer groups could initiate services dedicated to providing transportation to potentially social events or health-related opportunities for the significant proportion of older adults who find it difficult or impossible to drive or use public transportation . Findings related to low social support and increased risk of negative outcomes are consistently reported in the literature. For example, older adults reporting low social support who were exposed to extreme stress events such as natural disaster were at increased risk of post-traumatic stress disorder, depression, and generalized anxiety disorder . Of course, this was not a longitudinal study, and the direction of risk cannot be determined; it may be the case that those who are mistreated report lower social support and this factor results from, rather than causes, elder mistreatment. The same could be said for perceived poor health. However, it seems most likely that the relationship is bidirectional, with poor support or health both indicative of, and predictive of, mistreatment. Thus, the potential importance of these factors to negative outcomes in older adults, and by connection, their relevance to developing preventive interventions for both interpersonal violence and psychopathology appears great. --- Limitations This study is characterized by several limitations. First, all estimates of mistreatment and contextual variables were based entirely on self-report. Mistreatment events are often underreported in this age group, and thus the validity of estimates is less than perfect. A second major limitation was a failure to include some measure of cognitive functioning, either as a covariate or risk factor. Thus, these findings reflect responses of a cognitively intact, community-residing sub-population of older adults. By connection, generalization of these results to what may be the group most at risk of mistreatment, the cognitively impaired elderly, is not appropriate. We also did not include individuals who did not have a home phone , introducing the potential for bias. However, our data is nationally representative and is weighted on Census estimates, increasing its generalizability. Note: The level of the variable given first represents the reference value of the variable, which is also the level the variable hypothesized to be associated with increased risk. Confidence Intervals that do not cross the value 1.00 indicate increased or reduced risk for the reference value of the variable.

The prevalence of elder mistreatment with respect to race and ethnicity was examined in an unweighted sample of 5,777 participants (5,776 participants in weighted sample). Random Digit Dialing methodology was used to select a representative sample of community-dwelling older adults, and the survey was available in English and Spanish. Mistreatment types included emotional, physical, and sexual abuse. Race and ethnicity based differences were largely absent, and the only observed increase was for physical mistreatment among Non-White older adults; however this association was not sustained in multivariate analyses controlling for income, health status, and social support. Findings are in contrast to prior reports of increased risk of mistreatment in minority populations, and point to correlated and modifiable factors of social support and poor health as targets for preventive intervention.