Annual Research Review: A multilevel bioecological analysis of factors influencing the mental health and psychosocial well‐being of refugee children

Background This paper revisits the themes of an influential 1993 review regarding the factors shaping the mental health and psychosocial well‐being of refugees to take stock of developments in the evidence base and conceptualisation of issues for refugee children over the last 25 years. Methods The study deployed a systematic search strategy. This initially identified 784 papers, which was reduced to 65 studies following application of inclusion and exclusion criteria. We used a later iteration of Bronfenbrenner’s bioecological model of human development – the PPCT model – to consolidate evidence. Results We identify a range of risk and protective factors operating at individual, familial, community and institutional and policy levels that influence outcomes for refugee children. The dynamics shaping the interaction of these influences are linked to the life course principles of socio‐historical time and developmental age, proximal processes and child agency. Conclusions Actions at individual, familial, community, school, institutional and policy levels all have potential traction on mental health and psychosocial well‐being of refugee children. However, evidence suggests that greatest impact will be secured by multilevel interventions addressing synergies between ecological systems, approaches engaging proximal processes (including parenting programmes) and interventions facilitating the agency of the developing refugee child.


Introduction
Of the estimated global refugee population of approaching 26 million refugees, over a half are under the age of eighteen (UNHCR, 2018). The 'global refugee crisis' of the last decade has been marked by large-scale displacement of populations from conflict-affected and low-income settings and varieties of accommodation and resistance of these movements by other nations. Factors such as displacement experience, familial separation and acculturative stress all have clear implications for children's wellbeing and development outcomes in such circumstances. However, interpretation of the recent wave of research into these processes is usefully viewed from the perspective of lessons learnedand conceptual frames establishedin the course of multiple prior periods of population displacement with major implications for children.
In 1993, the second author completed a review of mental health in refugee populations with a developmental focus on behalf of Harvard Centre for the Study of Culture and Medicine (Ager, 1993). This was part of the preparation for the production of World Mental Health: Problems and Priorities in Low-Income Settings (Desjarlais, Eisenberg, Good, & Kleinman, 1995), one of the foundational works in establishing the field of global mental health. That review serves as a useful guide to the refugee situations of concern twenty-five years ago, the bodies of research literature related to these concerns andof particular interest herethe conceptual framing of that literature.
In the mid-1990s, the refugee displacements of central concern were related to the sequelae of the Vietnam War, the Cambodian genocide, proxy wars in southern Africa and the unfolding violence of the Balkans with the break-up of the former Yugoslavia. Accordingly, the research literature reviewed in Ager (1993) was heavily shaped by the acculturative processes of South-East Asian migrants to the United States (US), along with the rise of the refugee camp as a temporarybut increasingly protractedsetting for children's development.
There were two major themes structuring the Ager (1993) literature review: one regarding the analysis of risks and protective or ameliorative factors associated with the nested layers of resourcefrom familial process through religious or political affiliationshaping individual adaptation and development; the other, the sequential nature of the refugee experience through phases of preflight, flight, reception and (re)settlement. Although not explicitly cited, these principles very much reflect Bronfenbrenner's conceptualisation of developmental influence (Bronfenbrenner, 1979(Bronfenbrenner, , 1995, most particularly in the form of the later iteration of his PPCTperson-elaborates his established analysis of multi-layered contextual systems (i.e. micro-, meso-, exo-and macro-systems) with respect to the person, proximal processes and time. Bronfenbrenner sees persons as embodying a multiplicity of biological and genetic mechanisms which collectively impact on behaviour and development (Bronfenbrenner, 2005). Progressively more complex reciprocal enduring interactionsproximal processesoperate between the developing child and 'significant others', objects and symbols in their microsystem to serve as 'the engines of development' (Bronfenbrenner & Evans, 2000, p. 118). Within the PPCT model, time reflects both 'historical time' and 'developmental age'. A person's life course is embedded in and shaped by both the circumstances and events happening in 'the historical period in which one lives' and 'biological and social transitions based on culturally defined age, role expectations, and opportunities' (Bronfenbrenner, 1995, p. 641).
Reflecting on the suitability of this framing for studies of forced migration, Hayes and Endale (2018) note how the model's focus on 'examining multiple levels of context and change over time. . .can be especially useful in examining the complex issues surrounding the development of identity in newcomer children and adolescents ' (p. 285). It is thus the framing device for the current view (Figure 1), providing a coherent basis to assess our advance over the last 25 years in understanding of the factors and processes shaping the mental health and psychosocial well-being of refugee children.

Methods
A review protocol (available from the first author) was developed a priori specifying search strategy, search terms and keywords, inclusion and exclusion criteria, data extraction, data synthesis and analysis.

Search strategy and screening
A systematic search was undertaken in five databases -MEDLINE (EBSCO), CINAHL (EBSCO), PsycINFO (EBSCO) and SCOPUS (Pro-Quest) for peer-reviewed evidence published in English since January 2010 up to November 2019. Restriction by publication date was applied to capture the evidence reflective of the effects of recent political and war-driven displacement in Middle East, Asia, Sub-Saharan Africa and Latin America. Search terms and keywords were developed in consultation with a subject-specialist librarian. The following combination of terms were used: child OR minor* OR young OR adolescent OR youth OR teenage AND asylum seeker OR refuge* OR displaced person OR internally displaced person AND psychiatr* OR psycholog* OR psychosocial well-being OR psychosocial wellbeing OR mental OR mental health OR mental issue OR distress OR stress OR emotional stress OR emotion AND risk factor* OR stressor* OR trauma OR vulnerability factor* AND protective factor* OR adaptation* OR modifying factor* OR resilience OR develop* OR social support* OR coping* OR psychosocial support* OR recovery OR wellbeing OR well-being OR adjustment OR behavio?r OR outcome.

Selection criteria
Studies were included if (a) the primary aim was to explore risks (stressors) and/or protective factors/processes contributing to mental health and psychosocial well-being of refugee children and (b) study participants were refugee children (age range 0-18 years or mean age < 18 years) or caregivers sharing information about their refugee children. A risk factor was defined as a psychosocial adversity or event that would be considered a stressor to most people and that may hinder normal functioning (Masten, 1994). Protective factors were defined as exogenous variables whose presence is associated with desirable outcomes in populations deemed at risk for mental health (Werner, 1989). Protective processes were operationalized as dynamic processes operating in the family, peer group, school and community which tend to buffer the likelihood of negative outcomes (Benard, 1995). The WHO's definition of mental health was adopted: 'a state of well-being in which every individual realizes his or her own potential, can cope with the normal stresses of life, can work productively and fruitfully and is able to make a contribution to her or his community' (WHO, 2004). Only original, peer-reviewed published articles were included. Studies were excluded if (a) more than 50% of participating children and young people had mean age > 18 and where no segregated data by age were presented; (b) they focused principally on describing the feasibility or effectiveness of mental health-promoting interventions; and (c) they provided a literature review and no primary data. No restriction to a country was applied.

Data extraction
Data extraction was performed using a standardized data extraction form. Details were extracted on (a) the study design, (b) population demographics, (c) setting and length of stay, (d) methods of data collection, (e) factors intrinsic to the child and (f) factors related to ecological systems of potential influence in Bronfenbrenner's bioecological model (Bronfenbrenner, 1979(Bronfenbrenner, , 1995(Bronfenbrenner, , 2005. Reporting categories for the latter were developed iteratively through the process of data extraction, resulting in distinction of familial, community (social and physical), and institutions and policy systems.

Results
Initial database search identified 784 articles. On removal of duplicates, 519 titles and abstracts were screened for relevance resulting in 86 full-text articles retrieved for further eligibility assessment. Sixty-six articles corresponding to 65 individual studies met the inclusion criteria. Summaries of the included studies are presented in Tables 1 and 2. The majority of studies were conducted in highincome contexts: Scandinavia (10), Australia (10), the United States (7), mainland Europe (9), Canada (2) and the United Kingdom (2). Nineteen studies were focused in the Middle East: including Palestine (7), Turkey (6), Jordan (2) and Lebanon (5). Four studies were based in Africa and two in Asia. Fortyone studies (63.1%) employed quantitative, 19 (29.2%) qualitative and 5 (7.7%) mixed-method study designs, varying substantially in terms of sample size (range 6-2,074), participants' ages (children aged 0-23, caregiver aged 18-60), and methods of data collection (e.g. surveys, in-depth interviews, focus groups, observations, drawing). Thirteen (20%) studies reported on unaccompanied refugee children; eight (61.5%) of which presented quantitative data. Two qualitative articles used the same study participants, but different data collection methods (Veronese & Cavazzoni, 2019;Veronese, Cavazzoni, & Antenucci, 2018). Evidence supplied by these articles was partially overlapping and is therefore reported as a single study. Included studies describe factors and processes contributing to refugee children's mental health and psychosocial outcomes in relation to multiple adversities faced at the individual, familial, community, and institutions and policy levels. Argumentative synthesis of these descriptions is presented below.
Ethnicity & ethnic or national identity. While strength of ethnic and national identity varies amongst groups of refugees (d 'Abreu, Castro-Olivo, & Ura, 2019), this is frequently found to be a protective factor for refugee children. A stronger sense of ethnic and national identity is associated with less depression, anxiety and more resilience for Palestinian children residing in camps in Gaza (Aitcheson et al., 2017). Similarly, a strong sense of ethnic identity contributed to positive psychosocial well-being in resettled young refugees in Australia (Correa-Velez et al., 2010;Zwi et al., 2018). Somali adolescent girls presented with better mental health outcomes if having a stronger ethnic identity and felt more protected from discrimination when surrounded by peers sharing common ethnicity (Ellis et al., 2010). However, ethnic and national identity may interact with acculturation strategy. It was Somali adolescent boys who assimilated within mainstream culture in host cities in New England who presented with better mental health (Ellis et al., 2010). Being identified with a minority ethnic group can exacerbate anti-refugee discrimination by peers in the host community, with deleterious mental health consequences (Beiser & Hou, 2016).
Acculturation style. The minority of refugee children who are given an opportunity to resettle in a third country have to adjust to new belief system, culture, language and try to meet country-specific academic expectations (Ellis et al., 2010). For many children, these expectations present major additional 'acculturation stresses' (Eriksson & Rundgren, 2019;Oppedal & Idsoe, 2012;Patel et al., 2017). These stresses are generally conceptualised in terms of a cultural conflict between heritage and host cultures (Berry, 1992) which requires adaptation. In general, the higher the perceived level of acculturation stress, the more depressive symptoms children exhibit (Oppedal & Idsoe, 2012;Patel et al., 2017). Acculturation style mediates the relationship between acculturation stress and mental health outcomes (Beiser & Hou, 2016;Beiser et al., 2015;Oppedal & Idsoe, 2012;Thommessen, Corcoran, & Todd, 2015). Children adopting an integrative acculturation style, which assumes interest in and adherence to heritage culture alongside an active incorporation of values and attitudes from the host culture, are widely found to report better mental health outcomes (Beiser et al., 2015;Eriksson & Rundgren, 2019;Oppedal & Idsoe, 2012;Thommessen et al., 2015). Adopting an acculturation style based on separation or marginalization contributes to persistent challenges and poor mental health (Beiser & Hou, 2016;Beiser et al., 2015).
An integrative acculturation style requires instrumental and social competencies (Beiser et al., 2015;Oppedal & Idsoe, 2012). These competencies refer to the perceptions of self as able to perform schoolrelated tasks, and make and retain social relationships (Beiser et al., 2015). Perceptions of instrumental (e.g. fluency in host country language) and social competencies (e.g. forming friendships, getting along with others) are associated with increased selfesteem, self-efficacy, feelings of mastery and lower levels of perceived discrimination (Oppedal & Idsoe, 2012. These competences mitigate the adverse effect of cultural distance (i.e. the distinctiveness of culture of origin and that of resettlement) and social distance (i.e. low acceptance of relations with other ethnic groups) on mental health and promote child psychosocial well-being (Beiser et al., 2015;Oppedal & Idsoe, 2012. Host country language literacy in particular plays an important role in children's social and cultural competencies. Children obtaining fluency in the host country language report better functioning within new social networks and school systems (Eriksson & Rundgren, 2019). This can however put a strain on parent-child relationships; since children typically learn host country language and customs quicker than parents, parents may use their children as translators and 'cultural interpreters' (Lauritzen & Sivertsen, 2012).
Coping style. Psychosymptomology and emotional dysregulation vary according to children's coping styles (Aitcheson et al., 2017  . This is more pronounced in males, facilitated by higher neighbourhood attachment and peer support from the host community (Buchegger-Traxler & Sirsch, 2012). Alcohol abuse, transactional sex, night outs with 'sugar daddies' has been reported as prevalent in females living in camps (Stark et al., 2015) with clear risks for unintended pregnancies or early marriages (Bermudez et al., 2018). Engagement in risky behaviours is widely seen as a consequence of economic hardship (Bermudez et al., 2018).
However, children's narratives indicate that compromised parent-child relationships and parental inability to provide tangible resources may exacerbate such risks (Stark et al., 2015).

Familial factors
Parental education & employment. The impacts of parental education or employment status on child mental health are addressed by relatively limited number of studies with inconsistent results (Khan et al., 2019;Panter-Brick et al., 2015;Sapmaz et al., 2017). Positive association was established between maternal (but not paternal) literacy levels and Afghan children's distress levels (Panter-Brick et al., 2015). Maternal literacy is however rare in the Afghan context (Panter-Brick et al., 2015). It assumes certain family dynamics and many psychosocial pressures on children (e.g. schooling, employment) from literate mothers following the loss of maternal salaried employment and persistent poverty (Panter-Brick et al., 2015). In a study of refugees in Turkey children's psychopathology was linked with paternal education and paternal unemployment status (Sapmaz et al., 2017). No association was found between parental education and mental health of Rohingya children in Myanmar (Khan et al., 2019).  Sim, Fazel, et al., 2018). For example, while fathers may report the value of positive parent-child interaction, there may find it difficult to control their anger and irritability prompted by the uninterrupted flow of daily stressors (Van Ee et al., 2013;Sim, Fazel, et al., 2018). The range of reported stressed impacting parentchild interaction includes family traumatic histories; parental mental health problems and physical illness; financial hardship; perceived discrimination and abuse from host neighbours, employers, healthcare providers, and authorities; worries about the extended family in the country of origin; and uncertainty with residency permits or citizenship in a host country (Dalg ard & Montgomery, 2017;Van Ee et al., 2013;Sim, Fazel, et al., 2018).
Strained parent-child relationships, commonly described in literature as 'intergenerational conflict', have a negative effect on children's emotional wellbeing and psychosocial adjustment (Dalg ard & Montgomery, 2017;Van Ee et al., 2013). Communication problems are a widely reported source of distress in these circumstances (Afifi et al., 2019;Bermudez et al., 2018). Refugee children feel unable to openly discuss their feelings or report incidences of violence in the community or at school (Afifi et al., 2019;Bermudez et al., 2018). Bermudez et al., (2018) found that children's disclosure of sexual violence from adults in a camp in Rwanda rarely resulted in parental use of formal reporting and referral processes due to fears of shame, embarrassment and social rejection. Parental awareness of child abuse often led to little or no consequences for the abuser undermining children's sense of safety (Bermudez et al., 2018). At times, children avoid discussing their struggles and disclosing incidents of abuse due to fearing physical punishment from their parents (Bermudez et al., 2018) or due to concerns over parental mental health problems (Dalg ard & Montgomery, 2017) or marital conflict (Afifi et al., 2019).
Family cohesion. Family cohesion is understood as the emotional bond family members have with one another (Olson, Russell, & Sprenkle, 1983), involving openness, expressiveness, togetherness and communication. Family cohesion can act as a buffer regarding daily stressors and uncertain coping strategies and social support (Khamis, 2019). A stronger sense of family cohesion (Aitcheson et al., 2017) and open and free expression of negative feelings within a family (Khamis, 2019) have both been associated with reduced psychopathology in refugee children.
Marital cohesion refers to the degree of connectedness and emotional bonding between spouses (Olson et al., 1983). Social support from a spouse promotes positive parenting and reduces the risk of child maltreatment (Sim, Fazel, et al., 2018). Marital cohesion is, however, often at risk in situations of chronic adversity (Afifi et al., 2019). Sharing feelings of uncertainty or disclosing stressors to a family reinforces distress in children in the presence of marital conflict (Afifi et al., 2019). Poor family functioning and marital conflict are linked with child depressive symptoms, antisocial and delinquent behaviour (Afifi et al., 2019), which can persist over time (Sangalang, Jager, & Harachi, 2017).
Separation from a family. The presence of psychopathology in refugee children is associated with both current and previous nuclear family separation (Eriksson & Rundgren, 2019;Jani et al., 2016;Longobardi et al., 2017;Mace, Mulheron, Jones, & Cherian, 2014;McGregor et al., 2016;Thommessen et al., 2015) and parentless status (Khan et al., 2019;Zwi et al., 2018). Unaccompanied refugee children frequently speak of the emotional difficulty separation from family brings into their lives (Eriksson & Rundgren, 2019;Jani et al., 2016;McGregor et al., 2016). Connection or unification with the family is viewed as a key source of social support and emotional safety (Eriksson & Rundgren, 2019;Jani et al., 2016;McGregor et al., 2016). Feelings of missing, and being constantly worried about, family members left behind negative impacts on children's psychosocial well-being (McGregor et al., 2016;Thommessen et al., 2015). Only one study found no association between the whereabouts of parents, social supports and child mental health (Akiyama et al., 2013).
Physical factors. Poverty, poor hygiene and sanitation, overcrowding, food insecurity and lack of health care and support staff are reported frequently as key stressors linked to children's poor mental health, especially in those living in camp settings (Bermudez et al., 2018;Khamis, 2019;Lauritzen & Sivertsen, 2012;Meyer, Meyer, Bangirana, Mangen, & Stark, 2019;Nasıro glu et al., 2018;Panter-Brick et al., 2015;Sim, Fazel, et al., 2018). The severity of these problems variates by country context (Khamis, 2019;Nasıro glu et al., 2018;Sim, Fazel, et al., 2018). For instance, a higher prevalence of psychopathology amongst Syrian refugee children was found for those displaced in Lebanon compared to Jordan (Khamis, 2019). This was partially attributed to the more precarious living conditions faced by refugee children in the former setting, including poor sanitation facilities and overcrowded shelter, contributing to a prolongation of adverse experiences after displacement. Further, length of stay in a refugee camp in Bes ßiri district, Turkey located next to a river polluted by untreated sewage and with dysfunctional facilities predicted poorer mental health amongst Yazidi refugee children (Nasıro glu et al., 2018).
Narratives of children and their families frequently highlight the safety and security risks for children posed by overcrowded familial living arrangements and lack to access to economic resources (Bermudez et al., 2018;Meyer et al., 2019;Sim, Fazel, et al., 2018). Overcrowding may force girls to seek shelter with non-family members in unsupervised housing, placing them at risk of physical and sexual violence (Bermudez et al., 2018). Lack of access to economic resources and poverty perpetuates food insecurity, homelessness, involvement in risky behaviour, including sex work, and inability to plan for the future (Bermudez et al., 2018;Sim, Fazel, et al., 2018).

Institutions and policy factors
Prolonged detention and deportation. Children entering a host country for temporary settlement or Child and Adolescent Mental Health resettlement, especially those who are unaccompanied or without required documentation, must negotiate the legal system and advocate for their rights. A lengthy asylum process may result in children's prolonged and multi-site mandatory detention (Chase, 2013;Hanes, Chee, Mutch, & Cherian, 2019;Lauritzen & Sivertsen, 2012;Mace et al., 2014;Thommessen et al., 2015). Hanes et al. (2019) note that 97% of asylum-seeking children in Australia report experiencing prolonged detention (median 7 months, range 3-12.5 months) across multiple sites (median 2, range 1-3). Mandatory detention is strongly and consistently associated with poor mental health outcomes and behavioural problems (Hanes et al., 2019;Lauritzen & Sivertsen, 2012;Mace et al., 2014;Thommessen et al., 2015). Reports from children highlight the needs for prompt and fair asylum-application processing as uncertainties surrounding application outcomes and prospects of deportation are linked to elevated distress levels, worry, and anxiety (Chase, 2013;Thommessen et al., 2015). Deportation also has major mental health consequences (Chase, 2013;Kienzler, Wenzel, & Shaini, 2019). Children deported back to their countries of origin report peer discrimination, social association, issues with accessing employment, resources and professional help, economic hardship, housing problems and precarious living conditions (Chase, 2013;Kienzler et al., 2019), with immediate family the only source of emotional and practical support in times of their sadness, distress and emotional pain (Kienzler et al., 2019).
Access to health and support services. Refugee children generally present with complex medical and psychosocial needs (Hanes et al., 2019). They require quick access to health care and support services which is often lacking (Groark et al., 2011;Hanes et al., 2019;Meyer et al., 2019;Sim, Fazel, et al., 2018;Vervliet et al., 2014). Children who have had access to support services highlight the important role support and encouragement from staff play in helping them to overcome anxiety and worries while waiting for an asylum decision (Eriksson & Rundgren, 2019;Groark et al., 2011;Thommessen et al., 2015). Unaccompanied refugee children often keep regular contact with support staff after they leave detention centres (Groark et al., 2011;Eriksson & Rundgren, 2019).
Length of asylum and resettlement processes. Time spent in a host country of permanent resettlement is most commonly associated with decreased psychopathology (Khamis, 2019;Montgomery, 2010;Akiyama et al., 2013;Bettmann & Olson-Morrison, 2018;Correa-Velez et al., 2010). Montgomery (2010) report that the prevalence of mental health problems decreased from about threefourths to one-fourth at follow-up in Middle Eastern refugee children resettled in Denmark. However, mental health problems in unaccompanied refugee children resettled in the Netherlands (Jensen et al., 2014) and Belgium (Vervliet et al., 2014) persisted over time, linked to lengthy asylum procedures and resettlement stresses. Resettlement stresses brought by cultural differences (e.g. belief system, culture, language) are frequently reported as the main reasons contributing to instances of deterioration of the mental health of refugee children in Canada over time (Beiser et al., 2015).

Discussion
As noted earlier, we used a later iteration of Bronfenbrenner's bioecological model of human developmentthe PPCT model (Bronfenbrenner and Evans, 2000;Bronfenbrenner, 2005) -to consolidate evidence from the last decade and assess our advance in understanding of the factors shaping mental health and psychosocial outcomes in refugee children since the Ager (1993) review. The findings above indicate significant advance in documenting risk and protective factors associated with specific bioecological strata impacting on the child. However, in addition to characterisation of much of the evidence base on factors predicting psychological adjustment as 'anecdotal and impressionistic', the earlier review noted that 'the complex manner in which such influences may interact is poorly understood' (p. 23). This is thus potentially another important aspect of strengthened evidence. One of the core propositions of PPCT is the dynamic interaction of individual child factors and multi-layered contextual factors (see Figure 1). Another is the role of timein various formsshaping the child's interaction with these ecosystemic strata. In this discussion, we therefore develop analysis of the literature identified, and the understanding of the complex processes which shape refugee child mental health and well-being, by drawing upon these principles. Specifically, we consider the implications of Bronfenbrenner's understanding of life course principles (socio-historical time and developmental age), proximal processes and child agency (demand, resource, force characteristics) in determining outcomes (Bronfenbrenner, 1995(Bronfenbrenner, , 2005.

Time: historical time and development age
Bronfenbrenner gives a prominent place to life course principles in human development -'historical time' and 'developmental age' (Bronfenbrenner, 1995). The original (Ager, 1993) review considered the outworking of historical time in the lives of refugee children with reference to phases of the refugee experience: preflight, flight, temporary settlement and resettlement (for a minority, in a third, host country). This represents a temporal as well as geographical trajectory, although the emergence of widespread protracted displacement in countries of initial refuge (where, as Qasmiyeh (2020) has noted 'the camp is the incinerator of time') has rendered the phrasing 'temporary' problematic.
Many children continue to be exposed to violence, hunger, physical or sexual abuse, forced labour, homelessness during flight (Groark et al., 2011;Longobardi et al., 2017;Nasıro glu et al., 2018;Sapmaz et al., 2017;Vervliet et al., 2014). Those experiences are especially prominent in unaccompanied refugee children being frequently exploited by traffickers. In the study by Longobardi et al., (2017), all unaccompanied refugee minors (19) reported experiencing physical abuse (e.g. being insulted, hit with an object, punished with infliction of pain, tortured, forced to engage in dangerous activity) and more than half sexual abuse. These children reported up to a 6-month long migration journey (Longobardi et al., 2017).
Although recent literature has continued to document exposure to deprivations and stressors associated with the preflight and flight phases (and the psychological sequelae of these), the most marked development in the evidence base over the last 25 years is with respect to the stressors of settlement (either temporaryor protractedwithin camps or urban settings) and resettlement (most commonly through formal resettlement or asylum processes). There is strong evidence regarding the impacts of these (Ellis et al., 2010;Beiser & Hou, 2016;C ß eri & Nasiro glu, 2018;Groark et al., 2011;Montgomery, 2010;Patel et al., 2017;Thommessen et al., 2015;Vervliet et al., 2014). Host countries often fall short in providing adequate psychosocial, health care, educational and financial support to refugee children (C ß eri & Nasiro glu, 2018;Hirani, Cherian, Mutch, & Payne, 2018;Mace et al., 2014;Montgomery, 2010;Patel et al., 2017;Thommessen et al., 2015;Vervliet et al., 2014). Dissatisfaction and emotional tensions arising from uncertainly, unmet needs and expectations are linked to a range of persistent psychosocial problems (C ß eri & Nasiro glu, 2018; Groark et al., 2011;Hirani et al., 2018;Jensen et al., 2014;Montgomery, 2010;Sim, Fazel, et al., 2018;Zwi et al., 2018). Again, these experiences are more pronounced for unaccompanied minors, who are variably accommodated in the care system in different countries (Hirani et al., 2018;Jensen et al., 2014). For instance, a two-year follow-up assessment of unaccompanied refugee children's mental health resettled in Norway showed that symptoms of psychopathology were overall unchanged over time, with 11% scoring high on suicidal ideation (Jensen et al., 2014).
This axis of historical time interacts with developmental age. Younger children are generally at greater risk for developing psychopathology in response to war-related adversities (Aitcheson et al., 2017;Beiser et al., 2015;Lincoln et al., 2016). This is because infants and young children rely heavily on parents (or other caregivers) for food, warmth, shelter and protection (Smith, Cowie, & Blades, 2015). They also depend on early positive parent-child interaction for development of social skills and cultural meaning. Provisions of these resources are heavily compromised amongst refugee families, especially in those at temporary settlements. Evidence however suggests higher rates of PTSD and depression amongst older children (Buchegger-Traxler & Sirsch, 2012;Correa-Velez et al., 2010;Khamis, 2019;Nasıro glu et al., 2018;Panter-Brick et al., 2015;Sapmaz et al., 2017). Again, this trend is especially apparent in those who are unaccompanied (Hirani et al., 2018;Jensen et al., 2014;Longobardi et al., 2017).
The experience of older children often includes prolonged and cumulative adversities, including ongoing community violence, sexual abuse, parental neglect and repeated domestic violence, abuse of power by those in authority (C ß eri and Nasiro glu, 2018; Longobardi et al., 2017;Nasıro glu et al., 2018;Sapmaz et al., 2017). This type of repeated, at times life-threatening, experiences typically undermines basic trust in others and their social attendance resulting in breakdown of social interactions (proximal processes) which tend to protect against the development of psychosocial problems (C ß eri & Nasıro glu, 2018).

Proximal processes
Consistent with Bronfenbrenner's (1995Bronfenbrenner's ( , 2005 PPCT model, the reviewed evidence supports the centrality of proximal processes in determining refugee child mental health and psychosocial developmental outcomes. Proximal processes are enduring, reciprocal and evolving interactions between the developing child and people, especially 'significant others', symbols and objects in their immediate environment (Bronfenbrenner, 1995(Bronfenbrenner, , 2005. While peers and others in the child's immediate environment can be of relevance in this regard, this highlights the pertinence of focusing on parent-child interaction and parenting behaviours within the reconfigured family life of refugee children (Beiser et al., 2015;Fazel & Betancourt, 2018). Quality of parent-child relationships (proximal processes) can both enhance and inhibit a protective environment for children (Bermudez et al., 2018). A number of studies point to the effect of warmth, supports, responsivity and stability of parent-child interaction in determining positive mental health outcomes in refugee children (Bermudez et al., 2018;Bettmann & Olson-Morrison, 2018;Eruyar et al., 2020;McGregor et al., 2016;Nasıro glu et al., 2018;Oppedal & Idsoe, 2015;Zwi et al., 2018). Parents can increase children's sense of security and safety by providing emotional warmth and supports, material resources and advice on how to avoid risky situations (Bermudez et al., 2018). Responsive and stable parent-child interactions (proximal processes) are linked with post-traumatic growth and mitigating the impact of atrocities on children (Sleijpen at al., 2016).
According to emotional security theory, high levels of parental stress and inter-parental conflict threaten children's sense of security and psychosocial adjustment (Cummings & Davies, 2010). Refugee families high in interpersonal conflict may become overloaded with stress. Stress fosters more conflict and conflict elevates stress levels in parents. Accumulated tension from parental mental health and interpersonal conflict often spills over into parentchild interaction and manifests itself in the form of harsh or punitive parenting behaviours (proximal processes); ultimately, negatively impacting on child psychosocial developmental outcomes (Dalg ard & Montgomery, 2017;Eruyar, Maltby, & Vostanis, 2018, 2020Lauritzen & Sivertsen, 2012;Van Ee et al., 2013;Panter-Brick et al., 2015). Therefore, addressing mental health and psychosocial needs of refugee children requires thorough consideration of the mental health cascade across generations and the cumulative adversities that can negatively impact on proximal processes within a family system. These findings call for family-centred interventions targeting parental stress levels, promoting parental coping and positive parenting practices. Improvements in family cohesive relationships may support achieving optimal psychosocial and mental health outcomes in refugee children. Similar principles must be applied in the design of supportive services for unaccompanied minors.
Child agency: demand, resource, force characteristics Bronfenbrenner (1995Bronfenbrenner ( , 2005 proposes that individual factorsdemand (gender, race, ethnicity), resource (the mental and emotional resources formed through past experiences, skills, intelligence) and force characteristics (temperament, motivation, and persistence) represent child agency in interacting with time, contextual factors and proximal processes and shaping developmental outcomes. Consistent with this view, we identified consistent evidence for contextual and proximal processes within the ecology of refugee children vary as a function of child agency.
Nonetheless, in early resettlement many children are able to experience positive adaptational and existential outcomes following adverse experiences (Aitcheson et al., 2017;Mace et al., 2014;McGregor et al., 2016;Veronese et al., 2012). Children may become insightful, self-reflective on their emotional needs and functioning, experience heightened appreciation for life, spiritual and personal growth (Meyerson, Grant, Carter, & Kilmer, 2011).
Although children bring resource and force characteristics with them and may present as 'resilient' (Aitcheson et al., 2017;Mace et al., 2014;McGregor et al., 2016;Veronese et al., 2012), the way social and physical aspects of their environment respond (proximal processes) are an important influence on their psychosocial trajectories. For instance, in females residing in camps the adjustment difficulties, lower neighbourhood attachment, engagement in risky behaviour (e.g. transactional sex) is partly influenced by social and physical distancing caused by peer and adult discrimination (proximal processes), poverty and poor living conditions (contextual factors) (Bermudez et al., 2018;Buchegger-Traxler & Sirsch, 2012;Sim, Fazel, et al., 2018). On the other hand, unaccompanied refugee minors who get adequate support from peers and detention centre staff (proximal process) report lower levels of anxiety, worries and depression while waiting for lengthy asylum process (institutional factors) at resettlement (Eriksson & Rundgren, 2019;Groark et al., 2011;Thommessen et al., 2015). Refugee children therefore make better use of their adaptational abilities and may maintain a strong sense of self when adequately supported by 'significant others' (e.g. parents, peers, adults, support staff), symbols (e.g. belief system, language, culture), and objects (e.g. supportive policies and institutions) in their daily environment.

Implications for practice
The preceding analysis has identified empirical evidence of a wide range of factors having influence on the mental health and well-being of refugee children.
Interventionswhether focused on policy or programmingthat moderate risk factors or promote protective factors clearly stand to secure positive impact. However, grounding the preceding analysis within the PPCT frame suggests three wider principles shaping intervention planning.
First, the systems framing points to the potential for interventions at one systemic level to influence factors at other levels and, further, the potential value of identifying and synergising such linkages. Structural interventions in camp settings, for example in relation to shelter, are likely to have direct impacts on children's well-being. However, the benefits may also support community-, school-and household-based factors supporting children in this environment, especially if such benefits are actively anticipated and cultivated (Ager, Annan & Panter-Brick, 2013). An increasing number of studies are now adopting an ecosystems understanding of intervention of this form, this anticipation of systems linkages distinguishing it from a more siloed approach. The Welcoming Cities movement in the USA (Welcoming America, 2017) provides a vivid example of how policy intervention at the exosystemic level can facilitate change in multiple parts of ecosystem (including housing, healthcare, schooling, community relations and employment) into which a refugee child is resettling (McDaniel, Rodriguez, & Wang, 2019).
Second, while a diverse range of influences on child refugee mental health may be identified, PPCT analysis points to the crucial role of proximal processes, and thus the central relevance of interventions addressing parenting. The general upsurge in focus on parenting interventions of the last decade (Hodes and Vostanis, 2019) is to be welcomed. However, specific supportive interventions are needed which address proximal processes within refugee households under acute migratory or resettlement stressors. Psychoeducational interventions shaped by an understanding of processes of acculturation and the potential impacts of past traumatic exposure (e.g. Bjorn, Boden, Sydsjo, & Gustaffson, 2013;Fazel & Betancourt, 2018;Osman, Flacking, Schon & Klingberg-Allvin, 2017;Slobodin & de Jong, 2015) are likely to be particularly effective.
Third, Bronfenbrenner's formulation of the PPCT model was substantially driven by a concern to ensure children's development is not understood as solely socially determined but also reflecting the agency of the child. As noted above, our analysis suggests that there is an important place for child agency in intervention design for refugee children. We see significant value here in Ungar's formulation of resilience as 'the capacity of a biopsychosocial system (this can include an individual person, a family or a community) to navigate to the resources necessary to sustain positive functioning under stress, as well as the capacity of systems to negotiate for resources to be provided in ways that are experienced as meaningful' (Ungar, 2011). This understanding emphasises the value of interventions which strengthen capacity for navigation towards, and securing of, personal, familial and community resources. Strengthening such capacityin line with developmental ageshould be the central focus of preventive interventions regarding child refugee mental health and well-being. Noting the interdependencies discussed above, this will often be best accomplished in the context of multi-layered interventions addressing both contextual and individual factors influencing adjustment.

Limitations
Several limitations should be noted. Refugee children are a widely diverse group. Elaborations presented in this review do not distinguish between specific needs of high-risk refugee populations, for example, unaccompanied refugee minors, trafficked and undocumented children, former child soldiers and those with a history of torture, who might need more urgent and tailored mental health support interventions. We collate evidence from studies in widely differing social, economic and cultural contexts, including children in protracted camp settlements and those resettled in high-income host countries, within the same analytic frame. This allows us to observe trends in common across these diverse experiences, but should not be taken to underplay the significance of differences in these contexts. Evidence is collated from peer-reviewed articles indexed in four electronic databases in English. This may have introduced publication and language biases. Although these biases have received much less consideration in narrative reviews than systematic reviews and meta-analyses, their potential effects should be acknowledged. Most importantly, there are constraints associated with the use of a framework of risk and protective factors to organise literature search and data extraction.
Although we adopted clear definitions for these terms, we acknowledge that selected studies drew on a wide range of alternative theoretical framings for these constructs. Further, the relative impact on developmental trajectories of the risk and protective factors considered should not be implied: this has not been formally assessed andas indicated by the analysiswill, indeed, vary widely across contexts and time-course.

Conclusion
The study notes the burgeoning literature on mental health and psychosocial well-being of refugee children, significantly deepening our understanding of factors shaping experience and outcomes. A bioecological framing of the varied risks and protective factors influencing refugee children's well-being serves as a suitable frame for collating this evidence, especially with regard to processes of their interaction. Bronfenbrenner's later PPCT model serves to highlight the interaction of historical time and developmental age, the pivotal role of proximal processes and the influence of child agency in determining outcomes. Informed by this framing of the evidence base, mental health-promoting interventions should place greater emphasis on promoting positive proximal processes operating between an active, evolving refugee child and 'significant others' (parents, peers, adults, teachers and support staff), symbols (belief system, language, culture) and objects (policies and institutions) in their daily environment. These interventions targeting proximal processes are necessary if appropriate actions to address risks and stressors at individual, community, school, institution and policy levels are to have traction on mental health and psychosocial wellbeing.