A Task-shifting and Family-focused Approach Towards Mental Health Care for Youth Living in Refugee Camps

Background: In humanitarian emergencies with few specialist resources for mental health care, it is a practical and ethical necessity to allocate available resources to interventions that target youth most in need of support and address contextually relevant factors of inuence. Based on the ndings of an extensive epidemiological observational study with 230 Burundian refugee youth aged 7 to 15 years and both their caregivers in three refugee camps, we propose a multi-layered mental health service model that is characterized by task-shifting from professionals to non-professionals, close-meshed collaboration between different agencies within camps and inclusion of families at all stages. Discussion: The model prioritizes the identication of youth with clinically relevant mental health problems through extensive screening, who are then provided with trauma- and/or family-focused interventions depending on a more detailed assessment of needs. We emphasize the importance of incorporating evidence-based interventions and evaluating all model components and discuss caveats and limitations concerning the implementation of the model. Conclusion: The paper aims to sensitize researchers and practitioners in the eld of mental health care for youth in refugee and other humanitarian settings to the importance of conducting epidemiological assessments of the specic need for interventions and of contextually relevant intervention targets.


Introduction
Refugee children and adolescents living in refugee camps in low-and middle-income countries are at an increased risk of developing debilitating mental health problems, such as posttraumatic stress disorder (PTSD), internalizing problems including depression and anxiety and externalizing problems including aggressive and antisocial behavior, as a result of their exposure to violent con ict in their home countries and to ongoing hardships in the camps (Reed et Saltzman et al., 2003). The common idea underlying these approaches is that mental health and psychosocial care is provided on multiple layers, which can be illustrated in form of a pyramid. Although these layers are ideally put in place concurrently, they are hierarchically organized, from the provision of basic services and security at the bottom through broad-scale resilience-building activities for communities and low-level interventions provided by non-specialists to specialized services focusing on individuals suffering from severe mental health problems at the top of the pyramid (Inter Agency Standing Committee (IASC), 2007; Jordans et al., 2010). In this paper, we argue in favor of focused interventions targeted at children and adolescents with clinically relevant mental health problems over universal approaches addressing large numbers of youth in an indiscriminate manner for two reasons: a more e cient use of limited resources for mental health care and avoiding harmful effects of non-targeted interventions.
While available stepped-care models provide a comprehensive and broadly applicable framework for the planning and delivery of interventions to refugee and other con ict-affected children, their features and implementation also depend to a large extent on the speci c context, needs and resources (Jordans et al., 2010). A general imperative for any kind of intervention is that they should rst establish the need for an intervention in a given setting through an epidemiological assessment and assess contextually and culturally relevant risk and protective factors as intervention targets (de Jong et al., 2015). An obvious argument for this approach is to enable the allocation of scarce resources to those who are really in need of interventions (Stevens & Gillam, 1998).
Given the high prevalence rates of mental disorders often found among children and adults in most displacement settings in low-and middle-income countries (Morina et al., 2018;Vossoughi et al., 2018), large-scale and low-intensity interventions appear to make the most e cient use of limited available resources. However, epidemiological prevalence estimates highly vary based on the speci c context and methodological factors such as sampling and assessment methods, with lower rates observed in random samples assessed with structured clinical interviews (Kien et al., 2019). In settings with lower yet still considerable prevalence rates, a targeted approach may be more suitable to address those youth who are really in need of mental health care. Another less obvious argument for a priori epidemiological assessments of a population`s need and relevant factors of in uence refers to possible negative treatment effects for subgroups of children in areas of armed con ict (Jordans et al., 2016). For instance, in unstable and stressful settings, universal school-based programs may also undermine the natural recovery of children suffering from clinically relevant symptoms of PTSD, depression and anxiety and thus be harmful in fact (Ertl & Neuner, 2014;Tol et al., 2014).
In early 2018, we set out to conduct an epidemiological and observational cross-sectional study to assess the prevalence of mental health problems, i.e. the need for interventions, and relevant contributing factors among Burundian refugee families living in refugee camps in Western Tanzania. After brie y describing the overall study context we summarize the main ndings of the study, from which we then derive a mental health service model for Burundian refugee youth.

Overall Study Context
Having seen several phases of extreme inter-ethnic violence since its independence in 1962, including a long-lasting civil war from 1993 until 2005 (Uvin, Page 3/10 care, education and other services in the camps (Reliefweb, 2018). The International Rescue Committee (IRC), the main provider for mental health and psychosocial services in the three refugee camps, employs eight mental health professionals (four psychologists in Nduta, two in Nyarugusu and two in Mtendeli; no psychiatrists) for all Burundian refugees, implying a ratio of one psychologist for about 30 000 people. Currently, there are only limited and few speci c mental health interventions targeting children and adolescents in the camps, namely psychosocial support and socio-emotional learning groups at school, play therapy and counselling (personal communication).

Summary of study procedure and main ndings
The study was conducted between January and May 2018 in the three camps. We included family triads consisting of the mother or primary female caregiver, the father or primary male caregiver (in the following referred to as mothers and fathers) and the oldest child in primary school age, i.e. between 7 and 15 years. In each camp, we applied a combined systematic and random sampling approach (Scharpf et al., 2019), which ensured representativity for twocaregiver-households in the camps and provided epidemiologically relevant data. We conducted individual structured clinical interviews with mothers, fathers and children on their traumatic experiences, mental health problems and on factors potentially contributing to family members mentalhea < h. Theμ < i -∈ f or mantstructured ∫ erviewassessmentwasconsred → provm or evalatacompared → selfad min isteredque s mental health (Fazel et al., 2005). In addition, we conducted a small survey on childrens and parents awareness and use of existing mental health and psychosocial services in the camps as well as their coping resources. We took the following measures in order to increase the appropriateness of the assessment for the camp context and the cultural background of the sample: qualitative evaluations of the study instruments by members of the refugee communities in each camp who were also employed as research assistants, use of translators from the refugee community to increase participantsc omprehension of questions and a pilot assessment in the rst camp (Scharpf et al., 2019;Scharpf, Mkinga, Neuner, et al., 2020).
The assessment of youth revealed a one-month prevalence of 5.7% for PTSD and a prevalence of 10.9% for increased levels of internalizing and externalizing problems (Scharpf et al., 2019). According to mothers and fathers reports, levels of increased internalizing and externalizing problems were 15.9% and 11.5%, respectively. Interviews with parents revealed one-month prevalence of 32.6% for PTSD among mothers and of 29.1% among fathers, while 90.9% of mothers and 83.9% of fathers scored above the cut-off for general psychological distress within the past seven days (Scharpf et al., 2019).
While youthsl if etimeexposure → traumaticexperienceswasthestron ≥ stpredic → roftheirpsychopathology, higher ≤ velsof ⊥ hmothers and fathers psychopathologyadditionallycontributed → theirpsychopathology(Scharpf, Mk ∈ ga, Neu ≠ r, etal. , 2020). F or mothers, thisassociationwa s more insecure attachment representations of the mother-child relationship and higher levels of maltreatment by mothers as reported by youth. Moreover, higher levels of parental maltreatment were directly and indirectly (through youth spsychopathology l ∈ ked → m or eseveremem or ydeficits(Scharpf, Muel ≤ r, Masath, etal. , 2020), whichmayhavelongterm ¬ativecons s externalizing problems (Scharpf, Mkinga, Masath, et al., 2020). However, youth who reported to have higher quality friendships endorsed lower levels of PTSD symptoms and externalizing problems. The results of the survey on family members` use of existing mental health and psychosocial services as well as informal coping resources are displayed in Table 1.
) Table 1 Results of survey on the use of available mental health services in the refugee camps and informal sources of psychosocial support

Model Presentation
Based on these key ndings, we propose a mental health service model for Burundian families living in the refugee camps. In line with the Inter-Agency Standing Committee (IASC) Guidelines for Mental Health and Psychosocial Support (2007)  The base layer comprises all activities that aim at identifying those children who are suffering from severe mental health problems. This requires a broad approach and a close-meshed collaboration between different organizations and services, e.g. education, child protection, physical and mental health care.
The most important activity in this layer is screening children for mental health problems including PTSD symptoms, internalizing and externalizing problems, which can be conducted by trained para-professionals such as teachers, community mobilizers, nurses and social workers in their respective settings. It is crucial that screening instruments are locally validated (Hall et al., 2014). Youth who are screened positive can then be interviewed by a trained counsellor to establish a diagnosis and be referred to an appropriate intervention. For instance, Catani et al. (2009) followed such an approach in a treatment study with Sri Lankan youth who had been affected by war and tsunami: After an initial screening in schools, trained local counsellors conducted clinical interviews to assess the presence of PTSD among youth and delivered Narrative Exposure Therapy for children (KIDNET) or a meditation-relaxation intervention to those with a diagnosis.
The ndings of the survey further point to the importance of increasing youth saware ≠ ssofservices and reduc ∈ g riers → aesspartica ̲ rly ∈ parentswhoarema ∈ lyresponsib ≤ f or ∈ itiat ∈ gcontactwith or ganizatio understanding and make it easier for them to identify children suffering from problems. Moreover, working together with existing informal resources providing psychosocial support, e.g. traditional healers, religious groups or community elders, can be helpful in identifying and referring children in need of treatment.

Interventions
The identi cation of children and adolescents in need of intervention enables an effective and e cient allocation of available resources on the next layer of interventions. Our studies revealed youthstraumaexposure and familyrelatedfac → rs, i. e. ⊥ hparents psychopathology and maltreatment by parents, as promising targets for interventions. In view of these ndings, we emphasize the need for both trauma-focused and family-level interventions taking into account parents` well-being and aiming at reducing child maltreatment. Interventions that can be provided by non-specialist facilitators without cost-and time-intensive training may be most suitable. Most importantly, only evidence-based interventions should be included in the model, which need to be constantly evaluated (Fazel, 2018;Wessells, 2009).
With regard to trauma-focused interventions, narrative exposure therapy (NET) and its adaptation for children ( Based on a more detailed diagnostic assessment of children following identi cation, it is possible to tailor the intervention to the individual child sspec if icprob ≤ mareas. Acommone ≤ mentsapproachallows → comb ∈ ed ⇔ erenttreatmente ≤ ments ∈ af ≤ ξb ≤ ma ∩ erdepend ∈ g s needs and addresses not only PTSD symptoms stemming from prior trauma, but also emotional and behavioral problems related to daily stressors (Murray et al., 2018). Caregivers can be taught parenting skills in individual sessions. However, compared to the trauma-focused interventions described above, the evidence for such a common elements approach for refugee youth is much more preliminary. It has only been evaluated in a non-controlled study with refugee youth living in Somali refugee camps indicating decreases in PTSD symptoms, internalizing and externalizing problems as reported by youth and caregivers and improvements in youth-reported well-being (Murray et al., 2018).
Involvement of parents is only limited in the interventions described so far and a stronger focus on parenting may be warranted in order to effectively counter child maltreatment in the camps. In particular, our ndings suggest that the mother-child relationship may be an important target for the prevention of child maltreatment in the participating families. Therefore, the contextual adaptation and evaluation of existing relational interventions that have demonstrated effectiveness in reducing child maltreatment in Western samples may be promising (Toth et al., 2013;Valentino, 2017

Community-based prevention and resilience building
The layer at the top of the inverted pyramid comprises large-scale community-level psychosocial activities that promote children swellbe ∈ g and stren > hentheirresilience. Imp or tantly, theseshod ̲ focusonprevention and ¬conta ∈ e ≤ mentsai min gatreduc ∈ gme s and adolescents resilience in their social ecology at this stage (de Jong et al., 2015; Jordans et al., 2010), for example through sports contests, drumming and dancing sessions or praying groups (see Table 1). An overall positive effect of such joint activities is to strengthen peer relationships, which were related to better mental health in our study. A suitable setting may be child-friendly spaces, which are already implemented in the camps and provide a safe environment for children. A meta-analysis on the impact of child-friendly spaces in humanitarian settings in Ethiopia, Uganda, Iraq, Jordan, and Nepal observed an overall positive effect of these facilities on younger childrens (6 to 11 years) psychological well-being (Hermosilla et al., 2019). However, childfriendly spaces did not have an impact on adolescents` well-being and appeared to be ineffective in connecting younger and older children to wider community resources. Here schools may provide more appropriate settings to also engage families and communities as structured psychosocial activities can be combined with activities focusing on psychoeducation and community sensitization. For instance, Jordans et al. (2013) conducted a 2-session psychoeducation intervention delivered by lay community counsellors for groups of parents of children who had screened positive for emotional and behavioral problems at school in Burundi. The intervention group showed a short-term effect in reducing child-reported externalizing problems among boys compared to the control group.

Contextual factors
In order to achieve sustainable and comprehensive reductions of risks for and improvements of Burundian refugee childrens well-being, contextual factors related to living in the camps need to be addressed as well (Miller & Rasmussen, 2017). Our study ndings support the need for prevention of structural risk factors for childrens ongoing exposure to violence within the family and community, which constitute a source of continuous trauma and thus a signi cant mental health risk. Another study connected to this research project found that families lower household income was related to higher levels of mothers selfreported violence against children . This suggests that policies allowing refugees to work inside and outside the camps as well as livelihood programs teaching vocational skills may be fruitful to reduce economic and psychological strain on families and parents (Bermudez et al., 2018; Miller & Rasmussen, 2017). In a similar vein, a higher educational level of fathers was related to a lower child-reported use of paternal violence against children . This implies that programs which support parents in pursuing further education in the camps may also bene t families and children.

Discussion
Drawing on our epidemiological observational study with Burundian refugee families living in refugee camps, we emphasize the importance of considering both the speci c need for interventions as well as contextually relevant factors of in uence in order to make e cient use of scare specialist resources for mental health care for children in low-income humanitarian settings. We proposed a multi-layer mental health service model that takes into account the low prevalence of youth with severe mental health problems as well as the crucial role of family-level factors for affected youth and suggested possible interventions on each layer. In the following, we discuss concrete caveats and limitations with regard to the implementation of such a model.
While it is desirable that interventions are multi-modal to avoid addressing relevant risk factors for children`s mental health in a piecemeal fashion and overwhelming children and families through multiple different interventions at once, the lay providers of interventions should also not be overtaxed by having to learn many different treatment elements. Therefore, a specialized approach may be useful in which all providers are trained in activities related to identi cation, e.g. screening and psychoeducation, while they receive more specialized training in certain interventions, e.g. parent-or child-focused, and can be exibly consulted depending on the needs of a speci c child and family.
In general, our ndings suggest that fathers should equally be engaged in family-level and parenting interventions despite possible policy-and cultural-level Limitations of the described model predominantly re ect those of the original observational study, which may have affected study ndings, for example the use of instruments and cut-off scores that had not been validated in Burundian (refugee) samples as well as reporter biases such as over-and underreporting of symptoms (Scharpf et al., 2019;Scharpf, Mkinga, Neuner, et al., 2020). Moreover, as the proposed model was developed based on ndings from Burundian refugee families living in Tanzanian refugee camps, its generalizability to other populations and contexts is limited. However, while epidemiological need assessments in similar contexts are scarce and yielded highly varying prevalence rates of mental health problems among youth (Vossoughi et al., 2018), key intervention targets in our model such as traumatic experiences, parental mental health and family violence have been shown to be relevant factors for youths mental health also in other refugee camps (Scharpf, Kaltenbach, Nickerson, et al., 2020). Thus our model may also be applicable to other refugee camp contexts, which should however be tested in the future. From a pragmatic viewpoint, the proposed model could still be di cult to implement and sustain with limited nancial and specialist resources despite a strong focus on task-shifting to para-professionals (Jordans et al., 2010). As their training and supervision would make up a signi cant amount of the workload of available mental health professionals, the recruitment of additional specialists may be inevitable to provide su cient care capacities for all clients. Notwithstanding, improving and promoting refugee childrens and adolescents` mental health and adjustment is worth every effort, laying the foundation of a peaceful and productive society and healthy future generations.

Conclusions
A signi cant task mental health service providers in humanitarian emergencies, such as refugee camps, have to face is to allocate their limited resources in a way that will bene t those who are most in need of mental health care. Taking the example of Burundian refugee families living in Tanzanian camps, we argue that an epidemiological assessment of a target group´s needs and relevant factors of in uence are essential in deciding how to use available resources.
It is conceivable that it might appear more straight-forward and initially more time-and cost-effective to apply a service model that worked in another context without any a priori epidemiological assessment. Notwithstanding, this bears a risk of making an ine cient use of scarce resources and of undermining resilient trajectories through untargeted interventions. While the proposed care model recommends to provide interventions following a layered approach

Consent for publication
Not applicable Availability of data and materials Not applicable

Competing interests
The authors declare that they have no competing interests.

Funding
The empirical study this conceptual paper draws upon received funding from the North-South Cooperation of the University of Zurich [F-63212-13-01]. The funding body had no role in the design of the study, the collection, analysis, and interpretation of data and writing of the manuscripts.
Authors` contributions FS and TH developed the concept of the manuscript. FS wrote the rst draft of the manuscript and both FS and TH contributed to the nal version of the manuscript.