The database and grey literature search produced a total of 2359 records from which 13 reviews were included (for details of the review process see Additional file 1: Figure S1). Details of the included reviews are displayed in Table 1 (for excluded studies see Additional file 2: Table S1). Within the included reviews, 7 were systematic reviews of quantitative studies, 3 were systematic reviews of quantitative and qualitative studies and 3 were unsystematic narrative reviews. The reviews included studies up until 2017 and covered a diversity of countries, age groups and war settings, with pooled sample sizes ranging from 730 to 32,046.
The global coverage of the reviews is displayed in Fig. 1 and indicates that there are a number of countries and regions affected by conflict that have yet to have a study conducted on psychological interventions for children, predominantly in North Africa, south Asia and South America. Countries with brutal violent conflicts such as the Central African Republic, Syria and Somalia were not included in a single review. Some reviews covered countries not identified as conflict-affected within the last 9 years (according to the Global Peace Index 2007–2016), although this was largely due to sub-regional political violence such as in Poso, Indonesia (, reviewed in 15) or follow ups of historical conflicts such as the Spanish Civil War (, reviewed in 1).
The AMSTAR quality assessment results are displayed in Table 2. The assessment shows that 5 reviews were of low quality, 6 of moderate quality and 2 of high quality. Studies were most commonly marked down for not providing protocols, not providing a list of excluded studies, and not assessing publication bias.
Mechanisms of change
Out of the 13 reviews, only 6 referred to the mechanisms of change of their included interventions. Seven reviews were therefore not included in the analysis. In total, 16 core mechanisms (including 1 adverse mechanism) spanning the 4 tiers of the IASC model and different intervention aims were identified, and are outlined in Table 3. A quality assessment of these mechanisms’ supporting evidence rated 7 mechanisms as low, 5 as moderate, and 4 as high quality. Only one review  referred to two studies that statistically tested the mechanisms of change.
Creating safety and protection from harm
One review  cited the need to first and foremost protect children from harm, to create a protective environment and a sense of safety, in order to prevent further traumatisation, exploitation, and to promote wellbeing and mental health. Ager et al.  describe how ‘the building and strengthening of a protective environment for children vulnerable to abuse, exploitation and/or violence is paramount to effective [intervention].’ The evidence supporting this mechanism is low, however, as it is based primarily on program evaluations.
Two reviews [1, 14] cited the need for children to play in order to create a normal environment, to safely act out and explore traumatic memories and their meanings, to build relationships, using drama, music, role-playing and drawing, and to counterbalance stressful experiences. Apfel and Simon  describe how ‘interventions and programs that encourage and allow children to play, including playing out some of the traumatic events to which they have been subjected, may have a considerable impact on the child's ability to cope.’ The evidence supporting this mechanism is low, however, as it is based primarily on case or cross-sectional studies.
Community capacity building
Three reviews [6, 14, 17] cited community capacity building and the strengthening of community protective mechanisms for children as an important mechanism of change, through Child Friendly Spaces, greater community contact and reporting of violations of safety, an improved sense of community efficacy, stronger school systems and social networks, and good community cohesion. Apfel and Simon  describe ‘the tremendous importance of school in establishing and re-establishing some order and sanity in the lives of children traumatized by violence. School can provide the stabilizing framework in which the child's imaginative and cognitive skills can safely grow, or grow in relative safety.’ The evidence supporting this mechanism is low, however, as it is based primarily on program evaluations.
Increasing social support
Two reviews [1, 6] cited the importance of increasing social support, sourcing social support outside of immediate family, whose capacity may be stretched, and improving social skills to boost self-esteem, interpersonal deficits and access to social supports. ‘Resilient children have a knack,’ writes Apfel and Simon , ‘for turning to adults other than parents for guidance and resources if they cannot find such support in their own families.’ The evidence supporting this mechanism is again low, however, as it is based primarily on program evaluations and clinical experience.
Family and caregiver capacity building
Two reviews [1, 36] cited the need to support families, caregivers and practitioners in order to improve their ability to support children, through psychoeducation, dialogue, and through self-care. Caregivers are affected by the same war and violence affecting the children they care for, with the addition of containing the distress of children and their own childhood traumatic experiences. Statistical testing in two studies cited by Jordans, Pigott and Tol  provides high quality evidence for this mechanism.
Family and caregiver relationship strengthening
Three reviews [1, 14, 36] cited the strengthening of family and therapeutic relationships, of involvement in interventions, and of improved consistency of caregiving, particularly during periods of active conflict, for the long term wellbeing of children. This was also cited within the context of looking beyond traumatisation to the daily experience of children within the context of collectivist cultures, where family relationships are a core resource. Statistical testing in two studies cited by Jordans, Pigott and Tol  provides high quality evidence for this mechanism.
Engaging with values, traditions, religious and non-religious beliefs, and ideologies
Three reviews [1, 11, 14] cited the engagement in traditional, religious and political belief systems as important to enhance child wellbeing, by building hope and strength, connecting to culture, and restoring a sense of safety and normalcy. Also cited was the role of values like hope, strength, perseverance, forgiveness, honour and trust, and of culturally specific values such as Sumud in Palestine (meaning connection to the land, steadfastness, and the struggle to persist), Kwizerana in Rwanda (meaning family trust), Tarbia in Afghanistan (meaning a strong sense of morality), as well as customs such as cleansing rituals in Angola for the reintegration of former child soldiers into communities. Betancourt et al.  describe how ‘in many settings, traditional healing practices make critical contributions to social healing in the context of war. For instance, in Zimbabwe, Zezuru healers are known to engage family and community members in groups, draw out concerns over children's problems, facilitate reconciliation in and between families, and create a restorative climate.’ The evidence supporting this mechanism is low, however, as it is based primarily on qualitative studies or clinical experience.
Learning about the presenting problem, medication, and how to access services (psychoeducation)
One review  cited the mechanism of learning about symptoms of mental ill health and improving awareness about how to access services as a mechanism to improve mental health, especially when combined with skills building and counsellor contact. Betancourt  describes how ‘classroom-based programs that combine psychoeducation, skills building, and supportive counselor contact may be adequate to reduce distress in war-exposed youths living in low-resource settings.’ The evidence for this mechanism is moderate as it is based on controlled trials of interventions, but without specific testing of the proposed mechanism.
Learning stress management skills
One review  cited improving stress management skills as a key mechanism in promoting wellbeing, and preventing and treating mental ill health, including ‘relaxation techniques, good sleep habits… building safe settings, [and] setting positive goals,’ which function by reducing distressing symptomology of PTSD, as well as enhancing effective coping, increasing body and emotional self-awareness, and improving sleep. The evidence for this mechanism is also moderate as it is based on controlled trials of interventions, but without specific testing of the proposed mechanism.
Emotional regulation and bearing negative emotions
Two reviews [1, 6] cited the mechanism of improving emotional regulation, and the reduction of avoidance of negative or uncomfortable emotions, in order to promote wellbeing and prevent and treat mental ill health; by recognising, tolerating and responding to emotions as opposed to the natural tendency for temporary relief through avoidance, denial or suppression; as well as safely re-processing painful, shameful and overwhelming feelings. Cultural differences in the acceptability of expressing emotions were also cited however, as well as the adaptive mechanism of avoidance during ongoing emergencies, in order to concentrate on survival and defer emotional processing to a safer time. Apfel and Simon  describe the benefits of emotional flexibility, where there ‘is some ability to defer or defend against some overwhelming anxiety or depression when emergency resources are needed. This may mean compartmentalizing the pain and deferring the experience of overwhelming emotion until a time or situation when it is safer to experience it.’ The evidence supporting this mechanism is low, however, as it is based primarily on program evaluations and clinical experience.
One review  identified the process of ‘active problem solving’ as a positive mechanism for children’s mental health and psychosocial wellbeing as part of focused non-specialist interventions. The evidence for this mechanism is high as it has been statistically tested through mediational analysis.
Two reviews [1, 14] cited the process of altruism and helping others as a vehicle to promote wellbeing and prevent mental ill health, through an improved sense of purpose and increased internal locus of control, such as a preventative intervention for young children which encourages being a responsible caregiver for a toy dog. This mechanism is described by Apfel and Simon  as ‘learned helpfulness’ in contrast to the well-known phenomenon of ‘learned helplessness.’ They describe how altruistic acts create ‘a sense that "you may be helpless right now to stop a bomb from falling, but you are not helpless to deal with its human consequences"’ and support it with studies from Beirut, that found that ‘children instructed to use the interval between shellings to go out and bring food to an invalid relative, instead of using the time to watch television, did much better.’ The evidence for this mechanism is of moderate quality as it is based on controlled trials of interventions, but without specific testing of the proposed mechanism.
Adverse mechanism: Pathologising normal reactions
In the context of providing specialist support, one review  cited an adverse mechanism in which children’s wellbeing and mental health can be harmed by pathologising normal and adaptive responses to the extreme stresses of war environments. Apfel and Simon  state that ‘interventions specifically labelled as "psychological’, let alone "psychiatric," can alienate most of the people they are intended to help…survivors of terrible traumas such as the Holocaust or the Cambodian genocide have conveyed that they have already been labelled, categorized, and declared deviant, if not sub-human. These groups do not need any further psychiatric categorizing.’ The authors strongly recommend avoiding this by blending interventions into wider welfare programmes: ‘Combining psychosocial interventions with basic health and welfare interventions, therefore, tells both the clients and the providers that to be upset is expectable and that such responses are not deviant.’ The evidence for the adverse effect is of low quality however, as it is based on descriptions of clinical experience.
Trauma processing through narratives, exposure, dreaming or play
Four reviews [1, 6, 14, 36] described some form of trauma processing as a mechanism to treat traumatic stress, through narration or prolonged exposure to help to re-organize and integrate traumatic autobiographical memories. The technique to deliver this mechanism of change differs across interventions; through storytelling in KidNET in which children tell their whole life story, with detailed exploration of all traumatic memories; through imaginal exposure or in vivo exposure in CBT in which children retell specific traumatic events or face reminders of these events; through play re-enactment; or through dream work and guided imagery. Peltonen & Punamäki  describe the latter as techniques that enable a ‘rich, structurally coherent and healing symbolic process.’ The evidence for this mechanism is of moderate quality as it is based on controlled trials of interventions, but without specific testing of the proposed mechanism.
Restructuring unhelpful cognitions and appraisals
One review  describe the mechanism of altering and restructuring unhelpful, upsetting and unrealistic thoughts, interpretations and appraisals to treat and prevent traumatic stress, by both the correction of biased interpretations and the reframing of causal attributions (e.g. self-blame). Peltonen & Punamäki  describe how this process helps in ‘making sense of trauma…empowering coping skills and integrating of fragmented and intrusive thoughts and feelings into a more coherent experience.’ The evidence for this mechanism is again of moderate quality as it is based on controlled trials of interventions, but without specific testing of the proposed mechanism.
One review  cited the role of therapeutic rapport in treating mental ill health, specifically the development of a trusting therapeutic relationship and a safe environment for disclosure of traumatic experiences. In contrast, an adverse role was cited for therapeutic relationships which take on a moralistic stance. Jordans, Pigott and Tol  describe positive correlations in a mediation analysis for ‘counsellor demonstration of reflective involvement, the opportunity to express emotion, and the absence of moralistic behaviour.’ The evidence for this mechanism is rated as high as it is based on mediational analysis of the mechanism.
These mechanisms with their respective quality ratings are displayed in Fig. 2. Overall, the quality of evidence is poor, with few studies testing mechanisms statistically. High quality evidence was found only for family capacity building, relationship strengthening, problem solving, and therapeutic rapport. Mechanisms at lower levels of the IASC pyramid (basic services and security, and strengthening family and community support) such as protection from harm, play, and capacity building had the poorest quality of evidence. Trauma processing was the most cited mechanism, and was included at least once for each intervention type.
Subjective reflection and researcher bias
Researcher reflective notes showed three core considerations. Firstly, the utility of interventions (and intervention research) aiming to treat children for post-traumatic stress disorders was raised because of the reliance on post trauma literature and its application to populations experiencing ongoing and indefinite terror. Secondly, the eighth mechanism of engaging with traditions was worded carefully due to concern raised about cultural biases and norms around traditional healing, particularly as most reviews were conducted by authors based in European or North American countries. Finally, the difficulty in differentiating mechanisms from techniques and protective factors was noted. Reviews often highlighted this lacking detail and clarity on underlying processes in their included studies.