Background

Children and young people with experience of living in care represent a diverse population, with significant international variation in nomenclature and classification [1]. They can be defined as individuals who have had statutory involvement, whereby parental rights have been transferred to another adult. In some countries, such as the UK, there are specific mechanisms to support care entry, such as the issuing of Special Guardianship Orders [2]. Care can include a range of placement types, such as formal kinship care, foster care and residential care [3]. There are also variations in the identity of care-leavers, who are largely defined by their ongoing rights to statutory provision. For example in Germany, young people from a range of care placements are entitled to legal assistance until 21 years old while in England they are entitled to certain services up to 25 [3]. Globally, the estimation of children and young people in care has been challenging, with most recent efforts to establish the prevalence of individuals in institutional care reporting a range from 3.18 million to 9.42 million, depending on the methods and data sources employed [4].

While not a clearly defined population, evidence reports that care-experienced individuals generally have poorer mental health and wellbeing, and higher rates of suicide attempts, compared to non-care-experienced groups [5,6,7,8,9]. Individuals with a history of care have excess mortality in adulthood, attributable to non-natural causes of self-harm, accidents, and other mental health and behavioural risk [10]. Mental health problems incur substantial health and social care costs, largely due to the associated risk of placement instability and breakdown [11,12,13], which is concerning given increased financial pressures on social care systems [14].

There has been significant development in international intervention research to target reported issues. A number of literature and systematic reviews have synthesised the evidence base for social and healthcare approaches [15,16,17,18,19,20,21,22,23,24,25], with recent National Institute for Health and Care Excellence (NICE) reviews and associated guideline recommendations endorsing implementation of interventions centred on mentoring, positive parenting practices and system change to facilitate more efficient implementation [26].

Despite their contributions, there are two key limitations associated with extant syntheses, relating to both scope and methodology. The first limitation is a focus on a limited range of countries [26]; specific diagnosable conditions (e.g. depression) [20, 21]; discrete population subgroups (e.g. foster care) [17, 20, 27]; or single intervention packages (e.g. Treatment Foster Care) [27, 28]. Where reviews are inclusive of diverse outcomes, populations and intervention types, they tend to take an aggregative approach when presenting syntheses. Notably, there is limited differentiation between the evidence for interventions that operate in different parts of the social system.

This differentiation is imperative, as there is suggestion that interventions can be ineffective due to an over-reliance on individual-level approaches that are minimally disruptive [29], and there is a need to understand the evidence for structural interventions to guide the development of system-level approaches moving forward. Equally, with the advance of complex systems thinking perspectives in intervention research, there is increased recognition that an intervention’s functioning is dependent on its interaction with proximal and distal system characteristics [30,31,32,33,34]. As such, interventions operating in different parts of the system may be subject to different contextual influences and implementation challenges. We need to disentangle these complex interactions to inform effective intervention delivery in future.

There are a number of organising frameworks to help locate interventions in different parts of the social system, including the socio-ecological model, with versions originating from child development and public health research [35, 36]. There are broadly five domains of factors that influence outcomes, and which may be targeted for intervention [35]. These are as follows: intrapersonal, which is an individual’s knowledge, attitude and behaviour; interpersonal, which is an individual’s relationships and social network systems, including family and friendship networks; organisational, which is the formal and informal rules, ethos and characteristics of social institutions; community, which is the relationship between organisations and networks; and policy, which includes local, regional and national laws and policies.

The second limitation with existing reviews is that they tend to restrict syntheses to outcome evaluations, with scant attention paid to interventions’ programme theory, the context of evaluation, the process of implementation, acceptability or cost-effectiveness. Even recent comprehensive NICE reviews [26], which do include a range of evidence types, do not provide a clear overview of programme theories or the contextual factors that give rise to reported barriers and facilitators to intervention functioning.

Integration of these different evidence types is important in understanding how interventions operate and generate effects within their delivery context, and their potential transportability to other health and social care systems. This integrated approach to evaluation, which draws on a range of evidence, is recommended by a range of methodological guidance on intervention development, adaptation and evaluation [37,38,39,40]. As such, an evidence map and review that systematically charts the range of interventions targeting the mental health of care-experienced children and young people, in addition to the types of evidence currently generated, is important in identifying where there may be limitations in current intervention research and where it needs to be further strengthened [41].

The Care-experienced cHildren and young people’s Interventions to improve Mental health and wEll-being outcomes Sytematic review (CHIMES) was a complex systems informed, multi-method review that aimed to synthesise international evidence on programme theory, process evaluation, outcome evaluation, equity harms, and economic evaluation [42]. For the first phase, reported presently, we constructed a map of interventions and associated evaluations to chart key evidence gaps and clusters. It addressed the following review questions:

  • What are the targeted socio-ecological domains, theories and outcomes addressed in mental health and wellbeing interventions for care-experienced children and young people?

  • What are the types of evidence generated as part of intervention evaluations?

In charting the available evidence on interventions and types of evidence, the map informed the scope and feasibility of the second phase of the systematic review. For example, the map identified sufficient randomised controlled trials to conduct meta-analysis for relevant outcomes. The second review phase involved method-level syntheses for outcome evaluations, process evaluations, equity harms and economic evaluations. These were then integrated into an overarching review-level synthesis, where data from one synthesis (e.g. process evaluation) supported explanation of another synthesis (e.g. outcome evaluation) [43]. The third and final phase of the review entailed stakeholder consultation to reflect on the synthesis and prioritise interventions for future development and/or adaptation, evaluation and implementation.

Methodology

We generated an evidence map, drawing on systematic mapping guidance [44]. Evidence maps have some conceptual overlap with scoping reviews, but with clearer emphasis on stakeholder involvement in the early stages of the research process, a systematic search strategy, and the visual presentation of data [41]. As there is no standardised methodology for the reporting of evidence maps, we describe the process with reference to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Extension for Scoping Reviews (PRISMA-ScR) [45]. The review is registered with the PROSPERO registry of systematic reviews (reference number CRD42020177478).

Stakeholder consultation

We integrated a comprehensive programme of stakeholder involvement throughout the CHIMES review. Regarding the evidence map, we consulted with three key groups of stakeholders at the outset to refine and confirm the focus and scope of the review. These groups were as follows: (1) CASCADE Voices (young people’s research advisory group with care-experienced individuals up to the age of 25 years); (2) The Fostering Network in Wales Young Person Forum (group of care-experienced young people who provide advice and guidance to the charity on their programmes of work); and (3) All Wales Fostering Team Managers Forum (group of Local Authority and independent foster care providers). The central priorities of these groups were to focus on wellbeing and suicide-related outcomes, and to map interventions according to a socio-ecological schema. This latter priority related to stakeholders’ perception of a lack of structural interventions at the organisational and policy level, and a need to establish if this is a significant evidence gap and how it might be addressed moving forward.

Eligibility criteria

The inclusion parameters for the review were defined according to the Population, Intervention, Comparator, Outcome and Study Design (PICOS) framework:

Types of participants

Intervention participants could be care-experienced children and young people (\(\le\) 25 years old), or their proximal relationships, organisations and communities. Care could include in-home care and out-of-home care (foster care; residential care; and formal kinship care), and could be current or previous (e.g. care-leaver). The amount of time in care was not restricted. The following populations were excluded: general population; children and young people classified as being in need but not placed in care (e.g. having a Children in Need (CiN) plan or Child Protection plan); children and young people at the edge of care; care without statutory involvement; adoption; or unaccompanied asylum seekers and refugees.

Intervention

We defined interventions as an attempt to disrupt existing practices in any part of the social system (e.g. healthcare, social care, education, youth justice). They could operate across the following socio-ecological domains: intrapersonal; interpersonal; organisational; community; and policy. They could be mono-component or multi-component. There were no a priori criteria for implementation (i.e. delivery setting, delivery mode, delivery agent). Pharmacological interventions were excluded.

Comparator

For outcomes evaluations, a comparator had to be specified and could include: treatment as usual; other active treatment; or no specified treatment.

Outcomes

Interventions had to target one of three domains of primary outcomes: subjective wellbeing (in addition to life satisfaction and quality of life); mental, behavioural or neurodevelopmental disorders as specified by the International Classification of Disease (ICD)-11; and suicide-related outcomes (self-harm; suicidal ideation; suicide). Measurement could be dichotomous, categorical or continuous. Outcomes had to be obtained for the child or young person, but could be ascertained through clinical assessment, self-report or report by another informant. Excluded primary outcomes included substance misuse and eating disorders, which have some conceptual overlap with the eligible outcomes, but are large literatures that could form the basis of separate reviews. We mapped all secondary outcomes included in eligible study reports (e.g. physical wellbeing).

Study design

Different study designs were eligible according to the research question targeted. Study reports could describe an intervention’s programme theory; outcome evaluation (Randomised Contolled Trial (RCT) or non-randomised design); process evaluation that reported on context, implementation and/or acceptability (qualitative and quantitative design); and economic evaluation (cost-minimisation; cost-effectiveness; cost utility; or cost–benefit analysis).

Information sources and search strategy

We identified study reports from sixteen electronic bibliographic databases: Applied Social Sciences Index and Abstracts (ASSIA); British Education Index; Child Development & Adolescent Studies; CINAHL; Embase; Education Resources Information Center (ERIC); Cochrane Central Register of Controlled Trials; Cochrane Database of Systematic Reviews; Health Management Information Consortium (HMIC); International Bibliography of the Social Sciences; Medline; PsycINFO; Scopus; Social Policy & Practice; Sociological Abstracts; and Web of Science. We identified additional peer-reviewed studies and grey literature through searching websites of 22 relevant social and health care organisations. Searches were conducted May–June 2020 and updated April–May 2022. We contacted 32 subject experts and 17 third sector organisations for recommendations, particularly regarding grey literature and in progress studies. We screened relevant systematic reviews and conducted forward and backward citation tracking with included study reports. The search strategy was developed in Ovid Medline and adapted to the functionality of each platform (Supplement A). Searches were undertaken from 1990 to coincide with the ratification of the United Nations Convention on the Rights of the Child [46]. Study reports were restricted to higher-resource countries. They were not restricted by language.

Data selection

We uploaded retrieved citations to the Evidence for Policy and Practice Information and Coordinating (EPPI) Centre’s review software EPPI Reviewer version 4.0 for storage and management. Study titles were screened by one reviewer to identify clearly irrelevant retrievals, with irrelevant reports checked by a second reviewer. Title and abstracts were screened independently and in duplicate by two reviewers. Where there was a conflict on exclusion, the study report progressed to the next stage of screening. Full texts were screened independently and in duplicate with conflicts resolved through discussion or recourse to a third reviewer. An inclusion criteria proforma guided selection, which was tested and calibrated with a subset of retrievals. The same inclusion criteria were applied to study reports from databases and grey literature. Study quality or publication process (e.g. peer review) was not part of the inclusion criteria and was assessed as part of quality appraisal.

Data extraction

We coded eligible study reports for the evidence map according to country; publication date; intervention type; target population; intervention name; intervention characteristics; programme theory; evidence type; study design; and intervention outcome domains. Intervention characteristics were further coded in accordance with the Template for Intervention Description and Replication (TIDieR) Checklist for Intervention Development [47]. To support description of interventions, we extracted programme theory with a tool used in a previous systematic review [48]. Extraction domains were as follows: method or process for developing the theory; name of theory; discipline of theory; socio-ecological domain of theory; and description of theory.

Evidence map

Scoping review and systematic mapping methods supported the mapping of the evidence base [44, 49]. Following the coding of study reports, we constructed numerical and narrative summaries of intervention and evidence clusters and gaps, with accompanying infographics. For details on intervention characteristics, we produced a narrative summary and table describing the features according to extractable domains of the TIDieR framework. For interventions reporting on programme theory, we narratively summarised these according to the socio-ecological domains in which they operated and produced a summary table. For evidence types, we constructed a narrative summary and table.

Results

Study characteristics

A total of 15,068 unique study reports were identified. Of these, 888 were screened at full text, with 64 interventions being included that linked to 124 study reports (Fig. 1) [50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173].

Fig. 1
figure 1

PRISMA flow diagram

Full size image

Study reports were published between 1994 and 2022. Interventions were from twelve countries: USA (n = 77); UK (n = 22); Netherlands (n = 6); Belgium (n = 3); Australia (n = 3); Portugal (n = 3); Canada (n = 2); Ireland (n = 2); Israel (n = 2); Germany (n = 1); Spain (n = 1); Sweden (n = 1); and both the USA and UK (n = 1).

Intervention types

We classified interventions according to the socio-ecological domain or domains targeted (Fig. 2). As indicated, this was due to our assumption that interventions will interact with context differentially if they target different parts of the social system. The classification of interventions by socio-ecological domain was informed by information about the causes being targeted and the reported theoretical basis. Where the theory was not specified, we also drew upon reported information on the target population (e.g. individuals in a relationship with the care-experienced child) and delivery setting (e.g. a social care organisation). While interventions within each group had a shared target set of causes and theories, there was diversity in terms of activities. An overview of intervention characteristics is presented in Table 1.

Fig. 2
figure 2

Intervention type by socio-ecological domain

Full size image
Table 1 Description of intervention characteristics (N = 64) [174]
Full size table

The majority of interventions (n = 26) targeted the interpersonal domain. They primarily focused on the skills, knowledge and confidence of foster and kinship carers through training curricula and professional-delivered support. A small number of interventions promoted children and young people’s positive relationships with biological families, largely with the aim of facilitating reunification [54, 139, 169]. Elsewhere interventions provided opportunities to build relationships with peers [59, 142], trained mentors [99], clinicians [171] and wider social networks [72]. Where details on duration of delivery was specified, most interventions were delivered for 1 to 6 months (n = 17). Seven were delivered between 7 and 12 months.

Nine interventions targeted the intrapersonal domain, directly supporting care-experienced children and young people. Approaches included delivery of Cognitive and Affective Bibliotherapy [123], Cognitively-Based Compassion Training (CBCT) [96], Cognitive Behavioural Therapy (CBT) [74, 97] and mindfulness and yoga practices [89, 105]. These were delivered through a range of online and virtual modalities, including online tutorials and computer games [74, 97]. Where specified, interventions were primarily delivered over the course of 1 to 6 months, with only one intervention being delivered for a longer duration than 6 months [74].

A further fifteen interventions operated across the intrapersonal and interpersonal domains, combining both relationship-based components with skill and competency training for children and young people. For example, Fostering Healthy Futures (FHF) provided group-based curricula and mentoring by a trainee social worker [66]. Group-based activities could include creative or leisure tasks, such as drumming [124] or surfing [168]. For some of these interventions, relationships were fostered through animal-facilitated psychotherapy [158], specifically equine therapy [50]. Five interventions were delivered between 1 and 6 months, seven were delivered between 7 and 12 months and one was delivered between 13 and 24 months.

A further eight interventions primarily included intrapersonal and interpersonal targeting activities, but had a range of organisational- and community-based support to reinforce change mechanisms, support linkage to other interventions and optimise delivery. This included Keeping Foster and Kinship Parents Supported and Trained (KEEP) [53, 62, 71, 79, 81, 106, 140, 141, 149, 161], Multidimensional Treatment Foster Care (MTFC) [51,52,53, 58, 76, 78, 90, 128, 129] and Treatment Foster Care (TFC) [55, 57, 63, 91, 95, 119, 127, 176]. This group of interventions were delivered up to 6 months (n = 2), 7–12 months (n = 1), 13–24 months (n = 2), or delivery duration was not specified (n = 3).

There were a limited number of structural-level interventions: one had a focus on organisational culture and ethos [73, 151]; four considered the availability of community mental health and wellbeing provision [80, 115, 131, 147]; and one policy-level approach targeted the re-prioritisation and funding of placement types [150, 153, 157, 159]. Generally, the delivery duration of these interventions were not specified, although one was delivered for 12 months [131] and two for 3 years [73, 115, 151].

Programme theories

A subset of 13 interventions, with 24 study reports, included a clearly articulated programme theory (Table 2). These mapped onto three dimensions of programme theory: theories of change that explain the causal mechanisms through which an intervention is intended to bring about change; theories of implementation, which prescribe how an intervention will operationalise proposed change mechanisms; and context theories, which consider how system features interact with and are modified by the change mechanisms [177, 178].

Table 2 Overview of intervention programme theory (N = 13 interventions)
Full size table

Theories of change targeted different socio-ecological domains. Three interventions focused on intrapersonal theories [50, 64,65,66, 69, 70, 72]. Key theoretical approaches within this domain linked to Positive Youth Development [179] and resilience, emphasising the need for young people’s adaptive functioning and self-development so that they can enter prosocial relationships [64, 66, 69].

The majority of interventions foregrounded interpersonal theories of change (n = 12), which mapped onto three sets of causal mechanisms. First was to build a therapeutic environment that could be supportive of positive development and prosocial relationships [50, 56, 63]. Second was to develop parent and carers skills, knowledge and confidence, primarily through parenting curricula [51,52,53,54,55,56,57,58, 60,61,62, 67, 68]. Theories included Bowlby’s attachment theory [180], Social Learning Theory [181], Positive Youth Development [179] and resilience, which together emphasise the significance of positive attachments that provide opportunities for learning prosocial behaviours observationally through modelling and replication. Some interventions also re-orientated parenting practices according to coercion and operant conditioning, which encourage effective management of negative behaviour through positive reinforcement and non-harsh disciplinary methods [51, 52, 61]. Third, was to develop mentoring relationships [59, 64,65,66, 69, 101]. These also operated through attachment theory, Social Learning Theory [181] and Positive Youth Development [179].

One intervention included a theory that operated within the organisational domain [73]. The focus was on the transformation of organisational culture within the social care system to ensure its conduciveness with an attachment-based and trauma-informed ethos.

There was more limited inclusion of implementation and context theories. Two interventions operating across the interpersonal, organisation and community domains, focused on optimising delivery in a range of contexts [53, 58, 62, 73]. This included testing a ‘train the trainer’ approach and a structured scale-up model, where the learning from early implementation informed later delivery. We termed these implementation theories as ‘general system change’. One intervention included a context theory, mapping the wider system factors that could inhibit the functioning of an intervention’s parenting curricula. The study report termed this an ‘ecological context model’ [57].

Intervention outcomes

We mapped intervention outcomes according to the a priori outcomes specified by the review (Fig. 3). Outcomes were either theorised (e.g. study reports with theoretical descriptions) or empirically assessed (e.g. study reports with outcome evaluations).

Fig. 3
figure 3

Intervention primary and additional outcome domains

Full size image

Most interventions targeted mental, behavioural and neurodevelopmental disorders. Within this domain, interventions most frequently assessed outcome measurements of total social, emotional and behavioural problems (n = 48); socio-emotional functioning difficulties (n = 17); internalising problem behaviours (n = 22); and externalising problem behaviours (n = 26). There was a paucity of interventions that targeted subjective wellbeing (n = 11). Only four interventions targeted suicide-related outcomes, including suicidal ideation [124], self-harm [96, 151] and suicide attempt [59].

We inductively classified additional outcomes measured by evaluations. These were primarily child-level outcomes: relationships; additional health outcomes; health and social care service use; social care placements; education and employment; and offending and victimisation.

We classified fourteen study reports, linked to eight interventions, that considered potential equity harms in relation to intervention outcomes [54, 65, 69,70,71, 101, 122, 126, 130, 133, 140, 142, 170, 182]. Reported equity harms focused on children and young people’s personal characteristics (age; gender; ethnicity; baseline mental health status) and personal relationships (exposure to maltreatment; placement type; quality of relationship with caregiver; number of caregivers). Parent and carer-related equity harms were linked to personal characteristics (age; ethnicity; baseline mental health status; and drug and alcohol use) and personal relationships (relationship status).

Evidence types

We categorised study reports according to the type of evidence reported (Fig. 4). The evidence type linked to each intervention is further presented in Table 3. Twenty-four study reports described a programme theory [50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73]. Fifty process evaluations provided data on context, implementation and acceptability. Of these, we defined 27 as conceptually and/or empirically ‘thin’, whereby they provided limited description of intervention implementation and acceptability [60, 68, 77,78,79, 81, 82, 84, 85, 87, 89, 91, 94, 96,97,98, 101, 102, 106, 111,112,113,114,115, 117, 118, 183]. Meanwhile, 23 were considered conceptually and/or empirically ‘rich’, presenting detailed data and analysis of contextual characteristics that might structure intervention functioning through their influence on implementation and acceptability [72, 74, 80, 83, 86, 93, 95, 99, 100, 103,104,105, 107,108,109,110, 116, 119, 175, 182, 184,185,186]. This set of rich process evaluations had theoretical generalisability beyond the immediate evaluation context.

Fig. 4
figure 4

Cumulative rate of report and evidence type

Full size image
Table 3 Overview of intervention evidence types
Full size table

There were 86 outcome evaluations. Of these, 52 were randomised controlled trials (RCTs) and 34 were non-randomised evaluations. The majority of RCTs (n = 43) evaluated interventions (n = 31) that primarily targeted the intrapersonal or interpersonal domains [54, 59, 65, 67, 69, 70, 82, 85, 88, 89, 94, 96, 97, 101, 111,112,113, 117, 118, 120,121,122,123,124, 126, 130, 132,133,134,135,136,137,138,139, 142,143,144,145,146, 168,169,170, 172]. Five interventions that operated across the organisational, community and policy domains were evaluated via an RCT (n = 9 study reports) [71, 125, 127,128,129, 131, 140, 141, 182]. Of interventions evaluated through a non-randomised study, 21 interventions, with 22 study reports, targeted the intrapersonal and interpersonal domains [50, 60, 61, 68, 84, 102, 105, 114, 148, 152, 154,155,156, 158, 160, 162,163,164,165,166,167, 171]. Six interventions, with 12 evaluations targeted the organisational, community and policy domains [73, 115, 128, 147, 149,150,151, 153, 157, 159, 161, 182]. There were 14 study reports that provided moderator analysis or interaction effects that were relevant to assessing equity harms [54, 65, 69,70,71, 101, 122, 126, 130, 133, 140, 142, 170, 182].

There was one partial economic evaluation, which estimated the relative costs and consequences of a new intervention compared to the estimated costs of usual care [173].

Discussion

The CHIMES systematic review aimed to synthesise international evidence on interventions targeting the mental health and wellbeing of care-experienced children and young people. The first phase of the review, an evidence map of the available literature, is reported presently.

Mapping interventions by the socio-ecological domain targeted, the main cluster was intrapersonal and interpersonal approaches, often targeting children and young people’s skills and knowledge, or carers’ parenting practices. Some of these also combined organisational and community facing activities to optimise functioning and implementation. As identified in a range of systematic and practitioner reviews of parenting interventions for care-experienced children and young people [187,188,189,190], a couple of interventions were dominant in the map. These were the USA originated Multidimensional Treatment Foster Care (MTFC) [51,52,53, 58, 76, 78, 90, 128, 129] and its derivative Keeping Foster and Kinship Parents Supported and Trained (KEEP) [53, 71, 62, 79, 106, 140, 141, 149, 161], which provide intensive parenting training for foster and kinship carers, embedded in a wider system of support services. Overall, these interventions were under-described and under-theorised, but where specified they often draw on theories related to social modelling and prosocial developmental contexts [179, 181].

In contrast, there was a clear gap in structural-level interventions targeting organisational, community and policy drivers. This is significant given that risk factors for poor mental health in this population include a constellation of family and child welfare system-level factors, which are embedded in a wider context of community-level challenges, such as economic opportunity and socio-economic deprivation [191]. Equally, structural interventions were identified as a priority area for stakeholders who informed the scope and focus of the CHIMES review.

The map identified a wealth of interventions targeting mental health, behavioural and neurodevelopmental disorders, specifically total social, emotional and behavioural problems. Conversely, there was a dearth of interventions targeting subjective wellbeing and suicide-related outcomes, despite care-experienced young people reporting relative adversity in these areas compared to the general population [7, 8]. This reflects wider findings in the research evidence, with a recent review of suicide prevention interventions for children involved in child protection services also identifying a paucity of evidence-based approaches [192]. New interventions might be developed to target these outcomes, or existing approaches adapted if theoretically appropriate. To this end, there is a need to further develop the operationalisation of these constructs and understand the causes that should be targeted to leverage the most change [193]. The extant evidence base, while limited, suggests potential drivers of wellbeing that might be targeted. Primarily operating within the interpersonal domain, these include positive relationships with teachers and family [7, 194], and having available supports, notably material support [194]. Causal mechanisms for suicide-related outcomes are less evident, with current research tending to focus on identifying socio-demographic risk profiles within this population (e.g. age, ethnicity and maltreatment exposure) [6].

The evidence map has implications for future research. Presently the weight of available evidence is focused on outcome evaluation, although only a limited number consider the potential for equity harms [195]. Methodological guidance related to the development, adaptation and evaluation of interventions recommends the integration of outcome data with a clear understanding of the underpinning theory, explication of context, implementation and acceptability through process evaluation, and economic evaluation [39, 40, 196].

As indicated, there remains a lack of description of interventions’ programme theory, with less than a fifth of included interventions reporting a theoretical basis. This is imperative in knowing how interventions interact with system conditions in the generation of outcomes. Given that the evidence base is predominantly located in the USA to date, this means that there is currently a lack of knowledge about the implementation of different approaches in diverse contexts, cultures and countries. Understanding how programme theories function in the USA evaluation context can offer insight into the potential replicability of effects elsewhere. It can then support efforts to adapt interventions to different settings or population subgroups, or to identify where transportation may not be suitable and new approaches need to be developed [39].

Equally, while there continues to be expansion in the conduct of process evaluations, these tend to be conceptually and empirically thin, providing rudimentary summaries of reach and delivery. This is reflected in systematic reviews that currently synthesise process evaluation data, which largely detail barriers and facilitators to implementation [26]. Understanding of wider contextual characteristics, through conceptually and empirically rich process evaluation, is important from a complex systems perspective, which emphasises that intervention’s functioning is dependent on its interaction with both proximal and distal system characteristics [30,31,32,33,34].

There is also a paucity of economic evaluations, which reflects a wider issue identified in children’s social care research [197]. Failing to attend to the cost-effectiveness of interventions is a particular concern given extant issues around escalating costs across social care systems [14].

Beyond implications for evaluation research, there are also some initial suggestions for enhancing systematic reviews in the area of care-experienced populations. It is important that interventions are more comprehensively described in evaluations, preferably with the use of reporting guidance such as the TIDIeR Framework for intervention descriptions [47]. Systematic description of the complex system in which interventions are delivered, using frameworks such as the Context and Implementation of Complex Interventions (CICI) framework [198], will be particularly helpful in supporting future syntheses. In regard to the review process, future reviews of intervention evaluations might aim to map and synthesise all relevant types of evidence [47], particularly in relation to theory, equity and economic outcomes. This will help to identify where gaps continue and where good practice is emerging. Finally, reviews might take advantage of methodological progress in integrating complex systems perspectives into systematic reviews, to help understand the interaction of interventions with system features more fully [34, 199].

Review limitations

There are five central limitations associated with the evidence map. First, the literature around care-experienced populations can be challenging to identify, largely due to extensive international variations in terminology. As such, while the review searches were designed and tested to be sensitive, some study reports may have been missed. Second, the review was limited to studies conducted in higher-income countries, as classified by the Organisation for Economic Co-operation and Development (OECD). As a result, the review has limited generalisability to middle- and lower-income countries, and potentially higher-income countries that do not fall within this classification. Third, there was limited reporting of interventions and associated evaluations, which provided challenges in the cataloguing and mapping of study reports. There were further issues due to the under-specification of interventions’ programme theory. Fourth, study reports were aggregated to chart overarching evidence gaps and clusters. As a result, some of the diversity between interventions and countries is not fully described. Fifth, at the stage of evidence mapping, we did not quality appraise study reports. As such reporting of evidence clusters only reflects the quantity of interventions and evaluations.

Conclusion

The present evidence map describes intervention and evidence clusters and gaps in relation to mental health and wellbeing interventions for care-experienced children and young people. With the predominance of intrapersonal and interpersonal interventions from the USA, future development and adaptation might focus on structural-level theories and components, paying attention to how they function in different contexts. They might also focus on subjective wellbeing and suicide-related outcomes. Intervention research needs to integrate theory, outcome, process and economic evaluation to strengthen the evidence base.