Child Welfare Reform: A Scoping Review

While there have been ongoing calls to reform child welfare so that it better meets children’s and families’ needs, to date there have been no comprehensive summaries of child welfare reform strategies. For this systematic scoping review, we summarized authors’ recommendations for improving child welfare. We conducted a systematic search (2010 to 2021) and included published reviews that addressed authors’ recommendations for improving child welfare for children, youth, and families coming into contact with child welfare in high-income countries. A total of 4758 records was identified by the systematic search, 685 full-text articles were screened for eligibility, and 433 reviews were found to be eligible for this scoping review. Reviews were theoretically divided, with some review authors recommending reform efforts at the macro level (e.g., addressing poverty) and others recommending reform efforts at the practice level (e.g., implementing evidence-based parenting programs). Reform efforts across socioecological levels were summarized in this scoping review. An important next step is to formulate what policy solutions are likely to lead to the greatest improvement in safety and well-being for children and families involved in child welfare.


Introduction
Child maltreatment includes physical, sexual, and emotional abuse, neglect, and children's exposure to intimate partner violence [1,2]. It is a prevalent experience with many potentially serious negative consequences, such as injuries, developmental delay, anxiety and mood disorder symptoms, poor peer relationships, substance use and other risky behaviours [3][4][5][6][7]. Gilbert and colleagues [8] have specified two approaches to the response of child maltreatment: a child safety approach, where government-based agencies are responsible almost entirely for the response to child maltreatment; and a child and family welfare approach, where child welfare organizations respond to allegations of child maltreatment alongside referrals regarding other vulnerable children. The former involves a more investigative response whereas the latter involves more of a preventative and service-oriented response. At the time of publication, Gilbert et al. [8] noted that Canada, the United States, and Australia have taken a child safety approach, whereas New Zealand, the United Kingdom, and several western European countries have taken a child and family welfare approach.
In this article we refer to child welfare as a set of government and private services primarily designed to protect children from child maltreatment, encourage family stability, and, when necessary, arrange foster care and adoptions and child protection services (CPS) as a narrower set of services within child welfare that investigate allegations of child maltreatment [9]. Data from several countries suggests a high cumulative number of children are being investigated for experiences of maltreatment-in New Zealand approximately 25% of children were referred to CPS by age 18 [10] and in the United States 37.4% were referred by age 18 [11]. Certain groups are more likely to come into contact with CPS than others related to a complex array of factors, including colonialism, racism, discrimination, Int. J. Environ. Res. Public Health 2022, 19, 14071 2 of 24 and poverty [12,13]. For example, data from the 2019 First Nations Canadian Incidence Study of Reported Child Abuse and Neglect (CIS) reported on child welfare investigations involving First Nations and non-Indigenous children. This study found that, in Canada, First Nations youth are 3 to 4 times more likely to be the focus of a child protection report and 17 times more likely to experience an out-of-home placement related to a CPS investigation [12].
In general, the evidence regarding the effects of child welfare contact on children is absent or mixed [14,15]. It is challenging to assess the benefits and harms of child welfare interventions, including out-of-home care, as it is unclear if the differences between groups are reflective of the services or differences in a broad range of baseline factors, including socioeconomic status, caregiver educational status, immigration status, family risks, child welfare worker propensity to place children, and children's safety and well-being at the time of placement [16][17][18][19][20][21]. There has been increasing recognition of the high service demands on CPS and child welfare more broadly in high-income countries [22], as well as ongoing calls to reform child welfare so that it better meets children's and families' needs [23][24][25][26]. While much attention has been paid to the need for primary prevention of child maltreatment [27][28][29][30][31][32][33][34][35][36][37], to date there have been no comprehensive summaries of authors' recommendations for improving child welfare (secondary and tertiary prevention efforts involving child welfare).
This scoping review of reviews summarized authors' recommendations for improving child welfare. This review was guided by the following research question: What are authors' recommendations for improving child welfare across socioecological levels, including at the societal-level (e.g., policy), community-level (e.g., coordination of services), institutional level (e.g., child welfare initiatives), relationship level (e.g., ideal strategies to support children and families), and individual level (e.g., training)? A summary of authorproposed child welfare solutions across socioecological levels is important for policy and practice efforts to improve child welfare responses, including informing next steps for child welfare reform.

Materials and Methods
The present review followed principles of a systematic scoping review. According to Daudt et al. [38], a scoping review is a type of research synthesis that aims to "map the literature on a particular topic or research area and provides an opportunity to identify key concepts; gaps in the research; and types and sources of evidence to inform practice, policymaking, and research" (p. 8). While systematic reviews tend to have clear a priori inclusion and exclusion criteria, scoping reviews are guided by broader research questions related to mapping the literature base [39]. Unlike systematic reviews, scoping reviews also do not evaluate the quality of included studies. Most scoping reviews do not aim to synthesize the existing evidence base, but to map or chart the data within the evidence base. Systematic scoping reviews, alternatively, blend characteristics of systematic reviews and scoping reviews and require the following: an a priori protocol, descriptions about the objectives of the review, inclusion and exclusion criteria (at least the population, concept, and context), and methods for conducting the review [39]. Below we detail the proposed methods for the present systematic scoping review. An earlier version of this review was funded in a project to investigate child welfare reform in Canada, which led to the focus on high-income countries.

Inclusion Criteria
Systematic scoping reviews typically specify included populations, as well as the focusing 'concept' and 'context' of the review (see Table 1). In this scoping review we included reviews that addressed authors' recommendations for improving child welfare for children, youth, and families coming into contact with child welfare in high-income countries. Reviews that focused on child maltreatment only (e.g., etiology, risk and protec-tive factors, or health outcomes of child maltreatment) without discussing aspects of child welfare reform were not included.

Category Inclusion Criteria
Population Children, youth (0-25 years of age) and families with prior or current involvement with child welfare.

Context
Peer-reviewed reviews that summarized recommendations to improve child welfare in high-income countries. Reviews could include systematic reviews, meta-analyses, qualitative reviews (e.g., meta-syntheses), rapid reviews, scoping reviews, and narrative reviews.
Timeline 2010 to 2021 (when the search was conducted)

Systematic Search
The systematic search was conducted by an information professional (JRM). Index terms and keywords related to child welfare and out-of-home care (e.g., foster care, out-ofhome care, child protection investigation) and reviews (e.g., scoping review, meta-analyses) were used in the following databases: Medline (OVID), PsycINFO, Sociological Abstracts, and Social Science Citation Index (see Table 2 for an example search strategy). Databases were searched for results from the past 10 years (1 January 2010 to 4 June 2021, when the search was conducted). Titles and abstracts and full-text articles were screened by one author (JRM) for inclusion and a second author (CM) screened all excluded titles and abstracts and full-text articles to ensure relevant articles were not excluded. Differences in screening decisions were resolved by consensus. Table 2. Example of database search strategy. 2 (child * adj3 (welfare or aid)).tw, kw. (5015) 3 (child * protect * adj3 (service? or agenc * or organi?ation?)).tw, kw. (1560) 4 ("foster care" or "foster home?" or "residential care" or "kin care" or "kinship care"). tw. (6280) 5 (out-of-home adj3 (placement? or care)). tw. (894) 6 or/1-5 (12838) 7 exp child/ or exp infant/ (2552666) 8 (child * or girl or girls or boy or boys or infant * or baby or babies or toddler * or preschool * or pre-school * or "young person" or "young people" or teen * or adolescen * or youth *).tw. (2106256) 9 or/7-8 (3282693) 10 6 and 9 (9547) 11 meta-analysis/ or "systematic review"/ or review/ (2915611) 12 (review? or meta-analy * or metaanaly * or metasynthe * or meta-synthe * or (information adj2 synthesis) or (data adj2 synthesis)).tw. (1971484) 13 ((systematic or state-of-the-art or scoping or literature or umbrella) adj3 (review * or bibliographic * or overview * or assessment *)).tw. (512487) 14 "scoping study".tw. (339) 15 or/11-14 (3640612) 16 10 Table 3 indicates the types of reviews addressed by included articles. The majority of reviews were narrative reviews, followed by systematic reviews, scoping reviews, metaanalyses, integrative reviews, rapid reviews, and meta-syntheses. One mapping review was also included. Narrative reviews [42] "generally are comprehensive and cover a wide range of issues within a given topic" (p. 104) and tend to provide an overview of "background knowledge, evolving concepts and controversy" (p. 104) within a field. Systematic reviews [43] generally involve a summary of literature "that uses explicit and reproducible methods to systematically search, critically appraise, and synthesize [literature] on a specific issue" (p. 10). Scoping reviews [38], as discussed above, "map the literature on a particular topic or research area and provide an opportunity to identify key concepts; gaps in the research; and types and sources of evidence to inform practice, policymaking, and research" (p. 8). Meta-analyses [43] involve "the combination of data from several independent primary studies that address the same question to produce a single estimate like  Table 3 indicates the types of reviews addressed by included articles. The majority of reviews were narrative reviews, followed by systematic reviews, scoping reviews, metaanalyses, integrative reviews, rapid reviews, and meta-syntheses. One mapping review was also included. Narrative reviews [42] "generally are comprehensive and cover a wide range of issues within a given topic" (p. 104) and tend to provide an overview of "background knowledge, evolving concepts and controversy" (p. 104) within a field. Systematic reviews [43] generally involve a summary of literature "that uses explicit and reproducible methods to systematically search, critically appraise, and synthesize [literature] on a specific issue" (p. 10). Scoping reviews [38], as discussed above, "map the literature on a particular topic or research area and provide an opportunity to identify key concepts; gaps in the research; and types and sources of evidence to inform practice, policymaking, and research" (p. 8). Meta-analyses [43] involve "the combination of data from several independent primary studies that address the same question to produce a single estimate like the effect of treatment or risk factor" (p. 10). Integrative reviews [44] are a broad type of research review that allow for the "inclusion of experimental and non-experimental research in order to more fully understand a phenomenon of concern" (p. 547). Rapid reviews [45] are "a form of knowledge synthesis in which components of the systematic review process are simplified or omitted to produce information in a timely manner" (p. 2). Meta-syntheses [46] involve "the systematic review and integration of findings from qualitative studies" (p. 1). Mapping reviews [47] are like scoping reviews in that they "map out and categorize existing literature" (p. 94) but their results tend to be more visual compared to scoping reviews. Mapping review 1 1 Some reviews were both systematic reviews and meta-analyses, so a sum of the counts does not equal the total number of included articles (n = 433).

Population Focus
As is shown in Table 4, the majority of reviews focused on out-of-home care, followed by families (families, parents, children) involved with child welfare. Some articles also focused on child welfare organizations, child welfare professionals, and interdisciplinary initiatives (e.g., children's advocacy centres or family drug courts).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological theme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels). Community-relationships among organizations, institutions and informal networks Collaboration models, strategies, and components 40

Thematic Focus
Institution-characteristics and rules for operations Child welfare organizational policies, procedures, and overall environment 38 Child welfare workforce 17 Child welfare organizational performance and evaluation 12 Professional support for child welfare professionals 9 Interventions, services, programs and outcomes associated with different sectors • Child welfare 252 Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  • Personal characteristics 15 1 In the table the shaded headings indicate a level of the socioecological model [40], which was used to organize the themes. Bolded headings represent higher-order themes that were used to organize relevant sub-themes. The bolded headings are not associated with coded articles in the Excel file. 2 Most review articles had multiple themes, so a sum of the counts does not equal the total number of included articles (n = 433).

Societal-Level Thematic Focus
Some reviews addressed international or national factors affecting child welfare. These reviews addressed cross-country analyses of specific aspects of child welfare (e.g., development of family support services compared across countries); international influences, such as international actors influencing child welfare (e.g., international governmental organizations); and the impact of human rights treaties. National aspects of child welfare were also discussed, such as aspects of child welfare structure (e.g., history of child welfare, universal services versus targeted services, types of services offered at national level, funding streams, child welfare standards); national child welfare legislation and policies (e.g., mandatory reporting, COVID policies affecting child welfare, austerity policies); systemic disadvantage in child welfare (e.g., poverty, colonialism, disparity); and national actors influencing child welfare (e.g., courts, policymakers, child advocates).

Community-Level Thematic Focus
Some review articles addressed multi-and inter-disciplinary collaboration models, strategies, and components.

Institutional-Level Thematic Focus
It is notable that the majority of reviews addressed interventions, services, programs, and outcomes associated with child welfare or health. Reviews coded for child welfare interventions, services, programs, and outcomes included the following primarily child services and outcomes related to: placement (e.g., placement stability, breakdown, reunification, preventing re-entry); biological family (placement prevention, family functioning, contact, and programs directed at biological family, such as intensive family preservation or parenting programs); service usage (e.g., rates of service usage); participation (e.g., participation of children and families in child welfare services); transition from care (e.g., programs and services related to transitioning from care); safety (safety, recurrence, occurrence, peer violence in out-of-home care); and foster or kinship care as a service (e.g., comparing foster care as a service versus kinship care). Reviews coded for health interventions, services, programs, and outcomes addressed the following child services and outcomes: mental health (e.g., mental health diagnoses and disorders, eating and food-related problems, well-being, resilience); social health (e.g., attachment, peer relationships, identity development, social support); physical health (e.g., development, immunizations, medication management, physical activity, sexual health); and service usage (e.g., rates of service usage).

Individual-Level Thematic Focus
Some review articles addressed aspects of child welfare professionals and foster or kinship carers. Specifically, they focused on child welfare professionals' knowledge, skills, abilities, and needs; decision-making; personal characteristics; and the impact of their role (e.g., vicarious trauma). Similarly, for foster and kinship carers, articles discussed personal characteristics and knowledge, skills, abilities, and needs. Table 6 summarizes author-proposed improvements for child welfare across socioecological levels. Each level is discussed below. Given the number of articles addressing each theme, only a small sample of citations are listed after each theme below. Institution-institutional characteristics and rules for operations

Cross-disciplinary focus
• Important populations to serve better themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels). themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels). socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels). violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels). Families experiencing low socioeconomic status 20 els [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels). Children with complex needs (such as chronic illness, disability or sensory impairment and needs additional support on a daily basis) 15 Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Table 5 summarizes the thematic focus of included articles across socioecological levels [40]. As discussed above, the socioecological model helps to organize core themes in violence prevention, pointing to research trends and gaps. Each of the four levels of the socioecological model is shaded in Table 5 with examples of the level listed (e.g., "Society" themes tend to focus on international and national laws and policies whereas "Individual" themes tend to focus on knowledge, attitudes and skills of individuals). Within each socioecological level of themes, sometimes groups of sub-themes were found. For example, many articles focused on interventions or programs (e.g., parenting programs) and services (e.g., foster care) from different sectors (child welfare, health, education, research, justice, housing). Counts of themes are listed in Table 5; these counts indicate the number of articles that addressed the theme (e.g., in the Excel coding file, 40 articles are coded as discussing the theme "collaboration models, strategies, and components"). Headings associated with socioecological heme levels (e.g., "Society-laws and policy") and bolded sub-theme levels (e.g., "Child welfare professionals") in Table 5 have no associated counts as these lines are provided to help visually organize the thematic focus of included articles. Each level is discussed in more detail below. While relationships were a key factor mentioned in many articles, they were the primary focus of very few articles (and so are not coded in Table 5). Additional codes related to relationships are found in the next section (improvements across socioecological levels).  Training for healthcare, social service providers (including child welfare professionals), and foster carers

•
Training on unique needs of children/families involved in child welfare or out-of-home care 108 • Training on health topics 18 • Training on navigating child welfare and out-of-home care 12 1 In the table the shaded rows indicate a level of the socioecological model [40], which was used to organize the themes. Bolded headings represent higher-order themes that were used to organize relevant sub-themes. The bolded headings are not associated with coded articles in the Excel file. 2 Most review articles had multiple suggestions for child welfare reform, so a sum of the counts does not equal the total number of included articles (n = 433).

Individual-Level Suggestions for Improvements
At the individual level, the proposal most frequently identified by authors for child welfare improvement involved training of healthcare and social service professionals, including child welfare professionals, foster carers, and kinship carers. Training suggestions most often focused on the awareness and ability to respond to the unique needs of children and families involved in child welfare or out-of-home care [54,83,88,89,93,116,130,134,156,159,168,178,179]; training on health topics [48,56,57,153,178]; and training on how to navigate child welfare and out-of-home care and by extension how to support children and families who are navigating these systems [88,180].

Discussion
This scoping review sought to summarize reviews addressing children, youth, and families coming into contact with child welfare in high-income countries. A particular focus was given to authors' recommendations for improving child welfare across socioecological levels.
Over 20 years ago, Waldfogel [181] identified five main issues with the existing CPS system in the United States: 1.
under-inclusion (such as, inadequate support and protection for vulnerable children, particularly in response to sexual abuse and exploitation); 3. service capacity (for example, severe funding shortages; gaps in services); 4.
service delivery (such as, adversarial relationships with parents; discrimination towards low-income families); and 5. service orientation (for example, residual child welfare, or only providing support as a last resort when all other avenues fail; crisis intervention orientated).
Many of these issues were raised by authors of included reviews as concerns that continue to affect child welfare, including the over-representation of certain populations (e.g., Indigenous families, Black families, and families facing poverty), the under-inclusion of certain populations (e.g., children with disabilities experiencing maltreatment), service capacity concerns (e.g., underfunding and lack of political support for child welfare, difficulty in accessing services in communities), service delivery concerns (e.g., bias in professionals towards racialized, impoverished families), and service orientation concerns (e.g., a tendency towards reactive, additive protocols, priorities, and requirements in child welfare). A fundamental question to ask, then, is: Where to start with improving child welfare for children, families, and for the professionals whose role it is support them?
Below we organize our discussion by reform strategies across socioecological levels. Many of these themes correspond with policy and practice recommendations issued by a Lancet Commission regarding institutionalisation and deinstitutionalisation of children, including the need to improve children's outcomes and key elements of a national reform system [182]. Specific subthemes, such as the need for multi-and inter-disciplinary collaboration, a revaluation of workforce involvement and better evaluation of child outcomes, also correspond with reviews on public health approaches to child welfare [13,22] and a recent policy review on foster care [183].

Societal-Level Reform
A key concern for several review authors was the need for holistic policy reform. Authors [13] noted that there was "no ideal one-size-fits-all reform process" as "a variety of informed strategies to promote child, family and community health, well-being and social care are required across multiple services and programs, each with their own particular organizational and community context" (p. 5).
An important concern for a smaller subset of articles was systemic discriminationincluding racism, colonialism, ableism, and poverty-and how these factors put some families at increased risk of contact with child welfare. This is in comparison to most articles that focused on more 'malleable' factors such as individual factors (e.g., training), institutional factors (e.g., principles of good services), and community factors (e.g., need for increased collaboration) (discussed further below). As such, there tended to be a theoretical division in the included reviews regarding where improvement in child welfare would be most beneficial. Some authors focused on factors at the macro level, which are more difficult to shift (e.g., poverty) but would benefit a significant number of people, thus potentially reducing the need for child welfare involvement (primary prevention). Other authors focused on factors at the practice level; these factors can be more malleable (e.g., parenting skills) in the short-and medium-term to address the maltreatment the child is experiencing or had experienced in the past (secondary and tertiary prevention). Policy and decision-makers conducting child welfare reform must address how to meaningfully engage these two distinct groups who both have commitment and important ideas for improving child welfare.

Community-Level Reform
The most common proposal to improve child welfare identified by authors was the need for improved multi-and inter-disciplinary collaboration and coordination of services. While specific suggestions for improving collaboration and coordination across disciplines were sometimes offered [13,184,185], this theme was often not well described by authors.
An important proposal from several reviews was that the protection of children should involve more than just child welfare workers [13,22] and one review [13] noted that child welfare reform ideally required a "re-imagining of which workforce(s) are in scope, and re-crafting of the workforce knowledge, skills and values to ensure system capacity and practitioner capability to deliver properly targeted prevention strategies" (p. 8). Sectors that are in scope at different points of the child's life include education, healthcare, social services, housing, and justice [13,22].

Institutional-Level Reform
At the institutional level, review authors identified that there were a number of key populations who need to be served better in child welfare; they also consistently referenced a number of fundamental principles to effectively serve these populations.
A key concern for authors was how do so with limited resources and systemic barriers. For example, Font et al. [183] discussed how foster care reform efforts to date have been mostly "additive, layering more protocols, priorities, and requirements to foster care practice over time" (p. 19). One strategy to better serve populations referenced by several included reviews was the need to understand the components of services, interventions, and programming [94,186], and to effectively evaluate them, in order to improve their effectiveness. For example, improving foster care so it addresses children's well-being and safety (vs. process indicators, such as number of placements) requires (a) assessing what outcome measures accurately determine child safety and well-being, (b) discarding measures that are not clearly associated with child well-being and safety, and (c) evaluating programs and services with these measures, in order to determine which components are essential to child outcomes of safety and well-being [183]. It also ideally requires the use of measures that are not easily manipulated by overworked child welfare workers and administrators, such as data from health care, juvenile justice, and education (e.g., student honour role, sickness days, emergency room visits) [183]. This point resonates well with reviews that emphasized the need for integrative data systems [13,22].
A review by Herbert et al. [187] discussing components of child welfare services presented the collective program logic of multidisciplinary teams responding to child physical and sexual abuse. In delineating the assumptions of each program component, the authors [187] strived to move towards 'opening the black box' of program theory so that evidence-bases can be developed "not around programs or models, but around the common components that appear across many programs" (p. 9). These authors [187] argued that such work is necessary to evaluate the "explanation underlying the connection between team and centre activities and their intended outcomes" (p. 10). For example, does a "warm referral" (a facilitated introduction to another service provider) to therapeutic services increase engagement and completion of services offered by multidisciplinary teams, as well as improve child or family outcomes, as is assumed?
Aside from better evaluation of components of services, programs, and interventions, review authors also commented on the need for services to be tailored to and representative of those using the services. This especially involved an investment in culturally appropriate services and effectiveness research that accounts for diversity in racial, ethnic and cultural groups [67,77].

Relationship-Level Reform
Many of the reviews summarized within this scoping review discussed the importance of relational aspects of service provision, including comprehensive assessment, advocacy, referral, and support. More specifically, many authors noted that children involved in child welfare need comprehensive initial and ongoing assessment that includes, but is broader than mental health symptoms [187][188][189], as children may need referral to services and support before serious problems develop, including assessment and referral or support for their social-emotional well-being [148,188]; physical health [190], including developmental needs [99], dental care [191], sexual health [172], substance use or prenatal substance exposure [124]; and food-related behaviours [192]. Assessments of needs were recommended for various service entry and exit times, including entry into care, at regular intervals while in care [97], as well as exits from care [62] and transitions to adult mental health services [96]. Assessments of safety of children involved in child welfare were also of concern to review authors [193][194][195]. With respect to advocacy, review authors argued that healthcare and social service providers, including child welfare workers, should advocate for services, supports, and opportunities to increase the well-being and participation of children and families throughout and beyond their involvement with child welfare [96,134,[196][197][198]. As was noted above, support was broadly conceived by review authors to include a variety of activities to improve the lives and well-being of children and families involved in child welfare, including psychoeducation, enhancing support networks, and help to navigate the child welfare system, among others. Often, review authors simply noted that health and social services providers should "support" children and families.
In spite of significant theoretical differences informing child welfare research and summaries in this review, review authors [199,200] noted that one area of overlap is the importance of forming relationships during service provision. In social services there is an increasing focus on the need for services to be relationship-based [201]; however, in social services practice, including child welfare practice, there continues to be an emphasis on administrative processes, such as documentation, rather than relational processes [202]. Authors [203] have summarized a number of occupational risk factors faced by child welfare workers, such as "a personal history of maltreatment, inadequate support, high workloads, low salaries, long working hours, exposure to clients' trauma . . . violence and aggression from clients . . . political interference . . . and schedules which negatively impacted services rendered to clients" (p. 7). These risk factors and other factors across socioecological levels (e.g., demographic characteristics of child welfare workers; organizational support for workers, including peer support and supportive supervision) [204][205][206][207] lead to high staff turnover, which negatively impacts relationships with children and families. One of the key policy strategies raised by authors to reduce the occupational risk factors of child welfare workers was increased governmental funding for more workers, in order to reduce caseloads.

Individual-Level Reform
Blome et al. [208] argued that focusing on funding to reduce caseloads is important but noted that caseworkers still need to be skilled: "focusing on size may obfuscate other larger issues within the agency like the level of education and experience required for frontline positions" [203]. This scoping review made clear the-arguably impossible-variety of administrative and clinical skills that child welfare workers were expected to have. Authors from different disciplines (education, healthcare, child welfare, criminal justice) all concluded by suggesting how better training of child welfare workers was neededvery few authors concluded that their own discipline needed to better support children experiencing maltreatment. While many reviews recommended better training for child welfare professionals, very few reviews evaluated training strategies for child welfare professionals [142,209,210]. The unrealistic expectations towards child welfare workers by society underscores the need for a "comprehensive workforce needs analysis in collaboration with all relevant stakeholders (within universal, secondary and tertiary services), followed by development of a plan to address current and ongoing system and service user needs" [13]. As has been stated by several review authors (and stated several times in the present review), the important needs of children experiencing maltreatment cannot be served by child welfare alone-an investment of the skills and resources from many sectors is required.

Limitations, Future Research, and Implications
As this scoping review summarized published reviews in between 2010 and 2021, it is limited by the content of available reviews and excludes all articles that have not yet been included in published reviews. In our attempt to succinctly summarize the wide variety of author-proposed suggestions for child welfare reform, we have not critically appraised individual reviews or their empirical research. For example, a number of reviews spoke to the over-inclusion of families experiencing poverty in child welfare due to factors like implicit bias, but we have not critically examined individual studies that have sought to address if low-income families are referred to child welfare because of implicit bias. A recent commentary by Barth and colleagues [211] is an example of an accessible summary of individual studies seeking to address pressing child welfare reform queries. Given the sheer volume of reviews summarized, we have also not analyzed or assessed individual programs, services, or policy recommendations. For example, several reviews summarized specific child welfare services, including evidence-based programs to support children and families. We have not critically appraised these reviews or their included studies and as such were not able to make recommendations about the value of specific programs and services.
The 433 reviews summarized in this scoping review offer a variety of informed strategies to promote child, family and community well-being that have important and wideranging practice implications. Thematic analysis suggests that many of the issues plaguing child welfare have not changed in the past few decades. This suggests that while much evidence-based research has accumulated over the past decade in terms of services and supports for children and families involved in child welfare, as in other areas of violence prevention, the leadership, governance, and political will necessary to tackle child welfare reform is underdeveloped [212]. As such, future research and practice in the area of child welfare reform needs to move beyond identifying what should be done and examine how to undertake reform efforts with accompanying evaluation. For example, many reviews over the past decade have suggested that better collaboration between child welfare and other sectors is needed (the "what"), but it is less clear how this collaboration should be undertaken. In line with how collaboration might be undertaken, Luckock et al. [199] have summarized innovative models of practice at the interface between the United Kingdom's National Health Service and child and family social work. One relevant example of multisector collaboration cited by the authors [199] included "approaches to multi-agency team working in children's services [that] have emerged in which 'co-location' of health, social and other practitioners (especially police) is the preferred means of service integration at those points where 'early help' needs better alignment with a 'child protection' response" (p. 65). While the authors [199] noted that "developments of this kind are becoming widely established across England" (p. 65), their review found only two evaluations of these efforts, neither of which accounted for child and family well-being outcomes. As such, like Luckock et al. [199], future work should detail how child welfare reform can be meaningfully executed and these efforts should evaluate if and how such innovations improve the lives of children and families.
Future research is needed to assess the readiness of child welfare for any specific avenue of reform, including the readiness to evaluate its effectiveness at the level of child and family well-being. Such assessment, especially for widespread reform, can be challenging. For example, given that in Canada it is the responsibility of individual provinces and territories to implement child welfare services, it is politically and practically difficult to implement and assess the effectiveness of any national child welfare reform efforts. Important avenues of child welfare reform in Canada are seen with Indigenous efforts for child welfare sovereignty; however, these efforts are mired by funding and infrastructure (e.g., service delivery) issues [213,214]. Any particular avenue of reform, including the example of multi-sector collaboration discussed above, must therefore be tailored to the specific jurisdiction of interest and any particular contextual challenges.

Conclusions
This scoping review sought to summarize reviews addressing children, youth, and families coming into contact with child welfare in high-income countries by focusing on authors' recommendations for improving child welfare across socioecological levels. Several key strategies were discussed in this review, such as holistic policy reform that moves away from layered, additive, reactive protocols, priorities, and requirements; addressing underlying disparities and social determinants of health problems; meaningful engagement with children and youth involved in child welfare at all levels of decision-making; multiand inter-disciplinary collaboration, including a comprehensive workforce analysis to re-imagine what workforces are in scope; component analysis of all levels of child welfare services and evaluation of services with attention to child well-being and safety outcomes; and increased access to services in communities, including services that follow key principles (e.g., trauma-informed, comprehensive, tailored, relationship-based), accountability and better preventive services for populations that are overrepresented in child welfare (e.g., racialized groups and families experiencing poverty). As several authors noted, there is no one way to conduct a child welfare reform process. An important next step is to formulate what policy solutions are likely to lead to the greatest improvement in safety and well-being for children experiencing maltreatment and to rigorously evaluate these solutions at the level of child and family well-being.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The Excel file with all of the coding is available on Dryad Data Repository (https://datadryad.org/stash, doi:10.5061/dryad.jsxksn0dc). All other material and data used for this review are available in the paper and tables.