Interventions in child welfare: A Swedish inventory

In Swedish child welfare, there are no mandatory guidelines on what interventions to use. Local authorities are able to set their own criteria for implementing or designing interventions. We carried out a survey to identify interventions in use in Children's Social Services and Child and Adolescent Psychiatric Care in Sweden. A total of 102 interventions were stated to have been in use, with between 31 and 45 different interventions for each of the four different child welfare populations. Of the 102 interventions, 56 were designed outside Sweden and later imported. Only 27 interventions were supported with some kind of research evidence. About half of the interventions targeted the child. Possible implications for practice and research are discussed.

According to the Cambridge Dictionary (https://dictionary. cambridge.org),an intervention is the action of becoming intentionally involved in a situation in order to improve it or prevent it from getting worse.The likelihood of a desirable outcome from an intervention increases if there is scientific support to show that that specific intervention has been proven to be effective for a specific population, on specified outcomes, in comparison with an alternative (Shadish et al., 2002).The development of a new intervention can be a systematic process that involves several steps ending in a defined intervention with content that is specified (Fraser et al., 2009;Fraser & Galinsky, 2010).These authors are addressing scientists about the research-based development of interventions, not practitioners, about practice-based development.When the intervention has been tested in two or more effectiveness trials, can it be considered an evidencebased intervention (Flay et al., 2005;Gottfredson et al., 2015).The interventions developed locally in CWS are usually not based on systematic development, but designed by practitioner, in existing practice settings to solve practice-related problems with marginal supervision of outcomes and without burdening the financial budget expenditure (Sørensen & Sjoe, 2021;Wollter, 2020).
The Swedish welfare model is characterized by decentralization and powerful local authorities (Pollitt & Bouckaert, 2011).Sweden consists of 290 municipalities each with a CWS organization managed by a local municipal welfare board of politically appointed members who are mandated to ensure that children within their area receive support and protection (Liljegren et al., 2014).In most cases, social workers within the CWS organization offer, deliver or allocate interventions to both children and parents based in their assessment of need (National Board of Health and Welfare, 2020a).A case can come to the attention of CWS via mandatory reporting (schools, health service or police) or if parents have voluntarily applied for support.The Swedish child welfare legislation is an integrated part of the Social Services Act that includes a wide clientele (Ponnert, 2012), for example, those experiencing financial hardship, homelessness or substance abuse.
CAPS is part of the medical health care organization within each of the 21 regional councils in Sweden.A case can come to the attention of CAPS when parents have sought support under the Swedish Health and Medical Services Act.Within CAPS, several different professionals (e.g.psychologist, social workers and psychiatrist) can be involved in an assessment undertaken prior to referral for intervention.The responsibilities of CWS and CAPS overlap in complex ways rather than being clearly separated (National Board of Health and Welfare, 2014).If professionals working in CAPS suspect that a child under their care has a need that is due to social problems (e.g.neglect or abuse), CWS must be informed (Legislative Bill, Prop. 2002, 2012).
If CWS consider that the interventions provided by CAPS are sufficient, they are not obliged to intervene further.CWS and CAPS in Sweden are two organizations founded on different legal frameworks, but the child populations within these organizations have similar problems, and CWS and CAPS both leverage psychosocial interventions to meet the needs of children and families (e.g.Gustle et al., 2007;National Board of Health and Welfare, 2019).
Differences within the two systems stems from not only their legal mandate but also the category of professional working within the systems.Within the CWS, social workers with bachelor degrees in social work (Brante et al., 2019) is the primary professional category found while within CAPS psychiatrists are the primary professional category found.
Sweden has non-mandatory guidelines when it comes to what interventions to use in child welfare (Hessle & Vinnerljung, 1999).The study was based on four surveys administered to CWS and CAPS between 2016 and 2019 (Table 1).Only one survey, non-placement service, child abuse and neglect, included all 290 municipalities and all 35 child psychiatric units in Sweden.All the others had randomly selected samples.The randomized used a web-based research randomizer (www.randomizer.org).The sample size in each survey was limited by resources allocated for each survey.The foster care survey included 37% (n = 106) of all municipalities in the randomized selected sample, and all (n = 38) identified privately administrated foster care providers.The institutional care/group home for children with externalizing behaviours initially included a randomly selected sample of 160 institutional care/group homes.These were randomized from the list of 748 institutional care/group homes targeting children with externalizing behaviours provided by the IVO.Among these 160 institutional care/group homes, 28 had closed down its operations and 65 were at the time targeting others children regardless of the list.Therefore, we only account 67 randomly selected private and public provides in Table 1.The non-placement service for children with externalizing behaviours included 35% (n = 101) of all municipalities in the randomized selected sample and all 35 child psychiatric units.
The respondents were care managers of CWS and CAPS organizations.Among the private providers and the public institutional care/group homes, the superintendent of care was the respondent (see also Table 1).Each respondent was asked to state all interventions in use for the target group of children during the preceding 12 months.To simplify the process of answering, three of the surveys listed known interventions as identified in the social services' Method Guide database (Swedish National Health and Welfare Board; www.socialstyrelsen.se/utveckla-verksamhet/evidensbaseradpraktik/metodguiden/)and then provided space for respondents to add other interventions.In one survey (foster care), respondents were asked to list the interventions in use.Two reminders were sent to respondents, and when possible, a telephone reminder was conducted which ended in an interview with 12%-18% of the respondents in the different surveys.During the telephone interviews, it was stressed that we only sought information on interventions in use, not the individuals receiving the interventions.The time given to respond to the questionnaires was about 2 months.The average response rate over all surveys was 64%, with the highest rate of private care providers and child psychiatric units lowest.All our response rates were higher or close to average using written or web-based surveys (see Anseel et al., 2010).
Ethical approval was not required nor sought for this study as no individual was identified and no sensitive personal data (e.g.health status, political or religious opinion) was recorded in the surveys.

| SEARCH STRATEGY AND CODING OF INTERVENTIONS
All interventions that were listed in English or Swedish in the surveys were regarded as an intervention, irrespective of how general or detailed the description was.Initially, the listed names were used as search words in Google's search engine as a means of conducting an exploratory search for more information regarding the identified interventions (supplementary table).Some interventions may be known by different names, especially if they have been in use for many years.To address this possibility, the listed name was searched with various terms (e.g.listed names in Swedish were searched in English and vice versa).If a Swedish listed name clearly was a translation from an English name, it was merged into one intervention (e.g.Parent Management Training was only counted once if it was listed by two respondents, one with the name in English and one with the name in Swedish).If two listed interventions were found to be related, this was noted in the supporting information Table S1 (see, e.g., No. 15 and Sticky Taping,No. 85).
Each identified intervention was categorized according to origin, intended population and scientific support (Table 2 and supporting information Table S1).One challenge was to code interventions with a generic name.For example, Family Therapy (Table S1 All coding was done independently by two (MB and KS or MB and TÅ) with consensus agreement.
The origin of the intervention was primarily retrieved via webbased descriptions: • Swedish: The intervention was originally designed in Sweden or the intervention was modified in Sweden and given a new name.
• Imported: The intervention was designed outside of Sweden.
• Unknown: The intervention had a generic name which did not indicate a distinctive origin of the specific intervention.
The target group was derived from the description of the interventions: • Child: The child was involved in the intervention alone or in a group with other children with similar issues.
• Caregiver: The child's parents or foster parents were involved in the intervention alone or in a group.
• Combination: The child and their caregiver were involved in the intervention.
We did not systematically review and synthesize the research of The scientific support for each individual intervention was coded using the CEBC Rating Scale taxonomy (www.cebc4cw.org).We have only included evidence that directly or indirectly promoted the children's well-being or development (i.e. an intervention targeting the well-being of parents only was not included).The scale consists of six categories: • Well-supported by research evidence (1).At least two rigorous randomized controlled trials (RCTs) with non-overlapping analytic samples that were carried out in the usual care or practice settings have found the programme to be superior to an appropriate comparison programme on outcomes specified in the criteria for that particular topic area.In at least one of these RCTs, the programme has been shown to have a sustained effect of 1 year or more beyond the end of treatment.
• Supported by research evidence (2).At least one rigorous RCT in a usual care or practice setting has found the programme to be superior to appropriate comparison programme outcomes specified in the criteria for that particular topic area.In that RCT, the programme has been shown to have a sustained effect of at least 6 months beyond the end of treatment.
• Promising research evidence (3).At least one study utilizing some form of control (e.g.untreated group, placebo group, matched waiting list) has done one of the following: (1) established the programme's benefit over the control on the outcomes specified in the criteria for that particular topic area, (2) found it to be comparable on outcomes specified in the criteria for the topic area to a programme rated 3 or higher on this rating scale in the same topic area or (3) found it to be superior on outcomes specified for that particular topic area to an appropriate comparison programme.
• Evidence fails to demonstrate effect (4).Two or more RCTs with non-overlapping analytic samples that were carried out in usual care or practice settings have found that the programme has not resulted in improved outcomes specified in the criteria for that particular topic area, when compared to usual care.The overall weight of evidence does not support the benefit of the programme on the outcomes specified in the criteria for that particular topic area.
• Concerning practice (5).One or more of the following statements is true: (1) if multiple outcome studies have been conducted, the overall weight of evidence suggests that the programme has a negative effect on the target population being served or on outcomes specified in the criteria for that particular topic area; (2) there are case data suggesting a risk of harm that (a) was probably caused by the programme and (b) was severe and/or frequent; or (3) there is a legal or empirical basis suggesting that, compared with its likely benefits, the programme constitutes a risk of harm to those receiving it.
• Not able to be rated (NR).The programme does not have any published, peer-reviewed study utilizing some form of control (e.g.untreated group, placebo group and matched waiting that has established the programme's benefit over the control on outcomes specified in the criteria for that particular topic area or found it to be comparable with or better than an appropriate comparison programme on outcomes specified in the criteria for the topic area. We considered if the outcome directly affected the children's well-being or development.For example, the 12-step programme is described as essential component of the Alcoholics Anonymous by the CEBC and coded with Promising Research Evidence but is identified as targeting adult population and the coding is based in outcomes for adults.The 12-step programme was therefore coded NR (see Table S1, No. 95).
All searches and coding were conducted in April 2021.

| RESULTS
A total of 102 interventions were reported to be in use during the past year for one or more child welfare populations (supporting information Table S1).promising research evidence (Table 2).The remaining 73 were not able to be rated.None of the interventions were rated 'Concerning practice' or 'Evidence fails to demonstrate effect'.
About half of the interventions, 49, targeted the child, 21 the caregiver, and the remaining 32 the child and the caregiver.The scientific support was approximately the same in these three groups: Between 26% and 31% were rated as having promising research evidence or higher.
Of the 102 interventions, 68 were used in one studied population, 23 in two, eight in three and three in all four (Table 1 and supporting information Table S1).There was no obvious difference in scientific support depending on the number of populations with which the interventions were used.(Fraser et al., 2009;Fraser & Galinsky, 2010) and resources (Sørensen & Sjoe, 2021) The results show that a large majority (n = 81) targeted the child alone or together with the caregiver.The corresponding number targeting the caregiver is also high (n = 53).Independent of research evidence, there seems to be a large sample of interventions targeting both the child and the caregiver for professionals to choose among.

| DISCUSSION
The Swedish legislation in this area is mostly focused on individuals, which may partly explain why the individual child is prominent as an intervention target.

| CONCLUSIONS AND IMPLICATIONS
The results suggest five tentative conclusions and implications.However, it can also be negative, since some research indicates that home-grown interventions may be superior to imported ones (Löfholm et al., 2013;Sundell et al., 2016)   would enable best possible practice to be secured (Pålsson, 2020).

| METHODOLOGICAL CONSIDERATIONS
Given the methodological considerations of this study, all results need to be replicated in Sweden and elsewhere.One consideration is that the scientific evidence for the interventions is based on reviews primarily by the CEBC and to some degree complemented by SBU.
Those reviews are of varying age, some older and some more recent.
1 | INTRODUCTION In Sweden, a child or a child's family in need of psychosocial interventions can contact either Child Welfare Services (CWS) or Child and Adolescent Psychiatric Service (CAPS).Annually, about 35 000 children and their caregivers receive at least one non-placement psychosocial intervention and another 30 000 children are in foster care or group care provided by CWS (National Board of Health and Welfare, 2020b).In CAPS, about 130 000 children are receiving outpatient care at any given point during the year and another 3000 children receive in-patient care and are provided with interventions (Swedish Association of Local Authorities and Regions, 2020).These categories are not mutually exclusive.A child in foster care provided by the CWS can also be receiving non-placement psychosocial intervention by the CWS or out-patient care by the CAPS.The aim of this study is to investigate the interventions used by CWS and CAPS.
Local child welfare authorities (CWS and CAPS) have the freedom to choose which interventions to provide to clients.There is a governmental supervisory inspectorate, the Health and Social Care Inspectorate (IVO), which has a monitoring and oversight function to ensure that the provisions of the Social Services Acts and Health and Medical Services Acts are being met by local authorities (Pålsson, 2020).The focus of the inspectorate is on ensuring a basic level of care rather than auditing the content of care.There are seemingly a wide variety of interventions in the Swedish child welfare system.A survey by the National Board of Health and Welfare (2008) reports 129 different interventions delivered by the whole Swedish childcare (including preschools, child and adolescent habilitations and family centers).Most interventions were provided by youth clinics (n = 35), CWS (n = 57) and CAPS (n = 44).Further, the National Board of Health and Welfare (2009) described 90 interventions delivered by the municipal CWS to the general population of children.The aim of this study is to investigate the interventions in use by CWS and CAPS in a none-mandatory context.The study questions were as follows: 1.What interventions are used by these two authorities?2. What is the origin of the interventions? 3. What is the scientific support for the interventions? 4. Do the interventions target the child, the caregiver or both?
each identified intervention but instead focused on existing lists of the research support for child welfare programmes reviews of the effectiveness of the interventions.There are several organizations that synthesize research.Unfortunately, their results are not always agreeing.Some reasons for this are that they focus on somewhat different research questions, include different data bases and types of publications (e.g.include or exclude grey literature), the extent of publication years and rigour and transparency(Gough et al., 2020).We chose the California Evidence-Based Clearinghouse (CEBC; see also www.cebc4cw.org)list of synthesize research because of its focus on children and families served within the child welfare system, as well as its easy to access transparency reviewing research evidence.However, since CEBC includes research in the English language only, Swedish research was searched for in the database of the Swedish Agency for Health Technology Assessment and Assessment of Social Services (SBU; see www.sbu.se).This included a search of their database of evidence gaps.In addition to Swedish interventions, the SBU search also included English interventions that were not included in the CEBC.None of these searches (the CEBC, the SBU in Swedish, or the SBU in English) resulted in additional coding.When we did not find an intervention listed in CEBC or SBU, the Swedish National Board of Health and Welfare's (NBHW) database, the Method Guide for social services (sv.Metodguiden för social arbete; www.socialstyrelsen.se/utveckla-verksamhet/evidensbaserad-praktik/metodguiden/), was consulted.This database includes information on interventions in use in Sweden.This information was primarily used for the description of the intervention and coding of origin and target group.Thus, the search for the research evidence supporting the interventions identified by respondents of our survey was conducted in three steps.First, the CEBC was searched for each of the identified interventions.Second, if the intervention was not included in CEBC, SBU was searched.Third, if the intervention was not found in either the CEBC or SBU databases, NBHW's database was searched.No additional information on research evidence was retrieved from NBHW that was not already found in either CEBC or SBU.
The largest number of interventions were reported for children in foster care (n = 45), followed by non-placement service interventions for child abuse and neglect (n = 39), group care interventions for juveniles exhibiting criminal behaviours (n = 35) and nonplacement service interventions for juveniles exhibiting criminal behaviours (n = 31).Of the 102 interventions, 10 were assessment tools (assessment for selecting foster parents, attachment style interview, Kälvesten, On the Way, Protect, Bear cards, Emma dolls, Genogram, Signs of Safety and Sticky Taping).Some of these assessment tools are considered to have a therapeutic component.More than half of the interventions (n = 56) had a known origin outside of Sweden: 35 originated from the United States (e.g.Multisystemic Therapy, Treatment Foster Care Oregon, and Incredible Years), six from Australia (e.g.Team Parenting and Signs of Safety), six from the United Kingdom (e.g.Low Arousal Approach and Getting It Right For Every Child), five from Norway (e.g.Alternative to Violence and Child-Oriented Family Therapy), two from Canada (e.g.Connect) and two from the Netherlands (e.g.Marte Meo).Of the 56 imported interventions, 13 were well-supported or supported by research and 14 were rated as promising.The remaining 29 were not identified as being supported by research evidence in the CEBC, SBU or NBHW databases.A total of 22 interventions originated from Sweden.One was rated as having promising research evidence and one as supported by research evidence.The remaining 20 interventions were not able to be rated because they were not listed in the CEBC, SBU or NBHW databases.The last 22 interventions were of unknown origin.Several of these have a long history in social work (e.g.Counselling, Milieu Therapy and Network Therapy).None of these interventions were rated.Of all the 102 interventions, nine were well-supported by research, five were supported by research evidence, and 15 had . Importing interventions is arguably more expensive from a municipal budget perspective.Developing an intervention may take municipal resources (staff time) but importing an intervention takes cash.This makes it appear advantageous to import and implement established interventions.The 24 interventions of unknown origin all have a long history in social work and might have been developed in Sweden or elsewhere.The generic name of these interventions may be due to less systemization and greater flexibility.One can also wonder if an intervention is in use with a specific or proper name what would motivate people to not use that name instead of a generic.The third question dealt with the research evidence for the interventions.Of the 102 interventions, 27 were rated as having some scientific support.A cornerstone of evidence-based practice is the incorporation of scientifically supported systematic interventions into everyday practice(Thyer & Myers, 2011), often termed evidencebased intervention.Even if evidence-based practice has been promoted for several years in Sweden(Svanevie, 2011), according to our study, so far, it has not largely influenced the choice of interventions.Svanevie (2011) have thoroughly shown how the concept of evidence has been critically discussed and debated since it was introduced, which may have an influence on whether CWS and CAPS take the concept into account when implementing interventions.The fourth question deals with the focus of the interventions.
First, it seems plausible that the large number of interventions targeting the child welfare populations is a result of local authorities' adjusting to meet the needs of individuals within their geographical area of responsibility.Local authority areas vary in nature (e.g.rural or metropolitan) and to some degree in type and number of the child welfare population.However, why should similar problems require different interventions just because an individual lives in rural versus a metropolitan area, or density of immigrants?An implication for both research and society is to find out how much interventions need to be adjusted on the basis of context.Second, the large number of imported interventions shows that social work in Sweden is highly influenced by social work practice outside of Sweden.This can be positive, since it indicates that professional experience can be exchanged across different contexts.

Fifth
, there seems to be some confusion among social workers as to the definition of an intervention.Of the 102 interventions, 10 were assessment tools, five of them targeting the child and five the caregiver.If there are no commonly used criteria for describing interventions, almost everything can be regarded as a treatment.It seems important to establish the criteria for, for example, a systematic intervention, which could be a procedure marked by method, thoroughness, regularity, and transparent as to content.A societal implication would be to clearly declare a prominent use of systematic intervention and define variations in the concept of intervention.Focusing on content of interventions New research may have revised the rating of scientific evidence.It is also possible that other clearinghouses (e.g.Cochrane and Campbell collaboration) might have come to other conclusions on the evidence of specific interventions.A second consideration is that we have not validated the surveys that were used.It is possible that, for reasons of social desirability, respondents exaggerated the number of interventions in use or failed to produce a 'branded' name and instead added a vague descriptor (e.g.family therapy).Building on the second consideration, a third consideration is that we have categorized several well-known interventions (e.g.individual therapy) as having an unknown origin, because the respondent's vague descriptor, "Individual therapy," could be an umbrella concept for other 'branded' interventions (e.g."Dialectical behavioural therapy," "Interpersonal

Table S1 ,
No. 97)or Intensive home-based family treatment(Table S1, No. 46).If a respondent in the survey named an intervention with a generic name, it was categorized as unknown origin and lacking evidence.This approach might deflate the rate of interventions found to have scientific supporting but was considered the best alternative as it avoided speculation.Overview of the administered surveys Privately administered care providers as identified by Googles search engine or as known to the present authors.Ref.: Swedish Agency for Health Technology Assessment and Assessment of Social Services,(2017, 2018a, 2018b, 2020).
a a California Evidence-Based Clearinghouse Scientific Rating Scale.Abbreviation: CEBC, California Evidence-Based Clearinghouse.
To our knowledge, this is one of the first studies to investigate the variability of interventions provided by CWS and CAPS.Whether the findings are unique to Sweden or present a general phenomenon which may be similar in other countries needs to be investigated.
such as training and a licence fee.Importing interventions that lack research evidence does involve ethical considerations, given that research has found that interventions may be inefficient or even harmful (e.g.Welsh & Rocque, 2014).For this reason, evidencebased interventions should be preferred regardless of their origin.Furthermore, locally designed interventions should always be evaluated for efficacy, to avoid causing harm (cf.Primum non nocere).Fourth, given the heterogeneity of interventions in use for the same population, our results do not imply the existence of a professional consensus among Swedish social workers on desirable interventions.A fairly large proportion of the interventions were only mentioned by one or two municipalities.For instance, out of 80 randomly selected municipalities, 24 mentioned a total of 21 different interventions targeting a child placed in foster care (Swedish Agency for Health Technology Assessment and Assessment of Social Services, 2017).A similar finding emerged on interventions in group care: 53 institutions mentioned 33 different interventions, 15 of which were mentioned by only one or two institutions (Swedish Agency for Health Technology Assessment and Assessment of Social Services, 2018a).
Third, only 27 of the 102 interventions have research evidence found by CEBC, SBU and NBHW, indicating that evidence may not be an important criterion for the choice of child welfare interventions.One might perhaps expect that scientific support would be a salient criterion for imported interventions, which frequently come with distinct costs