A Randomized Controlled Study of Safer Kids: A Manualized Intervention to Prevent Child Abuse

BACKGROUND:

International studies have shown prevalence rates of child abuse globally, with around 20 percent of all children being exposed to violent or abusive parenting (Stoltenborgh, Bakermans Kranenburg, Ijzendoorn, \& Alink, 2013). A Swedish study showed that 15 percent of children living in Sweden sometime had experienced physical violence from a parent or close relative, from which half had been exposed several times (Annerbäck, Wingren, Svedin, \& Gustafsson, 2010). Apart from the suffering caused by the abuse in itself, does the experience of child abuse act as a risk-factor for future problems. It has for example been shown that children who have experienced child abuse run a greater risk for developing depression, anxiety disorders, eating disorders, conduct disorders and substance abuse problems (Norman et al., 2012).

Despite the apparent need for effective interventions to prevent child abuse, few well-designed studies of intervention effects have been published. In a recent meta-analysis including all randomized trials of all types of interventions to prevent or treat child abuse, only 23 studies were included (Euser, Alink, Stoltenborgh, Bakermans-Kranenburg, \& van IJzendoorn, 2015). The results of that analysis was also a disappointment, since the overall effect in terms of reducing or preventing child abuse was d=0.13, an effect that turned out to be non- significant after taking publication bias into account. However, a moderation analysis showed that interventions based on parent training (N=10) showed significantly larger effects than other types of interventions. In support of that result, another meta-analysis including studies of parent training interventions only, did show a significant overall effect of 11 percent absolute reduction in risk of reabuse (Vlahovicova, Melendez-Torres, Leijten, Knerr, \& Gardner, 2017). Thus, even if parent training interventions show some promise, the small to non-existent intervention effects found in the meta-analyses also point to the need of developing more effective interventions.

The application of manualized interventions is rare within the social services in Sweden, but there have been a few projects in recent years with the purpose of implementing interventions that previously have been evaluated in foreign studies. One example is the iRisk-project (Broberg et al., 2015). The main purpose of that project was to investigate the ability of regular service providers to implement the interventions fith fidelity and sustainability. The effects of the interventions were also measured, but only with pre-post design without a matched control group. Both parents and the practitioners appreciated the interventions. There were also positive effects on outcome measures between pre and post the interventions, but since no matched control group was included, no safe conclusion of intervention effects could be drawn. However, several challenges were also identified in terms of implementation and sustainability. First, it was found that it was harder than expected to recruit and engage caregivers in the interventions. Secondly, several of the included service units had difficulties getting started with interventions and many reported that contacts were terminated prematurely. A third conclusion was that several practitioners described difficulties in sustaining the work with the more time consuming interventions at their service unit. A final conclusion was that practitioners found it difficult to maintain continuity in the longer interventions, due to logistic challenges for the practitioner and the families.

In summary, most previously evaluated interventions demand considerable time and resources, and there have consequently been reports of challenges in terms of implementation. In view of the prevalence of child abuse, it is therefore apparent that many families that are investigated within the services never will get access to such interventions. Furthermore, a condition in several of the existing interventions is that the suspected caregiver acknowledges the abusive act that has been committed, either in the investigation that precedes the intervention, or as a part of the intervention itself. This may pose challenges in terms of stigma and motivation to engage in the intervention. In light of these challenges, it is important to develop interventions that a) readily can be implemented with fidelity and sustainability in regular social service units, b) is less demanding and time consuming - both to be able to serve a larger number of investigated families, but also to prevent fatigue and premature termination in participating families, c) that minimize stigma and therefore can engage reluctant caregivers.

It was with some of the above challenges in mind that the City of Stockholm developed the intervention Safer Kids (Tryggare barn in Swedish), which will be evaluated in the present study. The intervention is based on the Comet program (Komet in Swedish), which is an established parenting program based on social learning theory. Practitioners at several social service units participated in the development and necessary adjustments for the target group. The program was developed in 2013-2014 and has since been implemented within the Social services and is well perceived by practitioners and caregivers, but there is no published study of its effectiveness.

PURPOSE AND RESEARCH QUESTIONS:

The purpose of this project is to evaluate the effects of Safer Kids, that is offered within the social services to caregivers suspected for child abuse. Families who are reported to the social services will be recruited and randomized to either receiving intervention as usual (IAU) or IAU plus Safer Kids. The study will answer questions pertaining possible additional effects of supplementing regular interventions with Safer kids.

The specific research questions will be:

1. What is the additional effect of Safer kids on the occurence of/time to reabuse?
2. What is the additional effect of Safer kids on established risk-factors for child abuse/reabuse?
3. What is the additional effect of Safer kids on the well-being of the abused child?
4. What is the additional effect of Safer kids on the quality of the relationship between the caregivers and the child?
5. What is the additional effect of Safer kids on the frequency of caregivers who prematurely terminates contact with the social services?
6. What is the additional effect of Safer kids on quality of life and economic outcomes?
7. How is practitioner and caregiver adherence to the Safer Kids manual related to outcomes?
8. How are practitioners and parents experiencing the feasibility of Safer Kids?

METHOD:

Participants:

Families will be recruited through the social service units that take part in the research project. Service units from all parts of Sweden are invited to the project. The following inclusion criteria were applied at the start of the project: 1) at least one caregiver in the family is suspected to have perpetrated an abusive act towards a child, 2) at least one child in the family is between 3 and 12 years old and 3) there is a decision of a formal investigation of the matter. The exclusion criterium is that families already involved in an ongoing intervention within the social services at the time of the report of the suspected act will be excluded. The inclusion and exclusion criteria were revised early during the study. For revised criteria, see the section "Eligibility".

Design:

The study is a randomized controlled trial. Families who are recruited to the study from the units who take part of the project will be randomized to either the intervention as usual at the unit (IAU) or to IAU plus Safer Kids. In the later condition, practitioners trained in Safer Kids will offer that intervention. The service units reports to the research team when a caregiver has been recruited and an assistant conducts the randomization.

The participating caregivers will answer questionnaires before the intervention, after 4 months and after 7 months. If the child participates in meetings, the practitioners will also conduct a brief interview with the child at the same time-points. Caregivers will also be asked to answer a shorter follow-up survey after 18 and 30 months. After 4, 7, 18 and 30 months from the first assessment, the practitioners will also complete a report for the research team, based on records the unit keeps for every family they have served.

Intervention:

Safer Kids was developed in 2013-2014 by the City of Stockholm. It was a joint effort between practitioners at a selected number of social service units, and researchers and developers within the City of Stockholm. The content of the program was based on the Comet program, which is a regular parent training program based on social learning theory. The Comet program has shown medium to large effects on child conduct problems in previous randomized trials (Kling, Forster, Sundell, \& Melin, 2010; Stattin, Enebrink, Özdemir, \& Giannotta, 2015). The intervention is delivered individually to caregivers and offered immediately after the report of suspected child abuse have reached the social services. Thus, the question of guilt for the suspected acts is not addressed with the caregiver. Instead the program is presented as a preventive effort offered to all caregivers reported to the social services, regardless of whether an investigation is started and regardless of results. The advantage with this procedure is that the practitioner can start the work with the caregivers at a moment when motivation is higher, rather than at a later point after an investigation that may take time and result in fatigue. The first step of Safer Kids is 1-3 sessions with motivational interviewing. If the caregiver decides to continue after that phase, six individual sessions with general parenting content follows (e.g., child directed play, communication skills, praise, and anger management). The six parenting sessions are followed by 1-3 follow-up sessions depending upon the need in the family. Besides the meeting with the caregiver, Safer Kids also include two individual sessions with the child (without parents if possible).

Sample size calculation:

The effects of interventions on child abuse have generally been non-existent or small (Euser et al., 2015; Vlahovicova et al., 2017). However, the effects on indirect variables (i.e., risk factors) have been in the medium range (Chen \& Chan, 2016; Lundahl, Nimer, \& Parsons, 2006). If an intervention should have a practical meaning, the effect sizes at least have to be between small and medium according to Cohen's standards. It was therefore decided that the sample size should be large enough to at least be able to detect effect sizes in that range. A recruitment of at least 54 families result in a statistical power of 80% to detect effect sizes of d > .35 (Faul, Erdfelder, Buchner, \& Lang, 2009). This is however under optimal circumstances with no drop-out and expected heterogenity in the outcome measures. The goal is therefore set to 100 recruited families. It is expected that a rather large portion of included families will drop put, considering the high pressure and burden this group of families often experience, as well as the sensitive situation of being the subject of interventions within the social services.

Statistical analyses:

To examine possible differences between participants before the interventions, chi-squared tests will be used for categorical variables, and independent t-tests for normally or Mann-Whitney U-test for non-normally distributed continuous variables. Generalized Linear Mixed Modelling (GLMM) will be employed for analyses of most outcome, except for the analyses of time to reabuse (part of research question 1) and time to premature termination (part of research question 5). For those outcomes, survival analysis (Cox regression) will be used.