Exposure to potentially traumatic events and being bullied by peers or siblings in childhood constitute forms of victimization or revictimization, which may be associated with severe long-term effects on mental health [1-3], including anxiety, depression, or suicidality [4]. We include the experience of different forms of maltreatment (emotional, physical, and sexual abuse; neglect; and intimate partner violence) as well as bullying experiences in the definition of revictimization in this trial.

Given that childhood victimization leads to an increased vulnerability for subsequent revictimization in adolescence [5,6], findings highlight a strong need for evidence-based prevention programs targeting children with a history of maltreatment or bullying as a high-risk population for revictimization. Although all types of maltreatment in childhood were found to be associated with revictimization [7] and mental health, one study with adolescent girls in child welfare found that the emotional type of maltreatment showed the strongest link to revictimization in a cross-sectional study using self-report of the types of child maltreatment experienced [8]. However, when using a population sample, sexual maltreatment increased the risk for revictimization the most [7]. In a recent meta-analysis, Scoglio et al [9] identified the following risk factors in most studies on the association between sexual victimization and revictimization: risky sexual behavior, further maltreatment experiences in childhood, presence of posttraumatic stress disorder (PTSD), and emotional dysregulation. In contrast, protective factors have rarely been examined. Only parental caregiving was identified as a protective factor. Unfortunately, many of the included studies were cross-sectional, and it is not always clear which key mechanism is driving the link (eg, why the presence of PTSD is increasing risk).

Children are often placed in foster care because of early adverse experiences in their family of origin, including maltreatment, with the majority experiencing neglect and emotional maltreatment, followed by physical and sexual abuse [10]. These children usually show comprehensive problems in relationships, including foster parent–child relationship, relationship with siblings [11], and peer relationships [12]. Furthermore, many children in foster care are affected by PTSD or attachment disorders [13] and show externalizing problem behaviors, which increase parental distress in (foster) parents [14]. The consequences of maltreatment (eg, PTSD-related symptoms and cognitions, such as negative self-appraisal; maladaptive cognitions of others and the world, eg, concerning the reliability and trustworthiness of others; and threat of harm) have been discussed to be causally involved in the risk for revictimization, although direct mediating effects were only established for threat of harm [15].

Many parenting programs are designed to equip (foster) parents with strategies for increasing positive behaviors in their children and to support them in appropriately managing externalizing problem behavior [16]. Although parenting programs, in general, are a very promising approach to changing child behavior (via changed parenting behavior and reduced parental stress [17]), it has rarely been investigated whether parents may also help to lower the risk for revictimization in children with high risk due to adverse (early) childhood experiences. Warm and responsive parenting is associated with protective effects on children’s resilience to victimization [18], indicating that such parenting programs may be beneficial for coping with victimization and revictimization experiences. However, it is unclear whether parents could also be equipped with the knowledge and skills required to empower children with high risk for revictimization and thereby lower the risk for revictimizing experiences. Burke et al [19] outlined that parental support did not change the occurrence of victimization. Some authors also pointed out that some intervention components may be less effective than others in preventing or reducing child maltreatment experiences. Gubbels et al [20] concluded in their review and meta-analysis that “improving parental personal skills, improving problem-solving skills, and stimulating children’s prosocial behavior should not be the main focus of parental training programs for preventing and reducing child maltreatment” [20]. However, many of these have been identified as promising components to successfully change child externalizing behavior [21], suggesting that the 2 different outcomes may be the result of different pathways of change. Child externalizing problem behavior could be driven by engines different from the risk for revictimization. The intervention model theory of change is key to determining the best intervention components that are most likely to cause changes in the preferred outcome domain [22]. Furthermore, there is some evidence that also challenges the impact (foster) parents may have on the developmental adaptation of their children in care, and the dynamic and reciprocal processes between children and parents that build the foundation for many social learning–based parenting approaches have not been fully supported in a sample of children in foster care in the Netherlands [14]. Although the children’s behaviors affected the distress levels of the foster caregivers, the foster caregivers’ stress did not affect the children’s behaviors. Although this research group discussed a number of potential reasons for the lack of support for a transactional model (eg, foster parents may be expressing their distress less than biological parents, they could potentially give the child away, or children are less vulnerable to parental distress because they are accustomed to worse), this study may indicate that foster parent–child interactions may differ when children were maltreated in the past or at least the focus of the intervention may need to be shifted. For example, Burke et al [19] suggested teaching parents to “be more responsive and connected to their children when they are experiencing difficulties” [19] instead of equipping them with parent management skills more generally.

In sum, few evidence-based parenting programs for foster parents are available, and most of them include a package of intervention strategies [20] and require comprehensive training and parental participation (eg, in-home training as in Attachment and Biobehavioral Catch-up or Keeping Foster Parents Trained and Supported [23,24]). However, (foster) parent participation is challenging [25]. Furthermore, the intervention model and key drivers of change are not specifically tailored to the factors that put children with maltreatment experiences at a higher risk for future revictimization [26]. There is a clear need to identify behaviors and pathways that are responsible for revictimization [27] in this population.

We use the multiphase optimization strategy (MOST) framework [22] to prepare and optimize an intervention for foster parents. The intervention components to be tested in phase 2 of MOST (this factorial study) were developed based on the phase 1 results. Multimedia Appendix 1 [27-32] provides a brief summary of the phase 1 results.

We built our conceptual model (Figure 1) and selected the following domains as targets for the intervention (mediators in the conceptual model):

We combine these domains into 1 mediator called relationship-related risk.

We combine these domains into one mediator called relationship-related safety.

We developed promising intervention components and specified how we expect these to change the risks of revictimization experiences and their consequences.

The primary aim of this study is to examine the effects of each candidate component developed based on the conceptual model and delivered via an internet-based prevention program to the primary caregivers of preadolescent youth in care on revictimization in the form of conventional crime, child maltreatment, peer and sibling victimization, sexual victimization, witnessing and indirect victimization, and cyberbullying. The ultimate goal is to choose the candidates that best reduce and prevent revictimization from a set of 4 components with 2 levels each (on/off). The primary and secondary research objectives are presented in Textbox 1.

We will examine the hypotheses specified in Textbox 2, which are related to the main effects of the intervention components (3 content components and 1 adherence component).

We will also answer the following four questions to establish how well the conceptual model captures the relevant mechanisms of the intervention (mediation analyses, adapted from the study by Smith et al [56]): how well will the three mediators (relationship-related risk, relationship-related safety, and constructing an identity) predict the occurrence of revictimization, how well does each module content (Watch Out!, You are safe here, and Who am I and where do I belong?) and the coach evoke each of the 3 mediators, how much of the relationship between the intervention’s content components and revictimization is explained by the 3 hypothesized mediators, and how much variability between the content components and revictimization will remain unexplained.

Furthermore, the following variables were identified as potential moderators of intervention (component) efficacy based on the literature review in phase 1:

Taken together, we use the MOST framework [22] to optimize a web-based program for foster parents, which comprises 4 intervention components. To optimize the program, the primary aim of this study is to examine the efficacy of each component on the primary and secondary outcomes. This paper outlines the protocol (version 01) or the factorial trial, following the SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) guidelines for clinical trials (Multimedia Appendix 3 provides the SPIRIT checklist).

This study is conducted in accordance with the Declaration of Helsinki. This study was approved on September 1, 2021, by the University of Bremen Ethics Committee (no. 2021-09). Any changes to this protocol will be submitted to the institutional review board for notification and approval.

The EMPOWERYOU project (funded by the German Ministry of Education and Research, Project code FKZ 01KR1806D) aims to develop and optimize a web-based parenting program by using the MOST framework [22]. This study will use a 2×2×2×2 full factorial design by randomly allocating participants to 1 of 16 experimental conditions (Table 1). To estimate the main effects of the 4 intervention components and their interactions, data from all the experimental conditions will be used. For example, the effect of the component Relationship-related risk will be estimated by comparing the mean of the experimental conditions 1 to 8 with the mean of the conditions 9 to 16 (Table 1). Families in the first condition will not receive any experimental component but will be delivered the 2 basic intervention modules that every caregiver receives. Families will be informed that everyone may access all modules and that the sequence and timing of each module will be based on a randomization procedure.

Anticipated effect sizes were estimated from the reported effect sizes for victimization and revictimization interventions, which also included (1) facilitating risk detection skills; (2) social skills, such as problem-solving or conflict management; (3) skills to build healthy relationships; and (4) the ability to reflect on own expectations that relationships will include harm [28,60,61]. We expect small to moderate effect sizes between Cohen d=0.28 and Cohen d=0.46 for the main effects of specific intervention components (component 3: 0.46, component 4: 0.28, and component 6: 0.32). Thus, a sample size of 317 was determined as necessary to detect the smallest anticipated effect size of Cohen d=0.28. To detect this difference with the analysis of covariance as the suggested method for component selection by the developer of the factorial design (groups=2, df=1, covariate=1) with 80% power at α=.10 per intervention component or interaction, a sample size of 317 is required (calculated with FactorialPowerPlan SAS Macro provided by Dr Collins [62]). We will use an intention-to-treat approach, although only one family dropped out during the intervention period in the pilot study. In contrast, findings from the pilot study showed that 26% (4/15) of the recruited caregivers that registered for the website dropped out before the intervention. Therefore, we consider a dropout rate (before the allocation to condition) of 26%. Therefore, we aim to recruit a total of 429 families. Recruitment will be stopped as soon as we reach the sample size needed for the analysis (N=317).

Informed consent will be obtained from each participant (caregiver and child) on the web. Interested study participants can register on the website and then receive the information sheet and consent form on the web. The caregivers are informed about the objectives, study procedure, the rights and obligations of all those involved, and the procedure and data processing using the information and consent forms. If the participants have any questions about the documents or the study, they can call the study team at any time. The phone number will be clearly indicated on the website. Each caregiver is asked to provide active opt-in consent for their own participation and the participation of their child in care. Multimedia Appendix 4 provides the informed consent form in German language.

Each participant will be assessed on the web using SoSci Survey [66] 3 times, that is, at preintervention assessment (week 1), postintervention assessment (week 12), and 3-month follow-up assessment (week 24). Families will complete a brief questionnaire (5 minutes) on adverse events and child behavior during the past 2 weeks at the beginning of each module on the program website to ensure a closely monitored adverse event assessment. Multimedia Appendix 5 outlines the SPIRIT-recommended schedule of enrollment, interventions, and assessments. Families will receive a reimbursement of €30 (US $29.32 in the form of 2 vouchers: €20 (US $19.55) for caregivers and €10 (US $9.77) for the child) for their study participation after follow-up assessments.

To ensure data privacy, research data will be identified using pseudonyms and will be stored on 2 password-protected servers only accessible by approved study staff members. Personal data (ie, username, phone number, and email address) will be stored separately from research data. Personal data and the pseudonym codes will only be matched in a handwritten key code list to protect the confidentiality of data. Identifiable data (including the key code list) will be deleted 2 years after the end of the project. The research data will then be anonymized. The anonymized data set will be stored for 10 years and shared with other research teams upon request using a repository that will be chosen by the consortium (advised by the DSMB).

An independent DSMB will provide additional oversight on data safety, ethical procedures, and best clinical practices. A thorough data safety concept was developed and piloted in phase 1. This safety protocol outlines how adverse events will be identified, registered, acknowledged, and handled.