To investigate the effectiveness of the RA we use a Single-Case Experimental Design (SCED). We use this design for several reasons. In the first place, we learned from previous research on SofS (Hammink et al. 2015; Vink et al. 2017) that investigators did not succeed in including as many participants as necessary to conduct a Randomized Control Trial. Few families were referred and too few parents gave their consent for participation in the research. Furthermore, a SCED is used because of the high complexity and heterogeneity of the sample, the involvement of multiple partners, and innovativeness of the intervention (Maric et al. 2012; Norell-Clarke et al. 2011). The Single-Case Experimental Design (SCED) methodology allows us to investigate treatment effects on a case-by-case basis. SCEDs are an excellent opportunity to stimulate collaboration between clinicians and researchers, unifying research questions that emerge from clinical practice with research methodology to test these questions on a single-client level (Borckardt et al. 2008).
In a SCED study all participants undergo several conditions. In our study an AB-design will be used, as illustrated in Fig. 1. The A-phase is the baseline period between the referral and the start of the treatment (T0–T1) during which we conduct the assessments, and the families do not receive any treatment. Because of the nature of the problem and the need to provide treatment to these families as soon as possible, a naturalistic baseline period is used consisting of a regular waiting period of around 4–5 weeks. The B-phase concerns the period in which the families receive treatment (20 sessions spread over a period of 12 to 18 months) (T1–T2). The participants will receive pre- and post-treatment assessment (T1 and T2, respectively). Additionally, there are two follow-up assessments three and 6 months post-treatment (T3 and T4).
The current study will use both quantitative and qualitative assessment methods. The quantitative measures will be administered at five extensive assessment points (T0–T4) and consist mainly of questionnaires for parents and children, and one pictorial item. During the extensive assessment sessions, children and parents fill out the online questionnaires under the supervision of an independent assessor. In addition, short personalised questionnaires, so-called Idiosyncratic Assessments (IA) (5–10 questions online), will be filled out weekly during the baseline and treatment phases (Weisz et al. 2011). There are several multidisciplinary network meetings during the treatment phase during which parents, children, therapist and social network members rate the incidents of child abuse and the openness of communication about child abuse in the family on a Visual Analogue Scale (VAS-scale).
The qualitative measure, a semi-structured interview, will be administered post-treatment (T2). An independent assessor will interview parents and children.
Participants and Recruitment for the Study
The participants are members of fifteen families with children aged between 8 and 18 years who are referred to four different mental health care centers in the Netherlands specialised in child abuse. The professional initially involved with the family (i.e., child protection worker, neighbourhood team social worker, general practitioner) refers the families to one of the participating mental health care centers. It is possible that some of the families are court obliged to enrol in the program. Usually, the therapist conducting the intake will explore family’s interest to participate in this research project. If the family agrees, the therapist hands the case over to the researcher (second author) who provides extensive information about the research project and arranges the informed consent and the research assessments. If the family cannot come to the institution, meetings are organized close to where the family lives.
Inclusion criteria are: (1) there are specific signs of current child abuse, determined by more than one informant. Informants consist of family members, members of the social network of the family, doctors, teachers, police and other professionals; 2) one, or both, of the parents (partially) deny that they have maltreated their children; 3) parents are willing to participate in the intervention and research; 4) working towards safety within the family seems to be an achievable goal for all involved parties; 5) there is an existing safety network of at least one person; 6) the families’ case manager (i.e., the child protection worker) is able to attend the multidisciplinary social network meetings.
An exclusion criterium is: Families who do not have a permanent residence or do not speak Dutch.
This study has been approved by the Ethical Board of the Institute of Psychology (University of Amsterdam). Both parents and children (> 12 years old) need to provide their active, informed consent before participation in this study. Participation is entirely voluntary, and participants are free to withdraw from the study at any moment without an explanation and without any consequences for their further treatment at the mental health care center.
The Resolutions Approach (RA) Intervention
Ideally, professional help for abused children is organised in phases. In the first phase, the aim of the intervention is to create a safe environment for the child. If safety is established, in the second phase psychological trauma-related symptoms are considered. If indicated, a trauma-focused treatment can be delivered. In the final phase, plans for monitoring safety in future are organised (Vogtländer and van Arum 2016).
The aim of RA is to create a safe environment for the child through seven steps (Table 1). To this protocol, one additional technique from the SofS intervention (Turnell and Edwards 1999) - the safety card - is added. RA is conducted in an individual, family and social network context and consist of 20 sessions spread over a period of 12 to 18 months. In the research study described in this protocol, each family will go through the entire RA program in the same order (steps 1 to 7). The number of sessions and weeks spent on each step may vary and it depends on factors such as the severity of the suspected abuse and the motivation of the family members. Less motivated parents or parents who are in disagreement will take more time to come to an agreement with each other about each step of the intervention.
A noteworthy feature of RA is the elegant way the method deals with the denial of child abuse and neglect. Denial is seen as a complex interactional process on a continuum (Turnell and Essex 2006). In many cases, parents do not deny all of the accusations nor do they confess to all of them. Parents may deny accusations because confessing may have serious consequences such as losing their jobs, risking prosecution or the breakdown of relationships. In the context of RA, rather than arguing with parents about the past, professionals and families work together towards a safe home environment. This is done through signalling triggers for future abuse situations, preventing unsafe situations for the children, empowering the potentially safer parent and involving the social network of the family. All the professionals involved with the family and the family members work together in social network meetings towards the safety of the children.
Figure 1 and Table 2 display an overview of measures and assessment points in this study.
Primary Outcome Measures: Type and Frequency of Incidents of Child Abuse and Domestic Violence
Childhood Trauma Questionnaire-Short Form (CTQ-SF; Bernstein et al. 2003) is a self-report instrument designed to assess emotional and physical neglect, emotional and physical abuse, as well as sexual abuse as reported by the child aged 12–18 (CTQ-SF, 28 items on a 5-point scale). The English CTQ has good validity and is reliable, with a total α = .97, and subscale alpha’s range from .63 to .95 (Bernstein et al. 1997, 2003). The Dutch CTQ has good internal consistency, convergent validity and reliability (Thombs et al. 2009).
The Conflict Tactics Scale Parent-Child / Child-Parent (CTS-PC/CP; Straus et al. 1998) is a self-report instrument designed to assess the presence and degree of child abuse as rated by the child and the parent. It measures psychological and physical child abuse and neglect (21 items on an 8-point scale). Reliability indices range from low to moderate, and evidence for construct and discriminant validity of the subscales was found: ‘Overall physical assault on the child’ (α = .55), ‘Psychological Aggression towards the child’ (a = .60), and ‘Nonviolent discipline (α = .70) (Straus et al. 1998). For the child, the same items as in the Dutch Prevalence Study of Child Maltreatment (Euser et al. 2013) are used capturing being witness of psychological -, physical -, sexual violence and injuries due to parental fight.
The Revised Conflict Tactics Scale (CTS2; Straus 2014) is a self-report instrument measuring psychological -, physical -, sexual violence and injuries due to a fight with the (ex-)partner, completed by one parent about the other parent (78 items on an 8-point scale). Good construct validity, discriminant validity and internal consistency have been reported with α = .79 for assault victimisation and α = .81 for assault perpetration. The Dutch version has an acceptable to good reliability, with α = .82 for assault victimisation and α = .65 for assault perpetration (Straus and Mickey 2012).
Secondary Outcome Measures: Posttraumatic Stress Symptoms, Emotional and Behavioral Problems in Children, Parental Stress and Closeness of Child-Parent Relationship
The Revised Children’s Responses to Trauma Inventory (CRTI-R; Alisic and Kleber 2010) is a self-report questionnaire designed to measure child posttraumatic stress symptoms, completed by parents and children (34 items on a 5-point scale). Good reliability of the questionnaire has been reported with a total α = .91 (Alisic and Kleber 2010; Alisic et al. 2006).
Strengths and Difficulties Questionnaire (SDQ; Goodman 2001). The SDQ is a self-report questionnaire that assesses social-emotional and behavioral problems in children and adolescents (3–17 years) as reported by parents and children. The questionnaire consists of five subscales: emotional symptoms, conduct problems, hyperactivity/inattention, peer relationship problems and pro-social behavior (25 items on a 3-point scale). The total score on the SDQ has satisfactory validity and reliability, with α = .73 Goodman (2001). The Dutch SDQ has good validity and acceptable internal consistency, with α = .70 for the parent version and α = .64 for the self-report version (Goodman et al. 2000).
The Dutch Parental Stress Index-child report (Nijmeegse Ouderlijke Stress Index Kinderen; NOSIK; Brock et al. 1992; Prinzie et al. 2007) is a self-report questionnaire (25 items on a 6-point scale) assessing parental stress as rated by parents themselves. The Parental Stress Index has acceptable reliability, and content, concurrent, and construct validity (Loyd and Abidin 1985; Prinzie et al. 2007). The NOSIK has good reliability, with α = .95 (Brock et al. 1992).
The Inclusion of Other in the Self Scale (IOS; Aron et al. 1992) is a single-item, pictorial measure used to assess the closeness of the relationship between children and their parents/caregivers, as assessed by the child (1 item per parent, on a 7-point scale.) The child has to choose one image that best represents their relationship with the parent. The images consist of two circles, which increase in the degree of overlap, the more overlap, the more closeness. Convergent validity, discriminant validity and predictive validity of IOS were acceptable. The reliability was investigated using different formats of the response scale (circle and diamond); Chronbach’s alpha was .93, and test-retest reliability was r = .83 (Aron et al. 1992).
In addition to the extensive measurements at five assessment points, the parents and children complete a short Idiosyncratic Assessment (IA) on a weekly base during the baseline and treatment phases (T0-T2). The IA is a personalised questionnaire including 5 to 10 items representing top family problems as indicated during the intake and first assessment point (T0). Such a short questionnaire allows monitoring of the main family complaints on a regular base but also meets the needs of families and clients characterized by heterogeneity (Weisz et al. 2011). The IA items are derived from three questionnaires filled in during the T0: two primary outcome measures, measuring incidents of child abuse and domestic violence (CTS2, CTS-CP/PC), and one secondary outcome measure (NOSIK) measuring parental stress. The idea is to select the items on which the client reports the most problems, but, prior to the selection, client and therapist are also asked separately to identify the most distressing problems (types of child abuse) for him/her and the family environment. During this selection, the opinion of the participant and the therapist is also taken into account. The reason for this is to specify the most distressing problems for the participant and his/her environment. They can choose between types of abuse and domestic violence, which are derived from the subscales of CTS2 and CTS: (a) Parent-child: sexual-, emotional-, physical abuse, or injury; (b) Partner-partner: emotional- or physical abuse. It is possible to get biased results if the selection was based only on the least adaptively rated items on the questionnaire. This is mainly because the CTS2 and CTS-CP/PC measure the abuse on a frequency-scale. However, it is possible that a participant is more frequently abused emotionally, but that she/he finds the physical abuse worse. For the parental stress items, 1–3 items are always chosen related to the least adaptively rated items on the NOSIK.
Multidisciplinary Social Network Meetings Ratings
These meetings are planned approximately every six weeks during the intervention period (T1–T2). Parents, children and social-, and professional network members score the subjective safety of this family at the end of the multidisciplinary network meetings on a VAS- scale ranging from “0 = Recurrence of similar or worse abuse/neglect is certain” to “10 = Sufficient safety for the child for the case to be closed” (Turnell and Edwards 1999).
Mechanisms of Change
Specific attention will be given to the ratings of parental stress and openness of communication as clinical experience suggests that these constructs may be important mechanisms of change behind RA effects. Idiosyncratic assessments of parental stress as outlined above will be used to achieve this aim. Further, via a newly constructed VAS-scale, openness of communication about child abuse and safety will be assessed weekly via the following items: “In our family (parents + children) we talk about the concerns related to the safety of our children” and “We (our family = parents + children) talk with our extended family and friends about the concerns related to the safety of our children.” on a scale ranging from “0 = we never talk...” to “100 = we talk frequently with...”
Next to this, potential mechanisms will be assessed in the qualitative, semi-structured interview administered to the children and the parents at the end of the treatment. Qualitative data gathered at post-treatment (e.g., asking the client: “What aspect was the most helpful for you in stopping child abuse in your family?”) could provide unique insights into the process of the treatment (Dworkin et al. 2006). The goal of the interview is to acquire a better understanding of these potential mechanisms, and to discover other possible mechanisms involved in the remediation of child abuse for each family.
The qualitative approach allows us also to investigate the most effective RA components for each family, and also to examine the satisfaction with the treatment and the collaboration between professionals and the family. Interview utilized in Gumbleton’s study (1997) is serving as a starting point for the construction of our own interview. Main topics from our interview will include: (a) Has the safety of the children been improved and has the openness of communication about child abuse increased? (b) Did the family members reach their goals? (c) Which treatment elements helped in achieving the goals?
All of the therapists are trained and experienced in RA. To ensure treatment integrity, therapists mark the employed RA elements on a checklist after each session. If additional trauma treatment is conducted, this will be noted. Finally, all of the therapists will participate in regular supervision sessions provided by the first author.
To test the effectiveness of RA on primary and secondary outcome measures the following analyses will be conducted. First, Reliable Change Index (RCI; Jacobson and Truax 1991) will be calculated for each participant. The RCI facilitates the investigation of clinically meaningful change (i.e., as compared to a norm group) in the outcomes from pre- to post-treatment, and pre- to follow/up treatment. An RCI > 1.96 or < −1.96 (z-scores) indicates a statistically significant reliable change (Jacobson and Truax 1991). Second, a mixed-models method in SPSS (Maric et al. 2015) will be used to assess statistically-relevant within-person change in outcomes between the end of the baseline phase and the end of the treatment phase. This new-generation SCED data-analytic method requires less data observation points per phase (i.e., 4–5) to calculate between-phase differences than the more traditional SCED methods (Barlow et al. 2009).
Cross-lagged correlations will be used for the analyses of the mechanisms of change. Temporal relations and direction of changes in frequency and type of incidents of child abuse, level of parental stress and openness of communication about child abuse will be investigated. The qualitative data from the interviews will be analysed by the Thematic Analysis of Qualitative Data (Braun and Clarke 2006) using the software Atlas.ti (ATLAS.ti 2016).
Handling and Storage of Data and Documents
A personalised code is assigned to each participant in this study for the anonymous data storage. This code represents the participating mental health centre, family and the role of the participant (e.g., father, mother, oldest child). The participant fills in his/her personalised code at every extensive assessment point and this is conducted by the software Qualtrics. The idiosyncratic assessments are personalised and linked to the same code. The informed consent contains the name, email address, phone number and personalised code of the participant. The involved mental health care centers store the informed consent in protected files for administrative purposes. The copies of informed consent forms are securely stored at the University of Amsterdam (UvA), and only the second and the last author have access to them.
A method to enhance the generalisability of the SCED’s is the use of replication (Graham et al. 2012). Multiple SCED’s will be conducted. The aim is to achieve a scientific consensus about the effectiveness of the intervention considering the difference in participants, clinicians and settings. According to the guidelines of the Task Force of the American Psychological Association (1995) an intervention can be considered as well-established if a series of at least nine single case design studies demonstrate efficacy. We will aim at including at least 15 families taking into account possible dropouts.