Outcomes that matter

In relation to the outcomes they felt were important for maltreated children, one of the two groups felt that 18 of the 25 outcomes listed on the Q-sort cards were too specific to particular problems (such as anger or eating disorder) and would entail inappropriate generalisations, so they removed them. In this group, this left only seven outcomes to discuss, to which the young people themselves added an eighth. Despite differences in approach, there were clear similarities in the views of both groups. The items ranked most highly across the two groups were as follows:

Helping the person to:

• learn skills to handle life’s ups and downs (group 1)
• understand what being ‘treated badly’ is, and learning to recognise when things are not OK (group 2)
• learn ways to keep themselves safe (e.g. knowing when to report something and to whom to report it) (group 2)
• feel safe (groups 1 and 2)
• ‘bounce back’ if things in their life go wrong (groups 1 and 2).

Asking for, and receiving, help

The young people also viewed individuality as central to understanding what things might be important in making it easier for someone to ask for help, or might make some forms of help more acceptable than others. For example, two of the cards in the Q-sort were ‘person was still living with their family’ and ‘the person was no longer living with their family’. The young people pointed out, quite appropriately, that it was not easy to rank these cards one against the other because their importance would depend on whether the young person had experienced maltreatment within the family or outside it. One of the things that they emphasised in their discussions was the importance of choice, both in relation to starting and ending a service or therapy. Young people in one of the groups were unanimous in the view that the most important factor for a young person was the opportunity to meet a professional beforehand and to decide whether or not they would be happy to begin getting help from them. Furthermore, the group wanted the format of this preliminary meeting to be determined by the young person themselves, because some might want an opportunity to meet a potential therapist informally (e.g. meeting for a cup of coffee) without any mention of the therapy, whereas others might prefer a formal discussion of what to expect from the intervention. Clearly these alternatives are not mutually exclusive, but reflect different concerns. The opportunity to meet someone informally speaks to a concern about the likely quality of the therapeutic relationship. It may also be a proxy for choice and commitment, both of which may increase the likelihood of someone engaging with therapy, or staying with a course of treatment. A formal discussion of expectations can provide an opportunity to allay anxieties, to negotiate boundaries and to make an informed decision about the acceptability of what the therapist is offering.

Other things that mattered to young people included confidentiality (accessing help in ways that maintained their confidentiality) and trust in the help-giver. Making their own decisions about whether or not to get help, or at least being involved in that decision (rather than these decisions being taken by their parents or carers) was also ranked highly, along with not feeling judged or criticised – and, perhaps surprisingly, not having to worry about paying for the service. Clearly, many of these issues are of most relevance to older children and young people, such as those in the advisory groups, but, in terms of engagement with therapeutic services, they are probably salient to children of most age groups.

Things that get in the way

One group found it particularly difficult to identify a single factor that they regarded as the most significant barrier to young people accessing therapy. This group identified two things that they thought might deter young people from seeking help. The first was a worry that their situation was too complicated for anyone to be able to help with, and the second was that some children might think that those offering help would not believe them. Other factors identified were as follows.

The person does not:

• want to be seen as having mental health problems (group 1)
• know who to ask about getting help (group 2)
• think they need any help (group 1)
• trust the people/services offering help (group 2)
• think that the help available will work for them (group 1).

This is a percipient list from these groups of young people. Children whose families have been engaged with social services for reasons of maltreatment, some of whom may have been removed from home as a consequence, may well have ambivalent feelings about public services. It is not unusual for children to feel responsible for a family break-up, and some are blamed by their parents for the involvement of Child Protection Services. Together, these point to the importance of services anticipating, and addressing, feelings of stigma and concerns about the likely effectiveness of services, and ensuring that those who need help know where and how to access it. The barriers that can be created by parents who prevent children from accessing services was also identified as an issue.

Most of these concerns or issues surface in the studies included in our acceptability review, although few have been systematically investigated. We return to these issues after following a review of the included studies.

Study design

One^107,108 of the included studies was a randomised trial. The rest were uncontrolled studies.

Sample sizes

Sample sizes varied. They included a single case⁶²⁹ and four^{497,504,506,680} studies with very small samples of six^504,506,680 and 25.⁴⁹⁷ Five^{107,108,481,482,492,505} studies had samples numbering between 50 and 100, and four studies^630–633 had samples of > 100. McPherson et al.⁶³³ recruited 254. Fraynt et al.⁶³² had the largest sample of 562 children, but was a study based on data from a core data set in the USA.

Location

These were mostly US based (nine^{107,108,482,492,504,505,629–633} studies), with two Australasian studies (Australia,⁵⁰⁶ Philippines⁶⁸⁰), two European studies (UK,⁴⁸¹ The Netherlands⁴⁹⁷) and one South African study.²⁰⁰

Participants

The age ranges of children varied from 2.5 years to 25 years. Six^{492,497,504,505,633,680} studies reported on interventions that were solely treating children for sexual abuse history. Kolko 1996^107,108 focused solely on children who had been physically abused. The other studies recruited children with had experienced one or more types of maltreatment.

Interventions

Interventions covered individual, group- or family-based therapy. Lange and Ruwaard⁴⁹⁷ explored the impact and acceptability of a web-based version of cognitive/behavioural approaches with no face-to-face contact at all between therapist and client.

The number of sessions ranged from 8 to 34, with most interventions lasting 12 sessions (eight studies);^{107,108,481,492,497,504–506,632,680} three^629,631,633 studies did not specify the length of treatment. Interventions in the two Chasson et al.^482,630 studies comprised 20 sessions.

Fraynt et al.⁶³² reported variable attendance rates which differed in relation to ethnicity, the focus of treatment (family treatment vs. no family treatment), group treatment compared with no group treatment, and location of treatment (office vs. community).

Pre-treatment withdrawal

Lange and Ruwaard⁴⁹⁷ set out to explore the effects of an online treatment for young victims of sexual abuse. In light of significant pre-treatment withdrawal in an earlier uncontrolled study, the authors introduced a number of measures that they hoped would reduce this in the context of a controlled (within-subject baseline-controlled) study. In common with other online treatment studies, the previous study had experienced a pre-treatment withdrawal rate of 90%.

In Lange and Ruwaard⁴⁹⁷ there remained a high level (77%) of pre-treatment withdrawal (82 out of 106 applicants not excluded by the research team), despite the steps taken to minimise it, namely no randomisation, parental consent required only for children of < 16 years rather than < 18 years (Dutch law for RCTs of new interventions), raising the upper age level for participants from 18 to 25 years, and providing the alternative of a structured interview by ‘chat’ if they were reticent to answer screening questions on the telephone.

Data available to the researchers indicated that pre-treatment withdrawal was strongly correlated with biographic questions, suggesting that anonymity may be an important factor, although whether for treatment or as an artefact of the study is not clear. The lowest pre-treatment withdrawal was among the oldest group, among which 46% (19/41) of those aged ≥ 18 years started treatment. All but one of the eight adolescents aged 14–15 years (and who required parental consent) withdrew, and 12 of the 16 young people aged 16–17 years withdrew (75%). Once engaged in treatment, there were few subsequent dropouts. The authors conclude that ‘fear of losing anonymity is important for both young and old participants, whereas the fear of needing parental consent is more or less decisive for younger age groups’⁴⁹⁷ (≤ 16 years). Once engaged in treatment, there were few subsequent dropouts.

These findings need to be interpreted against the context of a very small study. One of the recommendations of the authors is to change the intervention from a therapist-led online treatment to a wholly ‘self-help’ model, which may resolve the anxieties about loss of anonymity, while not dealing with some of the legal dilemmas associated with professional accountability and so on.

Treatment engagement

Fraynt et al.⁶³² used regression analysis to investigate factors associated with treatment engagement in trauma-informed therapies. The paper⁶³² includes no detailed description of the therapies offered these children, but the interventions they identify as trauma-informed interventions include TF-CBT, cognitive–behavioural intervention for trauma in schools, CPP, and EMDR therapy.

Although the study⁶³² is set in the USA, its findings raise potentially important issues for the successful engagement of children and young people in therapy within the UK. Fraynt et al.⁶³² found that age, functional impairment and the receipt of group and community-based (as opposed to office-based) services were correlated with increased engagement. Younger children who received more group sessions, and children who received services in places other than the office, were more likely to engage in treatment, as were children with more functional impairments, although children with more impairments were also more likely to be deemed by their therapist to have dropped out of treatment involuntarily.

When these things were controlled for, ethnicity remained a significant predictor of engagement, with Spanish-speaking Latino clients being most engaged in treatment (an average of 34 sessions) and African American clients being least engaged (an average of 25 sessions). The authors hypothesise that because of their language preference, Spanish-speaking Latinos may be more likely to get a therapist of the same cultural background to themselves, which may enhance treatment engagement compared with African American families. Furthermore, they hypothesise that the latter may be less engaged in treatment because they may mistrust or have had negative experiences of mental health treatment services (p. 72). These findings underline the importance of addressing language and culture in the context of mental health services. They also highlight the importance of ensuring that services communicate relevance and sensitivity to families from minority ethnic groups, and to all families who may have found engagement with Child Protection Services itself a traumatic experience, leaving them reluctant to seek help or engage with available treatment.

Treatment completion

Regression analyses were used in four^{482,630,631,633} studies to examine treatment completion. Caregivers’ perceptions of the severity of abuse appear to be a common theme relating to treatment completion. Chasson et al.⁴⁸² reported that higher levels of depressive symptoms and feelings of intrusion during treatment were associated with dropout from TF-CBT. In a later study,⁶³⁰ in which they analysed data from the same group of children augmented with additional cases, the authors found that children who had been abused by another child (not by a parental/adult figure), or had experienced a single event and had not suffered a life-threatening or serious injury, were more likely to drop out than those children exposed to multiple, physical injurious abuse by an adult.

Eslinger et al.⁶³¹ found that the odds of dropout were greater for children with younger parents. The authors also found a relationship between age and dropout, with the odds of dropout being greater for older children. Children in foster care were more likely to complete treatment than those living with biological or adoptive parents, although the authors augur some caution as the parents in this study were more likely to be younger than foster carers. When children’s and parents’ scores were high for PTSD (which the authors interpret as ‘acknowledgement’) then the odds of children completing at least a moderate ‘dose’ of treatment were improved. Children who were the only victim in the family were also more likely to complete at least a moderate dose if their caregivers were involved in the treatment process. McPherson et al.⁶³³ also found that caregiver involvement was positively associated with treatment completion and achievement of mental health treatment goals among a sample of sexually abused children who were referred to a hospital-based children’s centre that provided assessment and therapy. This issue of how seriously the abuse is viewed will be discussed in more detail.

Children’s views

Four^{107,108,492,497,506} studies used data from rating scales to quantify children’s satisfaction levels. Scores indicated moderate to high levels of satisfaction, both with therapists and with CBT interventions. When studies also reported caregiver ratings, children’s reported levels of satisfaction were lower than those of their parents.

Lange and Ruwaard⁴⁹⁷ asked participants about their satisfaction with treatment in general and also specific aspects of treatment. Their participants rated the therapeutic alliance, and were asked questions about the nature of the online contact, whether or not they missed face-to-face contact with their therapists, and how they perceived the effectiveness of treatment. Participants generally expressed satisfaction with their online treatment, and, although 22% did miss face-to-face contact, all were highly satisfied with their therapists, and all but 2 of the 23 said that they would recommend the treatment to others. Significantly, although all modules were well received, the module that focused on the exposure was most highly rated.

This study⁴⁹⁷ (which used baselines as a source of historical control) found a steep drop in scores on the IES²⁷⁹ during the control phase, which Lange and Ruwaard,⁴⁹⁷ attributed to an effect of screening (which asked questions that required participants to focus on their trauma and current situations). They hypothesise that, in combination with the psychoeducation and expectation of treatment, this might have resulted in increases of awareness and hope.

The studies by Smith and Kelly⁵⁰⁶ and Kolko^107,108 used questionnaires to assess perceptions of treatment acceptability and treatment expectations. Four of the five participants in the Smith and Kelly study⁵⁰⁶ agreed that the programme was of high quality, it met their needs and they would recommend it to others. Children and parents in the study by Kolko^107,108 were asked to rate the overall acceptability of the key components of the interventions allocated to them. At the end of treatment children completed the 10-item Child Evaluation Inventory (CEI³⁰¹). Mean ratings suggested moderate to high acceptability of treatment at the outset of treatment, with all but one item scoring a mean rating of > 3 (out of 5). Ratings tended to be higher than those for FT in relation to participants interested in the material learned that session (4.6 vs. 3.8) and confidence that therapists could help minimise abuse potential (4.2 vs. 2.8).

The mean ratings for children’s responses on the CEI suggest moderate to high acceptability (25.1 for CBT, 22.6 for FT) and utility (15.1 for CBT, 14.0 for FT).

Barker and Place⁴⁸¹ and San Diego⁶⁸⁰ report qualitative data on children’s and young people’s views of their treatment. Most of the children were generally positive about the experience, but Barker and Place⁴⁸¹ described the children as having some difficulty in articulating what they found most and least useful about the therapy. The analyses of the progress of five young women through the course of therapy by San Diego⁶⁸⁰ illustrated the women’s reluctance at the start of treatment, initial loathing of re-experiencing trauma and being unsure that therapy would help, but becoming more positive as therapy progressed.

Caregivers’ views

Six^{107,108,481,492,504,505,629} studies reported caregivers’ views of treatment and provided both qualitative and quantitative data, using a range of data collection methods, from qualitative interviews to focus groups and evaluation forms. Details are provided in Table 13.

All used rating scales to collect data on satisfaction, and some on caregiver treatment expectancy. In five^{107,108,481,504,505,629} of these studies, parents were recruited by a convenience sample, as parents of children involved in treatment; it is unclear how the caregiver sample was selected in the Project Safe programme.⁴⁹²

Parents rated treatment satisfaction as ‘high’ in all studies that used treatment rating scales.^{107,108,492,504,505} The sample sizes in these studies varied from 6⁵⁰⁴ to 85.⁴¹ Some parents in the Barker and Place⁴⁸¹ study felt that the CBT intervention had ended sooner than expected.

Two^481,629 studies presented qualitative findings from interviews with parents and carers. Both studies^481,629 reported positive experiences of the interventions, including a clear understanding of treatment aims and their expectations for the therapy, appreciating having someone neutral to whom their child could talk;⁴⁸¹ Buschbacher’s single-case study⁶²⁹ found that the mother felt part of the therapeutic team that treated her son, although this study is particularly vulnerable to bias, as it was effectively (as its title indicates) a testimonial sought from a selected parent by a clinical team, one of whom conducted the interviews. The quantitative evidence presented supports the positive reports from the qualitative data.

In the Kolko^107,108 study, parents were telephoned between the third and the first treatment sessions and asked to answer 10 questions using a five-point Likert scale (e.g. how much did the counsellor listen to you?, how much do you like your counsellor?). At the end of treatment they completed a 16-item consumer satisfaction questionnaire. Responses indicated high levels of acceptability with both treatments (53.8 for CBT, 50.9 for FT) and overall satisfaction (27.7 for CBT, 27.4 for FT).

Staff views

Qualitative evidence of staff views was also presented by Barker and Place⁴⁸¹ and Buschbacher and Place,⁶²⁹ both of whom report positive findings. However, in both studies,^481,629 staff expressed concerns about resource constraints, or strain, that was felt to threaten the viability of the service. The Sunrise Project⁴⁸¹ relied on one worker, and staff referring children to the project worried about the security of this post, the overall lack of resources and the lack of potential to increase capacity. Concerns about capacity were also raised by Buschbacher.⁶²⁹

The importance of parents and caregivers in securing successful outcomes in therapy was generally recognised in all studies, and is a recurring theme that will be discussed in more detail.

Location

The studies by Cross et al.⁶³⁴ and Osofsky et al.⁵¹⁵ were set in the USA, whereas the studies by Sudbery et al.⁶⁶³ and Powell and Cheshire⁶⁶² were undertaken in the UK.

Study designs

The four^{515,634,662,663} studies vary widely in design.

Cross et al.⁶³⁴ used data from a multiyear, multisite, national evaluation of 15 sites providing services to children exposed to violence. They examined retention at 6 months post baseline, using logistic regression to analyse the characteristics of those retained in treatment.

Osofsky et al.⁵¹⁵ report on a multisite study of a pilot infant mental health programme. Interviews were used to obtain caregivers’ and therapists’ qualitative impressions of treatment.

The Sudbery et al.⁶⁶³ study is the only study of the three that presents children’s views of the intervention. As well as a focus group, semistructured interviews and survey methodology, case file analysis and organisational documentation were used, and two members of the research team were embedded within the organisation as participant observers.

The Powell and Cheshire⁶⁶² study was a pilot evaluation that used qualitative methods.

Sample sizes

Pooling data from 15 sites, the sample in the Cross et al.⁶³⁴ study was 1085. Osofsky et al.⁵¹⁵ recruited 75 mother–child dyads: 25 from each of three sites. The sample size in the Sudbury et al.⁶⁶³ study is difficult to ascertain, but the authors appear to have scrutinised the files of 113 children, conducted a focus group of eight young people and interviewed a further four. Powell and Cheshire⁶⁶² conducted semistructured interviews with four mothers and one grandmother.

Participants

Cross et al.⁶³⁴ reported an age range of 1–17 years, with the majority of participants aged between 3 and 7 years. Children in the Osofsky et al.⁵¹⁵ study were maltreated young children aged < 5 years, or young children at risk of maltreatment. Their mean age was 20.19 months (SD 10.91 months).

Sudbery et al.⁶⁶³ report on an older population with an age range of between 6 and 19 years, using purposive sampling to achieve a demographic mix of participants. Powell and Cheshire⁶⁶² conducted interviews with four non-abusing mothers and a grandmother of children aged 5–18 years who had been sexually abused.

Interventions

Services in the Cross et al.⁶³⁴ study differed across sites but all provided therapy to children, caregivers or both. Eight of the 15 sites provided a form of CPP,¹²⁹ often in addition to other services. Detailed information of the interventions is not reported but the reader is referred to other sources of information.

Osofsky et al.⁵¹⁵ describe a model of intervention designed to identify families with children at risk and provide clinical evaluation and treatment, with a view to enhancing children’s development. The treatment provided was CPP.

Sudbery et al.⁶⁶³ were focused on the holding therapy techniques used in a therapeutic residential setting for children, all of whom had been assessed as experiencing disordered attachment, some with an attachment disorder diagnosis.

The MOSAC Massage Programme (MMP) aims to equip mothers with simple massage routines that will ‘enable them to relax and calm their child, reintroduce positive touch in a safe environment, enable bonding/rebuilding of the mother-child relationship and work towards replacing memories of touch as fearful, painful, and distressing with memories of touch as loving, nurturing, and trusting’.⁶⁶²

Characteristics of treatment completers

Cross et al.⁶³⁴ examined five different predictors of study retention: demographics; violence exposure; child mental health; caregiver demographics; and engagement in intervention. Using logistic regression, they found, as in other studies, that those with older caregivers and those reporting higher levels of maltreatment were more likely to be retained. Physical health also had a relationship with retention, with those who rated their health as ‘poor’ or ‘fair’ also being more likely to maintain treatment. However, this study⁶³⁴ did not set out to examine treatment retention and thus the methodology falls somewhat short of exploring the factors related to treatment completion comprehensively, including motivation to change.

Osofsky et al.⁵¹⁵ reported that, of 129 child–caregiver pairs referred over 3 years, 75 were non-compliant from the outset or dropped out of treatment. Some families were court ordered and others were referred by child welfare or primary care providers. The authors observe that attrition is not surprising in samples in which substance abuse, parental mental illness or low functioning and homelessness are common. Of the 57 dyads that completed treatment, mother’s age at intake and maternal education (completed high school) were significantly correlated with treatment completion.

Acceptability

Location

The study⁶⁵⁶ was conducted in the USA.

Study design and sample size

Timmer et al.⁶⁵⁶ used social exchange theory as a framework within which to examine foster parents’ perceptions of their foster children, their relationships with them and their own functioning, comparing the views of 102 kin and 157 non-kin foster carers. The rationale for the study⁶⁵⁶ was that such perceptions might impact on the investment that foster carers make in their foster children, all of whom had been referred to PCIT on account of their behaviour problems.

Participants

The children were aged 2–8 years (M = 4.37 years) and were victims of multiple maltreatment. The majority were male (64%), cared for by women (95%) and one-third were Caucasian.

Intervention

The study⁶⁵⁶ report contains no detailed description of the intervention, but provides a brief outline of standard PCIT, that is, a two-phase therapy that begins with a focus on enhancing the parent–child relationships (CDirI) followed by a focus on enhancing child compliance (PDI). Both phases (which each last around 7–10 sessions) begin with a didactic component followed by therapist coaching, conducted by a ‘bug in the ear’ from a separate observation room.

Characteristics of treatment completers

Treatment completion and withdrawal metrics were reported. Dyads were considered to have completed treatment if they had mastered the relationship enhancement element of the programme, and could demonstrate sustained child’s compliance to commands and successful discipline approaches. If dyads did not attend treatment after the initial session, or if a parent chose to terminate therapy before the treatment goals had been met, then these were considered as early terminators. The average number of treatment sessions to treatment completion was 13.6 (SD 7.4). The average number of coaching sessions for dyads terminating early was 7.1 (SD 6.7). There were few demographic differences between the kin and non-kin carer groups.

Kin foster carers were significantly more likely to complete treatment than non-kin carers: 54.9% of kin carers completed treatment, compared with 36.9% non-kin carers completed [χ²(1, n = 259) = 8.09; p < 0.01].

Of the 145 foster carers who terminated treatment early, approximately two-thirds of kin and non-kin carers left treatment during phase 1 (CDirI). Of the rest, around half terminated treatment during the second phase (parent directed) and half never started. Of the early terminations, 40% were as a result of Child Welfare Services moving the child to a pre-adoptive home (more likely to happen in non-kin care settings). Some 43% were initiated by caregivers. In almost 28% of cases termination was triggered by the therapist and an 11% early treatment termination was categorised as ended by ‘other’ (not specified).

The results of binary logistic regressions suggest that kin caregivers with clinical levels of parental distress were more likely to stay in treatment than non-kin foster carers or kin caregivers scoring in the normal range on this indicator, although this effect was only marginally significant (p < 0.06). From their analyses the authors conclude that ‘parental distress’ explains some of the differences in overall attrition between kin and non-kin caregivers, and hypothesise that kin caregivers’ distress reflects their frustration and helplessness in the face of their foster children’s behavioural challenges and may motivate them to seek and continue with treatment.

Given that this was a study of foster parents, the authors note their concern about the relationship between elevated scores on the CAPI and early termination. If so, then this is yet another study that suggests that those who need help most are those who are least likely to access help or complete treatment. Similarly, those foster parents who avoided completing the Parenting Stress Index (PSI) were also more likely to leave treatment early, perhaps because they interpreted this as a measure of their mental health rather than that of their child, and therefore as a threat.

Location

Two^510,635 of the studies were based in the USA and one⁵¹¹ in the UK.

Study design

The study by Ducharme et al.⁵¹⁰ was a multiple baseline study of the effects of an intervention to help parents manage oppositional behaviour. The Golding and Picken⁵¹¹ study was a qualitative evaluation of two forms of group work. The Taban and Lutzker⁶³⁵ study was a study of parental satisfaction and acceptability of a parent training programme, exploring parental preference for different models of training.

Sample sizes

Sample sizes were small, with just 15 children from nine families in the Ducharme et al.⁵¹⁰ study and 44 children and 41 carers in the Golding and Picken⁵¹¹ study. In the study by Taban and Lutzker⁶³⁵ data were collected from the 31 parents provided with parent training in Project SafeCare.

Participants

Children in the Taban and Lutzker⁶³⁵ study ranged in age from birth to 5 years old (M = 4.9 years). The studies by Ducharme et al.⁵¹⁰ and Golding and Picken⁵¹¹ examined an older population aged between 3 and 12 years, and the foster carers in the Golding and Picken⁵¹¹ study were caring for school-aged children.

Participants in the Taban and Lutzker⁶³⁵ study were drawn mainly from the Latino population (68%). Golding and Picken⁵¹¹ suggest that the children fostered by their participants were white British (like the carers). No information is provided by Ducharme et al.⁵¹⁰

Children had been victims of physical abuse and witnessed domestic violence in the Ducharme et al.⁵¹⁰ study, and victims of physical abuse and neglect in the Golding and Picken⁵¹¹ study. Eighty per cent of those who received the parent interaction training in Taban and Lutzker⁶³⁵ were from a sample of maltreated children referred by the Department of Child and Family Services.

Intervention

Golding and Picken’s⁵¹¹ study compared two parent training interventions for foster carers. The first was based around the IY programme, with an additional psychoeducational component. This was compared with an intervention designed to develop emotional understanding and give skills in providing empathetic discipline, entitled Fostering Attachments. The IY intervention was delivered in 2-hour sessions over 9 weeks. Fostering Attachments was delivered in monthly, 2-hour sessions over an 18-month period.

In the Ducharme et al.⁵¹⁰ study, parents were taught ‘errorless compliance training’ in a five-session group format. Errorless compliance training is designed to improve children’s compliance while minimising non-compliance and associated risks of confrontation. The study⁵¹⁰ reported generalised improvements in compliance that were maintained at 6 months’ follow-up.

Project SafeCare was a 15-session programme that targeted home safety, infant and child health care, bonding and stimulation, and which included a parent–child interaction training component that was offered to those parents who needed it.⁶³⁵

Caregivers’ views

Using a five-point rating scale, mothers in the Ducharme et al.⁵¹⁰ study indicated a high degree of satisfaction with the intervention and therapist, and rated their children as being significantly more co-operative after treatment. However, as indicated above, almost half of the original sample did not complete treatment. Foster parents receiving the IY parent training intervention were more satisfied than those having the Fostering Attachments intervention in the study by Golding and Picken,⁵¹¹ with 84% saying that they found the programme to be very helpful, compared with 60% in the fostering attachment group. Only half of each intervention group stated that they had some increase in their understanding and confidence.

Participants in the study by Taban and Lutzker⁶³⁵ also reported positive feedback; parents reported high levels of satisfaction and training procedures and also rated staff highly. There are some limitations using these non-standardised measures.

Study design

The five^{526,533,621,673,678} studies examined family or multisystemic therapies. The Conran and Love⁶²¹ study was a case study, the Costa et al.⁶⁷⁸ study was an action research project, and the studies by Danielson et al.⁵²⁶ Tjersland et al.⁶⁷³ and Woodworth⁵³³ were uncontrolled studies (see Table 17).

Sample sizes

The FT/MST interventions all had small sample sizes, ranging from a single case⁶²¹ or samples of just eight⁶⁷⁸ and 10 families.⁵²⁶ The sample of Woodworth⁵³³ comprised 22 families completing treatment; Tjersland et al.⁶⁷³ had the largest sample of 31 families.

Location

The studies by Conran and Love,⁶²¹ Danielson et al.⁵²⁶ and Woodworth⁵³³ were US based, whereas the Costa et al.⁶⁷⁸ study was set in Brazil and the Tjersland et al.⁶⁷³ study was conducted in Norway.

Participants

Interventions in all five^{526,533,621,673,678} studies were directed at victims of sexual abuse. Danielson et al.’s⁵²⁶ sample had comorbid substance misuse. Woodworth et al.’s programme⁵³³ was for victims of incest. Participants in the Tjersland et al.⁶⁷³ study were referred by agencies concerned that a child aged < 18 years was being sexually abused by a family member. Participants in the Costa et al.⁶⁷⁸ study were extremely socially excluded, drawn from a population of settlement/dump dwellers with a high level of mobility.

Only Danielson et al.⁵²⁶ reported the age of participants (mean 15.0 years, SD 1.7 years).

Intervention

Conran and Love⁶²¹ did not provide any information about the FT intervention provided in the single case. Costa et al.⁶⁷⁸ briefly describes a Multifamilial Group Therapy intervention and Danielson et al.⁵²⁶ reported on RRFT, which was a combination of individual therapy and FT. RRFT was delivered by a university-based clinic and consisted of weekly 60–90 sessions over 14–34 weeks (M = 24 weeks, SD 8.0 weeks); participants were recruited through the university-based urban clinic, which specialises in adult/child trauma. The FT for suspected familial sexual abuse⁶⁷³ was delivered by one therapist, with a second therapist observing and reflecting on every session. It was anticipated that each family would be at crisis point at the start of treatment and this informed the development of the programme: narrative therapy techniques and relationship-building were used and could include confronting alleged perpetrators. The Multiple Family Incest Treatment Program⁵³³ provided support and therapy for all family members and included sibling support as one of its priorities, a group that they found particularly hard to reach.

Children’s views

Conran 1993⁶²¹ reports qualitative findings from a transcribed interview with the single female participant. Although not feeling forced to talk to the therapist, she said she found the two-way mirror uncomfortable at the start of treatment, but got used to it once treatment was established. She preferred IT and was ambivalent about group therapy. She also suggested that the therapist should take a more ‘child-like’ approach, by introducing games, jokes or tricks to engage the young person.

Costa et al.⁶⁷⁸ describe young people’s feelings of shame, anxiety, fear and pain and sadness; group therapy with other children with shared experience made it easier for the young people to talk about their problems.

The study by Danielson et al.⁵²⁶ was the only one to report results from a treatment satisfaction measure. Out of a total of 10 participants, nine completed ratings on the perceptions of the usefulness of treatment components: psychoeducation; coping/family communication; substance abuse; PTSD; healthy dating/sexual decision-making; and sexual revictimisation risk reduction. Each domain was rated positively. A total of 90% of participants completed all seven sessions; one participant ended treatment after five sessions.

Tjersland et al.⁶⁷³ collected data through observations from therapist sessions and follow-up interviews with children, mothers and alleged perpetrators of sexual abuse. In most cases, the abuse had not been substantiated at that time, and the majority of children were reluctant to discuss the abuse at follow-up interview. The reasons given for this reluctance included the following: they had been threatened by the abuser; they were afraid of upsetting their mother; and they feared not being believed. The majority of children displaying symptoms at the start of treatment had made progress by the end. Children expressed generally positive comments, and therapy observations reported positive exchanges between child and therapist.

Data were collected using interviews with 13 incest victims, 12 offenders and two siblings in the Woodworth⁵³³ study. Overall, three-quarters of victims found the group therapy to be helpful. The most commonly appreciated aspect of the groups was mutual support and the support provided by the counsellors. Two clients complained that counsellors left too soon. This may have been as a result of the use of interns to provide therapy (with shorter tenure), resulting in a negative impact on children who had formed bonds with them. Out of the 26 siblings, only three agreed to participate in the therapy. Reasons given for this 88% refusal rate were that mothers often refuse on behalf of non-victim children or that siblings were less convinced that therapy had benefits for them.

Caregivers’ views

Parents in the study by Costa et al.⁶⁷⁸ described how they felt their families were unprotected and vulnerable to further violence. Some expressed fears for their child’s future sexuality, including concern that the sexual abuse would negatively impact on the child’s sexuality (homosexuality) and how they might interact physically (i.e. sexually) with other children. The mothers in this action research study valued the group therapy with other families because it created an opportunity for them to talk to other women with similar experiences, but there was some criticism that treatment ended prematurely. Financial constraints impacted adversely on parents’ access to support – for most of the families the male perpetrator (and primary earner) had been removed from the home.

Tjersland et al.⁶⁷³ reported the conflicting interests expressed by participant mothers about treatment: they wanted help for their child but were concerned about revisiting the abuse by getting their child to talk about it. Concern for the alleged perpetrator was also observed: fear of criminal prosecution or negative reactions to the allegations; this was particularly relevant if an immediate family member (husband or son) had been implicated in the abuse. Some mothers felt vulnerable to being perceived negatively by the therapist because of their implicit role in the abuse, or acting in an over-protective way. In client satisfaction ratings, mothers were generally very contented with the treatment. In the Woodworth⁵³³ study, 83% of mothers found the group therapy to be helpful but they also favoured more direct confrontation with perpetrators and smaller group work.

Mothers were generally satisfied with the therapy received in the study by Woodworth 1991,⁵³³ with 83% feeling that it had helped them personally. Three-quarters of respondents in this study⁵³³ considered that the multiple-family group (several families meeting together for therapy) had been helpful. Comments indicated that some respondents thought that the group needed more guidance from the facilitators, and that offenders should have been confronted more, and some thought that the group was too large.

Staff views

Staff involved in the Costa et al.⁶⁷⁸ study recognised the limitations of the intervention in providing protection to vulnerable young people living in a potentially dangerous environment. They stressed the need for a wider network of support for these socially excluded families, ranging from the extended family of parents, grandparents to the social institutions responsible for their care and supervision during the investigation of child abuse. The process of dealing with criminal justice system can be humiliating and may have implications for the wider family network, by witnessing ongoing contact with the police, hospital staff, forensic teams and court officials.

Alleged perpetrators

Alleged perpetrators in the Tjersland et al.⁶⁷³ study were confronted about the abuse during treatment, one-third of whom were unaware of the suspicions prior to therapy starting. Reactions to the allegations presented elicited three different kinds of response: confirmation of the abuse; abuse was denied and the alleged perpetrator withdrew from the mother and child; and abuse was denied but the alleged perpetrator tried to maintain contact with the family. Six of the alleged perpetrators rated the treatment positively, and valued the objective role of the therapist. Those who were critical of the therapy (n = 2, an additional n = 2 were both contented and discontented) were unhappy that they had not been involved from the start and felt that the therapists had formed a coalition with the mother. At the end of treatment, conflicts associated with the question of abuse had been clearly reduced in 20 cases; three families were still facing significant conflict, with two cases brought to court.

In the Woodworth⁵³³ study, the offenders were by far the most positive in their satisfaction with the programme, with 88% saying that they were ‘strongly satisfied’ and 83% describing the therapy as ‘very helpful’ to them personally.

Study design

The study by Hyde et al.⁵³⁸ was related to a randomised trial.¹⁶⁹ Although the carers in the Rushton and Miles⁶⁶⁴ study were not randomised, the study was part of a trial in which the sexually abused adolescent girls for whom they cared were randomised, and the carers offered either a carers’ group or individual support.¹¹⁶ The study by Barth et al.¹⁶¹ was a COS with a control group. The Boisvert et al.⁶⁵⁸ study was an uncontrolled study, designed to investigate attrition rates amongst sexually abused children who were referred to mental health services.

Sample sizes

The total sample in the Barth et al.¹⁶¹ study was 27, with 15 foster carers assigned to the intervention group. Hyde et al.⁵³⁸ had a sample of 47 adolescents and their families and Rushton et al.⁶⁵⁸ had a sample of 65 carers. Boisvert et al.⁶⁵⁸ analysed data relating to 116 adolescents.

Location

The study by Boisvert et al.⁶⁵⁸ was based in Canada, whereas the Barth et al.¹⁶¹ study was based in the USA. The studies by Hyde et al.⁵³⁸ and Rushton et al.⁶⁶⁴ were conducted in the UK.

Participants and maltreatment

All participants had a history of sexual abuse. Participants in the Hyde et al.⁵³⁸ study were the youngest, ranging from 4 to 16 years. Both Barth et al.¹⁶¹ and Boisvert et al.⁶⁵⁸ report data relating to adolescents aged 12–17 years. Although Rushton et al.’s⁶⁶⁴ study was focused on carers, the children in the original trial were aged 6–14 years at recruitment.

Ethnicity was reported for the participants in the Barth et al.¹⁶¹ study (69% black people). Carers and mothers in the Rushton et al.⁶⁶⁴ study were largely white and UK born (75%), with another 10% being African Caribbean and 7% Mediterranean.

The studies by Barth et al.¹⁶¹ and Hyde et al.⁵³⁸ had a predominantly female sample. The Boisvert et al.⁶⁴⁷ study did not provide any data on gender.

Interventions

All of the interventions were group based, although Barth et al.¹⁶¹ also included some individual work, and Rushton et al.⁶⁶⁴ compared group-based support with individual support. Hyde et al.⁵³⁸ also incorporated some family network meetings.

Groups typically ran for at least 8 weeks and, for some young people, groups ran for around 20 weeks.^538,647 In the Rushton et al.⁶⁶⁴ study, treatment was planned for 30 weekly sessions for the girls, and the work with carers lasted for the same duration, but the authors note that it was not uniform because of limited resources. Birth and adoptive parents were usually seen weekly; foster carers were usually seen fortnightly.

Children’s views

The rating scale used to measure participant satisfaction is not described in Hyde 1995⁵³⁸ but, as indicated above, ratings were supplemented with qualitative interviews. Feedback was generally positive, but fewer than half the children were positive about talking about the abuse or felt that, as a result of the group work, they understood the origins of the abuse any better. Seventy-eight per cent said that they did find it useful for preventing further abuse and dealing with feelings of guilt. The helpfulness of treatment was generally rated higher by children than by their mothers.

Children in this study⁵³⁸ generally welcomed the opportunity provided by the groups to meet with others with similar experiences, but not all, and less than half of those interviewed felt positively about talking about the abuse.

Neither the Boisvert et al.⁶⁵⁸ study nor the Rushton et al.⁶⁶⁴ study report the views of children.

Caregivers’ views

A brief client satisfaction questionnaire was given to foster parents in the study by Barth et al.¹⁶¹ and the programme received high levels of endorsement for the group, but the length and intensity of the intervention were insufficient to observe any measurable changes of effectiveness.

Rushton et al.⁶⁶⁴ report that most carers were positive about the support provided to them (30% ‘very beneficial’, 48% ‘beneficial’). Mothers who were still in a relationship with the abuser were more likely to have negative or mixed views of the help provided to them and their daughter. Analyses indicate that those who reported positively on the help provided to them attended for an average of 8.8 months compared with those who had a mixed negative response, who attended just half of this time (4.3 months). This might mean that mothers who were not helped dropped out sooner, or that those who attended fewer sessions (perhaps for different reasons) did not receive enough help to find it beneficial. The authors note that both foster carers and adopters attended for significantly longer periods (M 9.2 months) than the birth parents (M 6.1 months), although, when levels of satisfaction were explored, birth mothers appeared to benefit more than foster carers, but the difference was not significant; few respondents said they received little benefit and foster carers were few in number.

The authors hypothesise that ‘mothers who clearly valued the support provided would probably have benefited from an independent professional listening to their difficulties and dealing with feelings such as guilt and anxiety’⁶⁶⁴ (p. 425), which, in turn, may have prevented deterioration in their relationships with their children. This study⁶⁶⁴ was primarily designed to examine the relationship between kinds of support to carers and outcomes for children hence the rather speculative reflections on what the help meant to carers themselves.

Study design

All but two^{171,636,637,639,676} studies were uncontrolled studies. The De Luca et al.¹⁷⁰ study was a COS, and the Monck et al.¹⁶⁹ study was a quasi-randomised trial.

Sample sizes

Sample sizes were all small, ranging respectively from just six and nine, respectively, in the studies by Grayston and De Luca¹⁷¹ and Ashby et al.,⁶³⁶ to a sample of 95 in the study by Monck et al.¹⁶⁹

Location

Three^636,637,639 studies were based in North America and two^170,171 in Canada. The study by Gustafsson 1995⁶⁷⁶ was conducted in Sweden, and the Monck et al.¹⁶⁹ study was UK based.

Participants

Interventions were delivered to children as young as 3 years, and up to 20 years. In five^{169–171,636,637} studies, all of the children had been sexually abused. Children in the Peled and Edleson⁶³⁹ were in treatment as a result of witnessing domestic violence. The participants in the study by Gustafsson 1995⁶⁷⁶ had suffered physical abuse, parental alcohol misuse and had witnessed domestic violence.

In two^170,636 studies the participants were sexually abused girls and in the study by Grayston and De Luca¹⁷¹ the participants were sexually abused boys. The remaining four^169,639,676 studies had mixed gender groups. Ashby et al.⁶³⁶ describe a population that was 100% American Indian, referred by tribal social services.

Intervention

Interventions comprised group activities (including art activities/circle time), abuse prevention skills, family reunification therapy and psychotherapeutic approaches. Interventions were delivered in group settings for children with similar abuse histories or, as in one study,⁶³⁷ a sibling/victim group setting.

In four^{170,171,636,639} studies treatment lasted between 10 and 12 weeks. Treatment in Monck et al.¹⁶⁹ could last up to 12 months. No information was available in the remaining two studies.^63,637

Children’s views

Using a child feedback questionnaire, Grayston and De Luca¹⁷¹ found that most children found the group helpful, enjoyed attending, liked feeling safe and were satisfied with the level of parental involvement. They had no suggestions for changes. No objective measures were used to assess satisfaction levels, and this is a limitation of this study.¹⁷¹

Participants in the studies by Ashby et al.,⁶³⁶ Monck et al.¹⁶⁹ and De Luca et al.¹⁷⁰ rated the programmes positively. On a scale of 1–11 (11 = outstanding), the group treatment programme in the Ashby et al.⁶³⁶ study scored 9.8 on average. Data from school counsellor reports saw positive behaviour change in school for 70% of participants. Children rated therapists highly in the Monck et al.¹⁶⁹ study and valued meeting others with similar experiences; they also valued being able to talk to their abuser. Some reported negative feelings about talking about the abuse, their family and the hospital location of the therapy. The majority felt that it was helpful in preventing further abuse, raising their self-esteem, understanding and feelings of guilt surrounding the abuse. There was mixed effects for relations with their family, planning for the future and understanding the origins of the abuse. The children in the De Luca et al.¹⁷⁰ study also reported similar positive and negative responses, plus some elements that frightened them, for example using puppets and the idea that abuse could recur.

Child-reported benefits of treatment identified in the study by Peled and Edleson⁶³⁹ included self-protection and strengthening self-esteem.

Caregivers’ views

Parents surveyed in the study by Baker et al.⁶³⁷ viewed the sibling group intervention positively, and, in the 2000 survey, parents’ mean score rating was 1.9 (on a scale of 1 to 4, with ‘4’ = unsatisfactory). Parents felt that their children had learnt how to deal with inappropriate advances (score 1.5) but helping the child to cope with stress was rated less positively (score 2.8).

Monck et al.¹⁶⁹ also report mothers rating the prevention of further abuse as helpful, but the intervention fared less well when trying to deal with issues including understanding why and accepting abuse has happened, resolving guilt relating to the family and managing the abused child. In the De Luca et al.¹⁷⁰ study, parents believed that the children liked feeling understood and having somewhere to talk about the abuse and someone to talk to, but would have liked to have received more feedback or observe the treatment.

Staff views

The importance of involving siblings in treatment is discussed by Baker et al.⁶³⁷ – siblings have a high risk of being abused too and often there are unresolved feelings of anger, jealousy and guilt, particularly if a family member is the perpetrator. On a purely practical level, involving all family members enables therapy to happen as no child-care issues arise. Sibling therapy also adds to the costs of the treatment, which may not be covered by the provider; it also has implications for rooms, materials and staffing.

Study design

The study reported by Haight et al.¹⁷⁵ was a randomised trial.

Baginsky⁶⁴⁰ conducted a review of the pattern of provision in the UK, the Netherlands and Italy and the reaction of young people who had or had not received services. The literature review was followed by interviews (both face to face and telephone), group discussions, questionnaires and letters to collect additional data.

Deb and Mukherjee⁶⁷⁹ used purposive sampling from four randomly selected shelters across Kolkata, India, and sourced a non-abused control group from local schools, which were also randomly selected. Both qualitative and quantitative data collection were used.

The other studies^{176,638,641–643,645,646,674,677} were uncontrolled designs, each using a purposive sample of those engaged in treatment. The studies by Fowler et al.^641,642 examined the acceptability of counsellor gender for treatment for sexual abuse. Kilcrease-Fleming et al.⁶⁴³ also investigated counsellor gender using a standardised rating scale of video-taped interviews analysing differences in verbalisation between male and female counsellors. Porter et al.⁶⁴⁵ used the Client Behavior System verbalisation measure to assess gender differences in counsellors. Thompson et al.⁶⁴⁶ used two semistructured interview guides to interview both mothers and youths. Scott ⁶⁷⁷ also conducted in-depth interviews with parents. Nelson-Gardell⁶³⁸ conducted focus groups. Overlien’s⁶⁷⁴ investigation of counselling provision in women’s refuges used a grounded theory approach to conduct face-to-face interviews while using age-appropriate schedules. Reddy et al.¹⁷⁶ relied on qualitative post-treatment feedback to assess intervention acceptability.

Kolko et al.⁶⁴⁴ conducted quantitative analyses to predict service use and Haskett et al.⁶⁵⁷ used regression analysis to investigate treatment entry.

Sample sizes

Sample size varied. Haight et al.¹⁷⁵ recruited 17 children from 10 families, and Scott⁶⁷⁷ recruited 15 children from 12 families. Nelson-Gardell⁶³⁸ recruited 34 participants, and four^{641,643,645,674} studies had samples of around 50. The largest study⁶⁴⁰ had 130 participants. The studies by Fowler and Wagner,⁶⁴² Haskett et al.,⁶⁵⁷ Reddy et al.,¹⁷⁶ Kolko et al.⁶⁴⁴ and Deb and Mukherjee⁶⁷⁹ all had between 70 and 100 participants.

Location

Most of the studies were USA based. One⁶⁷⁷ was set in Australia, one⁶⁷⁹ in India and another in Norway.⁶⁷⁴ Baginsky⁶⁴⁰ examined counselling provision for young people across three nations: the UK, the Netherlands and Italy.

Participants

All studies focused on participants with a sexual abuse history, apart from Overlien,⁶⁷⁴ who examined physically abused children who had witnessed domestic violence. The sample in Kolko et al.⁶⁴⁴ had also been subjected to neglect, and those in the studies by Haight et al.¹⁷⁵ and Reddy et al.¹⁷⁶ had been exposed to a range of abuse and neglect. Baginsky⁶⁴⁰ did not specify type of abuse, but sexual abuse recovery was included in the findings; therefore, an assumption has been made that at least part of the sample had been sexually abused.

Interventions

Four^{644,674,677,679} studies reported on IGT – concurrent with individual counselling. The group therapy included counselling with other sexually abused girls,⁶⁷⁹ family members^644,677 and play therapy within a women’s refuge setting.⁶⁷⁴

Five^{640–643,645,657} studies examined individual counselling, four of which specifically analysed counsellor gender preferences pre and post treatment.^{641–643,645} Haskett et al.⁶⁵⁷ examined factors associated with successful treatment entry for long-term counselling and, last, Baginsky⁶⁴⁰ reviewed varied counselling provision in Europe.

Counselling for sexually abused girls in Kolkata⁶⁷⁹ was based around basic support services, as many of these girls had been living on the streets and had been sexually exploited; provision included nutrition, safety and security, education and training, and medical care, as well as counselling. The intervention lasted between 2 and 3 months, and individual and group counselling were delivered on a weekly basis, with more if required.

Participants in the study by Nelson-Gardell⁶³⁸ had received therapy from a range of counsellors, the details of whom are not provided. In this study⁶³⁸ the researchers wanted to know what and whom the participants had found helpful in recovery.

Fowler and Wagner,⁶⁴² Kilcrease-Fleming et al.⁶⁴³ and Porter et al.⁶⁴⁵ all report a psychoeducation/psychological treatment programme, which lasted for six sessions. Adolescents in the study by Reddy et al.¹⁷⁶ were provided with CBCT. Haight et al.¹⁷⁵ describe the LSI for children who were living with parental methamphetamine misuse.

Haskett et al.⁶⁵⁷ describe the intervention as ‘long-term counselling’. Additional content of counselling intervention is not described in detail.

Young people in the study by Thompson et al.⁶⁴⁶ received a range of individual and group-based counselling services in a range of service settings. The families of these participants also received a range of other counselling services, as well as – for some – FT, drug counselling and inpatient services (mothers).

Characteristics of treatment completers

Only two studies^644,657 examined the characteristics of those who remained in therapy or dropped out.

Haskett et al.⁶⁵⁷ presented factors associated with successful treatment entry for long-term counselling in a convenience sample. A higher percentage of males attended, as did a higher percentage of white Americans, but attenders and non-attenders did not differ in parental education level, marital status or socioeconomic status (SES). Children in homes with telephones were more likely to attend the first session, as were those referred to a private centre. When mothers felt that the entire family needed counselling, attendance was also more likely.

In the study by Kolko et al.,⁶⁴⁴ children and parents were interviewed at study intake and at 4–8 months after receiving an initial service. Potential predictors of service use were computed using Pearson‘s correlations or chi-squared tests to determine the relationships between several key clinical characteristics. Four variables were found to be significant, and these were used to perform multiple regression analyses. Four variables predicted the number of services received at intake: white American children with lower levels of anxiety and parents with heightened distress and with more abusive experiences when they themselves were children received more services at intake. Three of these variables also predicted number of services at post-service assessment: white American child, parental distress and low child anxiety.

Acceptability

Study design

All studies^{647,648,665,675} were uncontrolled. In the study by Horowitz et al.⁶⁴⁷ the data analysed were collected as part of a longitudinal study of the psychobiological effects of CSA (Putnam and Trickett, 1987–1988⁷⁴⁸). The studies by Davies et al.⁶⁶⁵ and Jensen et al.⁶⁷⁵ used qualitative methodologies, and Lippert et al.⁶⁴⁸ reviewed case records with additional qualitative data collection.

Sample sizes

There were just four participants in the Davies et al.⁶⁶⁵ study and 15 in the Jensen et al.⁶⁷⁵ study. The other studies^647,648 had samples sizes of 81 participants⁶⁴⁷ and 101 participants,⁶⁴⁸ respectively.

Location

The studies by Horowitz et al.⁶⁴⁷ and Lippert et al.⁶⁴⁸ were US-based studies, the Davies et al.⁶⁶⁵ study was a UK study and the Jensen et al.⁶⁷⁵ study was set in Norway.

Participants

Sexual abuse history was the maltreatment experienced by children in three^647,648,675 studies, and in the Davies et al.⁶⁶⁵ study the four participants had been neglected/abused. The sample was 100% female in the studies by Jensen et al.⁶⁷⁵ and Davies et al.,⁶⁶⁵ whereas in the study by Horowitz et al.⁶⁴⁷ 60% of participants were female. Lippert et al.⁶⁴⁸ did not present a gender breakdown. Forty per cent of the participants in the study by Horowitz et al.⁶⁴⁷ were described as non-white and the entire sample in the Jensen et al.⁶⁷⁵ study was of Norwegian ethnic origin.

Intervention

Davies et al.⁶⁶⁵ and Jensen et al.⁶⁷⁵ describe individual psychotherapy, but details of intervention delivery are not reported in the studies by Horowitz et al.⁶⁴⁷ or Lippert et al.⁶⁴⁸ Children in the Davies et al.⁶⁶⁵ study had been in receipt of psychotherapy for between 4 months and 3.5 years, and in the Jensen et al.⁶⁷⁵ study weekly sessions were provided for a mean of 7.5 weeks.

Characteristics of treatment completers

Horowitz et al.⁶⁴⁷ reported that non-minority children received more therapy. Abuse variables were found to be powerful predictors of the total number of therapy sessions, and earlier onset predicted more sessions. Children who experienced higher levels of psychopathological disturbance also received more treatment. Family functioning did not predict level of treatment in the model.

Lippert et al.⁶⁴⁸ profiled those who failed to participate in treatment: 46% of the sample of 101 did not begin therapy and 54% had at least one therapy session (therapy initiators). Initiators of therapy were less likely to be ethnically black (33%) than decliners (50%), and were more likely to have been subject to maternal neglect (24%) than decliners (4%). Decliners were those whose first appointments were twice as long from the initial forensic interview following abuse report. Caregivers who declined treatment reported lower scores on the Self-Report Family Inventory (SFI) conflict, competence and expressiveness scales. Reasons for declining included ‘work conflict’ (50%); ‘inaccessible venue’ (40%); ‘child was symptom free’ (15%); ‘caregiver was busy’ (15%); and ‘caregiver wanted to forget about abuse or let their child forget’ (15%).

Acceptability

Interviews exploring children’s experiences of therapy were conducted in the study by Davies et al.⁶⁶⁵ Jensen et al.⁶⁷⁵ interviewed children and their caregivers, separately, at two different points in time: just after the last therapy session and 1 year later. Lippert et al.⁶⁴⁸ relied on parents’ accounts, and presented a profile of non-participation and data from case record reviews. Horowitz et al.⁶⁴⁷ collected data from therapists’ reports and ran multiple regression analyses to examine the correlates of therapy usage.

Children’s views

Young people ‘enthusiastically endorsed’ (p. 70) provision and found the outreach service to be very helpful. Staff were available by phone during evenings, which was valued. One-to-one support was considered to be the most important element and the youth-led support group was less favourably viewed.

Caregivers’ views

Parents found staff to be respectful and sensitive and identified the outreach service as unique. Parents felt that the parent-led group gave them information and coping strategies and appreciated hearing that they were not alone.

Staff views

Staff felt that the group filled services gaps and that no other agency provided similar support.

Location

One⁶⁷² study was based in the Netherlands and three in the USA.^649–651 The Gallagher and Green⁶⁶⁶ study was undertaken in the UK.

Study design

Cunningham et al.⁶⁴⁹ described the development of a measure of youth engagement that was suitable for use with young people in RTCs. In collaboration with staff from two RTCs, the research team established a programme logic model, which they used to develop a multidimensional measure of engagement, adapting items from existing measures of readiness to change and the therapeutic alliance. The tools were then piloted drawing on data from interviews with young people at four time points, interviews with primary caregivers at the first and last time points, questionnaires to the school, clinical and residential members of the young people’s treatment teams, and data from the client’s case files and school records. Confirmatory factor analysis using maximum likelihood estimation was the primary analytic method used for analysis, and informed subsequent modifications of the measure.

Leenarts et al.⁶⁷² examined motivation for change among girls in compulsory residential care, using a range of standardised measures of child maltreatment, trauma and treatment motivation, which they analysed in relation to motivation for treatment, using multiple linear regression analyses, and treatment dropout, using logistic regression.

Both Shennum and Carlo⁶⁵⁰ and Gallagher and Green⁶⁶⁶ used semistructured interviews. Both interviewed children who had previously lived in therapeutic residential care; the sample in the Shennum and Carlo study⁶⁵⁰ included some children still resident at the time of interview.

West et al.⁶⁵¹ used focus groups to explore the views of young people in a school that they attended under court order.

Sample sizes

A total of 154 adolescent girls participated in the Leenarts et al.⁶⁷² study and 130 young people were interviewed on four occasions by the researchers in the study by Cunningham et al.⁶⁴⁹ Shennum and Carlo⁶⁵⁰ recruited a sample of 80 young people and Gallagher and Green⁶⁶⁶ achieved a sample of just 16. Thirty-nine girls participated in focus groups in the study by West et al.⁶⁵¹

Participants

Characteristically, the children served by these interventions were described as ‘unfosterable’ or difficult to place, because of behavioural issues or physical and psychological conditions and/or disability.

The young people in the study by Gallagher and Green⁶⁶⁶ had experienced severe sexual, physical and emotional abuse and neglect, by one or more members of their family, and sometimes others, which had left them with significant problems of attachment. Previous placements had broken down, often because of challenging (including sexualised) behaviour. Some had experienced subsequent maltreatment in foster care.

Shennum and Carlo⁶⁵⁰ described residential facilities that were providing care for maltreated young people who were also presenting with behavioural and conduct problems.

Young people in the study by Cunningham et al.⁶⁴⁹ had a somewhat different profile in that the majority were in residential care (‘congregate foster care’) as the result of being in need of supervision (53%), and it is not entirely clear whether or not they had a history of maltreatment. A further 38% were adjudicated delinquents and 9% had been referred for reasons including abuse, neglect and special educational needs. It is possible that – with additional information – this study⁶⁴⁹ would fail to meet our inclusion criteria in respect of participants.

Those in the Leenarts et al.⁶⁷² study were in compulsory residential treatment facilities. All had experienced prior traumas, and their histories were characterised by several out-of-home placements (60%), homelessness (30%), police contact of family members (45%) and histories of physical or psychological problems of family members (62%).

The girls in the West et al.⁶⁵¹ study were maltreated girls who were involved in the criminal justice system.

Interventions

The therapeutic residential settings ranged from a compulsory treatment facility for severely traumatised girls^651,672 through residential treatment^649,650 to a small domestic-style setting described as a Therapeutic Children’s Home,⁶⁶⁶ where children lived in ‘families’ of three or four children with two adult staff acting in loco parentis. In this setting, the therapeutic model comprised three components: therapeutic parenting (to address attachment issues and a secure base), formal therapy sessions (based on play and expressive arts) and life story work.

The larger residential facilities provided a range of individual and group therapies as well as education. Typically, the environment was structured in ways designed to promote prosocial and to adaptive behaviour.^649,650,672

The treatment setting in West 2014⁶⁵¹ was a residential school, which offered a modified training curriculum (The Heart of Teaching and Learning: Compassion, Resiliency and Academic Success) and a ‘Monarch Room’ facility, which promoted emotion regulation and skills to de-escalate problem situations through problem-solving techniques, talk therapy and sensorimotor activities, and avoid student suspensions and expulsions, which are recognised as counterproductive.

Characteristics of treatment completers

One⁶⁴⁹ study specifically examined factors associated with engagement, but does not provide relevant data, as the study concerned the development of a measure appropriate to residential treatment settings. We discuss the issues raised by Cunningham et al.⁶⁴⁹ in our concluding discussion. One⁶⁷² study explored the factors associated with treatment engagement and dropout, and, given the risks associated with running away from residential care, this study – although conducted in the Netherlands – addresses an important UK-wide issue.

Leenarts et al.⁶⁷² report that several demographic variables predicted motivation for treatment, as assessed by the Nijmegen Motivation List 2 (NML-2⁷⁴⁹). The 34 items in this self-report questionnaire ask respondents to answer using a five-point Likert-type scale, ranging from one (‘not at all applicable’) to five (‘highly applicable’). The NML-2 generates three subscales: (1) preparedness to engage in treatment, (2) level of distress and (3) doubt about treatment. Data on dropout consisted of a total of five possible outcomes: ‘client left: runaway’; ‘judge did not extend stay’, ‘transfer to another facility’, ‘regular termination: end treatment’ and ‘stay not terminated: adolescent is still a resident’. Girls who terminated their stay by running away were identified as dropouts, that is, those who ran away and stayed away for > 14 days and, as a result, were discharged from the facility. In this study,⁶⁷² 23 girls (15%) ended their first uninterrupted stay by dropping out. One girl was transferred to another facility but dropped out after return, resulting in a total number of girls dropping out of 24. Girls with a non-Dutch ethnic background and a younger age reported significantly higher levels of distress and were more likely to engage in treatment.

Level of distress was predicted by a history of out-of-home placements when considering demographics only, and predicted doubt about treatment when considering demographics and childhood maltreatment. The authors point out that out-of-home placements and separating children from their parents may adversely affect their functioning. Out-of-home placements no longer predicted level of distress and doubt once emotional abuse, anxiety, depression and dissociation were taken into account. The authors conclude that the relationship between out-of-home placements and motivation is mediated by emotional abuse and trauma-related symptoms.

Emotional abuse was the type of maltreatment most strongly correlated with motivation to engage with treatment. Girls who reported internalising symptoms (anxiety, depression) were more likely to experience high levels of distress than those with fewer such problems. Girls with dissociative symptoms were more likely to have doubts about treatment. Adolescents are generally more willing to change their internalising problems than their externalising problems, and the authors point out that dropping out of treatment by running away may be attributable to externalising symptoms and antisocial behaviour. They go on to suggest that this is perhaps why the study⁶⁷² did not find a significant association between dropout and a history of child maltreatment. The authors suggest that as dropout often occurs when adolescents are on leave from residential care; future research should investigate whether or not going on leave adversely affects girls’ motivation for treatment and also the relationship between motivation to change and motivation for treatment.

Acceptability

Three^650,651,666 of these five^{649–651,666,672} studies specifically explored the view of children and young people.

Location

The studies by Staines et al.⁶⁶⁷ and Biehal et al.^145,146 were conducted in the UK. The Laan et al.⁶⁷¹ study was undertaken in the Netherlands.

Study design

Staines et al.⁶⁶⁷ used a prospective, repeated-measures design to investigate the supports and services provided to children and carers in an Independent Fostering Agency (IFA), and the relationship between these and children’s progress and placement outcomes over a 12-month period. They used questionnaires to obtain data from carers, children and social workers, at two time points (at the start of a placement and 1 year later). The included paper reports the views of foster parents.

Biehal et al.^145,146 undertook a small randomised trial of the effectiveness of MTFC, embedded in a larger, observational QEx case–control study (see Chapter 4 for details of the RCT).

Laan et al.⁶⁷¹ used data from case notes, together with data from a questionnaire completed by foster carers, to explore the characteristics of children included in an enhanced fostering programme, the content of counselling provided within the service, placement outcomes, and relationships between children’s characteristics and placement outcomes.

Sample size

The achieved sample in the study by Biehal et al.^145,146 was 219 participants (with 34 participants in the RCT). Laan et al.⁶⁷¹ examined case files for 78 children, and secured questionnaire data from 64 of the 78 foster parents. Staines et al.⁶⁶⁷ received completed questionnaires from 49% (221) of the IFA foster carers and 66% of the IFA social workers (299) at time 1 – when child was first placed. At time 2 – either 1 year following the start of the placement or when the placement ended, the team secured completed questionnaires from 50% (227) of foster carers and 69% (312) of the IFA social workers. For only 138 placements at time 1 and 80 placements at time 2, were completed questionnaires received from both IFA social workers and foster carers

Participants

Respondents in the study by Laan et al.⁶⁷¹ were foster parents looking after learning disabled children with challenging behaviour. It was the only study⁶⁷¹ of disabled children identified. Participants in the study by Biehal et al.^145,146 were children and young people in foster care, aged 10–17 years, who were showing complex or severe emotional difficulties or challenging behaviours, and whose placements were unstable, at risk of breakdown, or not meeting their assessed needs. Children in the study by Staines et al.⁶⁶⁷ were aged 5–14 years, who had been in a placement and provided by the IFA participating in the study for > 1 year. All of the children had been maltreated, with most having experienced more than one form of abuse.

Interventions

Biehal et al.^145,146 evaluated Multidimensional Treatment Foster Care for Adolescents (MTFC-A), described as a ‘wrap-around multimodal foster care intervention for children with challenging behaviour’.

The IFA in the study by Staines et al.⁶⁶⁷ incorporated a therapeutic approach to its service provision, which recognised the importance of individual therapeutic work with children, but focused the efforts of therapists on supporting foster parents and other staff within the agency. As the authors note, local authorities typically use IFAs for their more difficult-to-place children and this, together with the therapeutic focus, is why this study⁶⁶⁷ was categorised as one that was concerned with enhanced fostering provision.

Intensive Foster Care [Project Intensieve Pleegzorg (PIP): project for intensive foster care] was the focus of Laan et al.⁶⁷¹ PIP provided foster carers with intensive and specialised counselling by a counsellor who also had access to an educational psychologist, and a psychiatrist or psychotherapist from a multidisciplinary PIP support team.

Characteristics of treatment completers

In Laan et al.’s⁶⁷¹ analysis, intensive foster care placements were more likely to end prematurely for girls, for children with psychiatric problems and for children who had experienced neglect or sexual abuse in their biological family.⁶⁷¹ Staines et al.⁶⁶⁷ and Biehal et al.^145,146 do not present systematic data on this issue.

Acceptability

Of the three studies, Biehal et al.^145,146 was the only one to canvass children’s views of the intervention. All three studies considered the views of staff and caregivers.

Study designs

All five studies were qualitative studies. The study of art therapy⁶⁵¹ was primarily a descriptive account of a group for children in women’s refuges, but included analyses of children’s drawings and reports of the children’s written evaluations of their group experiences. The study of equine-assisted therapy⁶⁶⁹ incorporated participant observation, field notes and interviews (semistructured, ethnographic conversational plus unstructured interviews). Bannister and Gallagher⁶⁶⁸ investigated the case histories of children referred to the NSPCC. Hill⁶⁷⁰ examined case records, supplemented with 48 interviews with parents, therapists and children. Mishna et al.⁶⁶¹ interviewed children’s parents and professionals using semistructured interviews at 6, 12 and 18 months following the start of treatment. Both Burgon⁶⁶⁹ and Gilbert⁶⁵² wrote from the perspective of practitioner researchers.

Samples

Samples were extremely small: just seven children in the study by Burgon⁶⁶⁹ and six in the Bannister and Gallagher⁶⁶⁸ study. Mishna et al.⁶⁶¹ interviewed the parents, teachers and therapists of 11 children who were undergoing play therapy. Hill⁶⁷⁰ examined the cases of 13 children who were seen by four therapists. No sample size was available in the study by Gilbert.⁶⁵²

Setting

Bannister and Gallagher,⁶⁶⁸ Burgon⁶⁶⁹ and Hill⁶⁷⁰ conducted their studies in the UK. The studies undertaken by Gilbert⁶⁵² and Mishna et al.⁶⁶¹ were based in North America.

Participants

The seven children in the study by Burgon⁶⁶⁹ were in foster care; they had all experienced multiple abuse and presented with additional problems, such as school exclusion or involvement with youth justice. Those described by Bannister and Gallagher⁶⁶⁸ were children who had, themselves, been sexually abused and who were sexually abusing other children.

The children in the study by Gilbert⁶⁵² had been exposed to domestic violence exposure, and those in the study by Mishna et al.⁶⁶¹ included children who had been exposed to domestic violence, plus children who experienced serious verbal, physical and/or sexual abuse, neglect by parents or neglect prior to international adoption by their present parents. The children in the study by Hill⁶⁷⁰ had also been subjected to physical and/or sexual abuse and neglect.

Intervention

The length of treatment varied with each intervention and population but in three of the studies, the therapy could last up to 2 years.^661,669,670 The weekly group art therapy reported in the Gilbert⁶⁵² study ran for 8 weeks.

In the study by Bannister and Gallagher⁶⁶⁸ the treatment could last up to 8 months, although one child withdrew from treatment after 6 weeks, and, at the time of the study, some children were still receiving treatment.

Two interventions were based around creative activity.^652,668 The play therapy interventions described in the studies by Hill⁶⁷⁰ and Mishna et al.⁶⁶¹ also involved parents. Burgon⁶⁶⁹ helped to deliver an equine-assisted therapeutic intervention.

The intervention in the Bannister and Gallagher⁶⁶⁸ study drew on art, play and drama therapy techniques, and included an educative–behavioural intervention to treat offending behaviour; carers were involved as much as possible in the treatment. Two of the six children were seen for 3 months, purely for assessment, but the authors regarded assessment as intrinsically therapeutic.

Children’s views

The ethnographic study of equine-assisted therapy by Burgon⁶⁶⁹ presented some data on children’s experiences of this therapeutic approach. Working as practitioner–observer, Burgon⁶⁶⁹ identified some positive feedback from the seven children with whom she worked, interpreting these experiences as empowering for them. One young person expressed this as follows: ‘[the horse] kind of made me feel like, you know, I’m the queen of the world kind of thing because I was higher up’⁶⁶⁹ (p .171). Another child described how she learnt to deal with feelings of anger because she knew that she had to be calm around the horses in case she frightened them. Two other participants demonstrated how trying something new had helped them explore new opportunities with the confidence that they had gained from riding horses. These young people had gone on to begin training in an equine-related career.

Using written feedback from the final meeting of an art therapy group housed in a women’s refuge, Gilbert⁶⁵² gives positive examples of some children’s experiences, but the data are sparse and it is difficult to draw any conclusions about either acceptability or effectiveness for this group of children who had witnesses domestic violence. She notes that the children raised multiple issues of concern through the weekly art tasks, suggesting that the children were comfortable about doing this and that such group work might provide a fruitful platform for therapeutic work. Unfortunately, this was not the purpose of the group, and no information is provided that addresses the group’s effectiveness, which Gilbert⁶⁵² acknowledges as a significant gap.

Bannister and Gallagher⁶⁶⁸ report mostly positive views of the intervention but one child had found it difficult to discuss the abuse and felt that the intervention had not helped their own abuse to stop.

Caregivers’ views

The art therapy for child witnesses of domestic violence was timed to coincide with their mothers attending a therapy session (which meant that babysitters were not required). From a practical point of view, carers thought that this was helpful.⁶⁵¹

Hill’s⁶⁷⁰ study of parent–therapist interactions highlighted the importance of therapists thinking carefully about the parents’ needs, when and how it is appropriate to involve them in therapy, and when not, and the skills required to do so effectively. One parent described how betrayed she felt following the sexual abuse of her child, and how this had impacted negatively on her trust of all professionals. The ability to follow the parent’s lead at the start of the therapy was identified as important by one therapist, and Hill⁶⁷⁰ refers to this as ‘interactional expertise’ – valuing the expertise of others and combining it with professional expertise. It is also clear from Hill’s paper⁶⁷⁰ that parents also needed to be ‘taught’ how to engage with therapy and to recognise that they were part of the therapeutic process too.

Tensions were clear in the study by Mishna et al.,⁶⁶¹ who reported that it took parents some time to develop a relationship with the play therapists working with their children, typically around 1 year. The parents in this study⁶⁶¹ had a history of school failure, and were reluctant to engage with parent/teacher consultations; building a relationship with therapists who were school based was considered important in re-engaging the parents in a relationship with the school community.

In the study by Bannister and Gallagher,⁶⁶⁸ caregivers observed improved behaviour, but this was not sustained in all cases. No acceptability data per se were obtained from carers or children.

Staff views

Burgon⁶⁶⁹ described positive interactions with young people using horses as a means of initially communicating with them, with a shared goal of riding the horse safely and with enjoyment. She documented the growing confidence of the young people, who were very withdrawn at the start of the therapy, arguing that the children developed empathy and a strong bond with the animals with which they were involved.

The therapist delivering the art therapy intervention in the study by Gilbert⁶⁵² expressed frustration about the fluctuating membership of her group, with only two children regularly attending the group over an 8-week period. This is perhaps because it was an add-on intervention that was timed to coincide with the maternal therapy group.

The therapists in Hill’s⁶⁷⁰ study viewed parents as generally supportive, but in need of advice and support in how to deal with the complex difficulties that are associated with sexual abuse. Feelings of guilt and blame are common in parents of sexually abused children, and the therapists in this study⁶⁷⁰ described how they worked to develop parent confidence in their own parenting skills. Only three fathers were involved in this study,⁶⁷⁰ but the therapists interviewed described how they took a proactive approach in involving them in their child’s therapy.

Therapists in the study by Mishna et al.⁶⁶¹ identified some practical considerations. Staff agreed that delivering the play therapy intervention within a school setting facilitated the development of rapport with teachers and school administration staff, which was felt to be important to facilitate treatment. Further work was required to develop relationships with parents. During the first year, therapists reported difficulties with parents, which might be attributable to feelings of mistrust and guilt (e.g. as described by Hill⁶⁷⁰) or reluctance to deal with the school environment. Therapists describe how learning to engage and build trust with the parent ‘typically took up to a full year before regular contact and a degree of trust was established’ (p. 79).⁶⁶¹ There was also evidence of relationship strain between therapists and the teaching staff, which took time to resolve. Teachers valued therapist input into classroom behaviour management and also appreciated knowing more about the child’s family life. More experienced therapists were able to develop a more effective relationship with teaching staff, which, in turn, helped teachers to develop some empathy for the child’s family situation.

Study design

All three^653–655 studies analysed factors associated with treatment engagement and completion.

Sample sizes

Kolverola et al.⁶⁵³ analysed the records of 118 children and their caregivers.

Risser and Schewe⁶⁵⁵ collected data on 1365 children (and their caregivers) for whom services were sought from one of 12 sites between 2001 and 2010.

Murphy et al.⁶⁵⁴ used data on 928 youth from the National Child Traumatic Stress Network (NCTNS) Core Data Set (CDS).

Participants included children from birth to age 21 years who received assessment and treatment services from one of 56 community sites between 2004 and 2010.

Location

All of the interventions were USA based.

Participants

The participants in the study by Murphy et al.⁶⁵⁴ had suffered physical as well as sexual abuse. In the Koverola et al.⁶⁵³ study the children had been referred to outpatients with a history of intrafamilial violence, and those in the Risser and Schewe⁶⁵⁵ study had been exposed to violence, including domestic violence. Ethnicity was reported in all studies, with the majority being white American.

Intervention

Intervention type varied from individual therapy, group therapy and FT^653,655 to evidence-based mental health services for trauma-exposed children.⁶⁵⁴

Characteristics of treatment completers

Murphy et al.⁶⁵⁴ concluded that neither physical nor sexual trauma was directly related to the probability of treatment completion. Indirect associations were found: physical trauma was associated with hyperarousal, but hyperarousal did not predict treatment completion. Sexual trauma was significantly associated with higher avoidance symptoms, which, in turn, were associated with lower likelihood of treatment completion. Sexual trauma was also associated significantly with overall PTSD symptoms, which, in turn, were associated with treatment completion at a marginally significant level.

In the study by Risser and Schewe,⁶⁵⁵ children were categorised into groups based on whether or not they attended any therapy session after the intake, terminated prematurely from therapy or completed treatment. Results demonstrated that child emotional and behavioural problems at intake, general parent stress and income did not differ by treatment engagement. Type of violence exposure, parent–child stress and race differed by category of treatment engagement. Children exposed to both domestic violence and child abuse demonstrated higher rates of treatment completion and attended more sessions than children exposed to either domestic violence or child abuse. Caregivers with higher levels of parent–child stress were more likely to engage in treatment. White children in the full sample completed treatment at higher rates than minority children.

Koverola et al.⁶⁵³ found no differences in children’s race, gender or caregivers’ gender by treatment attrition type; 60.5% of the sample were legally mandated to participate in treatment, but whether treatment was court mandated or voluntary was not associated with a likelihood of engaging in, or completing, treatment [χ²(2) = 0.1; p = 0.95].

There were some baseline differences in compliers, dropouts and non-engagers in Koverola et al.⁶⁵³ The mean age in the three groups was significantly different [F(2,116) = 4.0; p = 0.02]. Non-engagers were older than both attriters and compliers (M = 11.1, 9.5 and 8.0 years). Attriters were more likely to be in the family of origin than non-engagers or compliers. Families referred for child abuse in this study were also more likely to comply with treatment than those who were referred for domestic violence (67% vs. 33%, respectively) [χ²(2) = 5.6; p = 0.06]. There were very few differences found among caregivers of attriters, completers and non-engagers. Caregivers’ self-reports revealed that caregivers of completers do not experience significantly higher levels of social support and daily stress than caregivers of attriters or non-engagers (F = 0.5; p = 0.63; F = 0.7; p = 0.52). In addition, they do not report higher levels of internalising or externalising behaviour problems or post-traumatic symptoms in their children (F = 0.5; p = 0.59; F = 2.9; p = 0.05; F = 1.8; p = 0.17). Differences were found in relation to psychological distress and parental distress. Caregivers reporting high child-related parental distress [F(2,32) = 3.9; p = 0.03] and high psychological distress [F(2,70) = 3.3; p = 0.04) were least likely to engage in treatment. No differences were found with regard to children’s self-report of post-traumatic symptoms or cognitive functioning (F = 0.3; p = 0.71; F = 0.7; p = 0.49). When treatment modality was examined, findings revealed that families were more likely to complete treatment successfully if they received multimodal treatment relative to individual or family-only treatment [χ²(2) = 7.6; p = 0.01].

Caregivers’ views

The majority of the mothers interviewed by Koverola et al.⁶⁵³ were reluctant to discuss the abuse because of fear that it had not happened. Some also felt vulnerable because of how therapists perceived their reactions – they either felt overprotective or too careless that abuse had happened. Most expressed positive opinions about treatment and felt that they understood their children better, had new ideas about possible solutions and valued the contributions made by the team.

Staff views

Conflicts associated with alleged abuse had been reduced in n = 20 cases in the study by Koverola et al.⁶⁵³ Alleged perpetrators who had engaged in therapy expressed relief being able to talk about it in a non-judgemental setting. Those who did not view the intervention positively felt that mothers had formed a coalition with the therapist.

Treatment completion and dropout

How one conceptualises treatment completion dictates how one defines premature treatment termination or dropout. This, in turn, can change the answers that research studies provide to questions about treatment acceptability. Even defining completion is not straightforward. Some researchers define it in terms of attending all sessions. Although one cannot assume that full attendance is equivalent to engaging with treatment (turning up at a lecture is not synonymous with listening or learning), it may be a reasonable indicator that someone is finding therapy helpful or expecting it to be helpful. It may indicate that someone is probably an active participant, although this may not be a reasonable assumption for children who might have no choice. In contrast, defining lack of success as lack of engagement is, at best, tautologous. Timmer et al.⁶⁵⁶ provide an example of the problem. In this study⁶⁵⁶ of PCIT, parents were considered to have completed treatment once they:

• had met the mastery criteria for the relationship enhancement element of the programme
• could demonstrate compliance commands from their children
• could successfully implement a discipline procedure, and
• could show maintenance of skills that they had acquired in the first phase of the programme.

In this study,⁷⁵⁶ failure to comply with all of these tasks was categorised as early termination, but this was essentially synonymous with programme effectiveness, thereby stacking the odds in favour of a positive result for the efficacy of the treatment being assessed.

Some studies define completion in terms of ‘dosage’, by calculating the average number of sessions attended and setting a minimum threshold. Theoretically, this is often argued on the basis of evidence from earlier studies suggesting that a minimum level of exposure is necessary for a certain intervention to have an effect.⁷⁵⁷ So, for example, Laan et al.⁶⁷¹ defined premature termination of intensive foster care placement as those placements that ended within 2 years of inclusion within the intensive fostering service. In this case, the authors acknowledge that this is ‘something of an arbitrary criterion’, based on the belief that fostering can postpone a residential placement and postponement of at least 2 years is deemed a success. The children in this study⁷⁵⁷ were predominantly learning disabled children, the remainder having serious physical impairments or health problems. All were in out-of-home placements as a result of maltreatment or inadequate parenting. Yet others use definitions such as ‘termination with therapist approval or mutual agreement’. Some might see these definitions of completion as a source of bias. Taken as a whole, these factors probably account for some of the contradictory messages that appear across studies, but perhaps make some of the consistent messages more important.

Although no intervention can demonstrate 100% effectiveness, the odds of success should arguably not be estimated without consideration of the factors that cause those it is designed to help to decline or drop out early. The reasons for both might reflect factors that are only indirectly linked to the intervention (relationship with therapist, service setting, time of day) but are important to know and understand.

Treatment engagement

Almost no study addressed the issue of treatment engagement, which is also subject to various definitions, some of which can run the risk of being used almost interchangeably with treatment success, that is, failure is explained in terms of failure to engage, when – in some circumstances at least – engagement might be a function of service or therapist attitude and behaviour.

Engagement can be used to refer to a variety of closely related things. Most usually it refers to someone’s commitment to the therapeutic process and active participation in it. As Cunningham et al.⁶⁴⁹ point out, it is ‘related to other concepts such as readiness to change, rapport, motivation, working alliance, and collaboration and compliance’ (p. 64). As such, measures of engagement (the focus of this study) have looked variously at clients’ motivation for, and expectations about, treatment; the client–therapist relationship; and client behaviour within therapy.

Expectations of treatment may act as facilitators or barriers to participation in treatment, for example believing that treatment will – or will not – be helpful, recognising the need for treatment compared with failing to see a problem to be resolved.

On the basis of the wider therapeutic literature, establishing rapport with the therapist or with care staff is probably a necessary prerequisite of achieving therapeutic change, but it may present particular challenges with reluctant, ambivalent or involuntary clients. Cunningham et al.⁶⁴⁹ noted that engagement by young people in the RTCs they studied required continual effort because the process of engagement was unstable and required constant ‘refreshing’. Few studies in this review of psychosocial interventions gave much attention to this aspect of effective interventions, with only a handful referring (in passing) to strategies such as motivational interviewing, and none to the wider challenges of engaging young people in therapeutic interventions.

What the study by Cunningham et al.⁶⁴⁹ emphasises is the importance of professionals thinking theoretically and strategically about what they need to do to facilitate engagement, and how this might vary with context (service setting, timing, voluntary/compulsory and therapy type) and indeed, influence service outcomes. For the population of seriously maltreated children, engagement may require particular thought and care, given the fracturing of trust and the damaged ability to form relationships that is so much a feature of these young people’s lives.