Introduction

Despite efforts to make services available to address the effects of abuse, affected children and families often have significant barriers that limit their capacity to engage with therapy. These barriers can include service-level barriers, such as difficulty finding the right service provider and qualifying for therapy under a particular funding scheme. Families with a disclosure of child sexual abuse may also have multiple complex issues such as homelessness, poverty, family and domestic violence, and alcohol and substance abuse issues that can present additional barriers to accessing services. These barriers reduce the proportion of affected children that may engage with and benefit from therapy services.

Commonly, informal systems of referral assume that children have a protective parent who is in a position to be able to advocate for and seek services for them. The lack of a planned system of referral at the point of disclosure means there is limited knowledge of the extent of the barriers that may exist for children and families to engage with therapy.

This review examines the factors that influence rates of engagement and completion of therapy following a disclosure of child sexual abuse to police or child protection authorities. In this paper, therapy refers to any program of therapy intended to reduce the effects of trauma following abuse. The type of therapy in each of the included studies is described in Appendix B. By identifying the factors that consistently, and across many studies, influence engagement or completion, this research aims to inform approaches to support and to increase engagement and completion of therapy.

Methodology

This paper reports on the results of the rapid evidence review described in the companion paper Rates of Therapy Use Following a Disclosure of Child Sexual Abuse. Appendix A provides an overview of the search strategy and extraction process that included studies examining the effects of abuse, child, perpetrator, family and response characteristics on engagement and completion of therapy. All studies included in the review of rates were examined by a single reviewer to identify whether studies looked at the effect¹ of any factors on rates of engagement or completion. This information was then included in the extraction template, separating out factors that significantly² affected engagement/completion (and the direction of the difference), and factors that were not associated with a significant difference.

Included studies were split across those reporting rates of engagement following contact with police or child protection authorities (investigation samples; see below for definitions), or those where children or families had contacted sexual abuse therapy services (therapy initiators). Completion studies were split between clinical samples (which had inclusion/exclusion criteria for eligibility) and community samples (where services were provided to any children without having to meet inclusion/exclusion criteria).

What are the findings of this review?

Generally, included studies were testing one of two quite different hypothesised arguments for why therapy engagement and completion rates varied. The first was that families with more complexity, with children that had longer histories of abuse, and parents with fewer economic and social resources would be more likely to drop out or not engage with therapy. The second was that families that were more supportive of their children, with more social resources, and where the abuse was a single incident and not part of a trajectory of persistent abuse may not engage or complete services out of concerns about stigmatising children or a sense that 'getting back to normal' may be the most appropriate way to support their children. The framing of the research questions in these studies differed considerably depending on the context of the research, and the types of samples these hypotheses were tested in.

1. Engagement with therapy factors

This section reviews five groupings of characteristics, which were examined in the studies for their effect on engagement: abuse, child, perpetrator, family and response characteristics (see Appendix A for inclusions); and includes two types of samples: post-investigation (studies that examined rates of commencing therapy following a police/child protection investigative response), and therapy initiator (studies examining rates of commencing therapy following a child/family making contact with service providers to obtain therapy for children/young people affected by sexual abuse).

Nine of the 49 selected studies examined factors that may influence the rate of engagement with therapy. The rate of engagement with therapy can be defined as commencing therapy (attending at least one session) in all but one study, where it was defined as attending three sessions (Deblinger, Stauffer, & Steer, 2001).

Abuse characteristics

Overall, abuse characteristics seemed to have a mild effect on engagement. Across five individual studies, four out of 11 comparisons of abuse characteristics on engagement with therapy were found to be significant; however, the direction of the findings (i.e. whether the factor was associated with an increase or decrease in engagement) was mixed, making it difficult to draw conclusions across study contexts. The findings seem to suggest that more complex abuse (e.g. more severe, more frequent, co-occurring neglect) was associated with increased engagement with therapy. Table 1 provides an overview of the abuse comparisons.

Among post-investigation samples, Anderson (2016) found that more severe abuse (i.e. penetrative contact offences vs non-penetrative contact offences) was associated with higher rates of engagement, although three other studies found no significant differences. Tingus, Heger, Foy, and Leskin (1996) found that children with abuse that involved multiple incidents (compared to single incidents) were more likely to engage, although two other studies found no difference. Cases in which neglect co-occurred with sexual abuse were found to be much more likely to engage, which was attributed to the effect of child protection safety plans that required parents to engage with therapy to retain care of their children (Lippert, Favre, Alexander, & Cross, 2008).

For therapy initiator samples, only one study tested whether different abuse characteristics influenced engagement (Oates, O'Toole, Lynch, Stern, & Cooney, 1994). Contrary to Tingus and colleagues (1996), they found that children with abuse that involved a single incident were more likely to engage than children with multiple incidents of abuse.

Child characteristics

Child characteristics for the most part did not affect engagement. Across six individual studies, the only significant effects found were differences in the rate at which white children in the United States engaged with therapy relative to minority groups (see Table 2). Three studies found white children were significantly more likely to engage with therapy than black and Latino/Latina/Hispanic children in the United States.

Perpetrator characteristics

Perpetrator characteristics also for the most part did not affect engagement (see Table 3). Oates and colleagues (1994) found a significant difference among therapy initiator samples in terms of the relationship with the perpetrator. Abuse by family members and extra-familial adults in regular contact were both found to be associated with an increased likelihood of engaging with therapy compared to perpetrators who did not have regular contact with the child. In contrast, four studies with post-investigation samples found no effect for the relationship to the perpetrator.

Family characteristics

From seven studies, five comparisons of family characteristics were found to be significant across 33 comparisons (see Table 4). Lippert and colleagues (2008) tested a variety of parental attitudes about their child and about therapy in interviews. While most had a non-significant effect, they found that parents who said they 'liked going places with their child' were less likely to engage in therapy. The researchers suggested that this seemed to reflect a tendency towards parent-centredness versus child-centredness among cases that engaged with therapy.

Lippert and colleagues (2008) also found that families that declined therapy were more uncomfortable about disclosing personal information to a therapist, and engagers were more likely to think therapy was for 'emotional help and change'. In a complementary finding, Hasket, Nowlan, Hutcheson, and Whitworth (1991) found that where a maternal caregiver thought the entire family needed therapy, they were more likely to engage.

Anderson (2016) found that parents who were designated as not supportive of their children following a forensic interview (as rated by observers) were nearly 10 times more likely to begin therapy. This may be due to the implied threat of child protection action for caregivers whom authorities could determine are not protective. Self-Brown, Tiwari, Lai, Roby, and Kinnish (2016) found an increased likelihood of engaging with therapy where more children lived in a household.

Response characteristics

Only four studies examined factors related to the response to the abuse; however, half of the comparisons were found to be significant in terms of engagement with therapy (see Table 5). Referrals to private practice therapy (Haskett et al., 1991), or additional counselling during or after the initial treatment (McPherson, Scribano, & Stevens, 2012) were found to significantly increase engagement. Tingus and colleagues (1996) found that where children had been removed from the home following their abuse, they were more likely to engage, although McPherson and colleagues (2012) did not find a significant difference on this factor. Tingus and colleagues (1996) also found that where both police and child protection agencies were involved in a case (other conditions were: child protection only and no agency involved), children were more likely to engage with therapy.

2. Completion of therapy factors

Twenty-seven studies tested factors that may influence the rate of completion of therapy following a disclosure of sexual abuse. This section reviews five characteristics: abuse, child, perpetrator, family and response (see Appendix A for inclusions); and includes two types of samples: clinical (studies that were limited to children that met criteria of clinical symptomatology), and community (studies including all children that presented for a service). The clinical samples represent studies with more tightly controlled conditions (i.e. procedures to monitor fidelity to the treatment model), usually as the study was undertaken to demonstrate the efficacy of the different therapies in optimal conditions.

The definitions of completion differed across the studies included (see Appendix A). Generally, clinical studies included criteria based on the delivery or completion of a program and the community studies involved a discharge from the service agreed to by the clinician and the client.

Abuse characteristics

Across the 33 comparisons conducted across 16 studies, none of the abuse characteristics were found to affect completion in clinical samples; however, many factors were found to be significant in the community studies (see Table 6). Three studies found that where the abuse included fewer incidents or a single incident, children were less likely to complete therapy (Chasson, 2007; Chasson, Mychailyszyn, Vincent, & Harris, 2013; Deblinger, Lippmann, & Steer, 1996). Findings were consistent for the effect of abuse severity on completion, with four studies finding that more severe abuse was associated with increased completion (Chasson, 2007; Chasson et al., 2013; Horowitz, Putnam, Noll, & Trickett, 1997; Mogge, 1999), although six other studies found no difference. Two studies found that longer abuse duration was associated with completion, although two other studies found no difference on this factor.

The direction of findings was mixed for the age of onset of abuse, with Horowitz and colleagues (1997) finding that earlier onset (i.e. age of the child when the abuse began) was associated with completion and Tavkar (2010) finding earlier onset was associated with dropping out of therapy.

Note: The reference Wamser-Nanney & Steinzor (2017) included two sets of analyses, which have been included separately on the comparison tables. Instances marked with an 'a' refer to the analysis based on clinician rated dropout.

Child characteristics

For child characteristics, 69 comparisons were made across 19 studies (see Table 7). For the clinical samples, across the 22 comparisons only one study found a significant difference. Celano, NeMoyer, Stagg, & Scott (2018) found that older children were less likely to complete therapy, although five other studies found no difference. All other comparisons in the clinical samples found no difference in terms of child gender, ethnicity and initial symptomatology.

For the community samples, across the 47 comparisons, a number of studies found significant differences. The effects of initial symptomatology on completion seemed to be mixed, partly due to the different definitions of completion. This may be because of the converse effects of: (a) children with more intense behavioural issues and trauma (based on psychometric instruments) being less likely to complete due to difficulty in getting them to sessions; (b) children with less severe trauma symptoms being less likely to complete due to parental concerns about stigmatising children. Five out of 16 studies found significant differences for rate of completion in terms of initial symptomatology.

Horowitz and colleagues (1997) and New and Berliner (2000) found that more severe symptoms were generally associated with higher rates of completion, whereas Barnett (2007) and Wamser-Nanney and Steinzor (2017) found the opposite. Tebbett (2013) found that moderate scores on the internalising problems subscale and high and low scores on the externalising problems subscale of the Parental Rating Scale Behavioral Assessment System for Children was associated with higher rates of completion. Barnett (2007) found that higher scores on the Social Behaviour Index (indicating a child's social competency) was associated with higher rates of completion. Barnett (2007) also found that on the Expectations Test (used to measure hypervigilance and expectations of future harm) that children who expected sexual abuse in their future were less likely to complete therapy, whereas children that did not know what to expect or had positive expectations were more likely to complete. Nine studies found no difference for rate of completion in terms of children's initial symptomatology.

In the community sample, Chasson, Vincent, and Harris (2008) measured symptomatology before termination (measures were completed bi-weekly during treatment) and found avoidance behaviour (measured on the Impact of Event Scale) was associated with dropout. Barnett (2007) also included some mid-treatment measures and found that a client's perceived improvement in symptoms was associated with completion.

Most of the community studies found no difference in terms of child age, gender or ethnicity, although Wamser-Nanney and Steinzor (2017) found that (in a United States context) minorities were less likely to complete, and New and Berliner (2000) found that younger children were more likely to complete.

Notes: The reference Wamser-Nanney & Steinzor (2017) included two sets of analyses, which have been included separately on the comparison tables. Instances marked with an 'a' refer to the analysis based on clinician rated dropout. Instances of Wamser-Nanney & Steinzor (2017) marked with a 'b' refer to the analysis based on completion of a minimum number of sessions (i.e. 12).

Perpetrator characteristics

The studies reviewed included 13 comparisons of perpetrator characteristics across 11 studies (see Table 8). Across the clinical studies, none of the comparisons were significant. Among community studies, around half found that the relationship to the perpetrator affected the rate of completion. The findings were generally consistent in showing that abuse by non-relatives was associated with lower rates of completion than abuse by a parent or parental figure (Chasson et al., 2013; Mogge, 1999; New & Berliner, 2000). Three other studies found no difference in rates of completion based on the relationship to perpetrator (Chasson, 2007; Reyes, 1996; Tavkar, 2010). Mogge (1999) found that where perpetrators remained in the home, children were less likely to complete therapy.

Family characteristics

Across 65 comparisons in 19 studies, 13 comparisons of family characteristics were found to influence completion rates (see Table 9). The most consistent result was the effect of parental attendance/engagement with therapy; all studies that included this as a variable found that increased parental involvement was associated with increased completion (Celano et al., 2018; Macias, 2004; McPherson et al., 2012).

Two studies (one clinical study and one community study) found that higher socio-economic status increased completion rates (Cohen & Mannarino, 1996; Wamser-Nanney & Steinzor, 2017); however, another seven studies found no difference. One out of four comparisons found that higher parental education positively influenced completion rates (Wamser-Nanney & Steinzor, 2017 - clinician-rated completion).

Nine studies measured family coping at intake in a variety of ways; however, only one study found that this affected rates of completion (Tavkar, 2010). Tavkar (2010) found higher parental distress, resentment associated with role as a parent, and the ability to identify behavioural and problem-solving strategies were all associated with lower rates of completion. However, two community studies looking at a parent's PTSD symptoms and psychopathology symptoms found no effect on completion rates (Self-Brown et al., 2016; Tebbett, 2013).

Single studies found that the absence of child protection history (Wamsner-Nanney & Steinzor, 2017) and lower rates of parental reported domestic violence (DeLorenzi, Daire, & Bloom, 2016) were associated with higher rates of completion. Out of four community studies that tested the effect of parental history of sexual abuse on completion rates, only Macias (2004) identified an effect and only among mothers who had an abuse history and perpetrated the abuse on their child.

Across the community studies, parental age (Tavkar, 2010), therapist ratings of caregiver support (Barnett, 2007), and participation in the therapy process (a therapist-rated checklist of the therapy process across both parents and children; Barnett, 2007) all had single comparisons that supported the influence of these factors on completion.

Notes: The reference Wamser-Nanney & Steinzor (2017) included two sets of analyses, which have been included separately on the comparison tables. Instances marked with an 'a' refer to the analysis based on clinician rated dropout. Instances of Wamser-Nanney & Steinzor (2017) marked with a 'b' refer to the analysis based on completion of a minimum number of sessions (i.e. 12).

Response characteristics

Only four of 17 comparisons of response characteristics found a significant effect (see Table 10). Cohen, Mannarino, and Knudsen (2005) found higher rates of dropout for non-structured supportive therapy compared with trauma-focused cognitive behaviour therapy. Celano and colleagues (2018) found that more assessment/diagnostic sessions at the start of therapy could result in higher dropout. Newman and Berliner (2000) found that delays to commencing therapy resulted in reduced dropout. McPherson, Scribano, and Steinzor (2012) found that referrals to other counselling services during or after the initial therapy was associated with higher rates of completion.

Notes: The reference Wamser-Nanney & Steinzor (2017) included two sets of analyses, which have been included separately on the comparison tables. Instances marked with an 'a' refer to the analysis based on clinician rated dropout. Instances of Wamser-Nanney & Steinzor (2017) marked with a 'b' refer to the analysis based on completion of a minimum number of sessions (i.e. 12).

Limitations to these findings

This review is limited in being able to draw conclusions on the effect of many factors due to the limited number of studies examining their effects on the engagement and completion of therapy. The number of different types of samples (e.g. investigation vs therapy initiators) in the included studies further limited the number of comparisons available. The review has focused on discussing findings where there appeared to be similar findings across multiple comparisons/studies.

The literature search was undertaken across the most common databases for social sciences and criminal justice research and included a pearling strategy to identify additional relevant studies (see Appendix A for more detail on the pearling strategy). However, a full registered systematic review could identify additional eligible studies. In addition, the search string included the term 'counselling', which may have limited the results by not including other variants on the word (e.g. counsellor, counseling).

Common to many evidence reviews, the studies identified in the search were overwhelmingly conducted in the United States (n = 45; 92%), with the other four studies conducted in Australia. While some caution should be taken in generalising the findings to other contexts, it is worth noting the diversity of approaches and resources across US states for responding to child sexual abuse (e.g. Herbert, Walsh, & Bromfield, 2018).

An important limitation of this review is that engagement and completion of therapy do not necessarily indicate improved outcomes for children and their families. While completion of an indicated service is important to the likelihood of obtaining benefits from the therapy, many other factors will influence the likelihood of reduced symptomology such as the effectiveness of the treatment model and the fidelity of the therapy to effective treatment models.

What are the implications for policy and practice?

The findings of this review were mixed with few clear and consistent findings across studies, making it difficult to draw conclusions about the key factors that impact therapy engagement and completion across contexts. The mix of findings points to the complex nature of these outcomes, how different contexts change the influence of these factors, and the interactive effects between different types of variables. Most consistently, two factors stood out:

1. Increased parental involvement in therapy was found to be associated with increased therapy completion (Barnett, 2007; Celano et al., 2018; Macias, 2004; McPherson, Scribano, & Stevens, 2012). Therapy models delivered to this population of children should incorporate a clear role for parents/caregivers. The presence of a parent/caregiver willing to be involved should be part of the therapy triage process, and resources should be put into attempting to build parent/caregiver's motivation to be involved (Lundahl, Kunz, Brownell, Tollefson, & Burke, 2010).
2. More severe and frequent abuse was associated with increased therapy completion (Chasson, 2007; Chasson et al., 2013; Deblinger, Lippmann, & Steer, 1996; Horowitz et al., 1997; Mogge, 1999). This combined with less consistent findings in terms of symptomology suggests children with trauma symptoms associated with less severe/frequent abuse may be at increased risk of not completing therapy. Monitoring symptoms and discussing progress on formal instruments may be a way to address the risk of these children not completing therapy.

Systematic monitoring of the needs and characteristics of children and families disclosing abuse to authorities is critical to addressing the barriers to access and maximising opportunities for earlier intervention in the management of trauma from sexual abuse. The complexity of findings across the included studies means that it is difficult to recommend factors that should be screened, monitored and controlled in the administration of therapy following a disclosure of child sexual abuse. However, this review highlights factors that service providers can consider monitoring to help develop strategies to inform approaches to referral and support at the point of disclosure depending on their service populations (see Appendix A). These strategies may focus on the service level, such as identifying ways to improve the client experience of referral or instituting a holding service while clients are on a waitlist.

Alternatively, strategies may focus on barriers within the family and on the motivation/hesitancy to engage with therapy among caregivers, as well as involve broader support and advocacy or even additional interventions to help stabilise families (e.g. Safecare®). Emerging evidence suggests that strategies addressing parental hesitancy about mental health services paired with strategies to streamline the referral, assessment, and waitlist for services can improve engagement and completion (Budde & Waters, 2014). Building motivation to engage with therapy through structured psycho-educational approaches seems particularly important for children who may not have trauma symptoms straight away.

References

• Allen, B., & Hoskowitz, N. A. (2017). Structured trauma-focused CBT and unstructured play/experiential techniques in the treatment of sexually abused children: A field study with practicing clinicians. Child Maltreatment, 22(2), 112-120.
• Allen, B., Tellez, A., Wevodau, A., Woods, C. L., & Percosky, A. (2014). The impact of sexual abuse committed by a child on mental health in adulthood. Journal of Interpersonal Violence, 29(12), 2257-2272.
• Ancha, A. J. (2003). Program evaluation of a time-limited, abuse-focused treatment for child and adolescent sexual abuse victims and their families (Psy.D.). Argosy University MI, Ann Arbor, United States.
• Anderson, G. D. (2016). Service outcomes following disclosure of child sexual abuse during forensic interviews: An exploratory study. Journal of Public Child Welfare, 10(5), 477-494.
• Barnett, C. W. (2007). Predicting treatment completion with maltreated children. PhD thesis. The University of Utah, Salt Lake City UT, United States.
• Budde, S. & Waters, J. (2014). PATHH: Improving access and quality of mental health services to sexually abused children in Chicago. Chicago, IL: Chicago Children's Advocacy Center. Retrieved from www.chicagocac.org/wp-content/uploads/2012/04/ChicagoCAC-PATHH-White-Paper-102414.pdf
• Celano, M., NeMoyer, A., Stagg, A., & Scott, N. (2018). Predictors of treatment completion for families referred to trauma-focused cognitive behavioral therapy after child abuse. Journal of Traumatic Stress, 31(3), 454-459.
• Chasson, G. S. (2007). Survival analysis of treatment attrition with child victims of trauma: The role of trauma severity. PhD thesis. University of Houston, Houston TX, United States.
• Chasson, G. S., Mychailyszyn, M. P., Vincent, J. P., & Harris, G. E. (2013). Evaluation of trauma characteristics as predictors of attrition from cognitive-behavioral therapy for child victims of violence. Psychological Reports, 113(3), 734-753.
• Chasson, G. S., Vincent, J. P., & Harris, G. E. (2008). The use of symptom severity measured just before termination to predict child treatment dropout. Journal of Clinical Psychology, 64(97), 891-904.
• Cohen, J. A., Deblinger, E., Mannarino, A. P., & Steer, R. (2004). A multi-site, randomized controlled trial for children with abuse-related PTSD symptoms. Journal of the Academy of Child and Adolescent Psychiatry, 43(4), 393-402.
• Cohen, J. A., & Mannarino, A. P. (1996). Factors that mediate treatment outcome of sexually abused preschool children. Journal of the American Academy of Child & Adolescent Psychiatry, 34(1), 1402-1410.
• Cohen, J. A., & Mannarino, A. P. (1997). A treatment study for sexually abused preschool children: Outcome during a one-year follow-up. Journal of the American Academy of Child & Adolescent Psychiatry, 36(9), 1228-1235
• Cohen, J. A., & Mannarino, A. P. (2000). Predictors of treatment outcome in sexually abused children. Child Abuse & Neglect, 24(7), 983-994.
• Cohen, J. A., Mannarino, A. P., & Knudsen, K. (2005). Treating sexually abused children: 1 year follow-up of a randomized controlled trial. Child Abuse & Neglect, 29(2), 135-145.
• Cohen, J. A., Mannarino, A. P., Perel, J. M., & Staron, V. (2007). A pilot randomized controlled trial of combined trauma-focused CBT and sertraline for childhood PTSD symptoms. Journal of the American Academy of Child & Adolescent Psychiatry, 46(7), 811-819.
• Cross, T. P., Jones, L. M., Walsh, W. A., Simone, M., & Kolko, D., Sczepanski, J. et al. (2008). Evaluating children's advocacy centres' response to child sexual abuse. Juvenile Justice Bulletin, 106, 1-11.
• Deblinger, E., Lippmann, J., & Steer, R. (1996). Sexually abused children suffering posttraumatic stress symptoms: Initial treatment outcome findings. Child Maltreatment, 1(4), 310-321.
• Deblinger, E., Mannarino, A. P., Cohen, J. A., Runyon, M. K., & Steer, R. A. (2011). Trauma-focused cognitive behavioural therapy for children: Impact of the trauma narrative and treatment length. Depression and Anxiety, 28, 67-75.
• Deblinger, E., Stauffer, L. B., & Steer, R. A. (2001). Comparative efficacies of supportive and cognitive behavioral group therapies for young children who have been sexually abused and their nonoffending mothers. Child Maltreatment, 6(4), 332-343.
• DeLorenzi, L., Daire, A. P., & Bloom, Z. D. (2016). Predicting treatment attrition for child sexual abuse victims: The role of child trauma and co-occurring caregiver intimate partner violence. Counseling Outcome Research and Evaluation, 7(1), 40-52.
• Fong, H. F., Alegria, M., Bair-Merritt, M. H., & Beardslee, W. (2018). Factors associated with mental health services referrals for children investigated by child welfare. Child Abuse and Neglect, 79, 401-412.
• Friedrich, W. N., Luecke, W. J., Beilke, R. L., & Place, V. (1992). Psychotherapy outcome of sexually abused boys: An agency study. Journal of Interpersonal Violence, 7(3), 396-409.
• Garland, A. F., Landsverk, J. L., Hough, R. L., & Ellis-MacLeod, E. (1996). Type of maltreatment as a predictor of mental health service use for children in foster care. Child Abuse and Neglect, 20(8), 675-688.
• Hartman, S. (2011). From efficacy to effectiveness: A look at trauma-focused cognitive behavior therapy in a community setting (Psy.D.). Pace University, Ann Arbor, MI
• Haskett, M. E., Nowlan, N. P., Hutcheson, J. S., & Whitworth, J. M. (1991). Factors associated with successful entry into therapy in child sexual abuse cases. Child Abuse & Neglect, 15(4), 467-476.
• Herbert, J. L., Walsh, W., & Bromfield, L. M. (2018). A national survey of characteristics of child advocacy centers in the United States: Do the flagship models match those in broader practice? Child Abuse & Neglect, 76, 583-595.
• Horowitz, L. A., Putnam, F. W., Noll, J. G., & Trickett, P. K. (1997). Factors affecting utilization of treatment services by sexually abused girls. Child Abuse & Neglect, 21(1), 35-48.
• Humphreys, C. (1995). Whatever happened on the way to counselling? Hurdles in the interagency environment. Child Abuse & Neglect, 19(7), 801-809.
• King, N. J., Tonge, B. J., Mullen, P., Myerson, N., Heyne, D., Rollings, S. et al. (2000). Treating sexually abused children with posttraumatic stress symptoms: A randomized clinical trial. Journal of the American Academy of Child & Adolescent Psychiatry, 39(11), 1347-1355.
• Koch, E. R. (2004). Factors associated with treatment outcomes among Hispanic survivors of childhood sexual abuse. PhD thesis. Fuller Theological Seminary, Pasadena CA, United States.
• Lane, W. G., Dubowitz, H., & Harrington, D. (2003). Child sexual abuse evaluations: Adherence to recommendations, Journal of Child Sexual Abuse, 11(4), 17-34.
• Lippert, T., Favre, T., Alexander, C., & Cross, T. P. (2008). Families who begin versus decline therapy for children who are sexually abused. Child Abuse & Neglect, 32, 859-868.
• Lundahl, B. W., Kunz, C., Brownell, C., Tollefson, D., & Burke, B. L. (2010). A meta-analysis of motivational interviewing: Twenty-five years of empirical studies. Research on Social Work Practice, 20, 137-160.
• Macias, S. B. (2004). The intergenerational transmission of abuse: The relationship between maternal abuse history, parenting stress, child symptomatology, and treatment attrition. PhD thesis. University of California, Santa Barbara, United States.
• Marx, T. (2004). Attrition from childhood sexual abuse treatment as a function of gender, ethnicity, and parental stress PhD thesis. University of California, Santa Barbara CA, United States.
• McPherson, P., Scribano, P., & Stevens, J. (2012). Barriers to successful treatment completion in child sexual abuse survivors. Journal of Interpersonal Violence, 27(1), 23-29.
• Mogge, K. L. (1999). Predictors of dropout in child sexual abuse treatment. PhD thesis. University of Memphis, Memphis TN, United States.
• Murphy, R. A., Sink, H. E., Ake, G. S., Carmondy, K. A., Amaya-Jackson, L. M., & Briggs, E. C. (2014). Predictors of treatment completion in a sample of youth who have experienced physical or sexual trauma. Journal of Interpersonal Violence, 29(1), 3-19.
• New, M., & Berliner, L. (2000). Mental health service utilization by victims of crime. Journal of Traumatic Stress, 13(4), 693-707.
• Oates, R. K., O'Toole, B. I., Lynch, D. L., Stern, A., & Cooney, G. (1994). Stability and change in outcomes for sexually abused children. Journal of the American Academy of Child & Adolescent Psychiatry, 33(7), 945-953.
• Reyes, C. J. (1996). Resiliency of young children: Self-concept, parental support, and traumatic symptoms after sexual abuse. PhD thesis. University of California, Santa Barbara CA, United States.
• Self-Brown, S., Tiwari, A., Lai, B., Roby, S., & Kinnish, K. (2016). Impact of caregiver factors on youth service utilization of trauma-focused cognitive behavioural therapy in a community setting. Journal of Child and Family Studies, 25, 1871-1879.
• Signal, T., Taylor, N., Prentice, K., McDade, M., & Burke, K. J. (2017). Going to the dogs: A quasi-experimental assessment of animal assisted therapy for children who have experienced abuse. Applied Developmental Science, 21(2), 81-93.
• Smith, D. W., Witte, T. H., & Fricker-Elhai, A. E. (2006). Service outcomes in physical and sexual abuse cases: A comparison of child advocacy center-based and standard services. Child Maltreatment, 11(4), 354-360.
• Tavkar, P. (2010). Psychological and support characteristics of parents of child sexual abuse victims: Relationship with child functioning and treatment. PhD thesis. The University of Nebraska, Lincoln NE, United States.
• Tebbett, A. A. (2013). Predictors of attrition in trauma-specific cognitive-behavioral therapy. PhD thesis. St. John's University, New York NY, United States.
• Tingus, K. D., Heger, A. H., Foy, D. W., & Leskin, G. A. (1996). Factors associated with entry into therapy in children evaluated for sexual abuse. Child Abuse & Neglect, 20(1), 63-68.
• Wamser-Nanney, R., & Steinzor, C. E. (2017). Factors related to attrition from trauma-focused cognitive behavioral therapy. Child Abuse & Neglect, 66, 73-83.
• Zaidi, L. Y., & Gutierrez-Kovner, V. M. (1995). Group treatment of sexually abused latency-age girls. Journal of Interpersonal Violence, 10(2), 215-227.

Appendix A: Therapy use

Search strategy

A search string was generated using the NVIVO auto-code feature with full text extracted from 15 target studies; this was used to identify key terms likely to identify other relevant studies (see Box A.1). The search string⁴ was run across Psychinfo, Embase, Medline, Proquest Social Science Premium Collection, and Proquest Dissertations and Theses Global. The results were screened in Covidence by title and abstract, and then by full record.

The search identified 2,374 individual studies for title and abstract screening, which was reduced to 205 studies by a single reviewer for full-text screening (see Figure A.1). These were studies that very clearly did not meet eligibility based on title and abstract. These 205 studies were assessed for eligibility, which identified 32 eligible studies. A total of 173 studies were excluded for the following reasons:

• The study was about the impacts of abuse rather than referral/engagement/completion of therapy (n = 42).
• The study involved an adult population (n = 32).
• The study design did not allow for data about referral/engagement/completion (n = 18).
• Sexual abuse was not analysed separately from other forms of maltreatment (n = 15).
• Study was about the characteristics or impacts of abuse rather than about the treatment of abuse (n = 12).
• The paper did not involve empirical research and was a conceptual or theoretical paper (n = 8).
• Full text could not be obtained, even when requested from the University of South Australia library services (n = 7).
• Other/miscellaneous (n = 11).

Pearling⁵ was undertaken across the 32 eligible studies. This involved reading through each of the articles and identifying any references to studies that may contain relevant information for the review. This process identified an additional 21 eligible studies. Some of the eligible studies (n = 4) were found to be an analysis of the same dataset. In these cases, information about the studies was combined into a single record in the extraction template. A total of 49 studies were found to be eligible and were extracted.

Figure A.1: PRIMSA flow diagram of systematic literature search

The extraction template included:

• study information: study ID number, authors, year, if the study shared data with any other included studies, title, country of study, publication type, study design
• sample information: sample type narrative description, whether caregiver consent to participate in the study was required, whether clinical symptoms were required to be included, whether the sample initiated therapy, if the sample was from a children's advocacy centre, if cases were required to be substantiated to be included, if a mixed abuse sample then what proportion included sexual abuse, if mixed age groups then what proportion was under 18, gender, ethnicity
• therapy characteristics: therapy narrative description, if caregivers were part of the therapy, definition of referral, definition of engagement, definition of completion
• data: total sample, referral data, engagement data, completion data
• independent variable narrative: significant differences, non-significant differences.

Five studies had relevant data on rates of referral, 19 studies had relevant data on rates of engagement, and 37 studies had relevant data on rates of completion of services. The reported rates for groups of studies are based on pooled proportions using a random effects proportional meta-analysis conducted in MedCalc (see below for meta-analysis methodology).

This research was undertaken to identify typical rates of referral, engagement and completion across different types of therapy services for children disclosing sexual abuse. The intent of this was to illustrate the typical proportions of children that are not benefiting from or not receiving the full benefit of therapeutic services to address harm from sexual abuse. We note that many survivors access therapy later in life; however, intervening early is critical to reducing the effects of trauma and the impacts of abuse across life domains.

Meta-analysis

Multiple proportional meta-analyses were conducted on the included studies to produce the pooled rates reported. Each of the rates was grouped based on characteristics that seemed likely to affect the referral/engagement/completion rates.

For referral rates, studies were grouped on whether they required cases to have been substantiated by authorities (i.e. police, child protection or some other authority), or whether abuse was suspected and still subject to investigation.

For engagement rates, studies were grouped into post-investigation - self and professional referral, post-investigation - specified professional referral, and therapy initiators. Two other groupings of studies were found that have been reported on (observational studies and children in care) but were not included as meta-analyses due to small numbers and these studies not being as relevant to the central questions of service access following disclosure. Post-investigation - self and professional referral refers to studies that examined engagement with therapy following an investigation regardless of whether a referral was made for a child. Post-investigation - specified professional referral refers to studies examining engagement related to a specific referral made by police, child protection or some other professional following an investigation. Therapy initiators refers to studies examining the rate of engagement for clients that have contacted specialist sexual abuse services; for these studies it generally is not known who made the referral.

For completion rates studies were grouped into either clinical samples, which required clinically significant symptomology to be included, or community samples, where services responded to clients regardless of if they met the threshold for clinically significant symptomatology. Clinical samples also tended to deliver a structured program of therapy as part of efforts to test the effectiveness of treatment, which also included screening out clients that may have ongoing issues in the home such as parental mental health, family and domestic violence, or parental substance abuse.

The proportional meta-analysis feature in MedCalc was used, reporting on the results of the random effects model, reflecting that the studies were heterogeneous in that rates were likely to be affected by the characteristics of the studies and samples. Publication bias testing was performed in Statsdirect 3.

The results of each of the meta-analyses are reported in Table A.1. The funnel plots for each of the tests are available from the author on request.

Note: * p = <.05

Appendix B: Therapy use

Notes: ^a Trauma-focused cognitive behavioural therapy; ^b Cognitive behavioural therapy