Cognitive Behavioral Therapy for Adolescent Offenders in Custody

Study 1

Evaluation Methodology

Design: A multi-site RCT design was used to test the effectiveness of the intervention and also assess the viability of offering an intervention to this population. Five trial sites were used; four Secure Children's Homes (SCH) and a Young Offender Institution (YOI). In practice the majority of recruits were from one of the SCHs and the YOI. The trial had two arms; CBT and TAU (treatment as usual). TAU consisted of any individualized interventions for emotional or mental health problems routinely offered at that specific trial site, and the specific interventions varied from site to site. At the largest trial site (the YOI), interventions included counseling and support from mental health nurses, who would have 5-10% of the population (approximately 400 young people) on their caseload at any given time. Due to the secure settings in which this research was conducted, ethical approval was obtained not only from the North West Multicenter Research Ethics Committee, but from management at each individual site.

Referral was made by key workers at the SCH or health care workers at the YOI on the basis of observed psychological or emotional problems. Inclusion criteria were that the young person currently recognize they had a problem and that they wanted help for this problem (i.e. subjective rating of problem severity and motivation to change). There were no exclusion criteria.

Staff at the trial sites identified 66 potential candidates. At recruitment interview eleven of these young people did not meet the inclusion criteria; either they did not think they had a problem or they did not want help for any problems they did have. Three were discharged to the community before randomization, two refused to participate and ten could not gain parental consent despite a personal willingness to participate in the trial. Forty young people went ahead into randomization. At baseline 22 participants were in secure care establishments and 18 were in young offender institutions. Originally, 19 participants were assigned to the intervention and 20 to the control group. Two young people dropped out of the study before the follow up interviews; one person from each of the CBT and TAU groups.

Participants were randomly allocated to the CBT (n=18) or TAU (n=20) group; the trial coordinator used an independent trials office at a local hospital for this purpose. The interventions offered differed between trial sites, as did the age range of young people they accommodated, consequently participants were streamed (or stratified) by age (under 16 years, 16 years or over) and by trial site but otherwise randomly allocated to either arm of the trial. The study researcher carried out all the assessments pre and post therapy, and was blind to the outcome of the randomization.

The intervention was delivered by the trial therapist, a mental health practitioner experienced in working with young offenders with mental health problems. The therapist worked with each trial site to identify the most appropriate space in which to meet with the youth within the site and maximize therapeutic results. Sessions were provisionally arranged on a weekly basis by negotiation with the young person and caretakers. However, gaining regular access to participants proved problematic at some sites due to operational and security issues at the centers and appointments would have to be rescheduled.

All participants were reassessed at a mean of 11.1 months follow up using the Youth Self Report, Salford Needs Assessment and Difficulties and Coping Profile. Follow-up had been planned for 6, 12 and 18 months post treatment but tracking the participants following release from custody proved difficult in many cases and disrupted the follow-up schedule. Consequently data from several time points were aggregated prior to analysis. In total, 64% of the sample (24 young people) was followed up within twelve months, however the time between posttest and follow-up varied between participants.

Sample: The sample was all male and the majority 39 (97.5%) was white. The mean age of the sample at the time of study was 15.58 years and the mean age at first offense was 11.28 years. Eighteen participants (45%) had a history of being in care. There were high levels of school exclusion, 38 out of 40 had been excluded from school. Just under two thirds (62.5%) of the sample met the SNASA criteria for an unmet mental health need severe enough to justify an intervention; 40% had at least two unmet mental health needs. Participants had an average IQ of 76.25, with nine participants having an IQ of less than 70.

Measures: Demographic information was collected from both a blind source interview and ASSET, an assessment of needs carried out on all young people in contact with the justice system. Demographic data included: gender, age, ethnicity, school attendance, offending behavior, living arrangements and family circumstances. Reliabilities for the scales are given where provided by the authors; reliabilites were not given for all scales.

The salford needs assessments schedule for adolescents(SNASA): SNASA is a structured interview specifically designed to assess the psychosocial needs of adolescents in domains including mental health problems, relationships and high risk behaviors, and has been previously been used to assess the needs of adolescent offenders. Test-rest reliability coefficients range from 0.73 to 0.85.

Structured assessment violence risk-youth (SAVRY): SAVRY is an evidence based risk assessment tool that combines historical (for instance, previous violent offending), social/contextual (peer delinquency) and individual (negative attitudes) risk factors, with protective factors (strong family bonds).

Youth self report (YSR): YSR is a self-report questionnaire for 12-18 year old adolescents with items assessing difficulties in several domains such as internalizing and externalizing problems that can be summed to give a total score measuring overall behavioral and emotional functioning.

Wechsler Abbreviated scale of intelligence (WASI): The WASI was designed to provide a brief and reliable measure of intellectual functioning for individuals aged 6-89 years. The WASI consists of four subtests: vocabulary, block design, matrix reasoning and similarities. These are scored to provide three measures of IQ, which are verbal intelligence, performance intelligence and full-scale intelligence. Scores are measured against a normative sample with a mean score of 100.

Difficulties and coping profile (DCP): This brief instrument was developed for the study, and uses broadly the same categories as the Mental Health Screening Questionnaire Interview for Adolescents (SQIFA). Unlike the screening tool, it is not an objective measure of problem severity; it was used is to detect any changes perceived by the young person (either in problem severity or ability to cope) as a result of the therapy.

Analysis: All analyses were carried out on an intent-to-treat basis, including all participants randomized in the trial who had completed follow-up measures. Baseline differences between CBT and TAU groups were examined using t -tests; variables that distinguished groups were used as covariates in the analysis. Cohens d effect sizes were calculated for all t -tests. Repeated measures analysis of variance (ANOVA) was conducted on scores assessed at follow-up. Statistical significance was set at p < 0.05. Repeated measures analysis was used as it can be run with relatively small sample sizes. Effect sizes for the repeated measures were also calculated. The effect size was calculated based on comparisons between the mean of the experimental group at T2 and the mean of the control group at T2 taking into account the correlation between the scores at T1 and T2. Using the effect sizes post-hoc power calculations are reported. Analyses were conducted using the Statistical Package for the Social Sciences (SPSS) version 14.

Outcomes

Treatment Fidelity: The trial psychologist provided specific training for three months prior to the study and weekly supervision during the study of the therapist to ensure treatment fidelity. The trial therapist also had weekly supervision from a consultant adolescent forensic psychiatrist to review participants' mental states and identify those meeting ICD-10 diagnostic criteria, and ensure effective case and risk management.

Baseline Equivalence: At baseline, 19 participants completed the CBT arm of the study and 21 were in the TAU group (routine care). At baseline (T1), there were no significant group differences between the CBT and TAU in terms of age or IQ. There were also no significant group differences in the proportions rated low, moderate or high on the assesment of the risk of youth violence. On the assessment measures there were no significant differences in total scores at baseline on the self report or needs assessment need and unmet need scores.

Outcomes:
Changes in needs (SNASA): On the needs assessment measure, there were no significant differences in the total number of needs at T1 and at follow-up (T2) for either the CBT or TAU groups. Despite these non-significant results the CBT group showed a decrease in the number of needs over time and the TAU group showed an increase in the number of needs. Despite non-significant findings, there was a larger reduction in unmet needs for the CBT group than the TAU group. An examination of each of the SNASA domains indicated that there were no significant group, time or interaction effects for mental health, violent behavior, risky behavior, education or relationships.

Changes in behavioral and emotional functioning (YSR scores): Based on Youth Self Report scores, on all occasions, the scores reduced between the two time points. The CBT group had a larger reduction in YSR External problem score at follow up than the TAU group, and this is supported by a large effect size (-0.87).

Difficulties and coping profiles questionnaire (DCP): Repeated measure analysis of variance of these domains did not show any significant effect of group or group × time interaction. On the coping questionnaire, there were a number of significant changes over time, but no group differences were found, suggesting that both groups improved similarly For example, scores for problem severity in relation to drug and alcohol problems, depressive symptoms, anxiety symptoms and self-harm and suicidal ideation decreased in both groups over time. Significant increases in coping ability over time were found across groups for drug and alcohol problems, depression, and anxiety. No significant changes were noted for problems associated with post-traumatic stress and aggressive behavior.

Dose Response: The relationship between compliance levels in CBT and levels of need at follow up was investigated using Spearman's correlation co-efficient. There were no significant correlations between the level of compliance and the total numbers of needs at follow up or between compliance levels and total numbers of unmet needs at follow up. There was also no relationship between compliance and YSR internal problems score, compliance and YSR external problems score, or compliance and YRS total score at follow up. The weak correlations imply that dose amount matters little for the outcomes.

No mediators were investigated.