Armelius BÅ, Andreassen TH. Cognitive-behavioral treatment for antisocial behavior in youth in residential treatment. Cochrane Database of Systematic Reviews 2007, Issue 4. Art. No.: CD005650.
Most clinicians would agree that the treatment of anti-social youth should ideally involve the family and community, as well as the individual. For youth in residential care however, this is rarely feasible. The typical treatment they receive, if any, is some form of cognitive-behavioural therapy (CBT). The object of this meta-analysis is to determine whether CBT is effective in reducing recidivism among conduct problem adolescents receiving out-of-home care (e.g., treatment centre, open or closed detention centres, group homes). Previous efforts either included studies of adult populations or adolescents living at home.
Electronic databases were searched for evaluations of CBT programs for youth with serious conduct problems. Studies were retained if (1) the treatment contained cognitive and behavioural components, (2) the comparison group either received no special treatment other than the standard services offered in a residential facility, or a less intensive therapy, (3) the study examined adolescents in residential settings, and (4) the outcome indicators included a measure of recidivism. Because the analysis was limited to well-designed studies, the original pool of 94 studies was whittled down to 12, more than half of which were not RCTs (i.e., randomized assignment to treatment and control groups). Given the widespread use of CBT for treating young offenders, the scarcity of quality outcome studies is in itself disconcerting.
Examined separately, the average study did not garner a significant difference between groups. However, when the results of the studies were pooled, the authors found a significant improvement at the 12-month follow-up; CBT was found to be 10% more effective in reducing recidivism. Such a small effect size might explain why significant effects were unlikely to be detected in a lone study. The pooled results did not garner significant treatment effects at the 6- and 24-month follow-up. A 6-month follow-up might be too brief a time frame to evaluate recidivism. The authors also warn that the finding of "no treatment effect at 24 months" should be interpreted cautiously because it was based on a relatively small number of studies. This limits the capacity of the meta-analysis to detect small effect sizes. Moreover the meta-analysis is more likely to be swayed by the results of a study with extremely divergent results. Publication bias might also have affected the findings.
This systematic review followed a prescribed and transparent method of retrieving, appraising and synthesizing empirical studies relevant to the research question. This study was vetted by both experts in the field of practice and by experts specializing in data synthesis methodologies. Systematic reviews of existing literature are increasingly being used to evaluate existing research evidence while addressing many of the biases inherent in narrative reviews. Systematic reviews differ from narrative reviews because they are more rigorous in information retrieval strategies; they follow an explicit and transparent criteria for appraising the quality of existing research evidence; they attempt to identify and control for different types of bias in existing studies; and they have explicit ways of establishing the comparability (or incomparability) of different studies.