Abstract: The Empirical Status of Interventions for Youth with Conduct Problems and their Families: An Overview of Meta-analyses (Society for Social Work and Research 14th Annual Conference: Social Work Research: A WORLD OF POSSIBILITIES)

12235 The Empirical Status of Interventions for Youth with Conduct Problems and their Families: An Overview of Meta-analyses

Schedule:
Friday, January 15, 2010: 4:00 PM
Pacific Concourse M (Hyatt Regency)
* noted as presenting author
Michael Vaughn, PhD , Saint Louis University, Assistant Professor, St. Louis, MO
Christy Litschge, MSW , University of Pittsburgh, Doctoral Candidate, Pittsburgh, PA
Background and purpose: Substance abuse and conduct problems, occurring during childhood and adolescence possess far reaching effects on future life-course outcomes. These problems are not only deleterious for children and youth, but also have implications for families and society. As treatments grow in number, meta-analytic reviews are useful methods for taking stock of findings. Meta-analyses collect studies pertaining to a given topic or research question and integrate findings across studies in order to arrive at a research summary. The goal of the present study was to identify and summarize treatment effects for children and adolescent conduct problems (e.g., aggression, violence, and substance abuse) based on accrued meta-analytic studies.

Methods: We selected meta-analytic studies of treatments for children and youth antisocial behavior according to eligibility criteria established a priori. Studies included were: a) investigations whereby the population or sample were antisocial children and youth (juvenile offenders, aggressive children and youth, court referred children and youth) or deviant behaviors (substance abuse, violence, aggression, delinquency, externalizing and disruptive behaviors) for the purpose of altering the child's or adolescent's behavior, even if the intervention targets parents; b) peer-reviewed, c) quantitative in nature, d) reported an effect size, and were e) published in English. Systematic search and coding procedures were used to identify and summarize studies from the time period of 1980-2007. Prior to final coding, the co-authors piloted a coding form for relevance and practiced until inter-rater agreement based on the kappa statistic was excellent ( > 0.75). We report the effect sizes for each investigation by outcome and treatment where available. In order to enhance interpretive value we report U3 scores for each of the effect sizes.

Results: A total of 26 meta-analytic reviews comprised of nearly 2,000 studies involving over 100,000 antisocial children and adolescents revealed substantial variation in effect sizes. Reviewed treatments were broad and varied but tended to cluster around five types: 1) individual behavioral/cognitive behavioral, 2) family, 3) multimodal, 4) group-based, and 5) a miscellaneous category (e.g., pharmacotherapy). Effect sizes varied and ranged from a low of d = -0.15 for humanistic therapy to a high of d = 1.15 for rational emotive therapy. The majority (45 percent) of effect sizes was small-to-medium in magnitude (d< 0.36). In terms of the distribution of effect sizes (N = 129), the mean is d = 0.44 (SD = 0.32). Family-based treatments for dually diagnosed adolescents across externalizing, internalizing behavior, and substance abuse outcomes (d=0.97, d=1.00, d=1.13, respectively) were comparable to individual cognitive problem-solving (d=0.99, d=0.80, d=0.92) across the same three outcomes.

Implications for Policy and Practice: Although there is considerable variation in effect sizes the results seem to demonstrate evidence for equifinality. That is, multiple types of treatments appear to be capable of working for the same problem. Further, these effects are sturdy across a number of interrelated outcomes. Practitioners who work with children and adolescents should be aware of the range of evidence-based treatments available for conduct problems. Limitations of taking stock of treatment outcome literature via meta-analyses are highlighted.