Identifying the Active Ingredients in Intensive Behavioural Intervention Programs for Children with Autism

Background: Research demonstrates that a subset of children with autism show significant improvements following Intensive Behavioural Intervention (IBI), while other children demonstrate more modest gains.  Given the heterogeneity of the population, variable outcomes are not unexpected.  This variability is likely attributable to a combination of child, family and treatment characteristics, the impact of which remain poorly understood.  Although several child and family characteristics have been linked to best outcome, research on treatment variables has focused nearly exclusively on quantitative treatment aspects, such as treatment intensity and duration.  Little exploration into qualitative treatment variables has occurred.  Despite the lack of systematic review, a variety of treatment characteristics are believed to be integral to effective IBI, such as well trained/highly supervised staff and parent involvement. However, more research needs to be done to precisely identify the “active ingredients” of treatment.

Objectives: The purpose of this research is to identify treatment variables associated with best outcomes, as well as identify variables linked to less optimal outcomes.  Further, a comprehensive review of the literature will be presented, including classification of studies based on methodological rigour (i.e., randomized controlled trial, comparison group, pre-post comparison).

Methods: This poster presents findings from an exhaustive review of the IBI literature, including analysis of efficacy and effectiveness studies conducted in the past 20 years. Studies were reviewed independently by two people and rated according to methodological rigour.  Studies were also independently coded on 13 treatment variables (e.g., model/amount of supervision, treatment setting).  Treatment effect sizes were determined for four outcome measures, cognitive and adaptive level, severity of autism and communication.  Associations between effect sizes (dependent variables) and treatment characteristics (independent variables) were then determined.

Results: Majority of studies used pre-post comparison designs with no control group or involved non-random assignment comparison groups.  Several studies provided insufficient information about treatment variables preventing detailed analysis on some treatment characteristics (e.g., type/fidelity of parent involvement).  Few studies reported on the impact of treatment on autism symptoms preventing an analysis of this outcome variable.  Treatment provided in "Home AND Community/School Settings" or "Integrated Settings" was associated with significantly larger cognitive and language outcome effect sizes.  Adaptive skills showed the reverse pattern with significantly larger effect sizes associated with children receiving treatment in "Home Only". Studies which measured treatment fidelity were associated with significantly larger IQ effect sizes.  While supervision conducted live with the child and therapy team was associated with larger effect sizes for cognitive skills, adaptive skills showed the reverse pattern (i.e., larger effect sizes associated with indirect supervision).  Finally, the largest IQ effect size was associated with frequent (i.e., <weekly supervision).

Conclusions: Although several treatment characteristics were associated with significant effect size differences for cognitive and adaptive level and language skills, the pattern was inconsistent. Further, variability in outcome data collected and descriptions of treatment characteristics limits analysis. Future research should include standardized outcome measures and report treatment characteristics in sufficient detail to enable cross study comparisons and effect size analysis.