Predictors of Outcome In a Large, Community-Based Intensive Behavioural Intervention Program

Predictors of Outcome in a Large, Community-based Intensive Behavioural Intervention Program

Background: Intensive Behavioural Intervention (IBI) for young children with autism is widely regarded as an efficacious treatment but much less is known about the effectiveness of this intervention in community settings (where it typically occurs).  The Provincial IBI Effectiveness Study (Perry et al., 2008) examined outcomes in the nine programs across Ontario, Canada, and showed that the majority of children made progress but that outcomes were highly variable, a finding reported in virtually all IBI research.  Thus, it is of great importance to try and determine what accounts for this heterogeneity.  

Objectives: The purpose of the study was to examine the importance of four predictors, measured at intake to IBI: age, IQ, adaptive behaviour level, and severity of autism.  Outcome variables were: autism severity, adaptive behaviour, cognitive level, rate of development during IBI, and categorical outcome (seven subgroups).  The second objective was to examine two subgroups further, children with the most optimal outcomes (Average Functioning) and the group who made little or no progress. 

Methods: This was a retrospective, file-review study of 332 children with an ASD aged 2 to 7 at entry to treatment, whose outcomes were described by Perry et al. (2008).  Measures included adaptive level using the Vineland Adaptive Behavior Scale (VABS), cognitive level using one of several cognitive tests (Mullen, Stanford-Binet, WPPSI, etc.), and autism severity using the Childhood Autism Rating Scale (CARS). 

Results: All four predictors were found to be related to children’s outcomes.  Age at entry to IBI was an important predictor but not in a simple, linear fashion.  Children who began treatment younger than 4 years, versus over 4, had significantly better outcomes on all variables (except VABS Motor).  However, a comparison of the age at entry of children subsequently classified in the seven outcome groups revealed that the Average Functioning group had been substantially younger at entry and none of the other groups differed from one another.  Cognitive level (available for a subset of children) was a strong predictor and, in regression analyses, accounted for the largest proportion of variance in outcomes.  Adaptive level was also a predictor of most outcomes.  Autism severity was a significant predictor but accounted for quite a small amount of variance.  Regarding the second objective, there was a fairly clear set of predictors for children who had achieved Average Functioning (they were younger, milder, and higher functioning) but initial variables were not found to be good predictors of children who had experienced poor outcomes.

Conclusions: All four predictors were found to be related to children’s outcomes to some extent.  The results have implications for eligibility decisions (especially related to the importance of beginning intervention early).  However, there was considerable heterogeneity that went unaccounted for, suggesting other child characteristics not measured here, and/or other factors such as treatment quantity/quality and family factors may also be important in accounting for children’s variable outcomes.