Effectiveness of Community-Based Early Intervention for Children with Autism Spectrum Disorder- a Meta-Analysis

Background: Research trials of early intervention programs for children with ASD generally produce medium-to-large gains, on average, compared to “treatment as usual,” in cognition, communication, social ability, and adaptive behavior (Hedges’s g 0.4 – 1.2). Almost all children with ASD receive their treatment through community-based services, however, and previous research has suggested that evidence-based interventions rarely make their way into community practice. Understanding the effects of community-based early intervention is the first step in developing strategies to improve wide-scale implementation of effective early intervention.

Objectives: To estimate the average effects and predictors of cognitive, social, communication, and adaptive behavior outcomes among children receiving community-based early intervention.

Methods: Studies of community-based early intervention for children with autism were identified through a systematic search of online databases, hand-searching relevant journals, and reviewing references. Community-based early intervention was defined either as “treatment as usual” control groups in randomized or quasi-experimental trials or in studies that explicitly examined treatment outcome in community settings. Changes in cognitive, communication, social, and adaptive behavior functioning from pre-treatment to post-treatment were assessed using standardized mean gain scores. Uncontrolled effect sizes (Hedges’s g) were calculated by dividing the mean change from pre-to post-treatment by the pooled standard deviation of the difference score. Overall effect sizes were estimated using random effects models. The Q-statistic and the I² index were used to examine heterogeneity of effect sizes. Moderators of interest included type of community EI program, year of publication, intervention duration, and total intervention hours, sample selection methods, age at intake, their effects were assessed using analysis of variance of mixed-effects models and meta-regression analyses.

Results: Forty groups from 29 studies met inclusion criteria (1496 participants, mean age 37.4 months, 84.9% male). There was significant improvement in each of the four domains; however, the gains were small. Hedges’s g ranged from 0.21 for adaptive behavior to 0.31 for communication outcomes, after removing outliers and correcting for publication bias. “Model” early intervention programs (e.g., those associated with universities and hospitals) were generally superior to other community early intervention program types across all four outcomes (Hedges’s g ranged from 0.4 - 0.5 for “Model” programs and from 0.01 – 0.3 for other community programs). Only communication outcomes demonstrated increasingly larger effect sizes in more recent years. Intervention duration and total intervention hours were negatively associated with effect sizes for communication and adaptive behavior outcomes. Studies in which participants were randomly assigned to receive community treatment had smaller effect sizes for social outcomes than studies utilizing other treatment group allocation strategies. Age at intake did not significantly moderate outcomes in any of the four domains.

Conclusions: These results indicate that there remains a large gap between research and community practice, and that for most outcomes community-based early intervention has not improved over time. Further research is needed to close the gap in outcomes observed in children receiving treatment in randomized trials and those observed in children receiving treatment in community settings.