Effects of Intensive Play-Based Toddler Intervention: A Multi-Site, Randomized Trials of the Early Start Denver Model

Background: Early detection is increasingly identifying ASD in the first two years of life, prioritizing interventions improving outcomes for infants and toddlers. A rigorous trial of the Early Start Denver Model (Dawson et al, 2010; Dawson, et al in press) demonstrated powerful effects on development, symptoms, and brain function using an approach that fused developmental, relationship-based, and applied behavior analysis concepts.

Objectives: Study objectives were (1) to attempt to replicate efficacy findings for ESDM using a multisite, randomized design and a younger, larger sample, and (2) to examine effects of biological, environmental, and child variables on outcomes. Hypothesis 1:  ESDM treatment will result in greater improvements in autism symptoms and developmental rates of progress than community treatment.  Hypothesis 2:  Individual child variables involving imitation and social attention, biological risk variables, and parent interaction styles will facilitate or modify child outcomes.

Methods: Ninety seven one year olds with ASD were enrolled at three different university sites, diagnosed, and randomized into ESDM or community intervention. Diagnostic and developmental evaluations were repeated at 12 months and 24 months after enrollment by assessors blind to assignment status. Children assigned to ESDM received 20 hours per week of 1:1 intervention in the home, and 3-4 hours per month of parent coaching for 24 months. Fidelity of treatment implementation and fidelity of assessment procedures were monitored quarterly. Children in the comparison group all received community interventions.   

Results: Data analysis by an independent data center (A. Fitzpatrick) will occur in December, 2012, after the final children have finished treatment.  (We are prevented from analysis by the DSMB until all children have finished treatment).  For Hypothesis 1: group comparisons on child outcome variables, we will use linear regression models to (1) predict the Year 1 and Year 2 outcome scores adjusting for the baseline score, the group assignment and possible confounders and to (2) test the main effect of group assignment. We will run the model in two ways: first adjusting only for the time 1 variables in the model for the time 2 and time 3 variables, and second involving adjustments for those variables in which there were initial significant differences between groups.

For hypothesis 2, we will use linear regression models to examine the effects of baseline age, imitation, social orienting, and joint attention, socio-economic status, number of biological risks, and parent interaction skills, in predicting cognitive outcomes and autism symptom reduction in the adjusted models.   

Conclusions: Very few high quality randomized controlled studies of early intervention in ASD have demonstrated efficacy, and the only such study involving toddlers with ASD involves the original trial of the Early Start Denver Model, which demonstrated very strong initial and long-lasting effects on developmental and autism outcomes (Estes et al, in review) as well as positive effects on brain function (Dawson et al, in press). However, independent replication is essential for demonstrating efficacy, and the findings from this study will have significant implications for the public sector involved in early intervention and clinicians as well as for neurodevelopmental scientists.