Prediction of Outcome after One Year of Early Intensive Behavioral Intervention

Background: During early intensive behavioral intervention (EIBI), which begins prior to the age of five years and involves 20-40 hours per week of treatment, some children with autism make large developmental gains. However, outcomes vary widely across children, and reliable predictors of differential response are unavailable for this expensive treatment.

Objectives: Examine predictors of outcome after one year of EIBI in community agencies
Methods: We conducted pretreatment and one-year follow-up assessments of 63 children with autism (53 males) who entered EIBI in publicly funded community agencies at age 1.96-4.85 years. Intervention fidelity was monitored via direct observation of discrete trial training (DTT). Predictor variables included motor stereotypy, social approach, preverbal communication (joint attention and social orienting), and imitation. Outcome variables were IQ (Mullen Scales of Early Learning), adaptive behavior (Vineland Adaptive Behavior Scales), and autism symptoms (Autism Diagnostic Observation Schedule revised algorithms). For each outcome measure, hierarchical multiple regression was performed, entering treatment sites, intake CA, and IQ before predictors.

Results: DTT procedures were implemented with 95% accuracy, indicating acceptable intervention fidelity. Participants’ IQ rose modestly from pretreatment, M(SD) = 59.95 (13.80), to the one-year follow-up, M(SD) = 65.33 (18.45), F(1, 62) = 11.88, p = 0.001, and autism symptoms decreased, pretreatment M(SD) = 18.16 (5.09), follow-up M(SD) = 16.44 (5.39), F(1, 62) = 16.44 (5.39), p = 0.002. However, adaptive behavior did not change significantly. Pretreatment CA predicted follow-up IQ (β = 0.35, p = 0.02) but not other outcome variables. Pretreatment IQ and motor stereotypy did not predict any outcome variable. Because social approach, preverbal communication, and imitation formed one factor, they were combined and predicted improvement in IQ (β = 0.40, p = 0.005, R = .43) and adaptive behavior (β = 0.51, p = 0.002, R = .44) above and beyond other variables, p < .02.  However, motor stereotypy did not predict any outcomes, including increase in independently assessed stereotyped behaviors. No variables successfully predicted autism symptoms.

Conclusions: Social approach, preverbal communication, and imitation may constitute one factor (social communication) that predicts follow-up scores on developmental tests for children with autism who receive EIBI.