A Population Level Study of the Effects of Early Intervention for Autism

Billions of dollars are spent each year on early diagnosis and intervention programs for autism. However, there is little reliable evidence about the effectiveness of these programs. Few studies that evaluate early interventions for autism use random assignment or quasi-experimental designs, and studies of the effects of early intervention programs have relied on small, selected samples that lacked power to detect even moderate associations. A recent meta-analysis by Spreckley and Boyd (2009) on the efficacy of applied behavior intervention in preschool children with autism found that compared with standard care, applied behavior interventions did not significantly improve the cognitive outcomes of the children in these programs. Using population-level data of all children with autism spectrum disorders who were born in the state of Florida between 1994-2002, we evaluate the effects of a free, statewide early diagnosis and intervention program for autism called Early Steps. Families can receive autism diagnoses from one of 18 Early Steps centers located around the state; we make use of distance to the nearest Early Steps center as an instrument for receipt of autism services prior to a child?s fourth birthday. The first stage is very strong: Children living in the same community as an Early Steps center at the time of birth are nearly twice as likely to receive early services as those In communities more than 30 miles away from a center. We use instrumental variables methods to determine whether early diagnosis and intervention impacts (1) short term outcomes, such as kindergarten readiness scores and attending kindergarten on time, (2) grade repetition, (3) significant behavioral problems, and (4) longer term cognitive outcomes, including elementary school test scores. Preliminary results show strong, significant effects of early intervention for autism by age four on attending kindergarten on time, and on third and fourth grade FCAT (Florida?s Comprehensive Assessment Test) scores. In addition, children who have had early intervention for autism by age four via Early Steps are significantly less likely to have a behavioral incident at school or to be suspended, and have fewer days of suspension than children with later diagnoses of autism. This study is the first population-level study of the effects of early intervention on autism. In addition, this is the first evaluation of a statewide free early diagnosis and intervention program for autism. Finally, this is the first study to examine the effects of early intervention for autism on school-based cognitive and behavioral outcomes. Thus, this study will hopefully lend insight into how policies that provide free treatment for autistic individuals can lead to a variety of positive developmental outcomes for these children.