Regression in autism: Prevalence and associated factors in the CHARGE study

Objective.-The aim of this study was to examine the prevalence of regressive autism and associated demographic, medical, and developmental factors by using 2 different definitions of regression based on the Autism Diagnostic Interview, Revised. Methods.-Subjects were aged 2 to 5 years, with autism (AU) or autism spectrum disorder (ASD) confirmed by standardized measures. Children with regression, defined as a) loss of both language and social skills or b) loss of either language or social skills, were compared with each other and to children with AU or ASD with no reported loss of skills on developmental and adaptive functioning. Parents reported on seizure, gastrointestinal, and sleep concerns. Results.-Fifteen percent (50/333) of the combined AU-ASD group lost both language and social skills; 41% (138/333) lost either language or social skills. No differences were found between the 2 samples of children with regression. Few developmental, demographic, or medical differences were found between the combined regression group and children without loss of skills, in both the larger AU-ASD sample and the more homogeneous AU-only sample. Children with regression had significantly lower communication scores than children without regression. Conclusions.-The prevalence of regression in a large sample of young children with AU and ASD varies depending on the definition used; requiring loss of language significantly underestimates the frequency of developmental regression. Children with regression performed significantly less well than those without regression on 2 measures of communication, but the clinical meaningfulness of these differences is uncertain because of the small effect sizes.