Toward a developmental operational definition of autism.

Traditional approaches to diagnosing autism emphasize delays in communication and socialization. Traditional diagnostic schemes typically list symptoms (e.g., lack of eye contact), but provide little guidance on how to incorporate information about developmental level in making a diagnosis. Because standardized measures of adaptive behavior can provide information about children's communication, socialization, and other behavior relative to their age, they may be useful tools for diagnosing autism. This study investigated the ability of the Vineland Adaptive Behavior Scales to identify children with autism. Vineland scores and measures of intellectual functioning were obtained for children with autism, PDDNOS, and other developmental disorders (DD). Discriminant function analyses indicated that the autism and combined nonautism (PDDNOS and DD) groups could be differentiated on the basis of socialization, daily living skills, and serious maladaptive behaviors. Socialization alone accounted for 48% of the variance in diagnosis. Using regression analyses derived from a large normative sample, adaptive behavior scores were predicted from chronological age (CA) and mental age (MA). Socialization scores in the autism group were substantially below the level predicted from CA or MA. An index derived from the ratio of actual to predicted socialization scores correctly classified 86% of both autism and nonautism cases. Findings suggest that comparison of obtained Vineland socialization scores to those predicted by CA or MA may be useful in clarifying the diagnosis of autism.