Rapid, Feasible Observational Paradigm for Confirmation of Autism Spectrum Disorders

Background:  Traditional methods for the clinical confirmation of autism spectrum disorders (ASD), which have been increasingly adopted in the U.S. as prerequisites for both service eligibility and research participation, are expensive and difficult to acquire consistently in public health settings.  They are predicated on categorical conceptualization of autistic syndromes, and are unfeasible for repeated-measures approaches to measuring response to intervention.  

Objectives:  This study represents a critical step in a series of efforts to substantiate a quantitative measurement system feasible for use in public health settings. The question addressed by this report is whether traditional paradigms implemented to establish clinical confirmation—given a strong level of suspicion of ASD engendered by the results of brief questionnaires encompassing developmental history and current symptomatology—might be achieved using a similarly brief observational paradigm that could be employed in general clinical and educational settings by clinicians and educators who are not specifically trained in the administration and scoring of extensive diagnostic rating scales. 

Methods:   In a sample of 64 subjects with a diagnosis of an autism spectrum disorder, we tested the ability of clinicians (including a school psychologist and a post-doctoral clinical psychologist) to discriminate level of severity of autistic symptomatology using specific adaptations of the Childhood Autism Rating Scale-2 (CARS-2).  Adaptations included 1) clinicians coded the CARS-2 based exclusively on a 15 minute videotaped observation (no other subject data was made available to the coders), in which  an examiner a) attempted to engage the subject in conversation or in simple imitative play; b) engaged in sustained symbolic interactive play with the subject; and c) initiated a transition to a sensory activity; 2) autistic severity indices were generated exclusively by the first 8 items of the CARS-2 which cover social impairments and stereotypic behavior; 3) the remaining items were used as a screen for the presence of co-morbid or alternate conditions, but did not figure into the total severity score for autistic impairment. 

Results:  Clinician’s quantitative severity ratings reliably captured levels of gradation in severity ascertained independently using the Autism Diagnostic Interview-Revised (ADI-R; correlation with ADI-r social impairment score, r=.60); and exhibited item-level and scale-level inter-rater reliability on the order of 0.85. CARS-2 scale scores extrapolated from the sub total for items 1-8 were well within the published range for autism spectrum conditions. 

Conclusions:  Standardized clinician ratings based on brief semi-structured observations of interpersonal behavior—without the need for extensive rater training—show tremendous promise for the diagnostic confirmation of ASD, especially when combined with rapidly-obtainable information on developmental history and current symptomatology in daily social contexts. Such observations could constitute a key component of a highly cost-effective strategy for diagnosis, and could facilitate the acquisition of repeated measures data, which is vital to the ascertainment of response-to-intervention in research, educational and public health settings.