Diagnosis of Autism Spectrum Disorder: a Study of Community Practice

Background: Longitudinal research studies have demonstrated experienced clinicians using standardized assessment measures can make a reliable diagnosis of ASD in children under age 3. Limited data are available regarding the sensitivity and specificity of these measures in community settings.

Objectives: To characterize the practices of a group of community clinicians who evaluate young children for developmental disorders and investigate the agreement between clinical judgment and results from standardized diagnostic measures in a community sample.

Methods: Clinical records for 145 children seen for possible ASD or social/language concerns were examined. The children (125 males and 20 females) were between 24 and 36 months of age (M = 30.2 months, S.D. = 3.2). An IQ estimate was available for 91% of the children (overall mean IQ= 71.64, S.D.=16). Classification results obtained from standardized diagnostic measures were compared with clinical final diagnosis. A standard method of record review based on the MADDSP protocol was also used to examine the diagnostic descriptions and the specific behaviors/symptoms that clinicians included in the reports. Reviewers using this method were blind to classification results from the diagnostic measures and clinical diagnosis. 

Results: During the 23-month period that was examined, the Autism Diagnostic Observation Schedule (ADOS) was used with 123 (85%) children and the Social Communication Questionnaire (SCQ) was adopted at a later time and used with 61 children (42%). Across the three diagnostic categories (autistic disorder, PDD-NOS, or nonspectrum), the ADOS had good sensitivity (.95) and specificity (.88). The SCQ had weak sensitivity (.64) but moderate specificity (.82). The Modified Checklist for Autism (M-CHAT) was used with only 27% of the children but had good sensitivity (.88) and weak specificity (.67). For children with a clinical diagnosis of autistic disorder, there was 80% agreement between the ADOS and SCQ or M-CHAT but only 55% for children with PDD-NOS had agreement between these measures. In 70% of cases without an ASD diagnosis, scores below the cutoff on both types of measures were obtained. There was agreement between clinician diagnosis, record review classification, and reviewer diagnosis across the majority of cases. The majority of discrepancies (70%) were between a clinical diagnosis of PDD-NOS and a record review classification of autistic disorder. Child factors, such as strengths and level of functioning, appeared to account for most of these discrepancies.

Conclusions: These results demonstrate that following best practice guidelines, standardized measures such as the ADOS and SCQ, can be successfully incorporated into clinical practice, have relatively good sensitivity and specificity, and worked well with a referred sample of two-year-olds. The standard record review approach allows for an examination of child characteristics that may influence diagnostic decisions.