Results of an Autism Screening Study Among Adults Receiving Community Mental Health Services

Background: Adults with ASD may be misdiagnosed, or their psychiatric comorbidities may be recognized without recognizing the ASD.  Studies report that 10% of psychiatrically hospitalized adults in US and 6% in the UK have undiagnosed ASD. Studies from the UK, Sweden and Taiwan find 0.6 ‐ 4% of adult patients with ASD in outpatient settings. The prevalence of ASD in adult US outpatient psychiatric settings is unknown. Accurately diagnosing ASD is challenging for community clinicians. ASD diagnostic tools are lengthy, most are validated only in children, and they require considerable training to administer correctly. Therefore they are not feasible for use with adults in community mental health settings. 

Objectives: 1) Estimate the prevalence of undiagnosed ASD in community outpatient settings; 2) compare clinical characteristics, treatment and assigned diagnoses of adults with ASD with those of adults with other psychiatric disorders; and 3) develop and test a brief tool for community mental health practice to screen for ASD in adults. 

Methods: Staff from three community mental health centers completed the Social Responsiveness Scale, Adult Version (SRS-A) and the Autism Spectrum Disorders in Adults Screening Questionnaire for their adult clients. Our research team conducted chart reviews to extract information found to discriminate adults with ASD from adults with other psychiatric disorders in prior studies. A subset of these clients, enriched for those with SRS-A scores >60, completed in-person evaluations. This evaluation included Module 4 of the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2), a brief clinical interview about ASD-related characteristics not assessed by the ADOS-2, and the Overview and Psychosis modules of the Structured Clinical Interview for DSM-IV-TR Axis I Disorders (SCID). Clinical case conferences were conducted for all participants to make a final determination of ASD. Item response theory was used to reduce the number of items on the SRS. Logistic regression and recursive partitioning were used to identify the most parsimonious set of elements needed to accurately screen for and diagnose ASD.

Results: 1134 adults were screened, 813 of whom were not receiving developmental services, and therefore counted towards our denominator. To date, 296 charts were reviewed and >58 clinical interviews were conducted (35 more expected, enriched for autism). Seventeen individuals were already diagnosed with ASD and an additional 21 had a mention of ASD in their charts. Two met research criteria for ASD, although 22 had scores on the ADOS indicative of ASD. 18 items on the SRS, selected through item response theory, had a .95 correlation with the SRS total score. Analyses examining the combined accuracy of the SRS plus additional items pulled from charts will be completed by the time of submission. 

Conclusions: Study results suggest that the prevalence of ASD among adults receiving outpatient psychiatric care is approximately .5%, although challenges in interviewing more severely impaired participants suggest that this is a lower-bound estimate. Traditional screening tools can be substantially shortened with no reduction of accuracy in this population, especially when combined with other data elements easily extracted from clinical charts.