Validitating ASD Instruments for Use In Screening and Prevalence Studies

Background: There are no tested methods for conducting studies of the epidemiology of Autism Spectrum Disorders (ASD) in adult general population samples. The Autism Diagnostic Observation Schedule (ADOS) is widely seen as the best standardised observational instrument for autism diagnosis, but it has never been tested with an adult general population sample.

Objectives: To assessed the validity and reliability of the ADOS module 4, designed for use with verbally fluent adolescents and adults, in an adult general population sample

Methods: A random probability sample of adults aged 16 or over was interviewed throughout England in a multi-phase general population survey. A 20 item screening questionnaire - the Autism-spectrum Quotient (AQ-20) was completed by 7,353 adults in phase one. A random subset of  respondents with a greater probability of scoring higher on the AQ-20  completed second and third phase data collections including face to face ADOS module 4 assessments (n=618); the probability of selection

increased with AQ-20 score. First and second phase data and qualitative clinical information were presented as vignettes to six experienced clinicians working in pairs to make consensus clinical judgements about autism spectrum presentation. The probability of respondents having an ASD was compared across the three phases of data collection.

Results: There was moderate agreement between clinical vignettes raters and ADOS findings, and a range of diagnostic cut points were identified: ADOS Total Score 7+ (non specific ASD) to 13+ (autism) with greatest agreement at 10+ (AUC = 0.82). Modelling of the presence of ASD using 57 DISCO (diagnostic interview for social and communication disorders) assessments suggested an ADOS autism threshold in the range of 10 to 13 with highest AUC at ADOS 10+-+/11+ (AUC=0.92-0.93). ADOS cases could not be reliably predicted using AQ-20 screening data only.

Conclusions: Findings indicate that it is possible to use the ADOS as a valid instrument in epidemiological research in general population adults.  Clinically recommended ADOS module 4 algorithm thresholds for ASD classification are comparable with existing ADOS validity data on children, with this community study indicating threshold scores of 7+ for broader spectrum and 10+ for definite cases. Further work on adult population screening methods is recommended