Diagnosing Autism Spectrum Disorder in community settings using the development and well-being assessment: validation in a UK population-based twin sample

BackgroundIncreasing numbers of people are being referred for the assessment of autism spectrum disorder (ASD). The NICE (UK) and the American Academy of Pediatrics recommend gathering a developmental history using a tool that operationalises ICD/DSM criteria. However, the best-established diagnostic interview instruments are time consuming, costly and rarely used outside national specialist centres. What is needed is a brief, cost-effective measure validated in community settings. We tested the Development and Well-Being Assessment (DAWBA) for diagnosing ASD in a sample of children/adolescents representative of those presenting in community mental health settings. MethodsA general population sample of twins (TEDS) was screened and 276 adolescents were selected as at low (CAST score<12; n=164) or high risk for ASD (CAST score15 and/or parent reported that ASD suspected/previously diagnosed; n=112). Parents completed the ASD module of the DAWBA interview by telephone or online. Families were visited at home: the ADI-R and autism diagnostic observation schedule (ADOS) were completed to allow a best-estimate research diagnosis of ASD to be made. ResultsDevelopment and Well-Being Assessment ASD symptom scores correlated highly with ADI-R algorithm scores (=.82, p<.001). Good sensitivity (0.88) and specificity (0.85) were achieved using DAWBA computerised algorithms. Clinician review of responses to DAWBA questions minimally changed sensitivity (0.86) and specificity (0.87). Positive (0.82-0.95) and negative (0.90) predictive values were high. Eighty-six per cent of children were correctly classified. Performance was improved by using it in conjunction with the ADOS. ConclusionsThe DAWBA is a brief structured interview that showed good sensitivity and specificity in this general population sample. It requires little training, is easy to administer (online or by interview) and diagnosis is aided by an algorithm. It holds promise as a tool for assisting with assessment in community settings and may help services implement the recommendations made by NICE and the American Academy of Pediatrics regarding diagnosis of young people on the autism spectrum.