Use of the Development and Well-Being Assessment (DAWBA) to Identify Autism Spectrum Disorder in a Community Sample of Adolescents

Background: Recent guidelines produced by the National Institute for Health and Clinical Excellence (UK) highlight an increased demand for diagnostic services for children and adults with Autism Spectrum Disorder (ASD) as a result of an increase in reported prevalence to at least 1% of the population (NICE clinical guidance 128 & 142, 2011 & 2012). ‘Gold standard’ diagnostic tools such as the Autism Diagnostic Interview (ADI-R) and Autism Diagnostic Observation Schedule (ADOS) are cost and time intensive and are therefore difficult to use in community settings. There is a need to develop ASD assessment tools that can be used in these settings. The Development and Well-Being Assessment (DAWBA) is an online package of questionnaires that can be administered online or by interview. It covers a range of psychiatric disorders and generates probabilities that an individual has each disorder. A clinician reviews all information to confirm diagnoses.  

Objectives: The aim of this study was to test the validity of the ASD module of the Development and Well-Being Assessment (DAWBA) in identifying cases of ASD in a community sample of adolescent twins. Information from the Social Aptitude Scale (SAS) and Strengths and Difficulties Questionnaire (SDQ), which are both included in the DAWBA package, was also used to help identify cases.  

Methods: The sample consisted of 285 adolescents selected from the Twins Early Development Study (TEDS) and included children who had been identified as being at risk of ASD (score of >= 15 on the Childhood Autism Spectrum Test (CAST) or ASD label given by healthcare or education professional), co-twins of children at risk of ASD, and low risk controls (CAST score < 12). Parents completed the ASD module of the DAWBA by telephone interview or online. Families were then visited at home and the ADI-R and ADOS were used by the research team to generate a consensus diagnosis of ASD.  

Results: DAWBA ASD symptom scores correlated highly with ADI-R algorithm scores (rho = .82, p < .001). Good sensitivity and specificity were achieved using computer generated probability bands (sensitivity = 0.86, specificity = 0.94). Using clinician reviewed DAWBA diagnosis improved specificity but sensitivity was lower (sensitivity = 0.77, specificity = 1.00). Sensitivity could be improved if other information in the DAWBA package (Peer Problems and Prosocial scores from the SDQ; SAS scores) was utilised alongside clinician reviewed diagnosis (sensitivity = 0.82, specificity = 0.97). Positive predictive value ranged from 0.97 to 1.00 and negative predictive value from 0.81 to 0.89.  

Conclusions: Results suggest that the DAWBA could be used to identify cases of ASD in samples of relatively straightforward cases of ASD and clear-cut non-cases. High confidence is warranted in positively identified cases, though the DAWBA appears to perform somewhat conservatively and misses some cases. Using SDQ and SAS scores can help to identify some of these missed cases. The DAWBA may therefore be useful in community settings, where many cases are likely to be straightforward and when it is impractical to use the ADI-R.