Adapting and validating the Autism Diagnostic Interview - Revised for use with deaf children and young people

Abstract

A Delphi consensus methodology was used to adapt the Autism Diagnostic Interview–Revised for the assessment of deaf children with suspected autism spectrum disorder. Each Autism Diagnostic Interview–Revised item was considered by a panel of nine international experts in terms of relevance and acceptability. Modifications were proposed and agreed by the expert panel for 45% of items. The pre-specified criterion for agreement between experts was set at 80% for each item. A first validation of the revised version, adapted for deaf children (Autism Diagnostic Interview–Revised Deaf Adaptation), was undertaken with a UK sample of 78 parents/carers of deaf children with autism spectrum disorder and 126 parents/carers with deaf children without autism spectrum disorder. When compared to National Institute for Health and Care Excellence guideline standard clinical assessments, the Autism Diagnostic Interview–Revised Deaf Adaptation diagnostic algorithm cut-off/threshold scores achieved a sensitivity of 89% (79%–96%) and specificity of 81% (70%–89%) for autism spectrum disorder. The alpha coefficients for each algorithm symptom domain ranged from 0.80 to 0.91, suggesting that the items had high internal consistency. Our findings indicate that the Autism Diagnostic Interview–Revised Deaf Adaptation is likely to be a useful measure for the assessment of deaf children with suspected autism spectrum disorder, although further research is needed. Lay abstract Autism assessment processes need to improve for deaf children as they are currently being diagnosed later than their hearing counterparts and misdiagnosis can occur. We took one of the most commonly used parent developmental interviews for autism spectrum disorder the Autism Diagnostic Interview–Revised and adapted it using international expert advice. Modifications were proposed and agreed by the expert panel for 45% of items; the remaining 55% of items were unchanged. We then tested the revised version, adapted for deaf children (Autism Diagnostic Interview–Revised Deaf Adaptation), in a UK sample of 78 parents/carers of deaf children with autism spectrum disorder and 126 parents/carers with deaf children without autism spectrum disorder. When compared to National Institute for Health and Care Excellence guideline standard clinical assessments, the Autism Diagnostic Interview–Revised Deaf Adaptation diagnostic algorithm threshold scores could identify those deaf children with a definite diagnosis (true autism spectrum disorder positives) well (sensitivity of 89% (79%–96%)) and those deaf children who did not have autism spectrum disorder (true autism spectrum disorder negatives) well (specificity of 81% (70%–89%)). Our findings indicate that the Autism Diagnostic Interview–Revised Deaf Adaptation is likely to prove a useful measure for the assessment of deaf children with suspected autism spectrum disorder and that further research would be helpful.