How Well Can we Diagnose Autism in Adults? Evaluating an Informant-based Interview: The Dutch Developmental, Dimensional and Diagnostic Interview - Adult Version (3Di-Adult).

Abstract

The current study evaluated a brief, informant-based autism interview: the Developmental, Dimensional and Diagnostic Interview - Adult Version (3Di-Adult). Feasibility, reliability and validity of the Dutch 3Di-Adult was tested amongst autistic participants (n=62) and a non-autistic comparison group (n=30) in the Netherlands. The 3Di-Adult consists of two scales based on DSM-5 criteria: A scale 'Social communication and social interaction' and B scale 'Restricted, repetitive patterns of behavior, interests or activities'. ROC curves were used to determine cut-off scores for the A and the B scale, using an ASD diagnosis made by an independent clinician as the criterion. Mean administration time was 42min. Internal consistency of the A scale (α = 0.92) and the B scale (α = 0.85) were good. Inter-rater reliability (ICCs = 0.99) and inter-rater agreement (ICCs ≥ 0.90) were promising. The 3Di-Adult showed good sensitivity (80.6%) and specificity (93.3%). Positive and negative predictive value were 96.2% and 70.0% respectively. Comparisons with the Autism-Spectrum Quotient-Short to investigate the convergent validity showed moderate, significant correlations with the 3Di-Adult in the total sample. Males, as compared to females, displayed significantly more autistic features on the 3Di-Adult. No relationship was found of the 3Di-Adult with education level, intelligence and age of the participants or informants. The feasibility and psychometric properties of the Dutch 3Di-Adult are promising, indicating that it can be a time-efficient, valid and reliable tool to use in diagnosing autism in adults according to DSM-5 criteria.