Comparisons Between the DISCO and the ADI-R and the ADOS

Background:  The Diagnostic Interview for Social and Communication Disorders is a standardized, semi-structured and interviewer-based schedule. The DISCO is constructed in a dimensional rather than a categorical way. Based on the broad-ranging nature of autism spectrum disorders, the DISCO items cover a very wide range of skills and features. Originally, the DISCO was designed to provide a comprehensive description of individual characteristics and needs, but the interview can also be used for categorical classification. For this purpose, computerized algorithms have been developed based on classification systems such as the ICD-10 and DSM-IV-TR.

Objectives:  To explore the convergent validity of the DISCO algorithms in comparison to ADI-R and ADOS.

Methods:  The DISCO interview schedule was administered by trained clinicians in the Netherlands, Sweden, and Belgium. In the Netherlands, the DISCO-11 and the ADOS (Module 1 or 2, revised algorithms) were administered from a sample of 115 children comprising 52 children with ASD (both with and without intellectual disability), 26 children with intellectual disability (non-ASD), and 37 typically developing children. In Sweden, the DISCO-10 and the ADI-R were administered from a Swedish sample of 57 children and adults with and without intellectual disability referred for neuropsychiatric assessment. In Belgium, the DISCO-11 and the ADOS (Module 3) were administered from 30 school-aged children without intellectual disability. 

Results:  The Dutch study reported a substantial agreement between DISCO-11 and ADOS (k = .69, p < .001). The correlation between raw total scores of the DISCO and ADOS algorithm was also high (r(107) = .87, p < .001). The correlation between DISCO and ADOS social/communication domain scores (r(107) = .87,. p < .001) was much higher than between the restricted/repetitive behaviour domain scores (r(107) = .64, p < .001), but were both significant. The Swedish study reported that the criterion validity of the DISCO was excellent when compared to the ADI-R. The ADI-R tended somewhat more to ‘over diagnose’ autism in relation to clinical diagnosis. Five cases with clinical ASD were missed by the ADI-R autism algorithm, but were all picked up with DISCO-10 algorithms. When the ADI-R thresholds for the broader category of ASD (Risi et al., 2006) were applied, four of the five were identified as ‘other ASD’. Preliminary findings for the Belgian study appear to support the foregoing results.

Conclusions:  The DISCO algorithms show good convergent validity in comparison to ADI-R and ADOS. Advantages over the ADI-R include valuable information of the broader autism phenotype and co-existing problems, relevant to both clinical practice and research.