17668
A Computerised Approach to Interviewing for ASD: Evidence for 3di's Value in Translation, and International Compatibility with DSM-5 Criteria
The Developmental, Dimensional and Diagnostic Interview (3di) is a computerised diagnostic parental interview, designed for deep phenotyping of the autism spectrum. Technically, the 3di offers patented advantages to facilitate the interviewing process. All variables (individual questions and summary scales) can be downloaded for statistical analysis. In line with revisions to the definition of ASD in DSM-5, the 3di question set and associated algorithms (formerly emulating ADI-algorithms and DSM-IV.TR) have recently been modified. There is considerable current interest in providing methodologies that will permit reliable recording of the ASD phenotype across languages and cultures; in May 2013 the WHO resolution Comprehensive and Coordinated Efforts for the Management of Autism Spectrum Disorderswas sponsored by more than 50 countries. The computerised structure of the 3di lends itself to efficient and reliable translation, and recent research has demonstrated its utility in this regard.
Objectives:
The objective of the technical demonstration will be, first, to provide evidence for the value of the 3di’s computerised interviewing procedure to facilitate both clinical assessments and research, based on the new DSM-5 criteria for ASD and Social Communication Disorder. The demonstration will encourage hands-on use of the interview and illustrate key features including data downloading procedures. Second, the demonstration will showcase technical innovations that facilitate translation into any language (with all text tailoring features intact), as well as the ease of exporting responses from any translation in a standard format for analysis (e.g. SPSS), thus permitting the compilation of cross-cultural databases containing high-density phenotypic data.
Methods:
In order to test the effectiveness of the 3di to emulate DSM-5 algorithms in translation, we recently derived diagnostic scores for each symptomatic subdomain, and tested them using Confirmatory Factor Analysis, in both an English and a Finnish version of the interview, on a total sample of 708 individuals with varying degrees of autistic trait severity. After testing the DSM-5 model in our Finnish ASD data, we proceeded formally to compare its fit in both UK and Finnish samples. We ran a series of ever more constrained models simultaneously in both translations, to conduct an increasingly rigorous and in-depth test of the DSM-5 model’s factorial invariance.
Results:
Factor loadings and estimates of fit of the DSM-5model were similar in both UK and Finnish translation. This provides strong evidence for the validity of the translation process into Finnish. Previous research has shown the 3di has equivalent high sensitivity and specificity in Thai translation as in English. Currently, Cantonese, Dutch, Finnish, Mandarin, Norwegian, Spanish, and Thai translations are available. Italian and Taiwanese translations are nearing completion, Arabic, Hindi, Swahili, and French translations are pending. Training courses in countries as diverse as the Netherlands and Argentina demonstrate high degrees of inter-rater reliability can be achieved in translation, with minimal hands-on instruction.
Conclusions:
Computerisation of the interview process for deep phenotyping of ASD facilitates translations that have excellent construct and predictive validity based on DSM-5 diagnostic criteria. Design innovations permit excellent inter-rater reliability to be achieved rapidly with minimal training, at low cost.