19187
Validation and Factor Structure of the 3Di Short Version in a DSM-5 Context

Friday, May 15, 2015: 11:30 AM-1:30 PM
Imperial Ballroom (Grand America Hotel)
G. Slappendel1,2, W. Mandy3, J. van der Ende1, F. C. Verhulst1, J. Duvekot1 and K. Greaves-Lord1,2, (1)Department of Child & Adolescent Psychiatry/psychology, Erasmus MC - Sophia Children's Hospital, Rotterdam, Netherlands, (2)Yulius Autisme Expertisecentrum, Rotterdam/Dordrecht, Netherlands, (3)Behavioural and Brain Sciences Unit, UCL Institute of Child Health, London, United Kingdom
Background:  Validated standardized instruments are important aids in the process of diagnosing Autism Spectrum Disorder (ASD). However, most validated standardized parent interviews currently available are time and resource intensive, which hampers regular use in clinical practice. The short version of the Developmental Diagnostic Dimensional Interview (3Di-sv) is a standardized, 45-minute interview that is increasingly used in clinical settings. Although this instrument has been validated against clinical diagnosis according to DSM-IV-TR criteria, more information is needed regarding its reliability and validity in the light of the DSM-5 conceptualization of ASD.

Objectives:  To extend the validation of the 3Di-sv by investigating its association with a diagnosis according to DSM-5 criteria, and test its underlying factor structure against a DSM-IV versus a DSM-5 conceptualization of ASD.

Methods:  Data were collected from 198 clinically referred Dutch children who showed significant ASD symptoms (SRS > 75). Sensitivity, specificity, Negative Predictive Value (NPV) and Positive Predictive Value (PPV) for the 3Di-sv classification were determined as compared to a best estimate DSM-5 diagnosis based on 3Di-sv and the Autism Diagnostic Observation Schedule, using the preliminary DSM-5 ASD criteria. Confirmatory factor analysis (CFA) of the 3Di-sv was used to test the model-fit of the instrument against a DSM-IV versus a DSM-5 model of ASD.

Results:  The 3Di-sv showed excellent sensitivity (.98) and NPV (.97), moderate PPV (.66) and poor specificity (.52). The DSM-5 model of ASD showed adequate to good fit (RMSEA = .049, CFI=.951, SRMR=.056), and outperformed the DSM-IV model of ASD.

Conclusions:  The 3Di-sv performs well when it comes to correctly identifying children with an ASD diagnosis according to the DSM-5. However, it also shows a relatively high rate of false positives. Possible reasons for this relatively high false positive rate will be addressed, including changes in the DSM criteria as compared to the DSM-IV, and related expected changes in ASD prevalence. CFA showed that the data best fit a DSM-5 model of ASD, however, the 3Di-sv does not fully include all symptoms described in the DSM-5. Future adaptations of the 3Di-sv will especially need to address the added criterion on sensory sensitivity, as well as changes in the definitions of existing symptom domains in the DSM-5.