Standardizing diagnostic assessment of autism spectrum disorder : assets and challenges

Abstract

The overall aim of this thesis was to study some previously under-researched psychometric properties of the standardized diagnostic instruments Autism Diagnostic Interview-Revised [ADI-R] and Autism Diagnostic Observation Schedule [ADOS] as well as the standardized Vineland Adaptive Behavior Scales, Second Edition [VABS-II] in order to improve current diagnostic practice.

Excellent interpersonal objectivity operationalized as interrater reliability of the ADI-R and the ADOS has been reported from research settings with highly trained, research reliable examiners. However, there are no studies from clinical practice despite the wide usage of both instruments in these settings. In studies I and II, the objectivity of the ADI-R and the ADOS was examined in clinical practice among clinical users. For the ADI-R, the objectivity for items (medians), domains and criteria exceeded G(q,k) (analogous to intraclass correlation) = .90 in all instances and was ĸ = .83 for classification. For modules 1-4 of the ADOS, items (medians) ranged from G(q,k) = .74 to .83, overall totals from .85 to .92 and classification was ĸ = .69.

Diagnostic validity of the ADI-R’s and the ADOS’ revised algorithms, separately and in combination, has previously only occasionally been reported for young children. In study III, the diagnostic validity of these instruments was examined in a large clinical sample of toddlers and young preschoolers. Diagnostic validity for the combined ADI-R and the ADOS yielded a sensitivity of 78% and specificity of 88% while the classification accuracy for the single use of the ADI-R and the ADOS in general was lower.

Finally, despite a multitude of studies reporting on the effect of the new DSM-5 symptomatology criteria of ASD on diagnosis rates, no studies have analyzed the effect of the novel DSM-5 impairment criterion. In study IV, this was examined in a subsample of the young children of study III. In accordance with the DSM-5 conceptualization of impairment, this was measured with the standardized VABS-II and different cutoffs of adaptive functioning were applied to operationalize impairment. We showed that, depending on the threshold (1, 1 1⁄2 and 2 SDs below the mean) used to operationalize impairment, 88%, 69% and 33% respectively of the children with a DSM-IV-TR ASD diagnosis fulfilled the impairment criterion compared to 91% for the DSM-5 symptomatology criteria.

To conclude, in terms of objectivity our results endorse the universal use of the ADI-R and the ADOS in everyday clinical practice and research for all age groups as the first choice diagnostic instruments for ASD. The combined use of the ADI-R and the ADOS yields better diagnostic validity than the separate usage of either instrument. Moreover, our results suggest that a strict application of the new DSM-5 impairment criterion might compromise early diagnosis for 12-67% of young children with a complete DSM-5 ASD symptomatology. The main asset of using standardized instruments is improved objectivity in assessing ASD symptomatology.