Ariane Cartigny, Corentin J Gosling, Georges Chatzis, Ophélia Godin, Ariel Frajerman, Elise Humeau, Paola Atzori, Sara Ramos Pereira, Pierre Ellul, Stephanie Antoun, Chema Sahnoun Derbel, Elodie Zante, Julien Pottelette, Natalia Robert, Virginie Pomies, Gabrielle Chesnoy, Delphine Weill, Romain Coutelle, Mario Speranza, Anouck Amestoy, FACE-ASD (FondaMental Advanced Centers of Expertise for Autism Spectrum Disorder) Groups, Sylvain Moutier, Marion Leboyer, Samuele Cortese, Richard Delorme
The ADI-R and ADOS-2, alone or combined, did not achieve clinically acceptable levels of both sensitivity and specificity. Their combination nonetheless supports context-dependent trade-offs between sensitivity and specificity, helping clinicians prioritize either case detection or avoidance of false positives. In conclusion, while the ADI-R and ADOS-2 are valuable assessment tools for individuals without ID, they cannot replace integrated expert clinical judgment.
OBJECTIVE: The Autism Diagnostic Interview-Revised (ADI-R) and the Autism Diagnostic Observation Schedule-Second Edition (ADOS-2) are widely regarded as gold-standard instruments for the diagnosis of autism spectrum disorder (ASD). However, evidence for their diagnostic accuracy in naturalistic clinical settings remains limited among children and adolescents. We assessed their accuracy, individually and in combination, within a real-world clinical sample aged 6-25 years without intellectual disability (ID).
METHOD: This retrospective study analyzed 798 participants referred for ASD assessment, reflecting high diagnostic prevalence typical of specialized settings: 336/415 children/early adolescents and 353/383 late adolescents received a final ASD diagnosis. Diagnostic accuracy of both instruments and their combination ('OR-rule' and 'AND-rule') was assessed against a multidisciplinary consensus clinical diagnosis.
RESULTS: The ADI-R showed fair specificity in children and adolescents (82.3% and 86.7%) but poor sensitivity (51.2% and 43.9%). Conversely, the ADOS-2 demonstrated fair sensitivity (89.6% and 80.5%) but poor specificity (50.6% and 66.7%). Allowing diagnosis by either ADOS-2 or ADI-R increased sensitivity at the cost of more false positives, while requiring criteria on both instruments increased specificity at the cost of more false negatives. Exploratory analyses indicated reduced diagnostic accuracy of the two instruments in adolescents and poorer ADI-R accuracy in children with higher IQs, and in female adolescents.
CONCLUSION: The ADI-R and ADOS-2, alone or combined, did not achieve clinically acceptable levels of both sensitivity and specificity. Their combination nonetheless supports context-dependent trade-offs between sensitivity and specificity, helping clinicians prioritize either case detection or avoidance of false positives. In conclusion, while the ADI-R and ADOS-2 are valuable assessment tools for individuals without ID, they cannot replace integrated expert clinical judgment.