DSM-5 based algorithms for the Autism Diagnostic Interview-Revised for children ages 4-17 years.

Area of Science:

• Psychiatry
• Developmental Psychology
• Clinical Psychology

Background:

• The Autism Diagnostic Interview, Revised (ADI-R) is a key tool for diagnosing Autism Spectrum Disorder (ASD).
• Existing ADI-R algorithms are based on DSM-IV criteria, with DSM-5 specific algorithms only available for toddlers.
• Large-scale validation of ADI-R algorithm sensitivity and specificity is limited.

Purpose of the Study:

• To develop and validate DSM-5-based algorithms for the ADI-R for children aged 4-17 years.
• To compare the diagnostic performance of these new algorithms against clinical diagnosis and original DSM-IV algorithms.
• To assess algorithm performance in individuals with and without phrase speech.

Main Methods:

• Utilized data from 2,905 participants (2,144 with ASD, 761 non-ASD) from clinical-research databanks.
• Selected ADI-R items discriminating ASD from non-ASD cases for DSM-5 algorithm development.
• Employed confirmatory factor analysis and ROC curve analyses to evaluate algorithm fit and diagnostic accuracy.

Main Results:

• The revised DSM-5-based ADI-R algorithms demonstrated adequate fit.
• Sensitivity of revised algorithms ranged from 77% to 99%, and specificity from 71% to 92%.
• Revised algorithms showed comparable or improved performance over original DSM-IV algorithms, particularly in specificity for individuals without phrase speech.

Conclusions:

• The original ADI-R algorithm has adequate diagnostic validity but lower specificity in non-phrase speech users.
• The newly developed DSM-5-based ADI-R algorithms offer comparable performance with enhanced specificity, especially for non-phrase speech users.
• These revised algorithms provide a DSM-5-compatible method for summarizing ASD symptoms.