Comparison of ADOS to ADOS-2 Diagnostic Classifications within the Autism Treatment Network

Background: The ADOS (a “gold standard” diagnostic tool for evaluating autism) is a semi-structured, standardized assessment comprised of 4 modules each targeting an individual’s expressive language and chronological age.  The ADOS assesses communication, social, and repetitive behaviors.  The recently published ADOS-2 allows for improved psychometrics and diagnostic algorithms and has expanded to include a Toddler module.  It can be administered to anyone suspected of having autism with a non-verbal mental age of at least 12 months to adulthood.  Clinicians in the Autism Treatment Network (ATN) were interested in evaluating whether the new algorithms change the classification results on the ADOS. On the previous version, an autism spectrum diagnosis required that the subject meet three cutoffs: social, communication, and social and communication combined.  The newly revised algorithm on the ADOS-2 requires that the subject only meet one cutoff: a combination of social affect and restricted and repetitive behaviors.

Objectives: The purpose of the current study was to compare the categorical designations across the ADOS and the ADOS-2 in a very large sample of individuals diagnosed with ASD.

Methods: The ADOS was administered to 5062 children between the ages of 2 and 17 years while participating in the ATN (2237 module 1s, 1197 module 2s, and 1528 module 3s). Coding from the original ADOS data were used to compute the total scores on the ADOS-2 and determine the individual’s new cut-offs. As Module 4 does not have a new algorithm, this module was not examined in the current study.

Results: Across modules, the percentage of children who met autism diagnostic cutoffs increased from 68.7% on the ADOS to 87.4% on the ADOS-2. The percentage of children who met cutoffs for autism spectrum decreased from 24.6% on the ADOS to 8.9% on the ADOS-2. The percentage of those that had not reached cutoffs on the ADOS but were still diagnosed with an ASD by the ATN clinician decreased from 6.7% to 3.8%. The diagnoses of 99% of children with autism did not change when using the ADOS-2 cutoffs, 25% with autism spectrum did not change, and 41% of those with no diagnosis did not change. Similarly, 71% of those with an autism spectrum diagnosis changed to autism, and 60% of those with no diagnosis changed to either autism spectrum or autism. In contrast, 1% of those with autism or autism spectrum lost their diagnosis, and.6% of those with autism moved to autism spectrum. Chi-squares were conducted on all modules combined and for each module separately.  The chi-square for all modules combined was highly significant, X² =2568.8, df=4, p<.0001.  Likewise, results were significant for each module independently:  Module 1 X² =1136.9, df=4, p<.0001; Module 2 X² =542.8, df=4, p<.0001; Module 3 X²=840.5, df=4, p<.0001.

Conclusions: These results indicate that the new ADOS-2 algorithms preserve caseness with respect to an ASD diagnosis. However, the results also indicate a clear shift to a classification that is more definitive (i.e., a shift to a higher level of severity) with regard to the presence of autism.