Calibrated Severity Scores for the Autism Diagnostic Observation Schedule-Toddler Module

Background:  Calibrated severity scores have been created for Autism Diagnostic Observation Schedule, 2^nd edition (ADOS-2) as a metric of the relative severity of autism-specific behaviors, as measured by this instrument (Gotham, Lord, & Pickles, 2009). Calibrated severity scores are available for modules 1-3 and soon will be available for module 4. The goal was to create a standardized score to compare autism symptoms within and across individuals, independent of age and developmental level. Calibrated severity scores for modules 1-3 had more uniform distributions across age and developmental level and were less influenced by demographic variables than ADOS-2 raw totals. A comparable metric is not yet available for the ADOS-2 Toddler Module (ADOS-T), used to assess toddlers between 12 and 30 months of age.

Objectives:  Standardize ADOS-T raw scores to approximate a metric of ASD-related symptom severity that is less influenced by developmental level, age, and other demographic variables.

Methods:  Analyses were conducted using ADOS-T and psychometric data for 360 children with ASD age 12 to 30 months, some with repeated assessments for a total of 823 cases. Following procedures informed by Gotham and colleagues (2009), the ASD sample was divided into groups based on chronological age and language level. Within each group, raw scores were mapped onto a 10-point severity metric, using percentiles of raw totals corresponding to ADOS-T classifications.

Results:  Distributions of severity scores showed increased comparability across the age/language groups compared to raw totals. Multiple linear regression analyses were performed separately for calibrated severity score and raw total to examine whether participant characteristics such as age and IQ would be less associated with severity scores than with raw scores. Only verbal IQ was a significant predictor for both raw total and calibrated severity scores. While verbal IQ accounted for 33% of the variance in ADOS-T raw totals, it accounted for only 19% of the variance in severity scores.

Conclusions: ADOS-T calibrated severity scores demonstrated reduced associations with verbal IQ. For modules 1-3, severity scores have demonstrated usefulness creating more homogeneous groups for clinical, genetic, and neurobiological research. The addition of ADOS-T calibrated scores allows comparisons to be made starting at 12 months, expanding the ability to identify trajectories of ASD symptoms potentially at the first signs of concern. Future studies should replicate calibrated severity scores in large independent samples. ADOS-T severity scores should be evaluated for validity in predicting outcomes (e.g., eventual ASD diagnosis, response to intervention).