Objectives: Describe differences in the RRBs based on the ADOS at different developmental levels across different diagnostic groups of autism, pervasive developmental disorders not otherwise specified (PDD-NOS), typical development (TD), and developmental delay (DD).

Methods: Participants were included from projects 1) one in which they entered the study at 12 months and were assessed by the ADOS–Toddlers, 2) one in which they entered the study at 24 months and were assessed by the Pre-Linguistic-ADOS, the previous version of the ADOS-T.

Results:  655 cases (mean age of 28.26 months) were categorized into 7 cohorts by chronological age.  A Poisson Generalized Estimating Equations (GEE) model was used with age, diagnosis at age 2, and full scale IQ scores as predictors, and the ADOS algorithm domain score for the RRBs as the outcome. Preliminary results revealed that the diagnosis, chronological age, and full scale IQ scores were significant predictors of the RRBs.  The estimated marginal means for children with autism were significantly higher than for children with DD and TD in all the cohorts.  Additional analyses for each item in the domain of the RRBs showed different trajectories and cross sectional differences.

Conclusions: Not surprisingly, children with autism show significantly higher RRBs scores than children with TD and DD.  Additionally, different trajectories and cross sectional differences for each item also reveal that lumping different types of RRBs together under one criterion of RRBs could be problematic.