Objectives: To assess the predictive validity of items from the Autism Observation Scale for Infants (AOSI) and Autism Diagnostic Observation Schedule (ADOS) at 18 months.
Methods: Prospective data were collected on 155 infant siblings of children with Autism Spectrum Disorder (ASD) and 73 low-risk controls using the AOSI and ADOS at 18 months. Infants were classified into 3 groups (ASD sibs, non-ASD sibs, controls) based on blind best-estimate diagnosis at age 3 years.
Results: Discriminant Function Analyses, using items identified through Fisher’s Exact Tests, yielded two discriminant functions for each tool. Discriminant Function 1 from the ADOS made a significant unique contribution to the model, χ2 (2) = 49.29, p < .001. Both AOSI Discriminant Function 1, χ2 (2) = 7.91, p < .01, and AOSI Discriminant Function 2, χ2 (2) = 11.69, p < .01, made significant unique contributions. At 18 months, ADOS items from the Social and Behavioural domains were most informative, while AOSI items measuring behavioural reactivity and motor control contributed additional information to ASD diagnoses at 36 months (all p’s < .05).
Conclusions: Findings highlight the importance of considering not only social-communication deficits, but also dimensions of temperament/state regulation and motor control when assessing toddlers with suspected ASD.