Examination of Developmental Sensitivity of Items on the Revised First Years Inventory Screener for Infants at Risk for a Later Diagnosis of ASD

Background:  The First Years Inventory (FYI) v3.1 is a newly revised parent-report questionnaire designed to identify infants at risk for a later diagnosis of ASD, expanding the target age from 12 months to 8-16 months. Each item on the FYI uses 5-point Likert scales (never to always) on items across two original domains (i.e., social communication; sensory-regulatory functions) and typical milestones. Due to potential maturational changes across this age range, the FYI items need to be assessed with respect to age norms and developmental sensitivity in order to develop a clinically useful scoring algorithm in the future. That is, a given screening item may be highly sensitive to ASD risk in a 15-month-old, but relatively insensitive to ASD risk in a 9-month-old. There is currently no “ASD-specific” parent-report screening tool for early risk markers in this age range that considers the effects of developmental sensitivity of individual items, which is a critical issue to address.

Objectives: The current study aimed to analyze a community sample of 5,960 infants aged 8-16 months to consider norms and developmental sensitivity of each FYI item across the domains of interest as a first step in creating a clinically more useful tool.

Methods: The FYI 3.1 was sent to ~40,000 families across North Carolina through the state birth registry. Each family received either an A or B version of the questionnaire to reduce time burden. There were 48 questions in each version with 27 core questions across the two versions, and thus 69 different items in total. We received 6,657 valid FYIs (response rate: 17%) and 5,960 were analyzed after excluding pre-term infants. Ordinal logistic regression analyses were conducted by item to examine the association between age and the 5-level ordinal response categories.

Results:  There were 10 items showing statistically significant associations with age and Cox & Snell’s pseudo R-square values ranging from 0.15-0.35, and all these items were in the domain of “social communication”. Among them, two core items within the construct of “social initiation” had the largest R-square values (above 0.30).

Conclusions: The results showed that the developmental factor should be taken into consideration for ASD screening items related to social communication, but less so for sensory-regulatory functions. The statistically significant associations between age and item responses indicated that age explained a relatively large proportion of the variability in these item responses. The current results are supported by a longitudinal research indicating that social engagement increases linearly within the first 5 years of life, with major strides during the first year (Feldman & Eidelman, 2009). Accounting for developmental sensitivity when scoring specific items on ASD screening tools may help decrease false-positive screens for those of younger age within a wider age window. Follow-up studies are being conducted to optimize risk scoring algorithms and maximize prediction to ASD outcomes at age 3.