Do Sensory Markers Improve ASD Screening Accuracy at 12-months?

Background: The drive to screen for autism spectrum disorder (ASD) earlier emerges from accumulating evidence indicating that intensive autism specific interventions provided prior to the age of three lead to better outcomes. Screening using markers from all ASD symptom domains, including social, communication, and sensory-behavioral, is controversial. Examining the impact of including sensory-behavioral markers (e.g., spinning objects, sensitivity to sound) on the accuracy of ASD screening can inform the design of efficient screening procedures.

Objectives:   (1) Validate First Year Inventory (FYI) norms in Israel; (2) Examine the contribution of sensory markers to the accuracy of ASD screening at follow-up.

Methods:   Participants were 589 families of 12-month old infants (52% boys) who completed the FYI, a norm-referenced ASD screening questionnaire, manually or electronically. A subsample of 84 families was followed at 13-months based on their meeting the FYI 95^th percentile risk status on the sensory-regulatory and/or social-communication scales. Follow-up assessments included the Autism Observation Scale for Infants (AOSI) and the Mullen Scales for Early Learning. Infants were referred for evaluation based on previously documented cutoffs for AOSI and Mullen scores as well as clinical judgment. A 30-month diagnostic follow-up is underway.

Results:   The Israeli version of FYI had a Chronbach’s Alpha of 0.75. The 95^th percentile risk cutoff of the current sample was 4.44 points higher than the US sample. The 13-month developmental status was compared between four groups of infants at FYI risk in (1) the sensory domain (n=19); (2) social-communication domain (n=22); (3) both sensory and social-communication domains (n=16); and (4) no risk (n=27; matched by gender and SES to risk groups). The Mullen Early Learning Composite Score was significantly lower for the social and both risk groups versus sensory and no risk groups. In Expressive Language, social and both risk groups had significantly lower t-scores than the sensory and no risk groups while in Receptive Language and Fine Motor only the both risk group showed lower scores than the no risk group. Total AOSI scores were higher for social versus sensory risk groups, and for the both risk group compared to sensory and no risk groups. Sensitivity and specificity were calculated for social and both risk groups relative to sensory and no risk groups combined. The outcome was defined by a clinical evaluation referral at 13-months (as a factor of AOSI and Mullen scores). Using the social domain risk only, sensitivity was 44% and specificity 69%, while using meeting risk in social and sensory domains yielded sensitivity of 73% and specificity of 89%.

Conclusions: Applying the FYI, a norm-referenced screener, requires adaptations of risk cutoff for use in Israel. Including sensory markers in ASD screening criteria rather than screening only for social risk was associated with higher sensitivity and specificity and in identification of infants with lower developmental status at 13-months. Screening criteria in infancy that takes into consideration type of marker together with frequency may lead to greater accuracy.