Comparison of Children with ASD at 3 Years of Age Ascertained from Community Screening in Primary Care Versus from a Prospective Infant Sibling Cohort

Background: Prospective research of infants at high-risk (HR) for ASD due to having an older sibling with ASD has provided critical information about early signs and developmental characteristics of ASD. HR infant sibling studies have shown that early signs are distinguishable at 12-18 months in infants later diagnosed with ASD suggesting that early detection is possible. However, the developmental level of infant sibling cohorts (Ozonoff et al., 2015) appears to be substantially higher than community samples ascertained from primary care using current autism-specific screening tools (Eaves et al., 2006; Robins et al., 2014; Stenberg et al., 2014), demonstrating a possible sampling bias toward missing higher-functioning toddlers with ASD. In a review of current evidence for ASD screening, priorities for future research included examining how broadband and autism-specific screening tools can be used together to improve screening accuracy and the need for large representative HR and low-risk (LR) samples to strengthen generalizability of findings (Zwaigenbaum et al., 2015). 

Objectives: To compare phenotypic presentations of children diagnosed with ASD at 3 years of age from two large prospective samples, one ascertained from community screening (CS) in primary care using a broadband screen for communication delay followed by autism-specific screening and the other ascertained from HR and LR infant siblings (IS). 

Methods: Children in the CS sample were recruited from over 10,000 children screened at 9-18 months in primary care settings using the Infant-Toddler Checklist (ITC; Wetherby & Prizant, 2002), a broadband screen of communication delay, followed by inviting positive screens and randomly selected negative screens for an ASD screening using the Systematic Observation for Red Flags (SORF; Wetherby et al., 2004). Children in the IS sample were recruited at 6-9 months of age and comprised of 175 HR and 111 LR infants. Children from both the CS (n=464) and IS (n=286) samples were invited for an evaluation to make a best-estimate clinical diagnosis using the ADOS, Mullen, and Vineland around 3 years of age; 204 CS and 32 IS children (30 HR, 2 LR) were diagnosed with ASD. 

Results: No significant differences were found between children with ASD ascertained from the CS (n=204) and IS (n=32) samples on any scale of the ADOS or Mullen. On the ADOS, the mean Social Affect total for both groups was 10.9 and the Restricted Repetitive Behavior total 4.2 and 4.0 for the CS and IS groups, respectively. The Mullen Early Learning Composite had a mean of 74.8 (SD=23.8) for the CS group and 76.4 (SD=18.9) for the IS group. T-scores on the four Mullen subscales were also equivalent. 

Conclusions: These findings demonstrate similar autism severity and developmental level in two large prospective samples, one ascertained from CS using a broadband screen for communication delay followed by ASD screening, and the other ascertained from an IS sample. The similar phenotypic expression on the ADOS and Mullen around 3 years of age provides initial support for the use of broadband and autism-specific screening tools together to improve screening accuracy and the generalizability of HR infant sibling research.