Differences Between Preschool Children with ASD Ascertained By Clinical Referral Versus Longitudinal Follow-up of Infants with an Affected Older Sibling

Background:  Early development in autism spectrum disorders (ASD) is increasingly studied using prospective ‘high-risk’ designs, focused on infants at risk of the disorder by virtue of having an affected older sibling. However, the comparability of children with ASD ascertained from these high-risk cohorts versus those ascertained by clinical referral from the community has never been examined, which is relevant to the generalizability of findings across groups.

Objectives:  To compare the sex ratio, symptom severity and adaptive functioning of preschool children with ASD ascertained from a high-risk longitudinal sibling cohort (HR-ASD) to those diagnosed following referral to a specialized assessment clinic (clin-ASD).

Methods:  The HR-ASD group (n=94) were identified as part of an ongoing multi-site prospective study of HR siblings. ASD diagnoses were established shortly after the child’s 3^rd birthday, using the Autism Diagnostic Interview –Revised (ADI-R), Autism Diagnostic Observational Schedule (ADOS) and expert clinical judgement based on DSM-IV-TR. The clin-ASD group (n=176) were identified from an inception cohort of clinically referred children with ASD participating in a multi-site longitudinal study. To ensure comparability by age, only children diagnosed at age 3 were included. Categorical variables (sex, language level) were compared using the chi-squared (χ²) test, and continuous variables (e.g., adaptive functioning, indexed by the Vineland Adaptive Behavior Schedule – II; VABS-II) were compared using ANOVA.

Results:  The clin-ASD group was slightly older than the HR-ASD group (41.2+3.9 vs 38.7+3.4 months, F_1,262=31.8, p<.001). Sex ratio differed between the HR-ASD (2.2:1) and clin-ASD groups (5.1:1) (χ²(1)=7.51, p=.006). ASD symptoms, indexed by the ADOS severity metric and ADI-R social, communication and behavioral subscales were less severe in the HR-ASD group (all ps<.001; effect sizes 0.54-0.86). There were marked differences in parent-reported language level (on the ADI-R), with phrase speech in 79 of 94 children (84.0%) in the HR-ASD group, compared to 63 of 166 children (38.0%) in the clin-ASD group (χ²(2)=51.5, p<.001). As well, the HR-ASD group had higher levels of overall adaptive functioning, as indexed by VABS-II Adaptive Behavior Composite Standard Scores (78.1+1.4 vs 71.8+10.7, F_1,241=16.1, p<.001). ANCOVAs incorporating sex and age (within the narrow range studied) as covariates did not detect any main or interactive effects for these variables, and HR-ASD versus clin-ASD differences remained essentially the same.

Conclusions:  Compared to clinically referred children with ASD, those ascertained from a HR cohort had less severe ASD symptoms, higher levels of language and adaptive functioning, and proportionately included more girls. These differences may reflect referral and ascertainment biases across the two cohorts, with important implications for generalizability of findings from HR infant studies. As well, children (and particularly girls) with ASD with phrase speech and higher levels of adaptive functioning who are identified by surveillance of high risk cohorts may be underrepresented among clinically referred preschoolers, due to later identification in the community. Surveillance strategies must take account of heterogeneity among children with ASD in order to ensure early identification of children across the spectrum.