Beyond ASD: Developmental Outcomes of High Risk Siblings

Background:  The Baby Siblings Research Consortium (BSRC) recently reported that 18.7% of high risk infant siblings of children with autism spectrum disorders (ASD) will themselves develop an ASD (Ozonoff et al., 2011). However, the three-year outcomes of high-risk siblings who do not have an ASD have not yet been well characterized.

Objectives: Describe three-year-old, non-diagnosed high-risk siblings with respect to ASD-related symptom severity and developmental functioning—and identify latent subgroups of high-risk siblings.

Methods:  This multisite BSRC dataset included 508 high-risk (HR) siblings with no ASD diagnosis and 324 low-risk (LR) controls ­(i.e., no known relatives with ASD) with no ASD diagnosis. Model building employed negative binomial regression analyses of Autism Diagnostic Observation Schedule (ADOS) severity scores that varied from 1-10 (Gotham, Pickles, & Lord, 2009). Latent class analyses were used to identify clusters of three-year-olds based on ADOS severity and Mullen Verbal (V) and Non-Verbal (NV) Developmental Quotients (DQs). The latent class analysis included the 447 HR siblings and 197 LR controls for whom requisite data were available.

Results:  

The ASD symptom severity scores of HR siblings (M=2.19, SD=1.76) were significantly higher than those of LR controls (M=1.64, SD=1.22), X²=13.99, df=1, p<.001. The verbal DQ of the HR group (M=104.28, SD=17.39) was significantly lower than that of the LR group (M=110.47, SD=15.97; t=3.81, p<.01). The non-verbal DQ of the HR group (M=107.50, SD=17.10) was significantly lower than that of the LR group (M=111.60, SD=15.36; t=2.76, p<.01. 

Evaluation of the latent class analysis using Bayesian Information Criteria indicated a best-fitting five cluster solution (see Table 1). High risk children were under-represented in Cluster 1 (low ASD symptom severity and high DQ). They were over-represented in Clusters 4 (low symptom severity and low DQ) and 5 (moderate symptom severity and moderately low DQ).

Table 1
	Measures				Status
Clusters	ADOS Severity	Verbal DQ	Non-Verbal DQ		Low Risk	High Risk	HR/LR Odds Ratios (95% CI)
1.  Low Symptom Severity, High DQ	1.2	119	122		49%	35%	0.56 (0.40-0.79)
2.  Low Symptom Severity,  Typical DQ	1.3	99	102		34%	30%	0.85 (0.59-1.21)
3.  Moderate Symptom Severity,  High DQ	4.3	113	113		11%	14%	1.32 (0.78-2.24)
4.  Low Symptom Severity, Low DQ	1.3	77	82		3%	8%	2.79 (1.16-6.73)
5.  Moderate Symptom Severity,     Moderately Low DQ	5.1	89	89		4%	13%	4.05 (1.82-9.03)
Note. Bolded Odds Ratios, p < .05.

Conclusions:  At three years, HR children without an ASD had higher levels of ADOS symptom severity, and lower levels of developmental functioning than LR children. They were more likely to occupy clusters characterized by lower levels of developmental functioning, and less likely to occupy a cluster characterized by higher levels of developmental functioning and low levels of symptom severity. Descriptively, two-thirds of HR children occupied Clusters 1 and 2, characterized by normative outcomes, whereas one third occupied Clusters 3, 4, and 5, characterized by elevated ADOS severity, lower developmental quotients, or both. These results suggest an early ‘broader autism phenotype’ in HR siblings characterized by ASD symptoms sub-threshold for diagnosis and/or developmental delays.