International Meeting for Autism Research: Finding the Trees In the Forest: Predictors of Typical and Atypical Outcome Based on ADOS-T Item Analysis at 12 Months

Finding the Trees In the Forest: Predictors of Typical and Atypical Outcome Based on ADOS-T Item Analysis at 12 Months

Friday, May 13, 2011
Elizabeth Ballroom E-F and Lirenta Foyer Level 2 (Manchester Grand Hyatt)
2:00 PM
S. Macari1, D. Campbell2, C. A. Saulnier3, K. Bearss4, F. Shic1 and K. Chawarska5, (1)Yale University School of Medicine, New Haven, CT, (2)Department of Statistics, Yale University, New Haven, CT, (3)Yale Child Study Center, New Haven, CT, (4)Yale University School of Medicine, New Haven, CT, United States, (5)Child Study Center, Yale University School of Medicine, New Haven, CT
Background:  

Prospective monitoring of high-risk infant sibling studies provides a window into the early course of autism spectrum disorder (ASD). Recent studies suggest that infants who go on to develop ASD begin to manifest some symptoms around 12 months of age. The patterns of performance at 12 months that are predictive of either typical or ASD outcomes later in infancy have yet to be identified.   

Objectives:  

To explore whether clinical outcome at 24 months could be predicted based on performance on ADOS-T items at 12 months.

Methods:  

A total of 79 infants (54 males) were assessed with the ADOS-T at 12 (n=79), 18 (n=78), and 24 months (n=79); 48 were at high risk for ASD (HR-ASD) and 31 had no familial risk factors (LR).  The ADOS-T is a new module of the ADOS designed for 12-30-month-old infants. At 24 months, a team of experts classified infants as having ASD (n=11), specific developmental delays (SDD), broader autism phenotype (BAP), subthreshold difficulties (SUB), and typical development (TYPICAL; n=28).  To predict the clinical outcome at 24 months, individual ADOS-T items were analyzed using classification tree modeling, an exploratory data analytic technique that makes predictions about group membership. Two types of outcome classifications were considered:  (1) TYPICAL (N=28) (children with no clinical concerns between 12-24 months) versus CONCERNS (n=43) (children with concerns ranging from mild language delays to ASD); and (2) ASD (n=11) versus NON-ASD (n=68) (both typical and atypical presentation but without ASD symptoms).  

Results:  

The TYPICAL group had a significantly lower ADOS-T Social Affect algorithm score at 12 months (mean=7.1, SD=4.0; F(1,77)=10.6, p<.005) than the CONCERNS group (mean=10.9, SD=5.5) and the ASD group (mean=18.2, SD=3.3)  Classification modeling aimed at separating TYPICAL from CONCERN outcomes at 24 months classified correctly 78% infants in the TYPICAL group based on only 3 items: showing, overactivity, and imitation.  The second analysis correctly classified 73% of the ASD group based on their performance on four items at 12 months: eye contact, level of language, ignore, and pointing.   A higher-functioning subgroup with ASD was predicted by a combination of inconsistent eye contact, no bids for attention during the ignore probe, and no pointing, despite the presence of emerging verbal skills.  A lower-functioning ASD subgroup was predicted by the combination of poor eye contact and lack of pointing at 12 months. 

Conclusions:

Our analysis suggests that the presence of robust and frequent showing behavior, good object imitation skills, and age-appropriate activity level at 12 months bode well with regard to developmental outcome in infancy.  However, limited eye contact, limited spontaneous overtures to others, and the absence of pointing were highly associated with ASD outcome at 24 months.  The findings suggest that item-level analysis of performance on the ADOS-T at 12 months can be extremely useful in predicting clinical outcomes at 24 months.

| More