Behavioral Signs of Autism In Premature Infants: Findings From the Autism Observation Scale for Infants and Autism Diagnostic Observation Schedule

Background: The Autism Observation Scale for Infants (AOSI) is a direct rather than parent report measure of early signs of autism spectrum disorder (ASD). It reliably distinguishes infant siblings who are later diagnosed with ASD from both non-ASD siblings and controls with no family history of ASD. Premature infants are also at increased risk for ASD. We previously found that 12-month AOSI scores of premature infants were similar to those of infant siblings later diagnosed with ASD (ASD-Sibling), although the specific items contributing to total scores differed between the groups. 

Objectives: To evaluate group differences (Premature versus ASD-Sibling versus Control) in AOSI scores at 18 months, determine which AOSI items distinguish the ASD-Sibling group from the Premature group, and examine whether, within each group, 24-month ADOS scores are correlated with AOSI scores at 12 and 18 months.

Methods: AOSI Total Scores (higher values indicate increasing deviation) and Total Marker Counts (items with non-zero scores) were computed at 12 and 18 months for 45 premature infants (<37 weeks gestation; no ASD diagnosis at 24 months), 49 infant siblings with ASD (diagnosed by age 3 by experienced clinicians blind to prior study data, based on the ADI-R, ADOS, and clinical judgment using DSM-IV criteria), and 75 control infants (none with ASD). All participants also completed the ADOS (Module 1) at 24 months.

Results: Kruskal-Wallis tests were significant for 18-month AOSI Total Score (H_(2,169) = 44.88, p<.001) and Total Marker Count (H_(2,171) = 49.74, p<.001). All groups differed on both Total Score and Total Marker Count (ASD-Sibling>Premature>Control). Compared to the Premature and Control groups, the ASD-Sibling group tended to have inconsistent response to name, decreased social babbling, inconsistent eye contact, extreme reactivity, and difficulty with transitions, the last three items of which were also more prevalent in the ASD-Sibling group at 12 months. In the Premature and Control groups, 12-month AOSI Total Score was correlated with 24-month ADOS Total Score (.45 and .32, respectively); 18-month AOSI Total Score was uncorrelated with the other two scores. ADOS scores ranged from 0-13 in the Premature group (4 were above the Autism cut-off and 10 were above the ASD cut-off), while scores ranged from 0-10 in the Control group (2 were above the ASD cut-off). By contrast, AOSI and ADOS scores were inter-correlated (.41-.64) in the ASD-Sibling group.

Conclusions: At 18 months, premature infants have more elevated AOSI scores than controls, although not as high as siblings with ASD. Importantly, there are specific AOSI items that distinguish siblings with ASD from premature infants and controls, suggesting that these markers may be more predictive of ASD. The correlation findings may indicate that the AOSI is a more effective ASD screening tool at 12 months than at 18 months when used with clinical populations other than infant siblings. We plan to ascertain diagnostic outcomes at age 3 in our premature group, particularly because over 30% scored above the ASD cut-off on the 24-month ADOS, in order to appraise how well the AOSI identifies premature infants with ASD.