ABSTRACT
OBJECTIVE: To determine how gestational age relates to research-identified autism spectrum disorder (ASD-R) in the context of perinatal risk factors. STUDY DESIGN: This is a population-based cohort study using the 1994-2000 Olmsted County Birth Cohort. Children included were born and remained in Olmsted County after age 3 years. ASD-R status was determined from signs and symptoms abstracted from medical and educational records. Cox proportional hazards models were fit to identify associations between perinatal characteristics and ASD-R. RESULTS: The incidence of preterm birth (<37 weeks' gestation) was 8.6% among 7876 children. The cumulative incidence of ASD-R was 3.8% (95% CI 3.3-4.2) at 21 years of age. Compared with children born at full term, the risk of ASD-R appeared to be increased for children born preterm with unadjusted hazard ratios (HRs) of 2.62 (95% CI 0.65-10.57), 1.68 (95% CI 0.54-5.29), and 1.60 (95% CI 1.06-2.40) for children born extremely preterm, very preterm, and moderate-to-late preterm, respectively. In a multivariable model adjusted for perinatal characteristics, the associations were attenuated with adjusted HRs of 1.75 (95% CI 0.41-7.40), 1.24 (95% CI 0.38-4.01), and 1.42 (95% CI 0.93-2.15), for children born extremely preterm, very preterm, and moderate-to-late preterm, respectively. Among children with maternal history available (N = 6851), maternal psychiatric disorder was associated with ASD-R (adjusted HR 1.73, 95% CI 1.24-2.42). CONCLUSIONS: The increased risk of ASD-R among children born preterm relative to children born full term was attenuated by infant and maternal characteristics.