42: Autism risk is increased in SGA and LGA infants


      The extremes of birth weight (low birth weight and macrosomia) have been associated with increased autism risk. This study was designed to determine whether being small or large for gestational age (SGA or LGA) also influences the risk for autism.

      Study Design

      This is a retrospective population-based cohort study. Within the 1991-2001 California birth cohort (n=5,979,605, database provided by the California Office of Statewide Health Planning and development), the annual 5th, 10th, 90th and 95th birth weight percentiles were calculated by gender and gestational age at birth. Children in the birth cohort later diagnosed with autism (n=20,206) were identified using a database provided by the California Department of Developmental Services. The associations between SGA birth weights (95%) and autism were calculated using the appropriate for gestational age group for reference (AGA, >10% to <90%). Univariate analysis (autism risk by birth weight appropriateness) was followed by a multivariate regression analysis (controlling for the effects of preeclampsia, hypertension, multiple gestation, birth order, race, inter-pregnancy interval, and diabetes). Effect modification was ascertained for gestational age strata using a Cochran-Mantel-Haensel test.


      The risk of autism was significantly higher in infants in the extreme SGA group (95%) group (aOR 1.12 (1.05, 1.19)). The increase in LGA risk was particularly pronounced in the 39-41 and 42+ week GA groups with roughly a 21% increase in risk for autism in the >95% percentile group.


      The extremes of birth weight percentiles (95%) are both associated with a 10% increase in autism risk, with the LGA effect most prominent in term and post-term infants. The bimodal pattern of risk likely reflects disparate pathologies, and should influence prenatal counseling, pediatric autism screening, and further research into the etiology of autism.