Prenatal exposure to extreme heat and autism in children

Increasing global temperatures have been associated with neural tube defects and neurodevelopmental delays. Effects of gestational temperature exposures on autism, another neurodevelopmental outcome with prenatal risk factors, have not been previously investigated. This study examined associations of weekly maximum and minimum temperature (T_max and T_min, representing daytime and nighttime temperatures respectively) on development of autism in children in a retrospective birth cohort study from Kaiser Permanente Southern California hospitals from 2001 to 2014. Autism diagnosis by age 5 was identified in electronic medical records with corresponding ICD codes. Weekly average T_max and T_min were estimated at the maternal residential addresses during pregnancy using the gridMET model. Cox proportional hazard models with nonlinear distributed lags were used to identify critical windows of exposure with hazard ratios (HR) comparing exposure to temperature at the 90th and 99th percentiles versus the 50th percentile. A total of 4076 children (80% male) in the cohort of 294,937 had autism diagnosis by age 5. Exposure to extreme T_min during gestational weeks 1-10 and 30-37 were associated with increased risk of autism, with HR (95% CI) 1.154 (1.040, 1.288) for exposure during weeks 1-10, and 1.132 (1.030, 1.246) for weeks 30-37 comparing the 99th percentile to the 50th percentile. No association was observed for T_max. Exposures to high nighttime temperature during early and late pregnancy were associated with autism risk in children, a result of concern in a warming world. Further research is needed to understand why daytime temperature was not associated with autism risk.