Prenatal Exposure to Ambient Heat and Autism in California, USA: A Distributed Lag Analysis of Weekly Gestational Windows

Abstract

Abstract Extreme heat is an increasing environmental health concern, yet its role in neurodevelopmental outcomes remains incompletely understood. Pregnancy represents a period of heightened vulnerability to environmental stressors, and heat-related physiological disruptions may influence fetal brain development. We examined associations between prenatal heat exposure and autism in a large, population-based cohort in California. We analyzed 2,013,431 singleton births in California (2015-2019), including 37,083 autism cases identified through linkage with Department of Developmental Services records. Prenatal heat exposure was assessed using daily average heat index, maximum temperature (Tmax), and minimum temperature (Tmin), assigned based on the maternal residential location. Logistic regression models estimated associations for full pregnancy and predefined gestational windows, and distributed lag nonlinear models (DLNMs) were used to evaluate weekly exposures across gestation. Models were adjusted for sociodemographic factors and for PM2.5. A higher pregnancy-average heat index was associated with increased odds of autism (odds ratio per 10°F increase = 1.14; 95% CI: 1.12, 1.17), with comparable positive associations observed for Tmin, a proxy of nighttime heat exposure, whereas Tmax showed little evidence of association. Window-specific analyses showed elevated associations during early and late pregnancy, with attenuation during mid-gestation. DLNMs revealed a consistent U-shaped temporal pattern across gestation. Associations were similar by sex but varied by socioeconomic status, with stronger relative associations observed in higher SES groups. Findings were robust across sensitivity analyses, including adjustment for ozone and season. Prenatal heat exposure, particularly sustained and nighttime heat, was associated with increased odds of autism, with evidence supporting multiple windows of susceptibility during gestation and potential modification by population-level differences in baseline heat exposure and acclimatization.