Air pollution and autism-like traits: sensitive periods and joint effects of exposure to sources and constituents of fine particulate matter.

Growing evidence suggests that exposure to air pollution during early life is associated with autism spectrum disorder (ASD). However, results have been inconsistent for autism-like traits and in low exposure settings. Here, we assessed sensitive windows of exposure to fine particulate matter (PM2.5), nitrogen dioxide (NO2), and ozone (O3) and joint effects of exposure to sources and constituents of PM2.5 on autism-like traits among 798 children in the Upstate KIDS cohort. Residential air pollution exposures were assessed with machine learning and land use regression models. Autism-like traits at age 10 were assessed via the ASD subscale on the Child Behavior Checklist. Treed distributed lag mixture model was used to identify critical windows of exposure to PM2.5, NO2, and O3 and partial linear single index model was applied to estimate contributions of sources and constituents of PM2.5 on autism-like traits. The median PM2.5 exposures were 8.36 and 8.00 μg/m3 for pregnancy and the first year of life. We did not observe critical windows of exposure to PM2.5, NO2, or O3 for autism-like traits. Exposures to elemental carbon (EC) PM2.5 during gestation and to EC and traffic PM2.5 during the first year of life were associated with higher odds of autism-like traits (prenatal EC adjusted odds ratio (aOR): 1.72, 95% CI: 1.04-2.83; first year EC aOR: 1.72, 95% CI: 1.21-2.46; traffic aOR: 1.31, 95% CI: 1.01-1.71). Our results suggest that exposure to motor vehicle traffic sources of PM2.5 may drive previously reported associations between total PM2.5 and autism-like traits.