Perinatal & Pediatric
Prenatal air pollution exposure and risk of autism: findings from ECHO cohorts Akhgar Ghassabian* Akhgar Ghassabian Aisha Dickerson Yuyan Wang Joseph M Braun Kristen Lyall Lisa A Croen Rima Habre Deborah H Bennett Jean A Frazier Hannah C Glass Stephen R Hooper Robert M Joseph Catherine J Karr Jiwon Oh Susan Korrick Kaja Z LeWinn Christine Loftus Craig Newschaffer Mike O’Shea Thomas G O’Connor Adam Szpiro Frederica Perera Heather E Volk
The influence of prenatal exposure to low-level air pollution on child neurodevelopment is unclear.
We used data from 7398 mother-child pairs from 39 cohorts (recruitment 2000-2016) in the Environmental influences on Child Health Outcomes (ECHO) to examine the association of prenatal exposure to fine particulate matter (PM2.5), nitrogen dioxide (NO2), and ozone (O3) with autism. An ensemble model integrating machine learning algorithms and predictors estimated daily exposure at residential addresses. Parents rated children’s autism-related traits at mean age 8.9 years (SD=3.3) using the Social Responsiveness Scale (SRS) and reported on physician-diagnosed autism spectrum disorder (ASD). We used quantile regression to examine associations at different levels of SRS (25th, 50th, and 75th quantiles) and logistic regression for ASD diagnosis. Models included 3 pollutants and were adjusted for sociodemographic and lifestyle factors, urbanicity, season, and year of birth. Models were run at census divisions and coefficients were pooled in a meta-analysis. We tested for modification by sex.
Average pregnancy exposure for PM2.5 was 9.0 μg/m3 (SD=2.6), 22.0 ppb (SD=9.0) for NO2, and 40.6 ppb (SD=5.9) for O3 (8hr max), with variations across divisions. The median SRS T score was 46 (interquartile range [IQR]=42, 52) and 451 children (6.1%) were diagnosed with ASD. Higher O3 was associated with higher SRS score on average (β per IQR increase in O3=1.11, 95%CI: 0.18, 2.04) and at the lowest quantile of SRS (β=0.95, 95%CI: 0.32, 1.58), but not at the highest quantile. Higher O3 was also associated with ASD (OR=1.43, 9%CI: 1.19, 1.72). There was heterogeneity across divisions for associations of PM2.5 and NO2 with SRS and for NO2 with ASD. Higher PM2.5 was associated with higher odds of ASD in girls only. Associations of O3 with SRS or ASD were stronger in girls.
Low-level air pollution exposure may be a risk factor for autism, even for traits in the non-clinical range.
