Association of PM2.5 Exposure on Birthweight: Racial and Ethnic Disparities

The few studies that have examined the impact of PM2.5 on reduced birthweight across different percentiles of the conditional birthweight distribution have produced equivocal findings, and only two assessed whether these associations varied by race/ethnicity or by racial/ethnic composition of the neighbourhood where mothers lived. We evaluated racial/ethnic differences in the association between prenatal PM2.5 exposure and birthweight across the birthweight distribution in a retrospective cohort study comprising 102,986 full-term singleton births in Harris County, Texas (2019-2020). Census tract-level daily PM2.5 concentrations were estimated using 'XGBoost-IDW Synthesis', and averaged exposures over pregnancy. Neighbourhood racial/ethnic composition was defined by whether ≥ 50% (i.e., majority) of non-Hispanic white residents lived in a census tract. Quantile regression models, adjusted for covariates, were applied to examine changes in birthweight [ β̂β̂ and 95% confidence interval (CI)] associated with an interquartile range increase in ambient air levels of PM2.5 at selected percentiles of the conditional birthweight distribution. Stratified analyses explored differential associations by maternal race/ethnicity and neighbourhood racial/ethnic composition. An inverted hook pattern was observed in the associations between prenatal PM2.5 exposure and reduced birthweight, with the strongest among infants born at the lowest ( β̂β̂ = -14 g, 95% CI: -20, -8; 10th percentile) and highest ( β̂β̂ = -11 g, 95% CI: -19, -4; 90th percentile) percentiles of the birthweight distribution, and a weaker association at the 75th percentile. In stratified analyses, the strongest association at the lowest percentile was observed among infants of Hispanic mothers or those living in neighbourhoods with less than a majority of non-Hispanic white residents. This study provides evidence that associations of prenatal PM2.5 exposure with reductions in birthweight varied among infants at the lowest, middle and highest percentiles of the conditional birthweight distribution; further, these associations varied by maternal race/ethnicity and neighbourhood racial/ethnic composition.