Associations between prenatal air pollution exposure and adverse birth outcomes: a siblings-matched case-control study
TL;DR
Prenatal exposure to NO₂, SO₂, and CO is associated with increased risks of PTB and SGA, and the importance of methodological refinement in environmental risk assessment to inform evidence-based air quality policies is highlighted.
Abstract
Previous studies linking prenatal air pollution to adverse birth outcomes may be confounded by unmeasured familial factors, potentially leading to biased effect estimates and misleading public health recommendations. This sibling-matched case-control study re-examined associations between six criteria air pollutants and preterm birth (PTB), low birth weight (LBW), and small-for-gestational-age (SGA). Associations were assessed using both a sibling-matched generalized linear mixed model (estimating subject-specific conditional effects) and conventional unmatched logistic regression (estimating population-averaged marginal effects) for comparison. We analyzed 194 284 mother-infant pairs (97 142 sibling pairs) from Nanning, China (2016–2022). Pollutant data were sourced from the high-resolution CHAP dataset. In sibling-matched analyses, first-trimester NO 2 exposure increased PTB risk ( OR = 1.004), while exposures to SO₂ ( OR f irst =1.009, OR second =1.009, OR third =1.010) and CO ( OR second =1.163, OR third =1.157) in specific trimesters were associated with elevated SGA risk. Some significant associations observed in unmatched designs (e.g., CO exposure across trimesters and PTB) were attenuated to non-significance after sibling matching. Furthermore, unmatched designs generally overestimated pollutant-PTB associations while underestimating pollutant-SGA links, with the largest between-model differences in effect estimates for CO. Prenatal exposure to NO₂, SO₂, and CO is associated with increased risks of PTB and SGA. Conventional unmatched study may lead to divergent effect estimates due to both unmeasured familial confounding and inherent mathematical differences between statistical models, underscoring the critical need to control for familial confounders in environmental epidemiological studies. Our findings highlight the importance of methodological refinement in environmental risk assessment to inform evidence-based air quality policies.