Mother-level random effect in the association between PM2.5 and low birth weight: A population-based pregnancy cohort in Israel

Abstract

Background/Aim: A growing body of literature reports associations between exposure to particulate matter with diameter ≤2.5 μm (PM2.5) during pregnancy and term low birth weight (TLBW). Although many studies accounted for clustering due to maternal region of residence, clustering due to inclusion of siblings in the cohort has not been sufficiently explored. We assessed the association between PM2.5 and TLBW in a cohort with high prevalence of siblings by using multilevel models accounting for geographical- and mother-level correlations. Methods: In Israel, we used Maccabi health services data to establish a population-based cohort of 437,370 singleton births reaching 24-42 weeks’ gestation and birth weight of 500-5,000 g (2004- 2016). Daily PM2.5 predictions from a satellite-based spatiotemporal model were linked to the date of birth and maternal residence. We generated mean PM2.5 values for the entire pregnancy and for each trimester of pregnancy. Associations between exposure and TLBW were modeled by using logistic regression model with random effects for locality of residence, administrative census area and mother.Results: Based on intra-class correlation coefficient (ICC) values and a comparison of the likelihood between models, the 2-level model clustered to the mother-level was the best one. A 10-μg/m3 increase in PM2.5 over the entire pregnancy was associated with TLBW when entering sequentially random effects of locality of residence, administrative census area and mother [Odds ratio, 95% confidence interval) = (1.13, 0.99-1.28), (1.13, 1.01-1.27), (1.16, 1.01-1.34) respectively]. The association accentuated to (1.22, 1.08-1.39) in the model with mother-level only. Estimates for TLBW for every 10-μg/m3 increase in the mean first, second, and third trimester PM2.5 were (1.02, 0.92 -1.12), (1.09, 0.99-1.21), and (1.17, 1.05-1.30), respectively.Conclusions: Consideration of mother-level variability in a region with high fertility rates provides new insights on the size of associations between TLBW and PM2.5.