Association between ambient PM2.5 exposure during pregnancy and risk of preterm birth: a retrospective birth cohort study in Chengdu, Southwest China
Prenatal PM 2.5 exposure has been associated with preterm birth (PTB), but evidence on exposure windows and population heterogeneity remain inconsistent. This retrospective cohort included 35,136 singleton births at a Chengdu hospital from 2015 to 2024. Satellite- and monitor-based PM 2.5 exposures were estimated for conventional and fixed gestational windows. Logistic regression adjusted for individual characteristics, meteorological factors, conception season, and district fixed effects, with cluster-robust standard errors by district and last-menstrual-period year-month. PTB was classified as earlier (28–33 weeks) or late (34–36 weeks), and exploratory subgroup analyses were conducted. Among 35,136 births, 2,622 (7.46%) were preterm. Mean whole-pregnancy PM 2.5 concentrations were 49.11 ug/m 3 for satellite data and 51.92 ug/m 3 for monitoring data. Per 10 ug/m 3 increase, the whole-pregnancy ORs were 1.089 (95% CI: 1.010–1.174) and 1.091 (95% CI: 0.996–1.195), respectively. Positive associations were observed during the first and third trimesters. For exposure during weeks 32–33, the ORs for late PTB were 1.063 and 1.069 using monitoring and satellite data, respectively; corresponding ORs for exposure during weeks 34–35 and PTB at week 36 were 1.084 and 1.095. Positive estimates occurred for late but not earlier PTB. Exploratory analyses suggested potential heterogeneity by ethnicity, parity, and migrant status. Prenatal PM 2.5 exposure was associated with PTB,with positive associations observed for the first and third trimesters and in the landmark analyses of weeks 32–33 and 34–35. Satellite and monitoring data yielded similar district-level estimates, while satellite data provided complete coverage and narrower confidence intervals. Reducing PM 2.5 exposure throughout pregnancy may help reduce PTB burden.