Maternal PM2.5 exposure triggers preterm birth: a cross-sectional study in Wuhan, China

Abstract

Background: Most of the studies regarding air pollution and preterm birth (PTB) in highly polluted areas have estimated the exposure level based on fixed-site monitoring. However, exposure assessment methods relying on monitors have the potential to cause exposure misclassification due to a lack of spatial variation. In this study, we utilized a land use regression (LUR) model to assess individual exposure, and explored the association between PM_2.5 exposure during each time window and the risk of preterm birth in Wuhan city, China.

Methods: Information on 2101 singleton births, which were ≥ 20 weeks of gestation and born between November 1, 2013 and May 31, 2014; between January 1, 2015 and August 31, 2015, was obtained from the Obstetrics Department in one 3A hospital in Wuhan. Air quality index (AQI) data were accessed from the Wuhan Environmental Protection Bureau website. Individual exposure during pregnancy was assessed by LUR models and Kriging interpolation. Logistic regression analyses were conducted to determine the association between women exposure to PM_2.5 and the risk of different subtypes of PTB.

Results: During the study period, the average individual exposure concentration of PM_2.5 during the entire pregnancy was 84.54 μg/m³. A 10 μg/m³ increase of PM_2.5 exposure in the first trimester (OR: 1.169; 95% CI: 1.077, 1.262), the second trimester (OR: 1.056; 95% CI: 1.015, 1.097), the third trimester (OR: 1.052; 95% CI: 1.002, 1.101), and the entire pregnancy (OR: 1.263; 95% CI: 1.158, 1.368) was significantly associated with an increased risk of PTB. For the PTB subgroup, the hazard of PM_2.5 exposure during pregnancy was stronger for very preterm births (VPTB) than moderate preterm births (MPTB). The first trimester was the most susceptible exposure window. Moreover, women who had less than 9 years of education or who conceived during the cold season tended to be more susceptible to the PM_2.5 exposure during pregnancy.

Conclusions: Maternal exposure to PM_2.5 increased the risk of PTB, and this risk was stronger for VPTB than for MPTB, especially during the first trimester.

Keywords: Air pollution; LUR model; Moderate preterm birth; PM2.5; Preterm birth; Very preterm birth.

Fig. 1

Flow chart illustrating the selection process for the cohort study population

Fig. 2

Geographical location of Wuhan in China (a). Spatial distribution of mean PM_2.5 estimations across Wuhan city from January 1, 2013 to August 31, 2015 (b). A scatter plot correlating the measured and predicted PM_2.5 values from 10 monitoring stations (c)

Fig. 3

Distribution of months of conception for participants and individual PM_2.5 exposure in the specific trimester. Notes: The height of the bar indicates the number of participants. The dot indicates the woman who exposure to a specific PM_2.5 concentration. The density plot on the right marginal of y-axis visualizes the distribution of the number of subjects over the PM_2.5 concentration intervals

Fig. 4

Adjusted odd ratios (95% CIs) of VPTB, MPTB, and PTB by entire pregnancy and trimester-specific exposure quartile. Logistic regression models were adjusted for maternal age, years of schooling, delivery mode, gravidity, parity, season of conception, gestational diabetes, gestational hypertension, and sex of baby. Abbreviations: OR, odds ratio; PTB, preterm births (<37 weeks); MPTB, moderate preterm births (32–37 weeks); VPTB, very preterm births (28–32 weeks); 1st, the first quartile; 2nd, the second quartile; 3rd, the third quartile; 4th, the fourth quartile