Exposures to air pollutants during pregnancy and preterm delivery.

Abstract

The association between preterm delivery (PTD) and exposure to air pollutants has recently become a major concern. We investigated this relationship in Incheon, Republic of Korea, using spatial and temporal modeling to better infer individual exposures. The birth cohort consisted of 52,113 singleton births in 2001–2002, and data included residential address, gestational age, sex, birth date and order, and parental age and education. We used a geographic information system and kriging methods to construct spatial and temporal exposure models.

Associations between exposure and PTD were evaluated using univariate and multivariate log-binomial regressions. Given the gestational age, birth date, and the mother’s residential address, we estimated each mother’s potential exposure to air pollutants during critical periods of the pregnancy. The adjusted risk ratios for PTD in the highest quartiles of the first trimester exposure were 1.26 for carbon monoxide, 1.27 for particulate matter with aerodynamic diameter less than or equal to 10 μm, 1.24 for nitrogen dioxide, and 1.21 for sulfur dioxide. The relationships between PTD and exposures to CO, NO2, and SO2 were dose dependent. In addition, the results of our study indicated a significant association between air pollution and PTD during the third trimester of pregnancy.

In conclusion, our study showed that relatively low concentrations of air pollution under current air quality standards during pregnancy may contribute to an increased risk of PTD. A biologic mechanism through increased prostaglandin levels that are triggered by inflammatory mediators during exposure periods is discussed.

Introduction

Preterm delivery (PTD) remains the leading cause of perinatal mortality and occurs in approximately 4–10% of pregnancies. Known risk factors for PTD include lower social class, less education, single marital status, low income, younger maternal age, low body weight, ethnicity, smoking, and poor housing, along with medical factors such as induction, premature rupture of membranes, infection, multiple pregnancy, intrauterine death, fetal and uterine abnormalities, and chorioamnionitis. Associations between ambient air pollutants and adverse pregnancy outcomes have also been reported. The ambient air pollutants of concern in these studies include carbon monoxide, nitrogen dioxide, sulfur dioxide, ozone, and particulate matter (PM). Because of differences in pathogenic mechanisms, the effects of air pollutants in these studies were analyzed separately for each perinatal outcome. Additional studies have reported an association between exposure to air pollutants during critical periods of pregnancy and PTD, although the biologic mechanism that mediates the link between exposure to air pollutants and PTD is not well understood. Previous studies for adverse pregnancy outcomes, however, had limited spatial and temporal information on pollution sources and concentrations.

The purpose of this study was to investigate the associations between air pollution and PTD in Incheon, Republic of Korea (Korea). The study has two main objectives. The first is to construct spatially and temporally explicit surfaces of atmospheric pollutants that serve as surrogates for potential exposure to air pollution, corresponding to the first, second, and third trimesters of pregnancy. A second, and the primary, objective of this study is to relate these exposure surfaces to PTD. The results of this study will provide a greater understanding of the effect of air pollution on PTD and the impact of potential exposure on critical periods of pregnancy, and suggest possible hypotheses about the biologic mechanism linking exposure to air pollutants and PTD.

Discussion

In our study, the highest ambient air pollution concentrations during the first trimester were significantly associated with elevated relative risks of PTD. Similar results were found for NO2 and CO during the third trimester. These results are generally consistent with the findings from China, the United States, Canada, and the Czech Republic. These studies reported significant associations between air pollution and PTD during early pregnancy (i.e., first or second month, first trimester) , late pregnancy (i.e., last month, last trimester, 7 days or 6 weeks before birth), or during both early and late pregnancies.

Our study has several strengths. First, this birth cohort study is population based and is less likely to suffer from selection bias than other studies. Second, the present study is one of only a few studies using a large sample size to assess the potential effects of maternal exposure to ambient air pollutants on PTD. A larger cohort size might have further improved this study; however, when this study was initiated, the 2003 birth cohort data were not available and the data before 2001 did not contain residential addresses. Third, birth records in Korea are generally accepted as complete, with reliable individual information on both parents and infants recorded on each certificate. Therefore, we were able to estimate the risks after controlling for the effects of potential confounding factors.

Finally, a more accurate exposure assessment for individual mothers was carried out in our study. Reliable measurements of daily SO2, NO2, CO, and PM10 concentrations were available from several air monitoring stations throughout Incheon, and our study used kriging methods to predict the spatial distribution of the air pollutants. The kriging method, unlike proximity models, uses real pollution measurements in the computation of exposure estimates. In situations where many monitoring stations exist, kriging methods are often preferred to other interpolation methods because they are fairly accurate in a variety of situations and avoid the artifacts that often result from the use of inverse distance weighted, spline, or global/local polynomials. Therefore, our assignment of exposure using monthly block kriging from air monitoring stations is one of the preferred methods.

This study also has several weaknesses. First, maternal smoking and environmental tobacco smoke are well-known risk factors for adverse pregnancy outcomes, but this information was not available from the birth registry. However, because most women in Korea are not likely to smoke during pregnancy, omission of this risk factor from the analyses is not likely to bias the results. Second, although our study attempted to decrease misclassification of individual exposures by enhancing exposure assessment through spatially and temporally explicit exposure models, the potential for misclassification of exposure due to the use of surrogate ambient air pollution data still exists. The only real way to avoid such potential misclassifications is to conduct personal exposure assessments, which are often not feasible.

Third, although we had access to a relatively high density of 27 air monitoring stations near and around Incheon and used block kriging to construct spatial exposure surfaces, the uncertainty of the predicted average concentrations for the dongs was not incorporated into the regression analyses. This is a common limitation of nearly all similar studies because error propagation is computationally difficult. Finally, because we could not geocode the residential addresses to point locations, the analysis is “ecologic,” meaning that the results associated with the dong level may not apply to individuals and that an analysis using different administrative units could produce different results.

Several hypotheses have been postulated to explain the mechanism of triggering PTD. One hypothesis suggests causality between uterine inflammation and PTD. The direct evidence that infection provokes preterm labor was first shown in an animal study. When group B streptococci were injected into the amniotic fluid in preterm rhesus monkeys, amniotic fluid cytokine concentrations increased, followed by production of the prostaglandins E2 and F2α, and finally uterine contractions. Similarly, in humans, preterm labor due to infection is thought to be initiated by cytokines, including interleukin-(IL)1, tumor necrosis factor, and IL-6, produced by macrophages.

Because IL-1β is not present in the membranes of term-laboring patients, it may be the unique mediator by which intrauterine infection induces preterm labor. Antenatal infection can trigger intrauterine inflammation, which then promotes preterm labor. In addition, periodontal disease may be an independent risk factor for preterm labor: Postulated mechanisms include translocation of periodontal pathogens to the fetoplacental unit and action of a periodontal reservoir of lipopolysaccharides or inflammatory mediators. Our inability to determine periodontal status of the mother is a potential confounding factor. Cyclooxygenase-2 inhibitor, developed as an anti-inflammatory drug, also has toxolytic effects. A similar inflammatory mechanism has been suggested for the effect of smoking on fetal growth retardation, PTD, and perinatal mortality. There are reports of increased blood viscosity and plasma fibrinogen during air pollution. It has been speculated that chronic exposure to high pollution levels may influence placental function. The placental dysfunction may lead to intrauterine fetal growth retardation. The effects of air pollution on pregnancy outcomes may differ with the timing of exposure, with early exposures likely to be important for pregnancy end points such as spontaneous abortion, intrauterine growth retardation, and birth defects. Intrauterine infection during pregnancy could also lead to brain damage of the developing fetus.

Recent studies suggest that antenatal infection and inflammation can increase the preterm infant’s susceptibility to develop chronic lung disease. It may be that exposure of the fetal lung to high concentrations of proinflammatory cytokines is the cause of this increased susceptibility. Photochemically produced gaseous products influence the toxic responses of cells, such as production of cytokines, in the absence of particles. PM10 is responsible for the production and the release of inflammatory cytokines by the respiratory tract epithelium as well as for activation of the transcription factor nuclear factor κB. Although fetal exposures to air pollution are probably much lower than exposure to the constituents of cigarette smoke, we propose that the biologic mechanism of PTD could be through increased prostaglandin levels that are triggered by inflammatory mediators during exposure periods.

The pathophysiology of CO may be more complex, involving hypoxic stress on the basis of interference with oxygen transport to the cells and possibly impairment of electron transport. CO can also affect leukocytes, platelets, and the endothelium, inducing a cascade of effects resulting in oxidative injury. CO may interfere with metabolic and transport function of the placenta and, after crossing the placental barrier, concentrate more in the fetus than in the mother. Neonates and fetuses are more vulnerable because of the natural leftward shift of the dissociation curve of fetal hemoglobin, a lower baseline PO2 (partial pressure of oxygen), and carboxyhemoglobin levels at equilibration that are 10–15% higher than maternal levels.

The causality between air pollution and risk of intrauterine growth retardation and decreased birth weight, birth length, and head circumference has been suggested through molecular epidemiologic studies where levels of DNA adducts are positively correlated with these outcomes. The DNA damage may occur through exposure to poly-cyclic aromatic hydrocarbons. Although this study identifies an association between air pollution and PTD, PTD may be less sensitive to air pollution, possibly because of the postulated multifactorial nature of this health outcome.

In this study, we observed air pollution levels critical to PTD in humans. These levels are important because they may be a good indication on how to protect fetuses against adverse effects from air pollutants. In Korea, the current annual air quality standards are 52.4 μg/m3 for SO2, 94 μg/m3 for NO2, and 70 μg/m3 for PM10. The CO standard over 8 hr is 10.4 mg/m3. Korea’s annual standard for air quality is certainly too high and does not prevent adverse pregnancy outcomes. Our study showed that statistically significant effects of PTD are seen below the air quality standards for CO and NO2 and potentially below the standards for PM10 and SO2. Our study may provide supportive evidence that reduction in the current air quality standards may improve pregnancy outcomes.

In conclusion, our study showed that relatively low concentrations of air pollution under current air quality standards during critical gestational periods may contribute to increased risk of PTD. Our results also suggest that fetuses in the early and late stages of development are susceptible to air pollutants. Further studies are needed to validate fetal susceptibility to air pollutants with more detailed information on personal exposures, confounders, and effect modifiers.