Ambient air pollutant PM10 and risk of pregnancy-induced hypertension in urban China

Background: The relationship between air borne particulate matter ≤10 μm (PM10) exposure and pregnancy-induced hypertension (PIH) is inconclusive. Few studies have been conducted, and fewer were conducted in areas with high levels of PM10. Methods: To examine the association between PM10 and PIH by different exposure time windows during pregnancy, we analyzed data from a birth cohort study conducted in Lanzhou, China including 8 745 pregnant women with available information on air pollution during pregnancy. A total of 333 PIH cases (127 gestational hypertension (GH) and 206 preeclampsia (PE)) were identified. PM10 daily average concentrations of each subject were calculated according to the distance between home/work addresses and monitor stations using an inverse-distance weighting approach. Results: Average PM10 concentration over the duration of entire pregnancy was significantly associated with PIH (OR = 1.12, 95%CI: 1.02, 1.23 per 10 μg m−3 increase), PE (OR = 1.16, 95%CI: 1.03, 1.30 per 10 μg m−3 increase), late onset PE (OR = 1.17, 95% CI: 1.03, 1.32 per10 μg m−3 increase), and severe PE (OR = 1.25, 95% CI: 1.06, 1.48 per 10 μg m−3 increase). Average PM10 during the first 12 gestational weeks was associated with the risk of GH (OR = 1.10, 95% CI: 1.00, 1.21 per 10 μg m−3 increase), and PM10 exposure before 20 gestational weeks was associated with the risk of severe PE (OR = 1.14, 95% CI: 1.01, 1.30 per 10 μg m−3 increase). Conclusions: We found that high level exposure to ambient PM10 during pregnancy was associated with an increased risk of PIH, GH and PE and that the strength of the association varied by timing of exposure during pregnancy.


Introduction
Ambient air contains a dynamic mixture of pollutants, including particulate matter (PM), nitrogen dioxide (NO 2 ), ozone (O 3 ), sulphur dioxide (SO 2 ), and volatile organic compounds. PM poses the greatest global air pollution threat, and therefore has received more attention [1]. According to the estimation of the 2010 Global Burden Disease [2], PM pollution causes more than 3.2 million deaths and 76 million disability adjusted life-years (DALYs) every year worldwide. Approximately one third of these deaths and DALYs occurred in China [3]. PM pollution has become a critical public health concern in China.
Pregnancy-induced hypertension (PIH), which includes gestational hypertension (GH) and preeclampsia (PE), is a major cause of maternal, fetal, and neonatal morbidity and mortality [4,5]. Furthermore, PIH could increase the risk of long-term cardiovascular diseases of both mothers and their children [6,7]. PIH became the second leading cause of maternal death in China, which accounted for 13.8% and 8.8% of total maternal deaths in urban and rural areas, respectively [8]. Several factors have been suggested to be associated with PIH risk, including maternal age older than 40 years, primiparity, diabetes, obesity and preexisting hypertension [9]. However, these factors cannot explain all PIH.
Inhalable particles are generally defined as less than 10 μm in aerodynamic diameter (PM 10 ). Studies have reported that exposure to inhalable PM can damage vascular endothelium, impair vascular reactivity, and accelerate atherogenesis [1]. Although inhalable PM has the potential to increase the risk of PIH, the association between ambient PM 10 pollution and PIH received inconsistent results. Some studies found a significantly positive association between PM 10 exposure during entire pregnancy and GH [10], while others reported no association between GH or PE and PM 10 during entire pregnancy [11] or PM 10 during the first trimester [12,13]. In addition, these studies were conducted in developed countries where the maximum PM 10 levels were under 50 μg m −3 [10][11][12][13][14][15]. Fifty eight percent of the urban population-concentrated in over 100 cities in China reported an annual average PM 10 concentration exceeding 100 μg m −3 in 2003 [16]. Given the inconsistent relationship between PM 10 and PIH and the paucity of studies in areas with high levels of PM 10 , we conducted a study in Lanzhou, China, where levels of PM 10 were routinely above 140 μg m −3 , to examine the association between PM 10 and PIH.

Study population
The study population has been described previously [17][18][19]. In brief, pregnant women who came to the Gansu Provincial Maternity & Child Care Hospital (GPMCCH) in Lanzhou, China for delivery in 2010-2012, who were 18 years or older with a gestational age of 20 weeks and without mental illness were eligible. A total of 10 542 (73.4%) women participated in the study. Subjects (N=1344) whose residence addresses during pregnancy were outside of Lanzhou city were excluded due to a lack of information on PM 10 exposures. Furthermore, women who gave multiple births, still birth, and/or birth defects, or who had chronic hypertension or chronic cardiovascular diseases were also excluded, which yielded a final sample size of 8 745. All study procedures were approved by the Human Investigation Committees at the GPMCCH and Yale University. After obtaining written consent, an in-person interview was conducted at the hospital by trained study interviewers using a standardized and structured questionnaire to collect information on demographics, reproductive and medical history, lifestyle factors, occupation, and residential history. The majority of women (84%) were interviewed within one to three days after delivery, while others were interviewed within 2 days before delivery. Information on maternal complications and birth outcomes were abstracted from the medical records.
Exposure assessment Data on ambient air pollutants were obtained from the Gansu Provincial Environmental Monitoring Central Station, which collects 24 h average concentration for PM 10 , SO 2 , and NO 2 through an automated data reporting system from four monitoring stations in Lanzhou [20]. The 24 h average PM 10 was measured for the period 1 April, 2009 to 31 December, 2012 for two stations (Huanghebei and Xigu stations), and 1 January, 2011 to 31 December, 2012 for the two additional stations (Xizhan and Tieluju stations). The monitors were located in the southern part of Lanzhou in the metropolitan area with high population density [20]. Though the distance from the participant's home and work addresses to the nearest monitors ranged from 0.1 to 88.5 km (mean: 5.0 km, median: 3.3 km), the majority (90%) of participants lived within 5.5 km from the nearest monitors. The coefficients of divergence between daily average observed PM 10 levels versus distance between monitor locations for all monitor-pairs were lower than 0.20, indicating less spatial heterogeneity [18]. Values from these monitors were used to represent community-level exposure for Lanzhou, to investigate the association between outdoor air exposure and PIH.
The exposure measurement for each subject's residences throughout pregnancy was described previously [18]. In brief, we used the earth online sharing website provided by Google (www.earthol.com) to obtain longitude and latitude coordinates for each subject's home and work addresses. The move-in and move-out dates, and work addresses were collected. We calculated daily PM 10 concentration at each subject's home and work addresses using (1) all four monitors with the inverse-distance weighting approach, (2) the two monitors in operation the full study period (April 2009 to December 2010) and inverse distance weighting, and (3) the nearest monitor.
For each subject we calculated the daily exposure levels during pregnancy by considering exposure time at home and work. Since the regular working hours are about 8 h d −1 , we used a time-weighted approach to calculate daily PM 10 on weekdays for each subject (i.e., two-thirds of exposure at home address and one-third exposure at work address). Weekend exposures were based on home address. Residential mobility was considered as time-weighted averaging to account for changes in residence during pregnancy. Finally, the daily exposures were averaged over four exposure windows based on a priori decisions: 1-20 gestational weeks, <13 gestational weeks, 13-20 gestational weeks, and entire pregnancy. Exposures for each subject to NO 2 and SO 2 were generated in the same manner as PM 10 .
Statistical Analysis PIH cases included both GH and PE conditions. GH was diagnosed as hypertension (measured twice with 6 h apart, 140/90 mmHg) manifested after 20 weeks of gestation without proteinuria. PE was diagnosed as hypertension concurrent with proteinuria (1+ on dipstick in two urine samples) after 20 weeks of gestation. PE was further classified as mild PE (raised blood pressure 140/90 mmHg and <160/ 110 mmHg, proteinuria 1+ and <2+ on dipstick in two urine samples, without symptoms of severity), severe PE (raised blood pressure 160/110 mmHg, proteinuria 2+ on dipstick in two urine samples, with symptoms of severity such as headache, blurred vision, epigastric burning pain, decreased urine output, decreased or absent fetal kick etc), early onset PE (EOPE, diagnosed before 34 weeks of gestation), or late onset PE (LOPE, diagnosed at or after 34 weeks of gestation). PIH was defined as a combination of GH and PE. Controls were those without GH or PE [19].
Chi-squared tests were used to compare the distributions of maternal characteristics between PIH and controls. Unconditional logistic regression models were used to calculate odds ratios (OR) and 95% confidence intervals (CI) for the associations between PM 10 exposure and risk of PIH, GH, PE, severe PE, mild PE, EOPE and LOPE by different exposure time windows. Average PM 10 levels in each exposure time window were treated as a continuous variable, and results were presented for the change in health outcome per 10 μg m −3 . Potential confounding variables included maternal age (<25, 25-29, 30-34, and 35 years), education (<college, college), family monthly income per capita (<2000, 2000-3999, 4000RMB, and unknown), smoking during pregnancy (yes or no), pre-pregnancy BMI [21] (<18.5, 18.5 −22.9, and 23 kg m −2 ), season of conception (spring (January-March), summer (April-June), fall (July-September), and winter (October-December)), nulliparous (yes or no), previous PIH (yes or no), vitamin supplement intake during pregnancy (yes or no), averaged temperatures and other pollutants (SO 2 and NO 2 ) in the time windows corresponding to the time frame used for PM 10 . Additional adjustments for alcohol consumption, infant gender, weight gain during pregnancy did not result in material changes of the observed associations and thus were not included in the final models (results not shown). All analyses were performed using SAS, version 9.2 (SAS Institute, Inc., Cary, NC).

Results
Of the 8 745 study subjects, 333 (3.81%) were diagnosed with PIH (127(1.45%) GH and 206 (2.56%) PE) (table 1). Of the 206 PE cases, 110 (53.40%) were severe PE and 22 (10.68%) were EOPE. Compared to controls, women with PIH were more likely to be older, less educated, lower family income, higher prepregnancy BMI, having previous PIH, multipara, and less likely to take vitamin supplement during pregnancy. There were no significant differences in the distribution of conception season and exposure to smoking during pregnancy between the cases and controls. The results of the multivariate logistic analysis on the association between PIH and these risk factors were presented in supplementary table 2. Table 2 presented average concentrations of PM 10 during different exposure time windows for the overall study population under different exposure estimation models. The different approaches yielded similar estimations of PM 10 exposure concentrations. Table 3 presented the results using exposure data obtained from all four monitors with the inverse-distance weighting approach. Average PM 10 exposure during the entire pregnancy was significantly associated with an increased risk of PE, LOPE, and severe PE (OR=1. 15

Discussion
Our study is the first to examine the effects of exposures to very high level of ambient PM 10 and the The results of previous studies assessing PM 10 and risk of PIH have been inconsistent. Vinikoor-Imler et al [10] conducted a study in North Carolina that included more than 222 000 women (12 085 PIH) and reported a 7% increased risk of PIH per 3.92 μg m −3 ) increase in daily concentration of PM 10 during entire pregnancy. Van den Hooven et al in Netherlands [14] including 7006 (250 PIH and 141 PE) study subjects found that per 10 μg m −3 increase in daily concentration of PM 10 during entire pregnancy was associated with a 72% increased risk of PIH but not PE. Consistent with these two studies, we also observed an increased risk of PIH associated with per 10 μg m −3 increase in PM 10 levels during entire pregnancy. Dadvand et al [11] conducted a study in Spain involving 8398 (103 PE) women and found no statistically significant association between PE and PM 10 exposure during entire pregnancy and different trimesters. Mobasher et al [12] conducted a case-control study among Hispanic women with 136 PIH cases and found no association between PIH and PM 10 exposure in any trimester. However, Lee et al conducted a cohort study in Pittsburgh including 1684 women (110 GH and 32 PE) and found that exposure to PM 10 during the first 20 weeks of gestation significantly increased systolic and diastolic blood pressures [15]. In Lee's later research in Pittsburgh, which enrolled another 34 705 women with 2078 GH, they found that PM 10 during the first 12 weeks of gestation marginally increased the risk of GH (OR=1.08, 95%CI: 0.98,1.20 per 7.7 μg m −3 increase in PM 10 ) [13], which was comparable to our study finding that per 10 μg m −3 increase in PM 10 during the first 12 weeks of gestation significantly increased the risk of GH.
Our study found that LOPE, not EOPE, was associated with average PM 10 exposure during entire pregnancy. While no study examined LOPE and EOPE seperately in relation to PM 10 exposure, Dadvand et al [11] found that PM 2.5 exposure in the third trimester was associated with LOPE. Previous studies have also demonstrated that EOPE and LOPE might have different pathogenesis [22,23]. EOPE appears to be more related to placental disorder. However, LOPE seems to be more linked to maternal constitutional factors [22,23]. We speculated that high PM 10 exposure during entire pregnancy might do harm to maternal cardiovascular constitution. We also found that PM 10 exposure during the first 20 gestational weeks or entire pregnancy could increase the risk of severe PE but not mild PE. While its mechanism needs further research, it might suggest that PM 10 plays a major role in PE progression not PE initiation.
All previous studies were conducted in areas with relatively low concentrations of PM 10 . The mean PM 10 exposure concentrations in previous studies [10,11,[13][14][15]  . A large percent of the world's population is exposed to high levels of air pollution and relatively few studies have been conducted in such areas. Our study is a timely effort to address this understudied issue.
Air pollution is a complex mixture of several pollutants. Other pollutants such as PM 2.5 , CO, O 3 , NO 2 , and SO 2 could also increase the risk of PIH. Given the available data in our study, we included SO 2 , NO 2 , and PM 10 into the multivariate models simultaneously and found that the associations with PM 10 were strengthened. We observed stronger associations with average PM 10 exposure in an entire pregnancy than in a specific trimester. A different time window of exposure during pregnancy yields different effects on PIH that are likely through different mechanisms [15]. Previous studies have demonstrated that long term exposure to air pollutants, such as high exposure during an entire pregnancy, may be related to the harmful effects on cardiovascular constitution [14]. Short term exposure to harmful pollutants, such as brief exposure during a specific trimester, may trigger harmful effects [11].
Strengths and limitations should be considered when interpreting the study findings. Diagnosis of Abbreviations: PM, particulate matter; SD: standard deviation a Calculated by using exposure data obtained from all four monitors. b Calculated by using exposure data obtained from two monitors in operation the full study period. c Calculated by using the nearest monitor.
GH/PE in our study was based on medical records not self-report, which minimized potential disease misclassification. Our study collected detailed information on demographic factors, smoking, occupation and medical histories, which allowed us to control for these important potential confounding factors. We considered both residential and work addresses as well as mobility during pregnancy when we calculated daily exposure levels of each pollutant. Information on work address and on residences throughout pregnancy is often not available in studies of air pollution and pregnancy outcomes. Individual exposures to ambient air pollutants in our study were based on monitoring stations rather than direct measurement (i.e., personal monitoring) for each subject due to feasibility concerns in such a large population-based study. PM 10 exposure in our study was much higher than in other published studies, and the findings from our study might not be generalizable to populations who live in the areas with low air pollution levels, although the results could be applicable to other regions of the developing world with similarly high levels of ambient air pollution. Previous studies have also reported a positive relationship between PM 2.5 and PE [12,14,15,25] which we were unable to investigate due to the lack of PM 2.5 monitoring. Therefore, our observed association between PM 10 and PIH may involve the effects of PM 2.5 on PIH.

Conclusion
The results of our study support the notion that increasing exposure to ambient PM 10 during pregnancy is associated with an increased risk of PIH, GH and PE, and the postive linear relationship sustains in high level exposures of PM 10  The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Yawei Zhang and Qin Liu had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors had final responsibility for the decision to submit for publication.