Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Dec 1.
Published in final edited form as: Environ Res. 2020 Sep 13;191:110201. doi: 10.1016/j.envres.2020.110201

Ambient Air Pollution and Risk of Pregnancy Loss among Women Undergoing Assisted Reproduction

Audrey J Gaskins a, Lidia Minguez-Alarcón b, Paige L Williams c,d, Jorge E Chavarro c,e,f, Joel D Schwartz b,f, Itai Kloog g, Irene Souter h, Russ Hauser b,h, Francine Laden b,c,h; EARTH Study Team
PMCID: PMC7658021  NIHMSID: NIHMS1629960  PMID: 32937174

Abstract

Accumulating evidence suggests that air pollution increases pregnancy loss; however, most previous studies have focused on case identification from medical records, which may underrepresent early pregnancy losses. Our objective was to investigate the association between acute and chronic exposure to ambient air pollution and time to pregnancy loss among women undergoing assisted reproductive technologies (ART) who are closely followed throughout early pregnancy. We included 275 women (345 human chorionic gonadotropin (hCG)-confirmed pregnancies) undergoing ART at a New England academic fertility center. We estimated daily nitrogen dioxide (NO2), ozone (O3), fine particulate matter <2.5 μm (PM2.5), and black carbon (BC) exposures using validated spatiotemporal models estimated from first positive hCG test until day of failure or live birth. Air pollution exposures were averaged over the past week and the whole pregnancy. Multivariable Cox proportional hazards models were used to estimate the hazards ratio (HR) for pregnancy loss for an interquartile range (IQR) increase in pollutant exposure. We tested for violation of proportional hazards by considering an interaction between time (in days) since positive hCG (<30 days vs. ≥30 days) and air pollution. The incidence of pregnancy loss was 29 per 100 confirmed pregnancies (n=99). Among pregnancies not resulting in live birth, the median (IQR) time to loss was 21 (11, 30) days following positive hCG. Average past week exposures to NO2, O3, PM2.5, and BC were not associated with time to pregnancy loss. Exposure throughout pregnancy to NO2 was not associated with pregnancy loss; however, there was a statistically significant interaction with time (p-for-interaction<0.001). Specifically, an IQR increase in exposure to NO2 was positively associated with pregnancy loss after 30 days (HR=1.34, 95% CI: 1.13, 1.58), but not in the first 30 days after positive hCG (HR=0.83, 95% CI: 0.57, 1.20). Overall pregnancy exposure to O3, PM2.5, and BC were not associated with pregnancy loss regardless of timing. Models evaluating joint effects of all pollutants yielded similar findings. In conclusion, acute and chronic exposure to NO2, O3, PM2.5, and BC were not associated with risk of pregnancy loss; however, higher exposure to NO2 throughout pregnancy was associated with increased risk of loss 30 days after positive hCG. In this cohort, later pregnancy losses appeared more susceptible to the detrimental effects of air pollution exposure.

Keywords: air pollution, assisted reproduction, miscarriage, pregnancy loss, in vitro fertilization

Introduction.

Miscarriage is the most common complication of pregnancy, affecting approximately 30% of conceptions (Wilcox et al. 1988) and 11-22% of all recognized pregnancies by 20 weeks of gestation (Ammon Avalos et al. 2012). Despite its already high frequency, data from the U.S. National Survey of Family Growth 1990-2011 suggests that the risk of pregnancy loss may be increasing- by at least 1% per year- even after accounting for changes in maternal- and pregnancy-related factors (Rossen et al. 2018). However, the cause of most miscarriages, including this apparent rise in incidence over the past two decades, is largely unknown and likely due to both intrinsic and extrinsic characteristics including demographic, genetic, hormonal, immunological, and environmental factors (Agenor and Bhattacharya 2015).

Ambient air pollution is of high interest given how universal and widespread exposure is among urban populations and its established link to adverse pregnancy outcomes such as preterm birth (Lamichhane et al. 2015; Shah and Balkhair 2011). A growing number of studies have documented that acute and chronic exposure to air pollution is related to increased risk of pregnancy loss; however, results have varied by pollutant and time period of exposure (Dastoorpoor et al. 2018; Di Ciaula and Bilancia 2015; Enkhmaa et al. 2014; Green et al. 2009; Ha et al. 2018; Hou et al. 2014; Leiser et al. 2018; Moridi et al. 2014; Perin et al. 2010a; Perin et al. 2010b). An important limitation of all but a handful of studies has been the reliance on case identification from medical records. This type of design tends to underrepresent early pregnancy losses because without prospective assessment of pregnancy, many losses go unrecognized. Moreover, even recognized early losses tend to be managed outside a hospital setting and are not always documented in the medical record. To date, the sole prospective study on exposure to ambient air pollution and risk of pregnancy loss (n=344 pregnancies) found that higher residential exposure to particulate matter <2.5 μm (PM2.5) and ozone (O3) throughout pregnancy was associated with increased risk of loss but acute exposures in the 1 or 2 weeks prior to loss were not (Ha et al. 2018).

To further explore this association and extend it to a subpopulation that may be more sensitive to air pollution, we utilized a prospective cohort of women undergoing assisted reproductive technologies (ART). Couples were closely followed throughout early pregnancy to investigate how both acute and chronic exposure to ambient air pollution influences pregnancy loss.

Materials and Methods.

Study population.

The Environment and Reproductive Health (EARTH) Study recruits patients presenting for infertility evaluation and treatment at the Massachusetts General Hospital (MGH) Fertility Center (2004-2019). All women planning to undergo infertility treatment who are 18-46 years are eligible to participate. Approximately 60% of eligible women who are contacted by the research staff participate in the study. The institutional review boards of the Massachusetts General Hospital and Harvard T.H. Chan School of Public Health approved this study and all subjects met with trained study staff before providing written informed consent,.

To be eligible for this analysis, ART cycles had to of ended by December 2015 due to the temporal constraints of the air pollution models. Of the 636 eligible ART cycles, we excluded 3 cycles from women residing outside of the US and 3 cycles missing data on the air pollutants. We then further excluded 285 cycles that did not result in a biochemical confirmation of pregnancy, bringing our final sample size to 345 ART cycles from 275 women. Two-hundred and nine pregnancies (81%) were the result of a fresh embryo transfer, 44 (13%) from a cryothaw embryo transfer, and 22 (6%) from a donor oocyte embryo transfer. For the analysis of black carbon (BC), we further excluded 6 women (9 ART cycles) who resided in states outside the range of this model.

Air Pollution Exposures.

Women reported their residential address at study entry for reimbursement purposes. These addresses were subsequently geocoded using ArcGIS and then linked to various spatio-temporal models of air pollution. For PM2.5, NO2, and O3, each woman was linked to the closest 1 x 1 km grid cell of the modeled daily exposure dataset. Ground-level PM2.5 concentrations were estimated using a validated hybrid model that uses satellite-derived aerosol optical depth measurements along with land use and meteorological variables (Kloog et al. 2014). NO2 concentrations were estimated using a validated model consisting of satellite-derived tropospheric NO2 vertical column densities, ground level NO2 concentrations from US EPA monitoring sites, land use variables, and meteorological data (Lee and Koutrakis 2014). Daily O3 exposure was predicted using neural networks that combine data from a chemical-transport model of ozone, satellite based ozone measurements, land use, and meterological variables (Di et al. 2017). For all models, land use terms included measures of population density, elevation, traffic, percentages of land use, normalized difference vegetation index, and point and source pollutant emissions. Meteorological variables included air temperature, wind speed, daily visibility, sea land pressure, and relative humidity. For PM2.5, temporally resolved data on planetary boundary layer were also incorporated. Daily residential exposure to BC was estimated at the home address using a validated spatio-temporal model based on support vector machine regression (Abu Awad et al. 2017). The inputs to this model included ambient BC measurements from hundreds of stationary monitors in Massachusetts, Rhode Island and southern New Hampshire as well as several spatial (e.g. proximity to transportation, topographical characteristics, neighborhood attributes) and temporal (e.g. temperature, wind speed, visibility, dew point, sea-level pressure, and humidity) predictors.

We assessed both acute and chronic exposure to air pollution. Because the underlying mechanisms linking air pollution and pregnancy loss have not been identified, our choice of exposure time windows was largely exploratory and informed by previous research. Chronic exposure was defined as the cumulative daily average exposure starting on the date of first positive human chorionic gonadotropin (hCG) until the date of loss or live birth. Acute exposure was defined as the daily rolling average of exposures in the 7 days prior to loss or live birth. As sensitivity analyses, we investigated the fixed, two-week time window between embryo transfer and hCG pregnancy test and we investigated the cumulative daily average exposure starting on the date of embryo transfer until the date of loss or live birth. We also truncated cumulative average air pollution exposure at 20 weeks to ensure a comparable length of exposure.

Pregnancy Loss.

The main outcome of interest was time to pregnancy loss as measured from the date of the first positive hCG to loss or live birth. Following embryo transfer, a quantitative serum β-hCG measurement is typically performed 17 days after oocyte retrieval or 12 days after a day 5 embryo transfer (in cryo-thaw cycles). A positive pregnancy test was defined as a serum β-hCG level >6 mIU/mL. Following a positive hCG, a transvaginal ultrasound is performed at approximately 6 weeks' gestation to confirm an intrauterine clinical pregnancy. For our analysis, any ART cycle that resulted in a positive hCG but did not end in live birth was considered a loss (n=99). This included biochemical pregnancies (n=40), ectopic pregnancies (n=3), induced abortions (which were all performed due to severe conditions incompatible with life, n=3), spontaneous abortions (n=48), and stillbirths (n=5). For comparison purposes, we followed guidelines from the American College of Obstetricians and Gynecologists to estimate gestational age using the following calculation: outcome date – date of transfer + 14 + cycle day of transfer (2014).

Covariate Assessment.

At enrollment, information on demographics (e.g. age, race,), medical history (e.g. parity, gravidity), and lifestyle characteristics (e.g. smoking status, education, employment) were collected on a study-staff administered and take-home questionnaire. Height and weight were measured using standardized protocols by trained research study staff to calculate body mass index (BMI) (kg/m2). Relevant clinical information including infertility diagnoses and stimulation protocol was abstracted from electronic medical records. Average daily temperature values were derived from the Parameter-elevation Regressions on Independent Slopes Model (PRISM) and averaged for the same time periods as the pollutants (PRISM 2019).

Statistical Analysis.

Descriptive statistics were calculated based on the first ART pregnancy a woman contributed to this analysis. We compared demographic, reproductive history, and clinical factors between pregnancies that resulted in a loss versus a live birth. Kruskal-Wallis tests and chi-squared tests were used to test for differences between these two groups for continuous and categorical variables, respectively. Spearman correlation coefficients were used to measure the strength of association between pollutants.

Multivariable Cox proportional hazards models were used to estimate the hazards ratio (HR) and 95% confidence intervals (CI) for pregnancy loss for an interquartile range (IQR) increase in air pollution pollutant exposure. In all of these models, a robust sandwich covariance estimate was used to account for the multiple cycles per woman. Exposure measures were treated as continuous and assumed to be linearly related to the log HR. Covariates such as age and treatment protocol were updated for each ART cycle. To test our assumption of linearity between air pollutants and log HR we included squared terms for the air pollutants in the final multivariable model. We also considered quartiles of exposure to explore other non-linear associations. We used inverse probability weights to control for potential selection bias introduced by restricting the analysis to ART cycles with a positive pregnancy test, as air pollution concentrations could influence the probability of succeeding until that point (Cole and Hernan 2008; Howe et al. 2016). To construct the weights, we used the original cohort (n=630 initiated ART cycles) to calculate the conditional probability of achieving a positive pregnancy test. The ART cycles that resulted in pregnancy in our present analysis received a weight inversely proportional to the estimated probability of not being censored. The weights were computed using a logistic regression model adjusting for age, BMI, smoking status, infertility diagnosis, and protocol. We tested for violation of proportional hazards by considering an interaction between days since positive hCG (<30 days vs. ≥30 days) and air pollution levels. A p-value for interaction was calculated using the likelihood ratio test comparing the model fit before and after the interaction term was included.

Confounding was evaluated using prior knowledge and descriptive statistics from our cohort through the use of directed acyclic graphs (Weng et al. 2009). Variables retained in the final multivariable models were maternal age, BMI, smoking status, race, education, employment outside the home, and treatment protocol. Sensitivity analyses were also done further adjusting for the other pollutants, season of conception, as well as average temperature. We explored effect modification by primary infertility diagnosis, smoking status, and female age by including an interaction term between these parameters and the air pollutant. We considered two-sided significance levels less than 0.05 as statistically significant.

Results.

The 275 women in our cohort were, on average, 35 years old and the majority were white (86%), highly educated (93% had at least a college education), and currently employed (96%) (Table 1). Among the 345 hCG confirmed pregnancies that occurred in our 275 women, 99 (29%) resulted in pregnancy loss (Figure 1). The median (IQR) time between positive hCG and loss among those cycles with losses was 21 (19) days. The median (IQR) gestational age at loss was 7.4 (2.7) weeks. Only 7 pregnancy losses occurred after 20 weeks gestation (2 therapeutic abortions and 5 stillbirths). All pregnancy losses occurred by 31 weeks gestation. Women who experienced a loss were, on average, older, less likely to have a college degree and be employed outside the home, and more likely to have undergone a flare or antagonist protocol and conceived in the summer (Table 1). The median (IQR) air pollution concentrations during pregnancy were 21.8 (18.2) ppb for NO2, 35.0 (12.1) ppb for O3, 8.8 (1.7) μg/m3 for PM2.5, and 0.5 (0.2) μg/m3 for BC (Supplemental Table 1). The Spearman correlation coefficients between pollutants ranged in magnitude from −0.17 for NO2 and O3 to 0.46 for NO2 and BC (Supplemental Table 2).

Table 1.

Baseline characteristics of 275 women in Environment and Reproductive Health (EARTH) Study (2005-2015) according to the outcome of their first ART pregnancy.

Number of Women Entire
Cohort
N=275		 Live
Birth
N=194		 Pregnancy
Loss
N=81		 p-valuea
Personal Characteristicsb
Age, years 35.1 ± 3.8 34.6 ± 3.8 36.2 ± 3.7 <0.001
BMI, kg/m2 23.8 ± 4.0 23.7 ± 3.5 24.1 ± 4.9 0.70
Never Smoker, n (%) 206 (74.9) 143 (73.7) 63 (77.8) 0.48
White, n (%) 235 (85.5) 169 (87.1) 66 (81.5) 0.23
Education level, n (%) 0.07
 Less than college 19 (6.9) 9 (4.6) 10 (12.4)
 College degree 92 (33.5) 67 (34.5) 25 (30.9)
 Graduate degree 164 (59.6) 118 (60.9) 46 (56.8)
Currently Employed, n (%) 264 (96.0) 189 (97.4) 75 (92.6) 0.06
Whole Pregnancy Air Pollution
Exposure
 NO2, ppb 27.0 ± 27.1 26.3 ± 22.4 28.8 ± 36.2 0.94
 O3, ppb 44.8 ± 38.9 44.6 ± 37.7 45.4 ± 41.8 0.84
 PM2.5, μg/m3 8.9 ± 1.7 8.9 ± 1.3 8.9 ± 2.4 0.21
 BC, μg/m3 0.55 ± 0.16 0.55 ± 0.17 0.54 ± 0.16 0.99
Whole Pregnancy Temperature 11.1 (6.3) 10.9 (4.2) 11.6 (9.6) 0.31
Exposure, °C
Initial Cycle Characteristics
Nulligravid, n (%) 166 (60.4) 118 (60.8) 48 (59.3) 0.81
Nulliparous, n (%) 236 (85.8) 169 (87.1) 67 (82.7) 0.34
Initial Infertility diagnosis, n (%)
 Male factor 95 (34.6) 72 (37.1) 23 (28.4) 0.32
 Female factor 83 (30.2) 58 (29.9) 25 (30.9)
  Diminished ovarian reserve 21 (7.6) 14 (7.2) 7 (8.6)
  Endometriosis 17 (6.2) 12 (6.2) 5 (6.2)
  Ovulatory 25 (9.1) 19 (9.8) 6 (7.4)
  Tubal 18 (6.6) 12 (6.2) 6 (7.4)
  Uterine 2 (0.7) 1 (0.5) 1 (1.2)
 Unexplained 97 (35.3) 64 (33.0) 33 (40.7)
Treatment protocol, n (%) <0.001
 Luteal phase agonist 180 (65.5) 141 (72.7) 39 (48.2)
 Flare or antagonist 55 (20.0) 26 (13.4) 29 (35.8)
 Cryo or donor egg cycle 40 (14.6) 27 (13.9) 13 (16.1)
Season of Conception, n (%) 0.13
 Winter 68 (24.7) 49 (25.3) 19 (23.5)
 Spring 68 (24.8) 53 (27.3) 15 (18.5)
 Summer 76 (27.6) 46 (23.7) 30 (37.0)
 Fall 63 (22.9) 46 (23.7) 17 (21.0)
Year of Conception, n (%) 0.64
 2005-2007 47 (17.1) 33 (17.0) 14 (17.2)
 2008-2010 110 (40.0) 82 (42.3) 28 (34.6)
 2011-2013 87 (31.6) 59 (30.4) 28 (34.6)
 2014-2015 31 (11.3) 20 (10.3) 11 (13.6)
Open in a new tab

Note: BC, black carbon; NO2, nitrogen dioxide; O3, ozone; PM2.5, particulate matter <2.5 μm.

a

P-value was calculated using Kruskal-Wallis tests for continuous variables and Chi-Square tests for categorical variables.

b

Data are presented as mean ± standard deviation or N (%).

Figure 1.

Figure 1.

Open in a new tab

Cumulative incidence of pregnancy loss by days since positive human chorionic gonadotropin (hCG) among 275 women (345 ART pregnancies) in Environment and Reproductive Health (EARTH) Study (2005-2015).

Exposures to NO2, O3, PM2.5, and BC in the past week were not associated with time to pregnancy loss (Table 2). Chronic exposures to O3, PM2.5, and BC throughout pregnancy were also not associated with pregnancy loss. While cumulative average exposure to NO2 throughout pregnancy was not associated with pregnancy loss, there was a statistically significant interaction with time (p-for-interaction<0.001). Specifically, an IQR increase in exposure to NO2 was positively associated with pregnancy loss after 30 days (HR=1.34 95% CI 1.13, 1.58), but not in the first 30 days following positive hCG (HR=0.83 95% CI: 0.57, 1.20). While the cut-off of 30 days was chosen a priori to correspond with a month after positive hCG and roughly 8 weeks gestation, when the interaction between NO2 and time was modelled continuously, the HR for the association between NO2 and risk of pregnancy loss became >1 (indicating a harmful association) at 15.6 days of pregnancy (6.7 gestational weeks) (Supplemental Figure 1).

Table 2.

Association between time-varying cumulative average and prior week exposure to nitrogen dioxide (NO2), ozone (O3), particulate matter <2.5 μm (PM2.5) and black carbon (BC) exposure and time to pregnancy loss (n=275 women, 345 ART pregnancies) among women in the Environment and Reproductive Health (EARTH) Study (2005-2015).

Adjusted HR (95% CI) per IQR Increasea
Main Effect <30 days ≥30 days
Number of Events/Pregnancies at Risk 99/345 74/345 25/271
Cumulative average exposure
  NO2 1.07 (0.87, 1.32) 0.83 (0.57, 1.20) 1.34 (1.13, 1.58)
  O3 0.98 (0.90, 1.06) 0.99 (0.91, 1.08) 0.93 (0.81, 1.06)
  PM2.5 0.80 (0.51, 1.27) 0.86 (0.53, 1.39) 0.62 (0.24, 1.58)
  BC 0.78 (0.52, 1.17) 0.72 (0.45, 1.17) 0.93 (0.56, 1.55)
Exposure in prior week
  NO2 0.98 (0.85, 1.14) 0.89 (0.69, 1.15) 1.05 (0.92, 1.20)
  O3 0.98 (0.91, 1.06) 1.00 (0.92, 1.08) 0.92 (0.76, 1.11)
  PM2.5 0.92 (0.78, 1.08) 0.95 (0.80, 1.14) 0.82 (0.61, 1.10)
  BC 0.90 (0.69, 1.17) 0.89 (0.65, 1.23) 0.92 (0.65, 1.31)
Open in a new tab

Note: BC, black carbon; NO2, nitrogen dioxide; O3, ozone; PM2.5, particulate matter <2.5 μm.

a

Models were adjusted for age, BMI, smoking status, race, education, current employment, and protocol. An IQR increase was 18 ppb for NO2, 12 ppb for O3, 2 μg/m3 for PM2.5, and 0.2 μg/m3 for BC.

To assess the robustness of our findings, we truncated cumulative average NO2 exposure at 20 weeks for all ongoing pregnancies to ensure a comparable length of exposure and the results were similar (HR=1.33 95% CI: 1.18, 1.50 for an IQR increase in NO2 after 30 days) (Supplemental Table 3). Results were slightly stronger when we began follow-up for the cumulative average at the day of embryo transfer (HR=1.40 95% CI 1.14, 1.71 for an IQR increase in NO2 after 30 days) rather than the day of positive hCG; however results were attenuated when we investigated the fixed, two week average NO2 exposure between embryo transfer and hCG (HR=1.19, 95% CI 1.12, 1.25 for an IQR increase in NO2 after 30 days) (Supplemental Table 3). Results were consistent after further adjustment for season of pregnancy, ambient temperature, and the other pollutants (Supplemental Table 4). Similar results were obtained after excluding ectopic pregnancies (n=3), induced abortions (n=3), and stillbirths (n=5), which may have different etiologies than spontaneous abortions. When we stratified on primary risk factors for pregnancy loss (e.g. age, smoking status, and infertility diagnosis), results were slightly stronger for women with a female factor infertility diagnosis although estimated association measures were imprecise (Supplemental Table 5).

Discussion.

In this prospective study of women residing in New England who conceived through the use of assisted reproduction, we found that cumulative average exposure to NO2 throughout pregnancy was associated with higher risk of pregnancy loss occurring 30 days or more following positive hCG (~8 weeks gestation). Earlier losses were not associated with NO2 exposure. Whole pregnancy and past week exposure to O3, PM2.5, and BC were not associated with pregnancy loss. Our findings suggest that chronic exposure to air pollution may be more detrimental than acute exposures and that later pregnancy losses, particularly those occurring after 8 weeks gestation, may be more susceptible to the adverse effects of air pollution exposure during pregnancy.

Our results suggesting a time interaction between NO2 and pregnancy loss are in concordance with findings from a cohort of pregnant women recruited from a health maintenance organization in California (n=4979 women) (Green et al. 2009). While this study likely missed many early pregnancy losses because women were recruited at an average of 8 gestational weeks, they found that higher exposure to traffic within 50 meters of the woman’s residence was more strongly associated with pregnancy losses between enrollment and 10 weeks gestation compared to later losses. Although we did not directly consider traffic as an exposure, spatial-temporal models of residence-based NO2 concentrations consistently show that traffic within 50 m of the residence is a strong predictor of NO2 concentrations (Clougherty et al. 2008; Hochadel et al. 2006). Moreover, the majority of our pregnancy losses which occurred after 8 weeks gestation (when we observed significant associations with NO2) happened prior to 10 completed weeks. Our findings are also supported by a large, multi-site study from the US (n=7403 women) which found that higher chronic exposure to NO2 between embryo transfer and end of pregnancy was associated with a reduced odds of live birth (Legro et al. 2010) and two case-control studies from Iran and China which found that women experiencing a clinical pregnancy loss prior to 14 weeks had higher chronic exposure to NO2 (defined as either during pregnancy or in the first month of pregnancy) compared to women with ongoing pregnancies (Hou et al. 2014; Moridi et al. 2014).

In contrast to our results, the only pre-conception study of air pollution and time to pregnancy loss (n=344 pregnancies), the Longitudinal Investigation of Fertility and the Environment (LIFE) Study, found no association between NO2 and pregnancy loss, regardless of timing (Ha et al. 2018). Whole-pregnancy exposure to O3 (HR: 1.12 per IQR, 95% CI 1.07, 1.17) and PM2.5 (HR: 1.13 per IQR, 95% CI 1.13, 1.24) were positively associated with pregnancy loss but exposure to NO2 only had a small, non-significant association (HR: 1.03 per IQR, 95% CI 0.98, 1.08). Discordant results for NO2 could be due to lower exposure to NO2 in LIFE (median 6.2 vs. 21.8 ppb) or differences in the two populations of women including average age (13% vs. 51% ≥ 35 yrs), underlying fertility status of women (no history of infertility vs. partners of a subfertile couple), and modality of conception (spontaneous vs. ART), all of which may alter the type and frequency of pregnancy loss (Bettio et al. 2008; Gray and Wu 2000; Ozawa et al. 2019).

While our study suggests that chronic exposure to air pollution may be more detrimental than acute exposures, other studies have documented acute effects. For instance, a case-crossover study which identified women presenting with a spontaneous abortion from the University of Utah Emergency Department found an increased risk of pregnancy loss with higher past week NO2 exposure (Leiser et al. 2019). A time-series study, which used information from all live births recorded at Beth Israel Deaconess Medical Center in Boston, MA (2000-2013) and all live births in Tel Aviv District, Israel (2010-2013), found an association between higher weekly NO2 exposure and decreased risk of live birth with the strongest associations occurring between gestational weeks 15-17 (Kioumourtzoglou et al. 2019). Moreover, while we found no association with NO2 and early pregnancy losses, a large ART study from Korea (n=4581 women) found that higher exposure to NO2 between embryo transfer and serum hCG test was positively associated with biochemical pregnancy loss (Choe et al. 2018). Given the limited studies and mixed findings regarding acute vs. chronic air pollution exposure and timing of pregnancy loss, further research is warranted.

Our findings that NO2 exposure during pregnancy was only associated with later losses may lend insight into the potential biological mechanisms at play. As gestational age at loss increases, the frequency of chromosome abnormalities decreases (Soler et al. 2017). This could suggest that environmental exposures such as NO2 are less relevant for embryos genetically incompatible with life. Alternatively, a different window of exposure- such as during gametogenesis in the male and/or female partner- may be more relevant for these early losses. Air pollution exposure is linked to oxidative stress-induced DNA damage (Lettieri Barbato et al. 2010), aberrant placental methylation and mitochondrial DNA content (Clemente et al. 2016; Kingsley et al. 2016), increased systemic and placental inflammation (Bobak 2000; Panasevich et al. 2009), and endothelial dysfunction (Wauters et al. 2013) which, in turn, may increase the risk of miscarriage (Christiansen et al. 2006; Germain et al. 2007; Gupta et al. 2007; Yin et al. 2012).

The limitations of our study are worth noting. Similar to previous studies we lacked personal monitoring of air pollution. We also lacked data on local mobility and daily activities. Thus, we relied on estimates of residence-based ambient air pollution exposure as a proxy of personal exposure which likely resulted in exposure misclassification. However, rather than relying on nearest monitor or regional average concentrations, the spatio-temporal models we used were specific to the woman’s home address which is particularly important for pollutants with small-scale spatial variation such as NO2 (Monn 2001). Although we had intensive monitoring of women throughout early pregnancy to document losses, the diagnosis date is only a proxy for when fetal demise actually occurred. These inaccuracies could be one reason why we found no association between acute exposures, which may be more sensitive to this misclassification, as opposed to chronic exposure. Residual or unmeasured confounding may still be biasing our associations despite our ability to control for many important variables. Our sole focus on women who achieved pregnancy following ART allowed us to routinely document very early pregnancy loss which is challenging to achieve in studies of couples conceiving without medical assistance; however, this also meant that all of our women were part of a subfertile couple, which potentially limits the generalizability of our findings. The majority of our women were also White and highly educated, which is typical of studies focusing on infertility clinic populations, but may limit the applicability of our findings to other race/ethnicities and socioeconomic status. Our incidence of pregnancy loss, however, was virtually identical to that in the LIFE Study, which only included women with spontaneous conceptions and no history of infertility. We also only included women undergoing ART at a single academic fertility clinic in Massachusetts. While this helped limit confounding due to differences in air pollution and clinical protocols across regions and treatment centers, the air pollution concentrations tended to be low. Therefore, it is possible we may have underestimated associations that may be present in other, more highly polluted regions. Strengths of our study included the standardized assessment of pregnancy following embryo transfer, which allowed us to document all early pregnancy losses, and our ability to account for potential bias associated with excluding women who were unable to achieve pregnancy possibly due to high air pollution exposure.

In conclusion, among our prospective cohort of women who achieved pregnancy following ART, higher average exposure to NO2 throughout pregnancy was associated with higher risk of pregnancy loss occurring after 8 weeks gestation or 30 days following a positive pregnancy test but not before. As the majority of previous studies lacked assessment of very early pregnancy losses, additional research is warranted, particularly in urban settings with higher baseline exposure to NO2 to confirm our findings and investigate potential biological mechanisms.

Supplementary Material

1

Highlights.

  • Using a prospective cohort of women who became pregnant using assisted reproduction, we investigated the association between acute and chronic exposure to ambient air pollution and time to pregnancy loss.

  • Higher exposure to NO2 throughout pregnancy was positively associated with pregnancy loss after 30 days, but not in the first 30 days after positive hCG.

  • Entire pregnancy and past week exposures to O3, PM2.5, and BC were not associated with pregnancy loss.

  • In this cohort, pregnancy losses occurring after 8 weeks gestation appeared more susceptible to the detrimental effects of air pollution exposure.

Acknowledgements.

The authors gratefully acknowledge all members of the EARTH study team, specifically Jennifer Ford, Ramace Dadd, the physicians and staff at Massachusetts General Hospital fertility center, and the study participants.

Funding: This work was supported by the National Institute of Health (grant numbers ES009718, ES022955, ES000002, and ES026648 and the United States Environmental Protection Agency (U.S. EPA) (grant numbers RD-834798 and RD-83587201). Its contents are solely the responsibility of the grantee and do not necessarily represent the official views of the U.S. EPA. Further, U.S. EPA does not endorse the purchase of any commercial products or services mentioned in the publication.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Competing Financial Interests: The authors declare they have no actual or potential competing financial interests.

References.

  1. [Anonymous]. 2014. Committee opinion no 611: method for estimating due date. Obstet Gynecol 124(4): 863–866. [DOI] [PubMed] [Google Scholar]
  2. Abu Awad Y, Koutrakis P, Coull BA, Schwartz J. 2017. A spatio-temporal prediction model based on support vector machine regression: Ambient Black Carbon in three New England States. Environ Res 159: 427–434. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Agenor A, Bhattacharya S. 2015. Infertility and miscarriage: common pathways in manifestation and management. Womens Health (Lond) 11(4): 527–541. [DOI] [PubMed] [Google Scholar]
  4. Ammon Avalos L, Galindo C, Li DK. 2012. A systematic review to calculate background miscarriage rates using life table analysis. Birth Defects Res A Clin Mol Teratol 94(6): 417–423. [DOI] [PubMed] [Google Scholar]
  5. Bettio D, Venci A, Levi Setti PE. 2008. Chromosomal abnormalities in miscarriages after different assisted reproduction procedures. Placenta 29 Suppl B: 126–128. [DOI] [PubMed] [Google Scholar]
  6. Bobak M 2000. Outdoor air pollution, low birth weight, and prematurity. Environ Health Perspect 108(2): 173–176. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Choe SA, Jun YB, Lee WS, Yoon TK, Kim SY. 2018. Association between ambient air pollution and pregnancy rate in women who underwent IVF. Hum Reprod 33(6): 1071–1078. [DOI] [PubMed] [Google Scholar]
  8. Christiansen OB, Nielsen HS, Kolte AM. 2006. Inflammation and miscarriage. Semin Fetal Neonatal Med 11(5): 302–308. [DOI] [PubMed] [Google Scholar]
  9. Clemente DB, Casas M, Vilahur N, Begiristain H, Bustamante M, Carsin AE, et al. 2016. Prenatal Ambient Air Pollution, Placental Mitochondrial DNA Content, and Birth Weight in the INMA (Spain) and ENVIRONAGE (Belgium) Birth Cohorts. Environ Health Perspect 124(5): 659–665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Clougherty JE, Wright RJ, Baxter LK, Levy JI. 2008. Land use regression modeling of intraurban residential variability in multiple traffic-related air pollutants. Environ Health 7: 17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cole SR, Hernan MA. 2008. Constructing inverse probability weights for marginal structural models. Am J Epidemiol 168(6): 656–664. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dastoorpoor M, Idani E, Goudarzi G, Khanjani N. 2018. Acute effects of air pollution on spontaneous abortion, premature delivery, and stillbirth in Ahvaz, Iran: a time-series study. Environ Sci Pollut Res Int 25(6): 5447–5458. [DOI] [PubMed] [Google Scholar]
  13. Di Ciaula A, Bilancia M. 2015. Relationships between mild PM10 and ozone urban air levels and spontaneous abortion: clues for primary prevention. Int J Environ Health Res 25(6): 640–655. [DOI] [PubMed] [Google Scholar]
  14. Di Q, Rowland S, Koutrakis P, Schwartz J. 2017. A hybrid model for spatially and temporally resolved ozone exposures in the continental United States. J Air Waste Manag Assoc 67(1): 39–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Enkhmaa D, Warburton N, Javzandulam B, Uyanga J, Khishigsuren Y, Lodoysamba S, et al. 2014. Seasonal ambient air pollution correlates strongly with spontaneous abortion in Mongolia. BMC Pregnancy Childbirth 14: 146. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Germain AM, Romanik MC, Guerra I, Solari S, Reyes MS, Johnson RJ, et al. 2007. Endothelial dysfunction: a link among preeclampsia, recurrent pregnancy loss, and future cardiovascular events? Hypertension 49(1): 90–95. [DOI] [PubMed] [Google Scholar]
  17. Gray RH, Wu LY. 2000. Subfertility and risk of spontaneous abortion. Am J Public Health 90(9): 1452–1454. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Green RS, Malig B, Windham GC, Fenster L, Ostro B, Swan S. 2009. Residential exposure to traffic and spontaneous abortion. Environ Health Perspect 117(12): 1939–1944. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Gupta S, Agarwal A, Banerjee J, Alvarez JG. 2007. The role of oxidative stress in spontaneous abortion and recurrent pregnancy loss: a systematic review. Obstetrical & gynecological survey 62(5): 335–347; quiz 353-334. [DOI] [PubMed] [Google Scholar]
  20. Ha S, Sundaram R, Buck Louis GM, Nobles C, Seeni I, Sherman S, et al. 2018. Ambient air pollution and the risk of pregnancy loss: a prospective cohort study. Fertil Steril 109(1): 148–153. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hochadel M, Heinrich J, Gehring U, Morgenstern V, Kuhlbusch T, Link E, et al. 2006. Predicting long-term average concentrations of traffic-related air pollutants using GIS-based information. Atmos Environ 40(3): 542–553. [Google Scholar]
  22. Hou HY, Wang D, Zou XP, Yang ZH, Li TC, Chen YQ. 2014. Does ambient air pollutants increase the risk of fetal loss? A case-control study. Arch Gynecol Obstet 289(2): 285–291. [DOI] [PubMed] [Google Scholar]
  23. Howe CJ, Cole SR, Lau B, Napravnik S, Eron JJ Jr. 2016. Selection Bias Due to Loss to Follow Up in Cohort Studies. Epidemiology 27(1): 91–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kingsley SL, Eliot MN, Whitsel EA, Huang YT, Kelsey KT, Marsit CJ, et al. 2016. Maternal residential proximity to major roadways, birth weight, and placental DNA methylation. Environ Int 92–93: 43–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kioumourtzoglou MA, Raz R, Wilson A, Fluss R, Nirel R, Broday DM, et al. 2019. Traffic-related Air Pollution and Pregnancy Loss. Epidemiology 30(1): 4–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kloog I, Chudnovsky AA, Just AC, Nordio F, Koutrakis P, Coull BA, et al. 2014. A New Hybrid Spatio-Temporal Model For Estimating Daily Multi-Year PM2.5 Concentrations Across Northeastern USA Using High Resolution Aerosol Optical Depth Data. Atmos Environ (1994) 95: 581–590. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lamichhane DK, Leem JH, Lee JY, Kim HC. 2015. A meta-analysis of exposure to particulate matter and adverse birth outcomes. Environ Health Toxicol 30: e2015011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Lee HJ, Koutrakis P. 2014. Daily ambient NO2 concentration predictions using satellite ozone monitoring instrument NO2 data and land use regression. Environ Sci Technol 48(4): 2305–2311. [DOI] [PubMed] [Google Scholar]
  29. Legro RS, Sauer MV, Mottla GL, Richter KS, Li X, Dodson WC, et al. 2010. Effect of air quality on assisted human reproduction. Hum Reprod 25(5): 1317–1324. [DOI] [PubMed] [Google Scholar]
  30. Leiser CL, Hanson HA, Sawyer K, Steenblik J, Al-Dulaimi R, Madsen T, et al. 2018. Acute effects of air pollutants on spontaneous pregnancy loss: a case-crossover study. Fertil Steril. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Leiser CL, Hanson HA, Sawyer K, Steenblik J, Al-Dulaimi R, Madsen T, et al. 2019. Acute effects of air pollutants on spontaneous pregnancy loss: a case-crossover study. Fertil Steril 111(2): 341–347. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Lettieri Barbato D, Tomei G, Tomei F, Sancini A. 2010. Traffic air pollution and oxidatively generated DNA damage: can urinary 8-oxo-7,8-dihydro-2-deoxiguanosine be considered a good biomarker? A meta-analysis. Biomarkers 15(6): 538–545. [DOI] [PubMed] [Google Scholar]
  33. Monn C 2001. Exposure assessment of air pollutants: a review on spatial heterogeneity and indoor/outdoor/personal exposure to suspended particulate matter, nitrogen dioxide and ozone. Atmos Environ 35(1): 1–32. [Google Scholar]
  34. Moridi M, Ziaei S, Kazemnejad A. 2014. Exposure to ambient air pollutants and spontaneous abortion. J Obstet Gynaecol Res 40(3): 743–748. [DOI] [PubMed] [Google Scholar]
  35. Ozawa N, Ogawa K, Sasaki A, Mitsui M, Wada S, Sago H. 2019. Maternal age, history of miscarriage, and embryonic/fetal size are associated with cytogenetic results of spontaneous early miscarriages. J Assist Reprod Genet 36(4): 749–757. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Panasevich S, Leander K, Rosenlund M, Ljungman P, Bellander T, de Faire U, et al. 2009. Associations of long- and short-term air pollution exposure with markers of inflammation and coagulation in a population sample. Occup Environ Med 66(11): 747–753. [DOI] [PubMed] [Google Scholar]
  37. Perin PM, Maluf M, Czeresnia CE, Januario DA, Saldiva PH. 2010a. Impact of short-term preconceptional exposure to particulate air pollution on treatment outcome in couples undergoing in vitro fertilization and embryo transfer (IVF/ET). J Assist Reprod Genet 27(7): 371–382. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Perin PM, Maluf M, Czeresnia CE, Nicolosi Foltran Januario DA, Nascimento Saldiva PH. 2010b. Effects of exposure to high levels of particulate air pollution during the follicular phase of the conception cycle on pregnancy outcome in couples undergoing in vitro fertilization and embryo transfer. Fertil Steril 93(1): 301–303. [DOI] [PubMed] [Google Scholar]
  39. Parameter-elevation Regressions on Independent Slopes Model (PRISM) Climate Group. Orgeon State University. Available: http://prism.oregonstate.edu/recent/ [accessed March 13th 2019].
  40. Rossen LM, Ahrens KA, Branum AM. 2018. Trends in Risk of Pregnancy Loss Among US Women, 1990-2011. Paediatr Perinat Epidemiol 32(1): 19–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Shah PS, Balkhair T. 2011. Air pollution and birth outcomes: a systematic review. Environ Int 37(2): 498–516. [DOI] [PubMed] [Google Scholar]
  42. Soler A, Morales C, Mademont-Soler I, Margarit E, Borrell A, Borobio V, et al. 2017. Overview of Chromosome Abnormalities in First Trimester Miscarriages: A Series of 1,011 Consecutive Chorionic Villi Sample Karyotypes. Cytogenet Genome Res 152(2): 81–89. [DOI] [PubMed] [Google Scholar]
  43. Wauters A, Dreyfuss C, Pochet S, Hendrick P, Berkenboom G, van de Borne P, et al. 2013. Acute exposure to diesel exhaust impairs nitric oxide-mediated endothelial vasomotor function by increasing endothelial oxidative stress. Hypertension 62(2): 352–358. [DOI] [PubMed] [Google Scholar]
  44. Weng HY, Hsueh YH, Messam LL, Hertz-Picciotto I. 2009. Methods of covariate selection: directed acyclic graphs and the change-in-estimate procedure. Am J Epidemiol 169(10): 1182–1190. [DOI] [PubMed] [Google Scholar]
  45. Wilcox AJ, Weinberg CR, O'Connor JF, Baird DD, Schlatterer JP, Canfield RE, et al. 1988. Incidence of early loss of pregnancy. N Engl J Med 319(4): 189–194. [DOI] [PubMed] [Google Scholar]
  46. Yin LJ, Zhang Y, Lv PP, He WH, Wu YT, Liu AX, et al. 2012. Insufficient maintenance DNA methylation is associated with abnormal embryonic development. BMC Med 10: 26. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1629960-supplement-1.docx (25.7KB, docx)

RESOURCES