Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Oct 1.
Published in final edited form as: Environ Res. 2014 Aug 28;134:198–204. doi: 10.1016/j.envres.2014.08.002

The effects of air pollution on adverse birth outcomes

Sandie Ha 1, Hui Hu 1, Dikea Roussos-Ross 2, Kan Haidong 3, Jeffrey Roth 4, Xiaohui Xu 1
PMCID: PMC4262551  NIHMSID: NIHMS620105  PMID: 25173052

Abstract

Background

Air pollution has been shown to have adverse effects on many health outcomes including cardiorespiratory diseases and cancer. However, evidence on the effects of prenatal exposure is still limited. The purpose of this retrospective cohort study is to evaluate the effects of prenatal exposure to air pollutants including particulate matter with aerodynamic diameter less than 2.5 micrometer (PM2.5) and ozone (O3) on the risk of adverse birth outcomes (ABOs) including term low birth weight (LBW), preterm delivery (PTD) and very PTD (VPTD).

Methods

Singleton births from 2004–2005 in Florida were included in the study (N=423,719). Trimester-specific exposures to O3 and PM2.5 at maternal residence at delivery were estimated using the National Environmental Public Health Tracking Network data, which were interpolated using Hierarchical Bayesian models.

Results

After adjustment for potential confounders such as demographics, medical and lifestyle factors PM2.5 exposures in all trimesters were found to be significantly and positively associated with the risk of all ABOs. Second-trimester exposure had the strongest effects. For an interquartile range (IQR) increase in PM2.5 during the second trimester, the risk of term LBW, PTD and VPTD increased by 3% [95% confidence interval (CI): 1–6%)], 12% (11–14%) and 22% (18–25%), respectively. O3 was also found to be positively associated with PTD and VPTD with the strongest effects over the whole pregnancy period [3% (1–5%) for PTD and 13% (7–19%) for VPTD for each IQR increase]. However, O3 was observed to have protective effects on term LBW. Results were consistent for multi-pollutant models.

Conclusion

PM2.5 has consistent adverse effects on ABOs whereas O3 has inconsistent effects. These findings warrant further investigation.

Keywords: air pollution, low birth weight, preterm delivery, birth outcomes

BACKGROUND

Adverse birth outcomes including preterm delivery (PTD), very preterm delivery (VPTD) and low birth weight (LBW) are associated with higher risk of subsequent morbidity and higher health care expenditure. Specifically, infants with these outcomes are more likely to have subsequent respiratory complications including both respiratory failures shortly after birth, and childhood asthma (Escobar et al., 2006; Sonnenschein-van der Voort et al., 2014). Additionally, they are associated with a higher risk of neurobehavioral problems (Aarnoudse-Moens et al., 2009). According to a nationally representative cross-sectional analysis using the Nationwide Inpatient Sample database from the Healthcare Cost and Utilization Project, hospitalization cost for infants with PTD or LBW in 2001 was 5.8 billion dollars, which represents almost 50% of the costs of all infant hospitalization in the United States (Russell et al., 2007). This analysis also reported that on average, infants with PTD or LBW have longer hospital stays with a mean length of stay of 12.9 days and a cost of approximately $15,100 compared to an uncomplicated birth with an average of 1.9 days in the hospital and a cost of $600 (Russell et al., 2007). Due to the serious impact of these two adverse birth outcomes, efforts towards their prevention remain a critical part of Healthy People 2020 (Heathy People 2020).

In recent years, there has been a growing interest in focusing on environmental determinants of adverse birth outcomes including air pollution. Air pollutants that have commonly been studied in relation to adverse birth outcomes are particulate matter with aerodynamic diameter of less than 2.5 micrometer (PM2.5) and ozone (O3). However, studies have yielded inconsistent results. For example, some studies have found that PM2.5 is positively associated with the risk of PTD (Geer et al., 2012; Hyder et al., 2014; Pereira et al., 2014). Others have found no difference (Fleischer et al., 2014; Rudra et al., 2011; Shah et al., 2011). Similar contradictory results have been found for O3 (Shah et al., 2011; Stieb et al., 2012; Vinikoor-Imler et al., 2013). Due to these inconsistent findings, further investigation of the relationship between these air pollutants and birth outcomes is warranted. Moreover, previous studies mostly relied on air pollution data from fixed monitor sites for exposure assessment, which likely lack spatial coverage as fixed monitors only provide the information about air pollution at the sites where the monitors are located. Thus, there are no air pollution data available at locations without monitors. In previous studies, some houses were located very far from sparsely located air monitors; therefore, relying solely on the closest monitor value may not accurately represent individual air pollution exposure. The interpolated air pollution data from statistical modeling can address this weakness by additionally taking into account meteorological patterns, emission and photochemical properties of pollutants. Therefore, the purpose of this retrospective cohort study is to use a more sophisticated exposure assessment based on Hierarchical Bayesian Modeling to determine the association between prenatal exposure to PM2.5 and O3 and the risk of adverse birth outcomes (ABOs) including term low birth weight (LBW), preterm delivery (PTD), and very preterm delivery (VPTD).

MATERIAL AND METHODS

Participants

Study participants included all singleton live births born in Florida from January 01, 2004 through December 31, 2005 identified from Florida Vital Statistics (FVS) (N=445,028). After excluding births that had addresses outside of Florida (n=4,672), missing address (423), unable to geocode (e.g. only PO Box available) (n=563), missing gestational age (n=937), multiple births (n=13,686), those with birth weight out of range (i.e. less than 500 and more than 5,000 grams) (n=903), and those with gestational age out of range (i.e. less than 140 days and more than 320 days) (n=125), 423,719 births remained for analyses.

Exposure assessment

Air pollution data was obtained from the US Environmental Protection Agency’s (USEPA) Hierarchical Bayesian Prediction Model (HBM) output. The HBM combines PM2.5 and O3 data from the EPA’s Air Quality System (AQS) and the gridded output from the Models-3/Community Multi-scale Air Quality Model (CMAQ), which is based on the National Emission Inventory and meteorological and geographical factors. The methodology for the HBM model is described elsewhere (McMillan, 2010). The HBM model output includes 12×12 km gridded estimates of PM2.5 (daily average) and O3 (daily 8-hr maximum) surfaces. For the purpose of this study, we extracted data for the state of Florida during the period 2003–2005.

To obtain exposure, we geocoded each mother’s residential address at delivery and overlaid this layer with the HBM output layer. Individual exposure during pregnancy was then estimated using daily concentrations in the grid in which the residential address falls. We determined pregnancy period and each trimester period by using gestational age given in the data by FVS. On Florida birth certificates, gestational age in weeks is typically determined by ultrasound measurements. When ultrasound is not available, fundal height—determined by clinical examination—or menstrual history is used to estimate gestational age. Exposures were calculated as daily concentrations averaged over each trimester. First, second and third trimester were defined as the first 13 weeks of gestation, week 14 through 26 and week 27 through birth, respectively.

Outcome assessment

The outcomes of interest were term LBW, PTD and VPTD; all of which are assessed using FVS. Term LBW is defined as a birth that occurred on or after the 37th week of gestation with weight less than 2,500 grams. PTD is defined as a birth that occurred before 37 weeks of gestation. VPTD is defined as a birth that occurred before 32 weeks of gestation.

Covariates

Covariates from this study come from FVS. They included infant’s gender (female or male), maternal age in years (continuous), gestational age in weeks (continuous), maternal education (<high school, high school graduate and/or some college, college graduate, graduate school), maternal race (White, Black, Hispanic, Asian/Pacific Islander, and Others), marital status (married or unmarried), prenatal care (yes or no), pregnancy tobacco use (yes <10/day, yes>10/day, quit and no), pregnancy alcohol drink (yes or no), maternal risk factors (yes or no), maternal infection (yes or no), maternal complications (yes or no), season of conception, urbanicity and year of birth. Season of conception was defined as warm (May through October) or cold (November to April). The presence of maternal risk factors was defined as whether or not the mother had previous ABOs, gestational/chronic diabetes or hypertension, or pre-eclampsia. Maternal infection was defined as whether the mother was diagnosed with an infection at the time of delivery or was treated during pregnancy for an infection. Maternal complications were defined as whether mother needed transfusion, had third or fourth degree perineal laceration, ruptured uterus, unplanned hysterectomy, admission to ICU or unplanned surgery. Since income was not available on the birth certificate, we obtained census block group level median household income information from the 2000 Census for each birth and categorized into quartiles. We also considered unemployment rate at the county level with two levels, high or low, defined as above or below the median.

Statistical analysis

T-tests and chi-square tests were performed to compare continuous and categorical characteristics for participants with and without ABOs. Univariate and multivariate logistic regression models were used to investigate the effects of PM2.5 and O3 on the risk of LBW, PTD, and VPTD. In the univariable models, we obtained the unadjusted effects of air pollutants. In the adjusted models, we selected confounders that have been found or reported to be associated with both the exposure and the outcomes to estimate the effects of air pollutants. Finally, multi-pollutant models were applied to estimate the effects after adjusting for their potential effects on each other. For all analyses, we compared births with a defined outcome with “healthy” births without any of the outcomes in this study. Data analyses were performed using SAS 9.3 (SAS Institute, Cary, NC). Statistical significance was set at p<0.05.

Sensitivity analysis

We also performed a capture area analysis by which we included only births within 5 miles of air O3 and PM2.5 monitor stations. We estimated individual exposure for each trimester using average daily concentrations from the nearest monitor.

RESULTS

Table 1 summarizes the characteristics of study participants by ABO status. Compared to the Controls (20.7%), the percentages of mothers with less than high school education were significantly higher among term LBW (27.8%), PTD(23.7%) and VPTD mothers (25.2%), respectively. Additionally, among the ABO groups, the percentages of mothers who were Black, lived in census block with income in the lowest quartile, unmarried, had no prenatal care, smoked or used alcohol during pregnancy, and had maternal risk factors (e.g. had previous ABOs, gestational/chronic diabetes or hypertension, or pre-eclampsia,) were significantly higher than Controls. Additionally, a higher percentage of term LBW infants were female (59.9% vs. 48.7%). However, the percentages of males were higher for PTD and VPTD. Unemployment status was not associated with any ABOs.

Table 1.

Characteristics of singleton births in Florida 2004–2005 (N=423,719) by adverse birth outcomes.

Characteristics Term LBW n(%) or mean(SD) PTD n(%) or mean(SD) VPTD n(%) or mean(SD) Controls n(%) or mean(SD)
9320 (2.42) p-value* 39082 (9.43) p-value* 5680(1.49) p-value* 375317
Maternal age (years) 26.7 (7.1) <0.001 27.6 (7.0) 0.0019 27.4 (7.3) 0.5857 27.4 (6.4)
Gestational age (weeks) 38.0 (1.0) <0.001 34.0 (2.9) <0.001 27.9 (2.6) <0.001 39.0 (1.1)
Mother education <0.001 <0.001 <.0001
 <High school 2560 (27.8) 9127 (23.7) 1397 (25.2) 77011 (20.7)
 HS grad/some college 4749 (51.6) 19616 (50.9) 2942 (53.1) 183225 (49.3)
 College degree 1504 (16.4) 7880 (20.4) 984 (17.8) 88270 (23.7)
 Graduate school 394 (4.3) 1925 (5.0) 215 (3.9) 23343 (6.3)
Maternal race <0.001 <0.001 <.0001
 White 3344 (36.4) 16479 (42.6) 1864 (33.3) 176354 (47.4)
 Black 2756 (30.0) 9642 (24.9) 1932 (34.5) 62419 (16.8)
 Hispanic 1881 (20.5) 7606 (19.7) 1155 (20.6) 83470 (22.4)
 Asian/PI 324 (3.5) 1026 (2.7) 124 (2.2) 10605 (2.9)
 Others 894 (9.6) 3952 (10.2) 522 (9.3) 39483 (10.6)
Census block group annual income <0.001 <0.001 <0.001
 First quartile (≤$29,643.00) 3026 (32.5) 11066 (28.3) 1892 (33.3) 92338 (24.6)
 Second quartile ($29,643.00–38,095.00] 2401 (25.8) 9748 (24.9) 1461 (25.7) 94213 (25.1)
 Third quartile ($38,095.00–49,457.00] 2152 (23.1) 9608 (24.6) 1320 (23.2) 94303 (25.1)
 Fourth quartile ≥$49,457.00 1741 (18.7) 8660 (22.2) 1007 (17.7) 94463 (25.2)
Urbanicity 0.0006 0.0020 0.0015
 Urban 8160 (87.6) 33958 (86.89) 4986 (87.8) 323992 (86.3)
County unemployment 0.7244 0.0463 0.9073
 High (≥ median (11%)) 6079 (65.2) 25225 (64.5) 3699 (65.12) 244140 (65.1)
Infant sex <0.001 <0.001 0.0128
 Female 5578 (59.9) 18565 (47.5) 2655 (46.7) 182823 (48.7)
Marital status <0.001 <0.001 <.0001
 Married 4408 (47.4) 21180 (54.2) 2631 (46.4) 224450 (59.8)
Prenatal care <0.001 <0.001 <0.001
 Yes 9055 (97.2) 37438 (95.8) 5224 (92.0) 370271 (98.7)
Tobacco use <0.001 <0.001 <.0001
 Yes <10/day 1519 (16.3) 5991 (15.3) 1031 (18.2) 51941 (13.8)
 Yes ≥10/day 1128 (12.1) 3058(7.8) 472 (8.3) 22611 (6.0)
 Quit 190 (2.0) 647 (1.7) 99 (1.7) 5952 (1.6)
 No 6483 (69.6) 29386 (75.2) 4078 (71.8) 294813 (78.6)
Alcohol <0.001 <0.001 <.0001
 Yes 69 (0.7) 192 (0.5) 32 (0.6) 1180 (0.3)
Maternal risk factors <0.001 <0.001 <.0001
 Yes 3611 (38.7) 18488 (47.3) 2836 (49.9) 114943 (30.6)
Maternal infection 0.006 0.7907 0.3025
 Yes 1951 (20.9) 7724 (19.8) 1093 (19.2) 74283 (19.8)
Maternal morbidity 0.0791 <0.001 <.0001
 Yes 1090 (11.7) 5066 (13.0) 960 (16.9) 41721 (11.1)
Season of conception 0.0040 0.8694 0.7374
 Warm 4382 (47.0) 18964 (48.5) 2743 (48.3) 182090 (48.5)
 Cold 4938 (53.0) 20118 (51.5) 2937 (51.7) 193227 (51.5)
Year of birth 0.5500 0.2007 0.1745
 2004 4548 (48.8) 19061(48.8) 2738 (48.2) 184325 (49.1)
 2005 4772 (51.2) 20021 (51.2) 2942 (51.8) 190992 (50.9)
Open in a new tab
*

Denotes p-values for t-tests (for continuous variables) and chi-square tests (for categorical variables) comparing the differences between those with outcomes and those without any.

Table 2 displays summary statistics for the distribution of prenatal exposure to PM2.5 and O3 by trimester and the whole pregnancy period. The mean daily average concentrations of PM2.5 during the first, second and third trimesters were 9.7, 9.9 and 10.2 μg/m3, respectively. Additionally, the mean daily average concentrations of O3 were 37.2, 37.6, and 37.4 ppb for the first, second and third trimester, respectively. There were significant but weak correlations between PM2.5 and O3 exposures during pregnancy periods with Pearson correlation coefficients ranging from 0.10 for the whole pregnancy to 0.39 during the first trimester.

Table 2.

Summary statistics for the distribution of pollutants during pregnancy by trimesters.

Pollutants Summary Statistics
Min P25 P50 P75 Max IQR Mean SD
O3 (ppb)
Trimester 1 20.4 33.1 36.5 41.0 56.2 7.8 37.2 6.0
Trimester 2 20.9 33.3 37.0 41.3 57.3 8.1 37.6 6.1
Trimester 3 18.5 33.1 37.0 41.1 69.2 8.0 37.4 6.1
All pregnancy 22.8 33.9 37.9 41.0 51.3 7.1 37.4 4.1
PM2.5 (μg/m3)
Trimester 1 2.6 8.2 9.6 11.0 26.6 2.6 9.7 2.1
Trimester 2 2.3 8.5 9.8 11.1 25.2 2.6 9.9 2.1
Trimester 3 1.8 8.7 10.0 11.5 44.3 2.8 10.2 2.3
All pregnancy 4.5 8.9 9.9 10.9 22.2 2.0 9.9 1.7
Open in a new tab

Abbreviations: P25: 25th percentile, P50: 50th percentile (median), P75: 75th percentile, IQR: Interquartile Range; SD: standard deviation.

Table 3 shows the unadjusted and adjusted odds ratio (OR) for the associations between prenatal exposure to PM2.5 and O3 and ABOs by trimester and the whole pregnancy period. In the unadjusted models, exposure to PM2.5 in all trimesters and the whole pregnancy significantly increase the unadjusted risk of term LBW. After adjustment for potential confounders, exposure to PM2.5 during the second trimester was significantly associated with a 3.4% higher odds (CI 0.7%–6.1%) of term LBW for each IQR increase in exposure but its effects during other periods were no longer significant (Table 3). On the other hand, in both adjusted and unadjusted analysis, O3 exposure during the third trimester and the whole pregnancy period had an inverse effect on risk of term LBW (Table 3). Each IQR increase in O3 exposure showed approximately 6% decrease in risk of LBW during the third trimester and a 6% decrease for the whole pregnancy period after adjusting for other covariates.

Table 3.

Unadjusted and adjusted odds ratio for the association of air pollutants and adverse birth outcomes for every interquartile (IQR)a increase in exposure.

Exposure period Term LBW PTD VPTD
Unadjusted Adjustedb Unadjusted Adjustedc Unadjusted Adjustedc
O3 (ppb)
Trimester 1 1.002 (0.976,1.029) 0.994 (0.967,1.023) 1.026 (1.012,1.039)* 1.012 (0.998,1.027) 1.022 (0.988,1.057) 1.054 (1.016, 1.093)*
Trimester 2 0.980 (0.954,1.007) 0.978 (0.950,1.007) 1.026 (1.012,1.040)* 1.023 (1.008,1.039)* 1.014 (0.980,1.050) 1.072 (1.033, 1.113)*
Trimester 3 0.954 (0.928,0.980)* 0.940 (0.912,0.969)* 1.003 (0.989,1.017) 0.993 (0.978,1.008) 0.963 (0.924,1.004) 0.984 (0.940, 1.030)
Pregnancy 0.959 (0.926,0.994)* 0.938 (0.900,0.977)* 1.039 (1.020,1.058)* 1.028 (1.007, 1.050)* 1.018 (0.973,1.066) 1.125 (1.067, 1.186)*
PM2.5 (μg/m3)
Trimester 1 1.037 (1.011,1.064)* 1.007 (0.981, 1.034) 1.046 (1.032,1.059)* 1.029 (1.016, 1.043)* 1.090 (1.055,1.126)* 1.063 (1.028, 1.098)*
Trimester 2 1.059 (1.032,1.087)* 1.034 (1.007, 1.061)* 1.139 (1.124,1.154)* 1.123 (1.109, 1.138)* 1.252 (1.214,1.291)* 1.215 (1.177, 1.253)*
Trimester 3 1.035 (1.010,1.061)* 1.005 (0.980, 1.031) 1.039 (1.026,1.052)* 1.026 (1.012, 1.039)* 1.038 (1.000,1.078) 1.010 (0.972, 1.049)
Pregnancy 1.050 (1.024,1.076)* 1.015 (0.989, 1.041) 1.071 (1.057,1.084)* 1.053 (1.040, 1.067)* 1.116 (1.082,1.151)* 1.082 (1.048, 1.117)*
Open in a new tab
a

Interquartile range for O3 in ppb: 7.8 for trimester 1, 8.0 for trimester 2, 8.0 for trimester 3, 7.1 for the entire pregnancy; for PM2.5 in μg/m3: 2.6 for trimester 1, 2.6 for trimester 2, 2.8 for trimester 3, and 2.0 for the entire pregnancy.

b

Adjusted for maternal education, ethnicity, marital status, maternal age, infant gender, prenatal care status, alcohol, smoking, season of conception, census group income, urbanicity, presence or absence of maternal risk factor, infection, PTD status, or co-morbidity.

c

Adjusted for maternal education, ethnicity, marital status, age, infant gender, prenatal care status, alcohol, smoking, census group income, urbanicity, presence or absence of maternal risk factor, LBW status, or co-morbidity.

*

Indicates statistical significance at p<0.05

For PTD, the unadjusted odds ratios indicate that higher exposure to both PM2.5 and O3 during the first trimester, second trimester and the entire pregnancy period were associated with higher risk. After adjustment for potential confounders, results remained consistent. Specifically, for each IQR increase in O3 exposure during the second trimester and entire pregnancy, the risk of PTD increases by 2.3% and 2.8%, respectively. Similar findings were observed for PM2.5, with specifically stronger association for exposure during the second trimester corresponding to a 12.3% increase in risk per IQR increase. The effects of PM2.5 and O3 were similar for VPTD.

Consistent findings were also observed when the models were adjusted for co-pollutant effects (Table 4). In the multi-pollutant models, O3 still has negative associations (protective effect) with term LBW but positive associations with PTD and VPTD, only for full pregnancy. Meanwhile, PM2.5 is still shown to increase risk of all outcomes.

Table 4.

Adjusted odds ratio for the association of air pollutants and adverse birth outcomes for every interquartile (IQR)a increase in exposure after adjustment for co-pollutant effects.

Pollutants Term LBWb PTDc VPTDc
O3 (ppb)
 Trimester 1 0.990 (0.960, 1.021) 1.012 (0.997, 1.028) 1.021 (0.983, 1.061)
 Trimester 2 0.960 (0.931, 0.990) * 0.990 (0.975, 1.005) 0.985 (0.949, 1.024)
 Trimester 3 0.936 (0.907, 0.965) * 1.001 (0.986, 1.016) 0.987 (0.943, 1.033)
 Pregnancy 0.930 (0.892, 0.970) * 1.034 (1.014, 1.055) * 1.075 (1.022, 1.131) *
PM2.5 (μg/m3)
 Trimester 1 1.013 (0.984, 1.042) 1.030 (1.015, 1.045) * 1.053 (1.016, 1.091) *
 Trimester 2 1.047 (1.018, 1.076) * 1.132 (1.117, 1.148) * 1.217 (1.177, 1.257) *
 Trimester 3 1.020 (0.994, 1.047) 1.030 (1.017, 1.044) * 1.020 (0.981, 1.061)
 Pregnancy 1.026 (1.000, 1.053) 1.052(1.038, 1.066) * 1.069 (1.035, 1.105) *
Open in a new tab
a

Interquartile range for O3 in ppb: 7.8 for trimester 1, 8.0 for trimester 2, 8.0 for trimester 3, 7.1 for the entire pregnancy; for PM2.5 in μg/m3: 2.6 for trimester 1, 2.6 for trimester 2, 2.8 for trimester 3, and 2.0 for the entire pregnancy.

b

Adjusted for maternal education, ethnicity, marital status, age, infant gender, prenatal care status, season of conception, alcohol, smoking, census tract income, urbanicity, presence or absence of maternal risk factor, infection, PTD status, or co morbidity.

c

Adjusted for maternal education, ethnicity, marital status, age, infant gender, prenatal care status, alcohol, smoking, census group income, urbanicity, presence or absence of maternal risk factor, LBW status, or comorbidity.

*

Indicates statistical significance at p<0.05

We also performed sensitivity analyses using exposures from the nearest monitor station for all births within 5 miles of monitor stations. In our sensitivity analyses using the same inclusion/exclusion criteria, there were 112,500 and 123,207 singleton live births living within 5 miles from O3 and PM2.5 monitor stations, respectively. Among those living within 5 miles from O3 stations, there were 2,569 cases of term LBW (2.51%), 10,193 cases of PTD (9.06%) and 1,499 cases of VPTD (1.33%). When repeating the same analyses but using closest monitors values as exposures, results remained consistent with main analyses. One exception is the inverse association between O3 and ABOs became insignificant. Specifically, after adjustment for covariates, an IQR increase in O3 exposures during the first, second and third trimester and the entire pregnancy were associated with 0.97 (95% CI: 0.91–1.03), 1.02 (0.96–1.08), 0.96 (0.91–1.02) and 0.97 (95% CI: 0.92–1.03) times the odds of having term LBW, respectively (Table 5).

Table 5.

Adjusted odds ratio for the association between air pollutants and adverse birth outcomes for every interquartile (IQR)a increase in exposure using closest monitor values among births within 5 miles from a monitor station.

Pollutants Term LBWb PTDc VPTDc
n(LBW/total) Adjusted OR n(PTD/total) Adjusted OR n(VPTD/total) Adjusted OR
O3 (ppb)
Trimester 1 2,569/112,500 0.966 (0.911, 1.025) 10,193/112,500 1.047 (1.015, 1.079)* 1,499/112,500 1.123 (1.041, 1.212)*
Trimester 2 2,569/112,500 1.017 (0.961, 1.075) 10,193/112,500 1.056 (1.023, 1.089)* 1,499/112,500 1.121 (1.037, 1.212)*
Trimester 3 2,569/112,500 0.960 (0.907, 1.016) 10,193/112,500 0.985 (0.956, 1.015) 1,499/112,500 0.972 (0.888, 1.063)
Pregnancy 2,569/112,500 0.974 (0.920, 1.031) 10,193/112,500 1.057 (1.023, 1.091)* 1,499/112,500 1.171 (1.079, 1.271)*
PM2.5 (μg/m3)
Trimester 1 2,887/123,207 1.027 (0.977, 1.080) 11,656/123,207 1.060 (1.037, 1.084)* 1,867/123,207 1.101 (1.047, 1.159)*
Trimester 2 2,887/123,207 1.054 (1.005, 1.106)* 11,656/123,207 1.149 (1.125, 1.173)* 1,867/123,207 1.226 (1.169, 1.285)*
Trimester 3 2,887/123,207 1.031 (0.988, 1.075) 11,656/123,207 1.052 (1.031, 1.073)* 1,867/123,207 1.031 (0.975, 1.090)
Pregnancy 2,887/123,207 1.048 (0.989, 1.110) 11,656/123,207 1.110 (1.081, 1.141)* 1,867/123,207 1.146 (1.077, 1.220)*
Open in a new tab
a

Interquartile range for O3 in ppb: 8.9 for trimester 1, 9.1 for trimester 2, 8.5 for trimester 3, 5.9 for the entire pregnancy; for PM2.5 in μg/m3: 2.4 for trimester 1, 2.5 for trimester 2, 2.5 for trimester 3, and 2.5 for the entire pregnancy.

b

Adjusted for maternal education, ethnicity, marital status, age, infant gender, prenatal care status, season of conception, alcohol, smoking, census tract income, urbanicity, presence or absence of maternal risk factor, infection, PTD status, or co morbidity.

c

Adjusted for maternal education, ethnicity, marital status, age, infant gender, prenatal care status, alcohol, smoking, census group income, urbanicity, presence or absence of maternal risk factor, LBW status, or comorbidity.

*

Indicates statistical significance at p<0.05

DISCUSSION

The purpose of this retrospective cohort study was to use a more sophisticated exposure assessment based on Hierarchical Bayesian Modeling to determine the association between prenatal exposure to PM2.5 and O3 and the risk of adverse birth outcomes (ABOs). Our results indicate that higher exposures to PM2.5 increase the risk of all ABOs. While it may have a protective effect on term LBW, O3 exposure has a negative effect on PTD and VPTD.

Our observation that O3 exposures increase the risk of PTD and VPTD are consistent with many previous reports. Other studies from different regions of the world have found that O3 exposures increase the risk of PTD (Hansen et al., 2006; Lee et al., 2013; Olsson et al., 2013). For example, among 120,755 singleton births during 1998 to 2006 in Greater Stockholm, Sweden, Olsson et al. 2013 observed an association between first trimester O3 and preterm birth (OR 1.04, 95% CI 1.01 to 1.08) per 10 μg/m3 increase in O3 (Olsson et al., 2013). This finding is similar to our data with 3% increase in risk during the first trimester. The association between pollution and specifically VPTD has received less attention. Most studies have focused on PTD. Since VPTD infants have higher rates of morbidity and mortality compared to PTD infants, it is also important to investigate the potential association between pollution within this particularly vulnerable group. Our study showed there is a stronger association between pollution and VPTD compared to PTD.

Unlike previous studies, we observed a protective association between O3 and term LBW during the third trimester and the entire pregnancy. This finding is inconsistent with previous studies with some reporting positive associations (Geer et al., 2012; Gray et al., 2014; Morello-Frosch et al., 2010) and others showing no association (Liu et al., 2003). To ensure that our results were not due to chance, we performed a sensitivity analysis for births with the same inclusion criteria to women who resided within 5 miles of monitor stations. Results remained consistent but the protective association between O3 and term LBW became insignificant perhaps due to a smaller sample size. Thus, the observed finding is unlikely to be explained by random error. One possible explanation of the main finding may be due to unselected confounders such as diet, which may account for these results. Other studies have shown that O3 can activate the antioxidant system, decrease tissue hypoxia, and increase the host immunity and characteristics of microcirculation and general health status of the exposed population (Elvis and Ekta, 2011). The protective properties of non-toxic dose exposure to O3 on health outcomes including cardiovascular complications, liver injuries, renal injuries, diabetes and its related complications, and radiation induced toxicity have been documented (Delgado-Roche et al., 2013; Ajamieh et al., 2002; Gul et al., 2012; Oztosun et al., 2012; Al-Dalain et al., 2001; Gultekin et al., 2013). However, we can only speculate that our findings reflect negative association of term LBW with O3 exposure. Finally, our O3 exposure mainly relied on the statistically modeled data or the data from the monitor close to residential area. These methods may result in misclassification, which could possibly account for the observed inverse association. Therefore, more research with personal exposure monitors may be needed to confirm this association.

Our findings regarding the negative effects of PM2.5 on ABOs are consistent with the literature. An extensive systematic review by Shah et al. 2011 found that exposure to PM2.5 increases the risk of both LBW and PTD (Shah et al., 2011). More importantly, our results were also consistent with two similar studies using Florida births by Salihu et al. and Mainolfi et al. (Salihu et al., 2012; Mainolfi et al., 2013). Specifically, using exposure based on Euclidean minimum distance from air pollution monitoring sites, Salihu et al. found that Florida women above the median exposure to particulate matter had a statistically significant 9% increase in risk of LBW, 5% increase in risk of PTD and 13% increase in VPTD (Salihu et al., 2012). Additionally, the fact that we observed the strongest effects for second trimester exposure is consistent with a recent study that found that the effects of criteria pollutants on adverse birth outcomes among 145,445 singleton live births in Hillsborough County, Florida were strongest in the second trimester from similar study periods (Mainolfi et al., 2013). According to the U.S. Census, many characteristics in this county are comparable to those of the state of Florida (US Census, 2013).

Though our results show relatively weak magnitude of association, they have an important clinical significance due to the omnipresent nature of air pollution. Although the exact mechanism between air pollution and adverse birth outcomes has not been well-established, some studies have suggested that air pollutants can invade the body through direct diffusion or active transport. Upon entrance, it may affect the body in several ways. The chief mechanisms underlying the association between air pollution and adverse birth outcomes have been suggested to involve oxidative stress and inflammation, changes in hemodynamic and rheological factors, endocrine disruption, and genetic and epigenetic changes (Ghio et al., 2012; Slama et al., 2008; Brook and Rajagopalan, 2012; Peters et al., 1997; Furuta et al., 2004; Takeda et al., 2004; Tran et al., 1996; Jafarabadi, 2007; Rubes et al., 2005; Somers and Cooper, 2009; Baccarelli et al., 2009; Barthauer, 1990). These changes are believed to ultimately affect maternal-fetal exchange in nutrients and oxygen, which subsequently affect fetal growth.

Our study offers some strength. While many existing studies relied on fixed site monitor stations, our study utilized the predicted PM2.5 and O3 estimates form the HBM model, which combined EPA’s AQS monitor measurements and the gridded output from the CMAQ models. These models predict air pollution concentrations at unknown locations using known samples after accounting for photochemical properties of the pollutants as well as meteorological factors. Consequently, this data has an increased spatial and temporal resolution compared to the monitored data.

Our study also has several limitations. First, the only address available is that at time of birth. However, many women may have lived at a different address during pregnancy. Therefore, we were unable to adjust our analyses for residential mobility of mothers during pregnancy. This assignment of exposure may lead to non-differential misclassification of exposure, which could bias the results towards the null. Additionally, a study assessing exposure misclassification due to residential mobility during pregnancy showed that the majority of pregnant women did not move throughout their pregnancy (Chen et al., 2010). Furthermore, among those who moved, they stayed close enough that their exposure estimates were similar to what they would have been if they had not reported relocation. Therefore, mobility is not likely to have profoundly affected our results.

Additionally, we did not have activities pattern data for these women. Therefore, we could not adjust our analysis for this potential source of misclassification. Women may reside at a certain address and work or have activities in other locations, which may have different exposures. Daily multiple locations may result in exposure misclassification. However, we have no reason to believe that activities pattern would result in differential misclassification. Thus, our result, if biased by activities patterns, is in a conservative direction.

Finally, since the resolution of our air pollution data is 12×12km, we were unable to account for variation at a smaller scale. Furthermore, despite the HBM model’s ability to incorporate more information (e.g. weather, emission) relevant for prediction, there are still limitations associated with its estimates. Specifically, this model relies on CMAQ outputs using monitoring observations. Therefore, the prediction error is likely to be larger in areas farther away from monitors. This error may introduce some bias as the populations living near monitors are potentially different from those living far from monitors (Bravo et al., 2012). However, we performed sensitivity analyses including only births within 5 miles from monitoring stations. The results were consistent suggesting that error associated with distance from air monitors did not differentially affect our results. Furthermore, in the main analyses, we adjusted for characteristics (e.g. urbanicity, unemployment, age, race, education, income), which Bravo et al. suggested could have affected our results. Therefore, our results are not likely confounded by these factors. Nevertheless, we cannot exclude the possibility of residual confounding and other sources of prediction error given that CMAQ model has the tendency to overestimate O3 and underestimate PM2.5 (Bravo et al., 2012).

CONCLUSIONS

We found that PM2.5 exposure increases the risk of adverse birth outcomes including term LBW, PTD and VPTD among singleton births born in Florida from 2004–2005. We also found that O3 increases the risk of PTD and VPTD while it decreases the risk of term LBW. The study adds to the existing evidence on the negative health effects of particulate matter. Furthermore, it supports the importance of minimizing population exposure to air pollution in order to reduce risks of adverse birth outcomes. Studies with personal exposure assessment will be needed to confirm these associations, especially for the inverse association between O3 and term LBW.

Highlights.

  • Prenatal exposure to PM2.5 may increase the risk of term LBW, PTD and VPTD.

  • The effects of PM2.5 on ABOs are strongest during the second trimester.

  • Prenatal O3 exposure may increase the risk of PTD and VPTD.

The protective association between O3 and term LBW needs further investigation.

Acknowledgments

FUNDING

The project described was supported by Grant Number K01ES019177 from the National Institute of Environmental Health Sciences (Xu) and a University of Florida Graduate School Fellowship (Ha). The research was approved by the Institutional Review Boards of the University of Florida and the Florida Department of Health.

The authors wish to thank the Florida Department of Health, Office of Vital Statistics for allowing us to link birth certificate data to EPA air pollution data.

Abbreviations

LBW

Low birth weight

PTD

Preterm delivery

VPTD

Very preterm delivery

PM2.5

Particulate matter with aerodynamics diameter less than 2.5 micrometer

O3

ozone

μg/m3

microgram per cubic meter

ppb

parts per billion

AQS

EPA’s Air Quality System

CMAQ

Models-3/Community Multi-scale Air Quality Model

HBM

Hierarchical Bayesian Prediction Model

Footnotes

The conclusions of this study are solely the responsibility of the authors and do not necessarily represent the official views of the funding agency or the data provider.

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 citable 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.

References

  1. Aarnoudse-Moens CS, et al. Meta-analysis of neurobehavioral outcomes in very preterm and/or very low birth weight children. Pediatrics. 2009;124:717–28. doi: 10.1542/peds.2008-2816. [DOI] [PubMed] [Google Scholar]
  2. Ajamieh H, et al. Similar protective effect of ischaemic and ozone oxidative preconditionings in liver ischaemia/reperfusion injury. Pharmacol Res. 2002;45:333–9. doi: 10.1006/phrs.2002.0952. [DOI] [PubMed] [Google Scholar]
  3. Al-Dalain SM, et al. Ozone treatment reduces markers of oxidative and endothelial damage in an experimental diabetes model in rats. Pharmacol Res. 2001;44:391–6. doi: 10.1006/phrs.2001.0867. [DOI] [PubMed] [Google Scholar]
  4. Baccarelli A, et al. Rapid DNA methylation changes after exposure to traffic particles. Am J Respir Crit Care Med. 2009;179:572–8. doi: 10.1164/rccm.200807-1097OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Barthauer L. A piece of my mind. Secret pain. JAMA. 1990;264:1457. doi: 10.1001/jama.264.11.1457. [DOI] [PubMed] [Google Scholar]
  6. Bravo MA, et al. Comparison of exposure estimation methods for air pollutants: ambient monitoring data and regional air quality simulation. Environ Res. 2012;116:1–10. doi: 10.1016/j.envres.2012.04.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Brook RD, Rajagopalan S. Chronic air pollution exposure and endothelial dysfunction: what you can’t see--can harm you. J Am Coll Cardiol. 2012;60:2167–9. doi: 10.1016/j.jacc.2012.08.974. [DOI] [PubMed] [Google Scholar]
  8. Chen L, et al. Residential mobility during pregnancy and the potential for ambient air pollution exposure misclassification. Environ Res. 2010;110:162–8. doi: 10.1016/j.envres.2009.11.001. [DOI] [PubMed] [Google Scholar]
  9. Delgado-Roche L, et al. Ozone oxidative preconditioning prevents atherosclerosis development in New Zealand White rabbits. J Cardiovasc Pharmacol. 2013;61:160–5. doi: 10.1097/FJC.0b013e31827a820d. [DOI] [PubMed] [Google Scholar]
  10. Elvis AM, Ekta JS. Ozone therapy: A clinical review. J Nat Sci Biol Med. 2011;2:66–70. doi: 10.4103/0976-9668.82319. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Escobar GJ, et al. Short-term outcomes of infants born at 35 and 36 weeks gestation: we need to ask more questions. Semin Perinatol. 2006;30:28–33. doi: 10.1053/j.semperi.2006.01.005. [DOI] [PubMed] [Google Scholar]
  12. Fleischer NL, et al. Outdoor air pollution, preterm birth, and low birth weight: analysis of the world health organization global survey on maternal and perinatal health. Environ Health Perspect. 2014;122:425–30. doi: 10.1289/ehp.1306837. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Furuta C, et al. Estrogenic activities of nitrophenols in diesel exhaust particles. Biol Reprod. 2004;70:1527–33. doi: 10.1095/biolreprod.103.024810. [DOI] [PubMed] [Google Scholar]
  14. Geer LA, et al. Ambient air pollution and term birth weight in Texas from 1998 to 2004. J Air Waste Manag Assoc. 2012;62:1285–95. doi: 10.1080/10962247.2012.707632. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gray SC, et al. Assessing the impact of race, social factors and air pollution on birth outcomes: a population-based study. Environ Health. 2014;13:4. doi: 10.1186/1476-069X-13-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gul H, et al. The protective effects of ozone therapy in a rat model of acetaminophen-induced liver injury. Environ Toxicol Pharmacol. 2012;34:81–6. doi: 10.1016/j.etap.2012.02.006. [DOI] [PubMed] [Google Scholar]
  17. Gultekin FA, et al. Effects of ozone oxidative preconditioning on radiation-induced organ damage in rats. J Radiat Res. 2013;54:36–44. doi: 10.1093/jrr/rrs073. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hansen C, et al. Maternal exposure to low levels of ambient air pollution and preterm birth in Brisbane, Australia. BJOG. 2006;113:935–41. doi: 10.1111/j.1471-0528.2006.01010.x. [DOI] [PubMed] [Google Scholar]
  19. Heathy People 2020, Objectives: Maternal, Infant, and Child Health. Vol. 2014. US Department of Health and Human Services; Washington DC: 2013. [Google Scholar]
  20. Hyder A, et al. PM2.5 exposure and birth outcomes: use of satellite- and monitor-based data. Epidemiology. 2014;25:58–67. doi: 10.1097/EDE.0000000000000027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Jafarabadi M. Episodic air pollution is associated with increased DNA fragmentation in human sperm without other changes in semen quality. Hum Reprod. 2007;22:3263. doi: 10.1093/humrep/dem329. author reply 3264. [DOI] [PubMed] [Google Scholar]
  22. Lee PC, et al. First trimester exposure to ambient air pollution, pregnancy complications and adverse birth outcomes in Allegheny County, PA. Matern Child Health J. 2013;17:545–55. doi: 10.1007/s10995-012-1028-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Liu S, et al. Association between gaseous ambient air pollutants and adverse pregnancy outcomes in Vancouver, Canada. Environ Health Perspect. 2003;111:1773–8. doi: 10.1289/ehp.6251. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Mainolfi MB, et al. Low-level exposure to air pollution and risk of adverse birth outcomes in Hillsborough County, Florida. J Occup Environ Med. 2013;55:490–4. doi: 10.1097/JOM.0b013e31828df013. [DOI] [PubMed] [Google Scholar]
  25. McMillan N, Holland DM, Morara M, Feng J. Combining numerical model output and particulate data using Bayesian space–time modeling. Environmetrics. 2010;21:48–65. [Google Scholar]
  26. Morello-Frosch R, et al. Ambient air pollution exposure and full-term birth weight in California. Environ Health. 2010;9:44. doi: 10.1186/1476-069X-9-44. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Olsson D, et al. Air pollution exposure in early pregnancy and adverse pregnancy outcomes: a register-based cohort study. BMJ Open. 2013;3 doi: 10.1136/bmjopen-2012-001955. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Oztosun M, et al. The effects of medical ozone therapy on renal ischemia/reperfusion injury. Ren Fail. 2012;34:921–5. doi: 10.3109/0886022X.2012.692752. [DOI] [PubMed] [Google Scholar]
  29. Pereira G, et al. Fine particulate matter and risk of preterm birth in Connecticut in 2000–2006: a longitudinal study. Am J Epidemiol. 2014;179:67–74. doi: 10.1093/aje/kwt216. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Peters A, et al. Increased plasma viscosity during an air pollution episode: a link to mortality? Lancet. 1997;349:1582–7. doi: 10.1016/S0140-6736(97)01211-7. [DOI] [PubMed] [Google Scholar]
  31. Rubes J, et al. Episodic air pollution is associated with increased DNA fragmentation in human sperm without other changes in semen quality. Hum Reprod. 2005;20:2776–83. doi: 10.1093/humrep/dei122. [DOI] [PubMed] [Google Scholar]
  32. Rudra CB, et al. Ambient carbon monoxide and fine particulate matter in relation to preeclampsia and preterm delivery in western Washington State. Environ Health Perspect. 2011;119:886–92. doi: 10.1289/ehp.1002947. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Russell RB, et al. Cost of hospitalization for preterm and low birth weight infants in the United States. Pediatrics. 2007;120:e1–9. doi: 10.1542/peds.2006-2386. [DOI] [PubMed] [Google Scholar]
  34. Salihu HM, et al. Particulate pollutants and racial/ethnic disparity in feto-infant morbidity outcomes. Matern Child Health J. 2012;16:1679–87. doi: 10.1007/s10995-011-0868-8. [DOI] [PubMed] [Google Scholar]
  35. Shah PS, et al. Air pollution and birth outcomes: a systematic review. Environ Int. 2011;37:498–516. doi: 10.1016/j.envint.2010.10.009. [DOI] [PubMed] [Google Scholar]
  36. Somers CM, Cooper DN. Air pollution and mutations in the germline: are humans at risk? Hum Genet. 2009;125:119–30. doi: 10.1007/s00439-008-0613-6. [DOI] [PubMed] [Google Scholar]
  37. Sonnenschein-van der Voort AM, et al. Preterm birth, infant weight gain, and childhood asthma risk: A meta-analysis of 147,000 European children. J Allergy Clin Immunol. 2014 doi: 10.1016/j.jaci.2013.12.1082. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Stieb DM, et al. Ambient air pollution, birth weight and preterm birth: a systematic review and meta-analysis. Environ Res. 2012;117:100–11. doi: 10.1016/j.envres.2012.05.007. [DOI] [PubMed] [Google Scholar]
  39. Takeda K, et al. Endocrine-disrupting activity of chemicals in diesel exhaust and diesel exhaust particles. Environ Sci. 2004;11:33–45. [PubMed] [Google Scholar]
  40. Tran DQ, et al. The anti-estrogenic activity of selected polynuclear aromatic hydrocarbons in yeast expressing human estrogen receptor. Biochem Biophys Res Commun. 1996;229:101–8. doi: 10.1006/bbrc.1996.1764. [DOI] [PubMed] [Google Scholar]
  41. US Census. Hillsborough County, Florida. Vol. 2014. US Census Bureau; Washington DC: 2013. [Google Scholar]
  42. Vinikoor-Imler LC, et al. Early prenatal exposure to air pollution and its associations with birth defects in a state-wide birth cohort from North Carolina. Birth Defects Res A Clin Mol Teratol. 2013;97:696–701. doi: 10.1002/bdra.23159. [DOI] [PubMed] [Google Scholar]

RESOURCES