Faiz, Ambarina S.a; Rhoads, George G.b; Demissie, Kitawb; Lin, Yongb; Kruse, Lakotac; Rich, David Q.d
Several studies have reported an association between ambient air pollution and adverse pregnancy outcomes including preterm birth,1–6 low birth weight,1,2,4,6–11 intrauterine growth restriction,1,2,12 and fetal death.8,13–15 Some of these studies1–6,9–13,15 evaluated the effects of ambient air pollutants during the entire pregnancy or during a specific trimester. A few have reported adverse pregnancy outcomes associated with short-term increases7,14 in the few days or weeks before the delivery.
In a previous study using the New Jersey fetal death certificate data, our group reported that increases in the mean concentrations of nitrogen dioxide (NO2) and sulfur dioxide (SO2) in the first trimester, carbon monoxide (CO) in the second trimester, and CO and SO2 in the third trimester15 were associated with increased risks of stillbirth. The purpose of the present study was to examine whether short-term increases in ambient air pollutants in the few days before delivery are associated with stillbirth. We hypothesized that a short-term increase in mean ambient air pollutant concentration could trigger stillbirth and that such an effect would be modified by maternal risk factors such as maternal age, maternal race and ethnicity, education, smoking, and prenatal care.16–19
We used a dataset of New Jersey fetal death certificates linked to their corresponding hospital delivery discharge records from 1998 to 2004 (maintained by the Division of Family Health Services, New Jersey Department of Health and Senior Services). For each fetal death, the residential address of the mother at the time of delivery was geocoded including X and Y coordinates. Approximately 90% of New Jersey files were successfully geocoded. We retained data on singleton fetal deaths to mothers residing in New Jersey at the time of stillbirths between 20 and 42 completed weeks of gestation (140–294 days of pregnancy) and weighing at least 500 g. We also retained data on maternal age (years), maternal race/ethnicity, education, prenatal care, smoking status, birth weight of the fetus (grams), and gestational age of the fetus (completed weeks of gestation). This study was approved by the University of Medicine and Dentistry of New Jersey Institutional Review Board and the New Jersey Department of Health and Senior Services Institutional Review Board.
We retrieved hourly concentrations of particulate matter ≤2.5 mm (PM2.5), NO2, SO2, and CO measured by the New Jersey Department of Environmental Protection from the United States Environmental Protection Agency Air Quality System website.20 We used measurements from the five monitoring sites for PM2.5 from 1 January 1999 to 31 December 2004, and nine monitoring sites for NO2, 14 monitoring sites for SO2, and 13 monitoring sites for CO from 1 January 1998 to 31 December 2004. Mean daily concentrations of air pollutants were calculated from hourly measurements if at least 75% of measurements were available in a 24-hour period. Otherwise the value was considered missing. For each stillbirth, we assigned mean pollutant concentrations from the closest pollutant monitoring station to the maternal residence. All stillbirths whose maternal residence was >10 km from a monitoring site for a specific pollutant were excluded from all analyses with that pollutant.
We retrieved hourly temperature and dew-point measurements from weather monitoring sites at the Atlantic City, Newark, Caldwell, Somerset, and Trenton airports from 1 January 1998 to 31 December 2004 and calculated mean apparent temperature as a measure of perceived temperature.21 The mean apparent temperature measurement from the weather monitoring site closest to the maternal residence was used for each stillbirth.
We estimated the relative odds of stillbirth associated with each interquartile range (IQR) increase in the mean pollutant concentration using a time-stratified case-crossover design.22,23 The case period for the study was defined as lag day 2 (ie, 2 days preceding the day of stillbirth). Control periods (3–4 per case) were selected by matching on weekday within the same calendar month. For example, if case period was the second Friday of a calendar month, then the control periods were all other Fridays of that same month. We chose lag day 2 as the case period because the fetus is not always delivered immediately after death; the average time between fetal death and spontaneous or induced delivery has been estimated to be 48 hours in the third trimester.24 We also estimated the relative odds of stillbirth associated with IQR increases in the mean pollutant concentrations 2 through 6 days before delivery of the dead fetus.
We used conditional logistic regression models to estimate the unadjusted relative odds of stillbirth associated with each IQR increase in the mean PM2.5 concentration 2 days before the stillbirth. We estimated the adjusted relative odds of stillbirth associated with each IQR increase in the mean PM2.5 concentration on lag day 2, after including a natural spline (4 df) for the mean apparent temperature on lag day 2. We then replaced the lag day 2 mean PM2.5 concentration and the lag day 2 mean apparent temperature with the mean PM2.5 concentration and mean apparent temperature for lag days 2 and 3. We reran these models with successively longer time intervals to estimate the relative odds of stillbirth associated with PM2.5 concentrations and apparent temperatures (mean of lag days 2 through 4, 2–5, and 2–6). We repeated these analyses for NO2, SO2, and CO. Odds ratios (ORs) and 95% confidence intervals (CIs) were scaled to the IQR of each pollutant for each lag period.
To examine effect modification of the pollutant/stillbirth association by maternal education (less than equal to high school and more than high school), prenatal care (at any time during pregnancy or never), and self-reported smoking (yes vs. no), we added an interaction term (eg, PM2.5 * maternal education) to the model. To examine effect modification of the pollutant/stillbirth association by maternal age (<25 years, 25–34 years, ≥35 years) and race/ethnicity (white non-Hispanic, black non-Hispanic, other non-Hispanic, and Hispanics), we included an interaction term for each level of the variable in the same model. To evaluate the potential confounding effect of other pollutants on the pollutant/stillbirth association, we ran single-pollutant and two-pollutant models on subjects who resided within 10 km of the monitoring sites of each of the two pollutants. For example, to estimate the relative odds of stillbirth associated with each IQR increase in CO concentration and each IQR increase in PM2.5 concentration at the same lag time, we ran a two-pollutant model including both PM2.5 and CO concentrations in the same model. We then ran single-pollutant models with PM2.5 and CO separately. From these models, we compared the relative odds of stillbirth associated with a pollutant from the single-pollutant model to that of the two-pollutant model (with a common sample size).
Although the average time between fetal death and spontaneous or induced delivery has been estimated to be 48 hours in the third trimester,24 the median time has been reported to be 22 to 38 hours for all cases of stillbirth in all trimesters.25 Therefore, we examined whether our pollutant/stillbirth relative odds estimates changed substantially if we used lag day 1 mean pollutant concentration as our case period. We also included lag day 1 in our moving average pollutant concentrations (ie, means of lag days 1–6 vs. means of lag days 2–6). Second, we evaluated whether our relative odds estimates associated with increased CO and NO2 concentrations (pollutants with substantial spatial variability) were robust to our definition of the study population (ie, defined by distance from the maternal residence to pollutant monitoring site). We restricted our analysis to those stillbirths with a maternal residence within 5 km from the CO or NO2 monitoring stations and repeated all analyses. All analyses were performed using SAS v9.2 (SAS Institute, Cary, NC) and R package version 2.13.1 (The R Foundation for Statistical Computing).
More than half of stillbirths occurred in women younger than 25 years or 35 years or older. Black non-Hispanic women contributed the highest proportion (38%) of stillbirths, with about 25% occurring in Hispanics. Women with stillbirth generally had a high school education or less (51%). About 32% initiated their prenatal care after the first trimester, with about 9% having no prenatal care. About 70% of stillbirths occurred before 37 weeks of gestation, 50% before 33 week of gestation. Approximately 70% of stillbirths had a birth weight <2500 g and 2–3% had a birth weight ≥4000 g (Table 1). The distributions of pollutant concentrations for the case and control periods are shown in Table 2.
The cases of stillbirth were evenly distributed in the four seasons for all pollutants, with seasonal variability in the mean concentrations of the pollutants ranging from 12.6 to 17.9 µg/m3 for PM2.5, 10.2 to 12.7 ppb for NO2, 4.2 to 8.3 ppb for SO2, and 7.6 to 9.6 ppm for CO. Apparent temperature showed much larger variability within seasons, ranging from 1°C during winter to 25°C during summer (Supplementary Table, http://links.lww.com/EDE/A679).
The only pollutant with strongly correlated values on successive lag days was CO (r = 0.91 for lag day 2 vs. 3; r = 0.52 for lag day 2 vs. 6). We also assessed correlations between pollutants on lag day 2. NO2 was moderately correlated with SO2 (r = 0.51) and CO (r = 0.59). SO2 was moderately correlated with CO (r = 0.49). The correlations between other pollutants for the mean concentrations on lag day 2 were low (r = 0.31 to 0.37). The mean apparent temperature on lag day 2 was moderately correlated with PM2.5 (r = 0.45) and inversely correlated with NO2, SO2, and CO (r = −0.28 to −0.07) (Table 3).
In unadjusted analyses, we found increased relative odds of stillbirth associated with IQR increases in the mean CO, SO2, and NO2 concentrations on lag day 2, lag days 2 to 3, lag days 2 to 4, and lag days 2 to 5. We also found increased relative odds of stillbirth associated with IQR increases in the mean concentrations of CO and SO2 on lag days 2 to 6, although the increase in relative odds was small for SO2. There was also a small increased relative odds of stillbirth associated with IQR increases in mean PM2.5 concentrations on all lag days (Table 4). After adjusting for mean apparent temperature, effect estimates were little changed (Table 4).
The relative odds of stillbirth associated with each pollutant appeared independent of the other pollutants. The relative odds of stillbirth associated with individual pollutants were similar in two-pollutant and single-pollutant models on the same subjects (Table 5).
Last, we evaluated whether maternal risk factors modified the associations, focusing on lag day 2. There were no clear pattern of increasing or decreasing risk within categories of maternal age, maternal race, and maternal education. The associations of pollutant concentrations with stillbirth were stronger among those with some prenatal care and among nonsmokers (Table 6).
When we defined lag day 1 as the case period, the relative odds of stillbirth were reduced. For example, for SO2 on lag day 1 the OR was 1.04 (95% CI = 0.95–1.13) compared with 1.11 (95% CI = 1.02–1.22) on lag day 2. This pattern of attenuated risk was similar for other pollutants (data not shown).
Our relative odds estimates for CO and NO2 concentrations on lag day 2 were not changed after restricting to only those stillbirths with a maternal residence within 5 km from a monitoring station (CO: OR = 1.21, 95% CI = 1.00–1.47; NO2: OR = 1.05, 95% CI = 0.89–1.24) (Table 4). Patterns were also similar for other moving average NO2 and CO concentrations (data not shown).
We found increased relative odds of stillbirth associated with increases in the mean concentrations of NO2, SO2, CO, and PM2.5 in the immediate few days preceding stillbirth. These risks appeared independent of ambient temperature changes. Non–time varying characteristics of the mother, such as demographic characteristics, maternal health history, socioeconomic status, maternal residence, neighborhood characteristics, etc, could not have confounded these associations because these variables were controlled by the case-crossover design. We did not observe effect modification with maternal age, race/ethnicity, or maternal education. However, the associations of pollutants with stillbirth risk were stronger among the relatively low-risk women with some prenatal care and among nonsmokers.
Our findings of increased stillbirth with increases in NO2, SO2, CO, and PM2.5 concentrations in the immediate few days before delivery are broadly consistent with our previous study, in which we reported 13% to 26% increases in the risk of stillbirth associated with IQR increases in 1st, 2nd, and 3rd trimester mean NO2, SO2, and CO concentrations.15
Our findings are also similar to results of several recent studies that reported acute effects (from 6 weeks to 2 days before birth) between short-term increases in ambient air pollution and preterm birth.7,14,26–28 Only a few studies8,13,14 have examined the association between ambient air pollution and stillbirth, and the findings are inconsistent. Bobak and Leon8 found an increased risk of stillbirth in a study in Czech Republic associated with each 50 µg/ m3 increase in the annual mean concentrations of NO2 (OR = 1.21, 95% CI= 0.89–1.64), but Landgren13 did not find any association between stillbirth and levels of air pollution in Swedish municipalities. However, both these studies were ecological by design, and long-term time trends, season, day of week, temperature, and other potential confounding factors were not adjusted in these studies. Our findings are consistent with the results of a study conducted in Sao Pãulo, Brazil, by Periera et al,14 in which they used a time-series design and reported an association between daily counts of intrauterine mortality and NO2, SO2, and CO concentrations after adjusting for weather and seasons. These associations were observed at short time lags (not greater than 5 days before delivery), which is similar to our findings.
An acute association between short-term increases in ambient air pollution and the risk of stillbirth may suggest that ambient air pollutants can acutely compromise the fetus, leading to stillbirth in the next few days. Our previous and current findings collectively provide evidence for harmful effects of ambient air pollutants on the growing fetus. It may be possible for these pollutants to cross the uteroplacental barrier and trigger hypoxic or immune-mediated injury, causing irreversible damage to the fetus leading to fetal death. However, the biological mechanisms causing fetal demise are not understood for either long-term or short-term effects of these pollutants.
We used a time-stratified case-crossover design, which is the recommended referent selection method for a case-crossover design and controls by design for confounding by time trends (weekday, long-term time trend, and season) and any interaction between them.29–31 There was a strong short-term autocorrelation in our exposure data for CO (autocorrelation in 6-day period) but not for other pollutants. This autocorrelation may have biased our results toward the null, in that it suggests that the exposure of the case period is similar to that of the control periods. As discussed by Janes et al32 and Mittleman,33 bias due to incorrect use of the conditional maximum likelihood was minimal in our study because the time-stratified referent selection strategy divides the time period a priori into fixed strata (ie, months), which allows control/referent periods to be matched to the case period by day of the week within that calendar month, thereby minimizing possible overlap bias.
Although our study had several strengths including a statewide dataset of linked stillbirths and hospital discharge data, there are a few limitations that should be considered. First, we assigned mean daily pollutant concentrations to each pregnancy based on the closest pollutant monitor within 10 km of the maternal residence at birth, regardless of the time spent at other locations, time spent indoors versus outdoors, etc., resulting in exposure error (ie, difference between ambient concentration and true pollutant exposure). However, this exposure error was not likely to be different for each subject’s case and control periods, therefore resulting in underestimates of the relative odds of stillbirth associated with increased pollutant concentration. Second, the true date of fetal death was not known, but it was estimated using the date of delivery recorded on the fetal death certificate. Because this is estimated to be, on average, 48 hours after the fetal death,24 matching air pollution data to the case and control periods based on this delivery date likely resulted in some exposure error as well. However, this misclassification would occur for both case and control periods, likely resulting in bias toward the null.
In summary, we found an increased risk of stillbirth associated with short-term increases in mean concentrations of NO2, SO2, CO, and PM2.5 in the previous a few days. Further studies (with a larger sample size, better measurement of the stillbirth date/time, and improved exposure assessment methods) are needed to confirm these findings and to investigate the biological mechanisms underlying these associations.
1. Dejmek J, Selevan SG, Benes I, Solanský I, Srám RJ. Fetal growth and maternal exposure to particulate matter during pregnancy. Environ Health Perspect. 1999; 107:475–480
2. Bobak M. Outdoor air pollution, low birth weight, and prematurity. Environ Health Perspect. 2000; 108:173–176
3. Ritz B, Yu F, Chapa G, Fruin S. Effect of air pollution on preterm birth among children born in Southern California between 1989 and 1993. Epidemiology. 2000; 11:502–511
4. Wilhelm M, Ritz B. Local variations in CO and particulate air pollution and adverse birth outcomes in Los Angeles County, California, USA. Environ Health Perspect. 2005; 113:1212–1221
5. Ritz B, Wilhelm M, Hoggatt KJ, Ghosh JK. Ambient air pollution and preterm birth in the environment and pregnancy outcomes study at the University of California, Los Angeles. Am J Epidemiol. 2007; 166:1045–1052
6. Maroziene L, Grazuleviciene R. Maternal exposure to low-level air pollution and pregnancy outcomes: a population-based study. Environ Health. 2002; 1:6
7. Sajiv SK, Mendola P, Loomis D, et al. A time series analysis of air pollution and preterm birth in Pennsylvania, 1997–2001. Environ Health Perspect. 2005; 113:602–606
8. Bobak M, Leon DA. Pregnancy outcomes and outdoor air pollution: an ecological study in districts of the Czech Republic 1986-8. Occup Environ Med. 1999; 56:539–543
9. Ritz B, Yu F. The effect of ambient carbon monoxide on low birth weight among children born in southern California between 1989 and 1993. Environ Health Perspect. 1999; 107:17–25
10. Maisonet M, Bush TJ, Correa A, Jaakkola JJ. Relation between ambient air pollution and low birth weight in the Northeastern United States. Environ Health Perspect. 2001; 109:(suppl 3)351–356
11. Chen L, Yang W, Jennison BL, Goodrich A, Omaye ST. Air pollution and birth weight in northern Nevada, 1991-1999. Inhal Toxicol. 2002; 14:141–157
12. Rich DQ, Demissie K, Lu SE, Kamat L, Wartenberg D, Rhoads GG. Ambient air pollutant concentrations during pregnancy and the risk of fetal growth restriction. J Epidemiol Community Health. 2009; 63:488–496
13. Landgren O. Environmental pollution and delivery outcome in southern Sweden: a study with central registries. Acta Paediatr. 1996; 85:1361–1364
14. Periera LAA, Loomis D, Conceicao GMS, et al. Association between air pollution and intrauterine mortality in Sao Paulo, Brazil. Environ Health Perspect. 1998; 106:325–329
15. Faiz AS, Rhoads GG, Demissie K, Kruse L, Lin Y, Rich DQ. Ambient air pollution and the risk of stillbirth. Am J Epidemiol. 2012; 176:308–316
16. Willinger M, Ko CW, Reddy UM. Racial disparities in stillbirth risk across gestation in the United States. Am J Obstet Gynecol. 2009; 201:469.e1–469.e8
17. Fretts RC, Usher RH. Causes of fetal death in women of advanced maternal age. Obstet Gynecol. 1997; 89:40–45
18. Fretts RC. Etiology and prevention of stillbirth. Am J Obstet Gynecol. 2005; 193:1923–1935
19. Faiz AS, Demissie K, Rich DQ, Kruse L, Rhoads GG. Trends and risk factors of stillbirth in New Jersey 1997-2005. J Matern Fetal Neonatal Med. 2012; 25:699–705
21. Zanobetti A, Schwartz J. The effect of particulate air pollution on emergency admissions for myocardial infarction: a multicity case-crossover analysis. Environ Health Perspect. 2005; 113:978–982
22. Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol. 1991; 133:144–153
23. Levy D, Lumley T, Sheppard L, Kaufman J, Checkoway H. Referent selection in case-crossover analyses of acute health effects of air pollution. Epidemiology. 2001; 12:186–192
24. Gardosi J, Mul T, Mongelli M, Fagan D. Analysis of Birthweight and Gestational Age in Antepartum Stillbirth. BJOG. 1998; 105:524–530
25. Genest DR, Williams MA, Greene MF. Estimating the time of death in stillborn fetuses: I. Histologic evaluation of fetal organs; an autopsy study of 150 stillborns. Obstet Gynecol. 1992; 80:575–584
26. Glinianaia SV, Rankin J, Bell R, Pless-Mulloli T, Howel D. Particulate air pollution and fetal health: a systematic review of the epidemiologic evidence. Epidemiology. 2004; 15:36–45
27. Xu X, Ding H, Wang X. Acute effects of total suspended particles and sulfur dioxides on preterm delivery: a community-based cohort study. Arch Environ Health. 1995; 50:407–415
28. Liu S, Krewski D, Shi Y, Chen Y, Burnett RT. Association between gaseous ambient air pollutants and adverse pregnancy outcomes in Vancouver, Canada. Environ Health Perspect. 2003; 111:1773–1778
29. Lumley T, Levy D. Base in the case-crossover design. Implications for studies of air pollution. Environmetrics. 2000; 11:689–704
30. Navidi W. Bidirectional case-crossover designs for exposures with time trends. Biometrics. 1998; 54:596–605
31. Bateson TF, Schwartz J. Control for seasonal variation and time trend in case-crossover studies of acute effects of environmental exposures. Epidemiology. 1999; 10:539–544
32. Janes H, Sheppard L, Lumley T. Case-crossover analyses of air pollution exposure data: referent selection strategies and their implications for bias. Epidemiology. 2005; 16:717–726
33. Mittleman MA. Optimal referent selection strategies in case-crossover studies: a settled issue. Epidemiology. 2005; 16:715–716