Abstract
Prenatal exposure to ambient air pollution has been associated with preterm birth in several studies. Associations between air pollution and gestational or pre-existing diabetes have been hypothesized but are not well established. We examined the association between air pollution exposure in pregnancy and gestational diabetes and whether the association between air pollution and preterm birth is modified by diabetes (gestational or pre-existing) in a highly polluted area of California.
Birth certificates and hospital discharge data from all singleton births from 2000-2006 to women living in four counties in the San Joaquin Valley of California were linked to criteria air pollution and traffic density measurements at the geocoded maternal residence. Air pollutants were dichotomized at the highest quartile and compared to the lower three quartiles.
Logistic regression models were adjusted for maternal race-ethnicity, age, education, payment of birth expenses, and prenatal care. There were consistent inverse associations between exposure to air pollution during the first two trimesters and gestational diabetes (statistically significant odds ratios (OR) less than 1). When stratified by any diabetes (gestational or pre-existing), associations between air pollution exposure during pregnancy and categories of preterm birth (20-27, 28-31, 32-33, 34-36 weeks) were generally similar with few exceptions of exposures to carbon monoxide (CO) and particulate matter <2.5 microns (PM2.5). Those with diabetes and exposure higher levels of CO (in first trimester or entire pregnancy) or PM2.5 (in first trimester) had higher risk of extremely preterm birth (20-27 weeks) compared with those without diabetes.
The associations between traffic-related air pollution and gestational diabetes were in the unexpected (“protective”) direction. Among those with any diabetes, associations were stronger between CO and PM2.5 and extremely preterm birth.
Keywords: air pollution, preterm birth, pregnancy, diabetes
Introduction
Prenatal exposure to ambient air pollution has been associated with preterm birth in several studies [1–4]. In a previous investigation of the current cohort, preterm birth was associated with increased exposure to particulate matter <10 (PM10) and 2.5 microns (PM2.5) during pregnancy. The associations were strongest with exposure in the second trimester, particularly for early preterm births (<28 weeks gestation) [5].
It has been hypothesized that adult exposure to air pollution may be associated with Type 2 diabetes and both the experimental and epidemiological evidence in support of this hypothesis are robust [6, 7]. Long-term air pollution exposure may both decrease insulin-dependent glucose uptake leading to insulin resistance and impair β-cell function resulting in reduced insulin secretion [8]. Upstream mechanistic pathways linking air pollution exposure to insulin resistance and β-cell dysfunction that have been suggested by experimental animal data include oxidative stress and systemic inflammation [7]. A systematic review and meta-analysis of ambient air pollution in adults and diabetes studies in Europe and North America observed an 8-10% increase in odds of diabetes for a 10 μg/m3 increase in PM2.5 or nitrogen dioxide (NO2) [9].
Several studies have investigated the associations between air pollution and diabetes in pregnancy. One study in Taiwan found associations between PM2.5 and oral glucose tolerance tests, a screening test for gestational diabetes, during pregnancy [10, 11]. Another study found preconception and early pregnancy nitrogen oxides (NOX) exposure were associated with gestational diabetes [12, 13]. Exposure to PM2.5 and ozone (O3) during pregnancy was associated with gestational diabetes in a study in Florida. A prospective cohort in Boston, Massachusetts found an association between traffic-related air pollution and impaired glucose tolerance, but not gestational diabetes [14]. In a retrospective study also in Massachusetts, no consistent association between air pollution and gestational diabetes was observed except when stratified by maternal age <20 years, where they found a 1.36 higher odds of gestational diabetes for each IQR increase of second trimester PM2.5 [15].
It is plausible that the association between air pollution and preterm birth may be stronger among those with gestational or pre-existing diabetes, given their common associations. There have been a few studies that have examined diabetes (either gestational or pre-existing) as a potential effect modifier of the association between air pollution and preterm birth. A study in Ontario, Canada found associations between PM2.5 and NO2 and preterm birth among those with pre-existing diabetes [16, 17]. In a study of over 1 million births in Taiwan, associations between O3 and preterm birth were stronger among those with gestational diabetes [18, 19]. We are also interested in the possibility that diabetes is on the pathway between air pollution and preterm birth – i.e., that diabetes mediates the air pollution – preterm birth effect.
We examine the association between air pollution exposure in pregnancy and gestational diabetes and whether the association between air pollution and preterm birth is modified and/or mediated by diabetes in a highly polluted area of California.
Methods
Study Population
Birth certificates from all 2000-2006 births to women living in the four most populated counties in the San Joaquin Valley of California (Fresno, Kern, Stanislaus and San Joaquin) were obtained from the California Department of Health. The four study counties included 329,650 births in 2000-2006. Exclusions were multiple births (n=8,373), those missing file numbers (n=262), those with gestational age missing or <20 weeks or >42 weeks (n=45,726), and those with birth weight missing or <500g or >5000g (n=762). Completeness of pollutant assignments was 80% for CO, 94% for NO2, 93% for PM10, 93% for PM2.5, and 96% for traffic density. The study population included 262,182 births with measurements for at least one of these pollutants. Furthermore, we linked the birth records with Office of Statewide Health and Planning (OSHPD) maternal and infant hospital discharge data, with 98.61% of successfully linked (n=258,522). We also removed 6317 births with any missing adjusted covariates. Therefore, we included 252,205 births for further analysis.
Preterm birth was defined by gestational age at birth as determined from the last menstrual period on the birth certificate. The maternal residence at birth street address locations obtained from birth certificates were geocoded to X and Y coordinates with ArcGIS software (ESRI, Redlands, California). Addresses were corrected with ZP4 software (Semaphore Corporation, Aptos, California).
Ambient air quality data were acquired from U.S. Environmental Protection Agency’s Air Quality System database (www.epa.gov/ttn/airs/airsaqs). Daily metrics of the following pollutants were calculated: carbon monoxide (CO), nitrogen dioxide (NO2), particulate matter ≤ than 10 μm (PM10), and PM ≤ than 2.5 μm (PM2.5). These data were used to create averages for each trimester of pregnancy, entire pregnancy average, and the last 6 weeks of pregnancy. The station-specific daily air quality data were spatially interpolated using inverse distance-squared weighting [5]. Data from up to four air quality measurement stations were included in each interpolation. Traffic density was calculated from distance-decayed annual average daily traffic volumes within a 300m radius of geocoded maternal residences [20]. Further details on exposure assessment were published previously [5, 21].
Variables from birth certificates included in analyses were: maternal age (<20, 20-24, 25-29, 30- 34, >35 years), maternal race (White, Hispanic, African-American, Asian, other), maternal education (no high school, some high school, some college, bachelors or other degree), parity (0, >1), prenatal care (initiated in first trimester), Medi-Cal (Medicaid) or other government program payment of birth costs, infant sex, year (2000-2006), season of conception, and maternal county of residence (Fresno, Kern, Stanislaus, San Joaquin).
This research was approved by institutional review boards from the University of California, Berkeley, Stanford University, and the California State Committee for the Protection of Human Subjects.
Statistical Analysis
The pollutants for each exposure period were dichotomized at the highest quartile and compared to the lower three quartiles. A sensitivity analysis was also performed comparing the highest to the lowest quartile of each exposure. First, second, and third pregnancy trimesters were defined as gestational weeks 1-13, 14-26, and 27 to birth, respectively. Additionally, we calculated metrics for the last 6 weeks of pregnancy (birth minus 42 days). Exposure periods of the term births were truncated to match the same period as the comparison period-length of the preterm births and the last 6-week exposures were matched on gestational age between preterm and term births.
For the first main analysis, the outcome was diagnosis of gestational diabetes (ICD-9 code: 648.8), obtained from OSHPD - the hospital discharge records of the mother. Those with pre-existing diabetes were excluded.
In the second main analysis, preterm birth categories (20-27, 28-31, 32-33, 34-36 weeks) were compared to term births (37-42 weeks) and diabetes (either gestational or pre-existing diabetes; ICD-9 code: 648.0, 250) was considered as an effect modifier. We chose to combine these two conditions because we hypothesize that the potential susceptibility to air pollution among the two groups may be shared. Although age may be a factor (and is adjusted for in the model), women with gestational diabetes are at risk for diabetes later in life. Additionally, those with diabetes are no longer at risk of developing gestational diabetes, and our aim was to include them as a potentially susceptible group. We performed a stratified analysis to explore the association between the pollutant and preterm birth by diabetes and no diabetes. Secondly, we created an interaction term (exposure × diabetes) to add into the model and used Wald’s method to assess the multiplicative interaction.
Models for both analyses were adjusted for the covariates maternal race, age, education, payment of birth expenses/insurance type, and prenatal care. We performed a sensitivity analysis with additional adjustment for season of conception and limited preterm births to those that were spontaneous (preterm labor or premature rupture of membranes) as opposed to medically indicated. Additionally, we removed women from the analysis who had been diagnosed by either pre-existing or pregnancy-induced hypertension, or pre-eclampsia to disentangle any potential association attributable to hypertension.
To evaluate the question of whether diabetes mediates the relationship between PM2.5 and extreme preterm birth (20-27 weeks gestation), we employed the four step method of Baron and Kenny [22]. It included the following regressions: (1) preterm birth ~ (B) PM2.5 + covariates; (2) diabetes ~ PM2.5 + covariates; (3) preterm birth ~ diabetes + covariates; (4) preterm birth ~ (B1) PM2.5 + (B2) diabetes + covariates) and provides a calculation of an indirect effect by either B-B1 [23] or B*B2 [24].
All analyses were performed with SAS 9.4 (Cary, NC).
Results
The study population was majority Hispanic, followed by white, non-Hispanic (Table 1). More than half had birth expenses paid by Medi-Cal (public insurance) and 5% were diagnosed with gestational diabetes. The distribution of the pollutants (CO, NO2, PM10 and PM2.5) and traffic density by exposure periods (Table 2). The medians did not change considerably across pollutants and the distribution was wider for the shorter exposure periods (i.e., trimester averages compared to entire pregnancy). Air pollution was dichotomized at the highest quartile cut off: 0.60 ppm for CO, 19.47 ppb for NO2, 42.65 μg/m3 for PM10, 20.72 μg/m3 for PM2.5, and 45.85 for traffic density. The correlations between each of the pollutants and exposure periods are presented in the Appendix.
Table 1.
Distribution of covariates by gestational age in births in the four most populous counties in San Joaquin Valley, California, 2000-2006.
| Covariate | Gestational age in weeks (%)* |
Total | ||||
|---|---|---|---|---|---|---|
| 37-42 n=223,417 |
34-36 n=21,225 |
32-33 n=3702 |
28-31 n=2550 |
20-27 n=1311 |
N=252,205 | |
| Spontaneous preterm birth † | 35.8 | 47.9 | 59.1 | 69.3 | 41.0 | |
| Maternal age (years) | ||||||
| <20 | 13.3 | 15.2 | 16.9 | 20.3 | 20.8 | 13.6 |
| 20-24 | 28.9 | 28.7 | 28.5 | 28.3 | 26.4 | 28.9 |
| 25-29 | 27.7 | 25.6 | 23.3 | 22.9 | 21.4 | 27.3 |
| 30-34 | 19.4 | 18.2 | 18.1 | 15.3 | 18.7 | 19.2 |
| >35 | 10.8 | 12.3 | 13.2 | 13.2 | 12.7 | 11.0 |
| Race/ethnicity | ||||||
| White, non-Hispanic | 30.6 | 26.0 | 23.2 | 24.1 | 23.0 | 30.0 |
| Asian | 7.4 | 8.7 | 8.3 | 9.4 | 7.9 | 7.6 |
| African-American | 4.9 | 6.7 | 8.0 | 9.1 | 9.2 | 5.1 |
| Hispanic | 55.7 | 57.3 | 59.2 | 56.1 | 58.4 | 55.9 |
| Other | 1.3 | 1.3 | 1.3 | 1.3 | 1.5 | 1.31.5 |
| Education | ||||||
| <High school | 12.1 | 12.5 | 13.9 | 11.7 | 11.1 | 12.1 |
| High school | 53.2 | 57.6 | 59.2 | 62.8 | 60.8 | 53.8 |
| Some college | 21.5 | 19.9 | 18.6 | 17.7 | 19.8 | 21.3 |
| College degree | 13.2 | 10.0 | 8.3 | 7.8 | 8.4 | 12.8 |
| Medi-Cal | 53.4 | 60.2 | 64.5 | 64.0 | 60.2 | 54.3 |
| Prenatal care in 1st | 82.1 | 78.3 | 73.9 | 68.4 | 67.3 | 81.4 |
| trimester | ||||||
| Low SES | 17.3 | 20.2 | 23.5 | 22.6 | 21.5 | 17.7 |
| Diabetes diagnoses | ||||||
| Gestational diabetes | 4.9 | 6.2 | 6.1 | 5.3 | 3.1 | 5.0 |
| Pre-existing diabetes | 0.7 | 1.4 | 1.6 | 1.5 | 1.1 | 0.8 |
| Hypertension diagnoses | ||||||
| Pregnancy-induced | 3.9 | 8.5 | 12.3 | 11.8 | 6.4 | 4.5 |
| Pre-existing | 1.0 | 1.9 | 3.1 | 3.0 | 2.5 | 1.1 |
| Cesarean section | 25.4 | 29.1 | 35.6 | 43.9 | 41.0 | 26.1 |
| First born | 35.3 | 33.0 | 34.3 | 35.8 | 41.0 | 35.2 |
| County | ||||||
| Fresno | 32.7 | 35.2 | 35.0 | 33.1 | 37.0 | 32.9 |
| Stanislaus | 23.2 | 24.8 | 26.7 | 25.2 | 23.8 | 23.4 |
| Kern | 25.5 | 23.0 | 21.1 | 23.1 | 22.1 | 25.1 |
| San Joaquin | 18.7 | 17.0 | 17.2 | 18.6 | 17.1 | 18.5 |
| Year | ||||||
| 2000 | 13.2 | 12.5 | 12.5 | 12.8 | 15.7 | 13.1 |
| 2001 | 13.4 | 12.6 | 11.8 | 13.2 | 12.5 | 13.3 |
| 2002 | 13.7 | 13.0 | 13.3 | 12.3 | 12.4 | 13.6 |
| 2003 | 13.9 | 13.9 | 13.8 | 13.3 | 13.4 | 13.9 |
| 2004 | 14.4 | 14.5 | 14.7 | 14.2 | 15.0 | 14.4 |
| 2005 | 15.1 | 16.0 | 15.3 | 16.8 | 15.0 | 15.2 |
| 2006 | 16.5 | 17.6 | 18.6 | 17.4 | 16.1 | 16.6 |
Percentages may not equal 100 owing to rounding.
Spontaneous preterm birth identified as those births <37 weeks with preterm premature rupture of membranes (ICD-9-CM code 658.1 or birth certificate complication of labor/delivery code 10), those with premature labor (ICD-9-CM code 644), or the use of tocolytics (birth certificate complication/procedure of pregnancy code 28).
Percentage of total was calculated among preterm, i.e. 11791/28788*100=41.0%
Table 2.
Distribution of pollutant averages across each exposure period - median (interquartile range)
| CO (ppm) | NO2 (ppb) | PM10 (μg/m3) |
PM2.5 (μg/m3) |
Traffic Density |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Entire Pregnancy | 0.49 | 0.18 | 17.26 | 4.14 | 35.93 | 11.55 | 17.05 | 6.39 | 16.51 | 45.23 |
| 1st trimester | 0.49 | 0.28 | 17.42 | 6.43 | 34.86 | 16.37 | 15.18 | 13.45 | ||
| 2nd trimester | 0.47 | 0.29 | 17.01 | 6.63 | 34.75 | 16.59 | 14.41 | 12.77 | ||
| 3rd trimester | 0.45 | 0.29 | 16.88 | 6.68 | 34.81 | 17.47 | 13.91 | 12.4 | ||
| Last 6 weeks | 0.45 | 0.29 | 16.63 | 6.94 | 34.05 | 17.38 | 13.34 | 12.2 | ||
Air pollution and gestational diabetes
There was a consistent inverse association between exposure to traffic-related air pollution during pregnancy and gestational diabetes, excluding those with pre-existing diabetes (Table 3). In general, there was approximately a 5-10% decrease in risk of gestational diabetes comparing above to below highest quartile of exposure to CO, NO2 and PM2.5. Pollutant exposures were limited to the first two trimesters, prior to diabetes screening and diagnosis, which generally occurs in weeks 24-28. When comparing the highest to the lowest quartile of exposure, the results were similar and in some cases the associations were slightly larger (data not shown).
Table 3.
Association between air pollution and gestational diabetes comparing above versus below highest quartile of exposure (N=252,205)
| Pollutant | Exposure Period | Odds Ratio (95% Confidence Interval) |
|
|---|---|---|---|
| Unadjusted | Adjusted* | ||
| CO | 1st Trimester | 0.87 (0.83-0.91) | 0.91 (0.87-0.96) |
| CO | 2nd Trimester | 0.90 (0.86-0.94) | 0.92 (0.88-0.96) |
| NO2 | 1st Trimester | 0.87 (0.83-0.91) | 0.91 (0.87-0.95) |
| NO2 | 2nd Trimester | 0.87 (0.83-0.91) | 0.94 (0.90-0.98) |
| PM10 | 1st Trimester | 0.90 (0.86-0.93) | 0.94 (0.90-0.97) |
| PM10 | 2nd Trimester | 0.87 (0.84-0.90) | 0.92 (0.89-0.96) |
| PM2.5 | 1st Trimester | 0.98 (0.94-1.02) | 1.01 (0.97-1.06) |
| PM2.5 | 2nd Trimester | 0.94 (0.90-0.98) | 0.96 (0.92-1.00) |
| Traffic Density | Entire pregnancy | 0.97 (0.93-1.01) | 0.99 (0.95-1.03) |
Adjusted for maternal race, age, education, payment of birth expenses/insurance type, prenatal care (and for pollutants, the alternative trimester of exposure)
Highest quartile cut offs: 0.60 ppm for CO, 19.47 ppb for NO2, 42.65 μg/m3 for PM10, 20.72 μg/m3 for PM2.5, and 45.85 for traffic density
Air pollution and preterm birth among those with pre-existing and gestational diabetes
When stratified by diabetes status (preexisting or gestational versus neither), the associations between air pollution and preterm birth were generally very similar across strata with few notable exceptions (Table 4). Five estimates were statistically different with cut-off of p<0.05 for the Mantel Hanzel chi-square. The association between PM2.5 during the first trimester of pregnancy and very early preterm birth (20-27 weeks) was considerably stronger among those with diabetes (aOR=2.15; 95% CI: 1.24, 3.73) compared with those without diabetes (aOR=1.07; 95% CI: 0.95, 1.20). A similar pattern was observed for CO during the entire pregnancy and first trimester and very early preterm birth. The association between PM10 and late preterm birth (34-36 weeks) was observed among those without diabetes (aOR= 1.11; 95% CI: 1.07, 1.14) and not among those with diabetes. Mantel Hanzel chi-square p-values were noted for those with p<0.02 (Table 4).
Table 4.
Association between air pollution and preterm birth, stratified by diabetes (N=N=252,205)
| Gestational Age | Pollutant Exposure | Exposure period | Adjusted* Odds Ratio (95% Confidence Intervals) |
P-value <0.2‡ | |
|---|---|---|---|---|---|
| With Diabetes† (N=14,493) | Without Diabetes (N=237,712) | ||||
| Entire pregnancy | 1.12(0.98-1.28) | 1.12 (1.08-1.16) | |||
| 1st trimester | 1.06 (0.93-1.19) | 1.05 (1.01-1.08) | |||
| CO | 2nd trimester | 0.98 (0.87-1.12) | 1.02 (0.98-1.05) | ||
| 3rd trimester | 0.88 (0.77-1.00) | 0.98 (0.94-1.01) | 0.123 | ||
| Last 6 weeks | 0.84 (0.74-0.95) | 0.96 (0.93-1.00) | 0.042 | ||
| Entire pregnancy | 0.99 (0.87-1.12) | 1.07 (1.04-1.11) | |||
| 34-36 weeks | NO2 | 1st trimester | 0.96 (0.86-1.08) | 1.03 (1.00-1.06) | |
| 2nd trimester | 0.97 (0.86-1.09) | 1.03 (0.99-1.06) | |||
| 3rd trimester | 0.92 (0.82-1.04) | 0.99 (0.96-1.02) | |||
| Last 6 weeks | 0.90 (0.80-1.02) | 0.99 (0.96-1.02) | 0.154 | ||
| Entire pregnancy | 1.01 (0.90-1.15) | 1.09 (1.05-1.12) | |||
| PM10 | 1st trimester | 0.96 (0.85-1.08) | 1.11 (1.07-1.14) | 0.029 | |
| 2nd trimester | 1.01 (0.90-1.14) | 1.09 (1.05-1.12) | |||
| 3rd trimester | 1.00 (0.89-1.12) | 1.04 (1.01-1.07) | |||
| Last 6 weeks | 0.93 (0.82-1.04) | 0.98 (0.95-1.01) | |||
| Entire pregnancy | 1.19 (1.05-1.34) | 1.23 (1.19-1.27) | |||
| PM2.5 | 1st trimester | 1.04 (0.93-1.16) | 1.05 (1.01-1.08) | ||
| 2nd trimester | 1.05 (0.941.18) | 1.05 (1.011.08) | |||
| 3rd trimester | 0.86 (0.76-0.97) | 0.94 (0.91-0.97) | |||
| Last 6 weeks | 0.89 (0.79-1.00) | 0.99 (0.96-1.02) | 0.114 | ||
| Traffic Density | Entire pregnancy | 0.96 (0.85-1.09) | 1.05 (1.01-1.08) | ||
| Entire pregnancy | 1.07 (0.79-1.44) | 1.17 (1.08-1.27) | |||
| CO | 1st trimester | 0.99 (0.75-1.30) | 0.95 (0.88-1.03) | ||
| 2nd trimester | 0.99 (0.75-1.31) | 0.98 (0.90-1.06) | |||
| 3rd trimester | 0.89 (0.67-1.18) | 1.03 (0.95-1.12) | |||
| Last 6 weeks | 0.82 (0.62-1.10) | 1.03 (0.95-1.12) | 0.136 | ||
| Entire pregnancy | 1.04 (0.79-1.37) | 1.12 (1.03-1.21) | |||
| NO2 | 1st trimester | 1.12 (0.87-1.44) | 1.00 (0.93-1.07) | ||
| 2nd trimester | 1.01 (0.78-1.31) | 1.03 (0.96-1.11) | |||
| 32-33 weeks | 3rd trimester | 0.91 (0.70-1.18) | 1.03 (0.96-1.11) | ||
| Last 6 weeks | 0.88 (0.68-1.15) | 1.00 (0.93-1.08) | |||
| Entire pregnancy | 0.95 (0.71-1.26) | 1.13 (1.04-1.22) | |||
| PM10 | 1st trimester | 1.01 (0.77-1.33) | 1.19 (1.11-1.29) | ||
| 2nd trimester | 1.16 (0.89-1.51) | 1.10 (1.02-1.19) | |||
| 3rd trimester | 1.02 (0.78-1.32) | 1.09 (1.01-1.17) | |||
| Last 6 weeks | 0.98 (0.75-1.28) | 1.01 (0.94-1.09) | |||
| PM2.5 | Entire pregnancy | 1.35 (1.03-1.76) | 1.46 (1.36-1.58) | ||
| 1st trimester | 0.97 (0.75-1.25) | 1.03 (0.96-1.10) | |||
| 2nd trimester | 1.09 (0.84-1.41) | 1.08 (1.00-1.16) | |||
| 3nd trimester | 0.90 (0.69-1.17) | 0.93 (0.86-1.01) | |||
| Last 6 weeks | 1.04 (0.80-1.35) | 1.05 (0.98-1.14) | |||
| Traffic Density | Entire pregnancy | 1.19 (0.92-1.56) | 1.09 (1.01-1.18) | ||
| Entire pregnancy | 0.96 (0.63-1.44) | 1.16 (1.05-1.28) | |||
| CO | 1st trimester | 0.75 (0.51-1.12) | 1.01 (0.92-1.11) | 0.140 | |
| 2nd trimester | 0.93 (0.63-1.36) | 0.98 (0.89-1.08) | |||
| 3rd trimester | 0.79 (0.52-1.19) | 0.98 (0.88-1.08) | |||
| Last 6 weeks | 0.94 (0.64-1.36) | 0.98 (0.89-1.08) | |||
| Entire pregnancy | 0.92 (0.64-1.33) | 1.12 (1.02-1.23) | |||
| NO2 | 1st trimester | 0.88 (0.63-1.23) | 1.13 (1.04-1.23) | 0.168 | |
| 2nd trimester | 1.13 (0.82-1.57) | 0.98 (0.90-1.07) | |||
| 3rd trimester | 0.79 (0.56-1.14) | 0.94 (0.86-1.03) | |||
| 28-31 weeks | Last 6 weeks | 0.97 (0.69-1.36) | 0.96 (0.88-1.05) | ||
| Entire pregnancy | 0.96 (0.66-1.41) | 1.01 (0.91-1.11) | |||
| PM10 | 1st trimester | 1.26 (0.89-1.78) | 1.11 (1.01-1.22) | ||
| 2nd trimester | 1.17 (0.82-1.66) | 1.02 (0.93-1.13) | |||
| 3rd trimester | 0.78 (0.53-1.14) | 0.96 (0.87-1.06) | |||
| Last 6 weeks | 0.91 (0.64-1.30) | 0.94 (0.86-1.04) | |||
| Entire pregnancy | 1.27 (0.89-1.81) | 1.37 (1.25-1.50) | |||
| PM2.5 | 1st trimester | 0.89 (0.63-1.25) | 1.09 (1.00-1.19) | ||
| 2nd trimester | 1.22 (0.88-1.70) | 1.06 (0.97-1.16) | |||
| 3rd trimester | 1.13 (0.80-1.60) | 0.89 (0.81-0.98) | 0.166 | ||
| Last 6 weeks | 1.17 (0.83-1.63) | 1.04 (0.95-1.14) | |||
| Traffic Density | Entire pregnancy | 0.82 (0.57-1.19) | 1.02 (0.93-1.12) | ||
| Entire pregnancy | 2.21 (1.20-4.08) | 1.07 (0.93-1.23) | 0.026 | ||
| CO | 1st trimester | 1.89 (1.04-3.44) | 0.99 (0.87-1.13) | 0.038 | |
| 2nd trimester | 0.76 (0.37-1.53) | 1.01 (0.89-1.15) | |||
| 3rd trimester | NC | NC | |||
| Last 6 weeks | 0.83 (0.42-1.65) | 1.04 (0.91-1.18) | |||
| Entire pregnancy | 1.56 (0.87-2.80) | 1.21 (1.07-1.37) | |||
| NO2 | 1st trimester | 1.39 (0.80-2.42) | 1.04 (0.93-1.17) | ||
| 2nd trimester | 0.77 (0.41-1.44) | 1.02 (0.90-1.15) | |||
| 3rd trimester | NC | NC | |||
| Last 6 weeks | 0.60 (0.31-1.16) | 1.06 (0.94-1.19) | 0.099 | ||
| 20-27 weeks | Entire pregnancy | 1.32 (0.72-2.42) | 1.15 (1.01-1.30) | ||
| PM10 | 1st trimester | 1.73 (0.98-3.06) | 1.18 (1.04-1.33) | 0.175 | |
| 2nd trimester | 0.83 (0.43-1.59) | 1.04 (0.92-1.18) | |||
| 3rd trimester | NC | NC | |||
| Last 6 weeks | 0.54 (0.26-1.11) | 0.89 (0.79-1.02) | 0.197 | ||
| Entire pregnancy | 2.44 (1.39-4.29) | 1.58 (1.40-1.78) | 0.119 | ||
| PM2.5 | 1st trimester | 2.15 (1.24-3.73) | 1.07 (0.95-1.20) | 0.014 | |
| 2nd trimester | 1.32 (0.74-2.36) | 1.07 (0.94-1.20) | |||
| 3rd trimester | NC | NC | |||
| Last 6 weeks | 2.56 (1.60-4.09) | 1.75 (1.57-1.94) | |||
| Traffic Density | Entire pregnancy | /1.54 (0.95-2.50) | /1.19 (1.06-1.33) | ||
NC, not calculated
Adjusted for maternal race, age, education, payment of birth expenses/insurance type, prenatal care
Gestational and pre-existing diabetes
P-value <0.2 for Wald’s Chi-squared test for interaction
Highest quartile cut offs: 0.60 ppm for CO, 19.47 ppb for NO2, 42.65 μg/m3 for PM10, 20.72 μg/m3 for PM2.5, and 45.85 for traffic density
Mediation of diabetes in the association between PM2.5 and extreme preterm birth
The results of the mediation analysis did not show that diabetes mediated the relationship between PM2.5 and extreme preterm birth. The third regression of preterm birth on diabetes was not significant and the calculation of the indirect effect was near zero (0.0009).
Sensitivity Analyses
When analyses were conducted comparing the highest quartile to the lowest quartile (rather than the lower three quartiles) of pollutant measures, the observed results were not meaningfully different. Similarly, results did not differ when we restricted the outcome to include only spontaneous preterm births nor when we removed women with hypertension or pre-eclampsia from the analysis.
Discussion
Our large population-based study observed inverse associations (in the unexpected direction) between air pollution exposures during pregnancy and gestational diabetes. One explanation might be that a subset of the population at risk for gestational diabetes and exposed to high levels of air pollution resulted in miscarriage – and therefore removed from study observation. These results add to the existing inconsistency across studies examining air pollution exposures during pregnancy and gestational diabetes [10, 12, 14, 15].
We did find evidence of effect modification by diabetes in the relationship between CO and PM2.5 and very early preterm birth. This finding is consistent with the few previous studies [16, 18]. though our results were not consistent across multiple pollutants (PM10, NO2) nor multiple categories of preterm birth (32-33 and 34-36 weeks). Furthermore, the first trimester and the entire pregnancy period were more often statistically significant for these associations.
Births on the early end of the very early preterm birth category (20-27 weeks) may have not had the opportunity to receive gestational diabetes testing, therefore leading to possible misclassification of the modifier. However, most women have the test between 24-28 weeks and even earlier if there is high glucose in the urine earlier in pregnancy to capture those at risk of GDM. The confidence intervals surrounding estimates of odds of preterm birth given air pollution exposure among those with diabetes (~5%) are less precise than the those among those without diabetes owing to the sample size. This may have decreased our ability to discern a difference in the associations between other air pollutants and preterm birth by diabetes status.
Our previous analyses of these data excluded women with diabetes to determine the effect of air pollution on preterm birth. The current analysis provides a potential pathway by which air pollution may affect some proportion of preterm birth. Several mechanisms of action have been hypothesized to explain the pathway by which toxicants may affect adverse reproductive outcomes including preterm birth and gestational diabetes [25]. Oxidative stress has been identified as the most relevant with evidence from increased levels of lipid peroxidation products and inflammatory cytokines in response to air pollution exposure [26, 27].
Although our study sample was large and population-based, it was not designed specifically to investigate diabetes and identification of diabetes was ascertained via the medical discharge record; however, we do not expect that any errors in identification of diabetes were likely related to air pollution levels, thus, resulting in non-differential misclassification. We did not have access to additional relevant clinical data such as oral glucose tolerance test glucose levels or treatment of diabetes during pregnancy. Regardless, our study included detailed air pollution exposure assessment at the precise geocoded residence over a wide geography and across several years. The levels of several pollutants (PM2.5, PM10 and ozone) caused the San Joaquin Valley to be in nonattainment during this period. In other words, the levels of these pollutants are higher than acceptable according to the Clean Air Act.
In conclusion, our study observed an inverse association between air pollution exposure and risk of gestational diabetes and a stronger effect of CO and PM2.5 and CO on very early preterm birth among those with diabetes, compared to those without. This may help identify populations that are particularly vulnerable to the detrimental effects of air pollution. Future research could examine this relationship in additional studies and examine additional maternal morbidities, such as pre-eclampsia, to explore potential mechanisms by which air pollution affects preterm birth.
Highlights.
Prenatal air pollution may be associated with diabetes
Prenatal air pollution may be more critical for certain at-risk populations
Prenatal air pollution among those with diabetes may increase risk of preterm birth
Research is needed to identify vulnerable populations for air pollution exposure
Acknowledgements:
Support for this study came from NIEHS (R21 ESO14891, P20 ES018173, P01ES022849, R00ES021470), and the March of Dimes Prematurity Research Center at Stanford University. This publication was made possible by US EPA STAR Grant RD83459601 and RD83543501. Its contents are solely the responsibility of the grantee and do not necessarily represent the official views of the US EPA. Further, the US EPA does not endorse the purchase of any commercial products or services mentioned in the publication. We thank Bryan Penfold of Sonoma Technology, Inc. for traffic data processing and traffic density estimation.
Abbreviations:
- Aor
adjusted odds ratio
- CI
confidence interval
- CO
carbon monoxide
- PM10
particulate matter less than 10 μm
- PM2.5
particulate matter less than 2.5 μm
- NO2
nitrogen dioxide
Appendix
Appendix Table 1.
Pearson Correlation of pollutant averages across entire pregnancy (EP), each trimester (T1, T2, T3) and the last 6 weeks of pregnancy (L6W).
| CO | NO2 | PM10 | PM2.5 | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| EP | T1 | T2 | T3 | L6 W | EP | T1 | T2 | T3 | L6 W | EP | T1 | T2 | T3 | L6 W | EP | T1 | T2 | T3 | L6 W | |
| CO EP | ||||||||||||||||||||
| CO T1 | 0.46 | |||||||||||||||||||
| CO T2 | 0.84 | 0.17 | ||||||||||||||||||
| CO T3 | 0.46 | −0.43 | 0.32 | |||||||||||||||||
| CO L6W | 0.31 | −0.41 | 0.11 | 0.93 | ||||||||||||||||
| NO2 EP | 0.75 | 0.25 | 0.67 | 0.42 | 0.31 | |||||||||||||||
| NO2 T1 | 0.47 | 0.82 | 0.34 | −0.38 | −0.41 | 0.59 | ||||||||||||||
| NO2 T2 | 0.74 | 0.02 | 0.83 | 0.47 | 0.29 | 0.87 | 0.34 | |||||||||||||
| NO2 T3 | 0.27 | −0.42 | 0.15 | 0.81 | 0.80 | 0.57 | −0.20 | 0.45 | ||||||||||||
| NO2 L6W | 0.14 | −0.37 | −0.02 | 0.71 | 0.79 | 0.45 | −0.21 | 0.26 | 0.93 | |||||||||||
| PM10 EP | 0.57 | 0.07 | 0.46 | 0.49 | 0.43 | 0.48 | 0.12 | 0.47 | 0.42 | 0.38 | ||||||||||
| PM10 T1 | 0.57 | 0.59 | 0.52 | −0.13 | −0.21 | 0.42 | 0.63 | 0.39 | −0.20 | −0.25 | 0.59 | |||||||||
| PM10 T2 | 0.50 | −0.26 | 0.58 | 0.60 | 0.48 | 0.46 | −0.13 | 0.61 | 0.49 | 0.39 | 0.82 | 0.28 | ||||||||
| PM10 T3 | 0.04 | −0.21 | −0.20 | 0.52 | 0.61 | 0.07 | −0.28 | −0.08 | 0.57 | 0.64 | 0.60 | −0.13 | 0.38 | |||||||
| PM10 L6W | 0.00 | −0.04 | −0.26 | 0.34 | 0.51 | 0.03 | −0.13 | −0.18 | 0.43 | 0.59 | 0.46 | −0.11 | 0.21 | 0.86 | ||||||
| PM2.5 EP | 0.76 | 0.34 | 0.70 | 0.30 | 0.17 | 0.59 | 0.41 | 0.64 | 0.11 | 0.01 | 0.70 | 0.72 | 0.57 | 0.08 | 0.01 | |||||
| PM2.5 T1 | 0.26 | 0.87 | 0.04 | −0.50 | −0.47 | 0.10 | 0.71 | −0.08 | −0.50 | −0.42 | 0.10 | 0.62 | −0.27 | −0.18 | −0.03 | 0.42 | ||||
| PM2.5 T2nd | 0.65 | 0.08 | 0.87 | 0.19 | 0.01 | 0.54 | 0.30 | 0.73 | 0.05 | −0.10 | 0.50 | 0.58 | 0.61 | −0.21 | −0.27 | 0.77 | 0.05 | |||
| PM2.5 T3rd | 0.28 | −0.50 | 0.20 | 0.86 | 0.81 | 0.29 | −0.45 | 0.40 | 0.69 | 0.61 | 0.53 | −0.11 | 0.62 | 0.56 | 0.36 | 0.37 | −0.49 | 0.17 | ||
| PM2.5 L6W | 0.15 | −0.46 | 0.00 | 0.79 | 0.86 | 0.18 | −0.47 | 0.21 | 0.67 | 0.66 | 0.44 | −0.20 | 0.47 | 0.62 | 0.52 | 0.21 | −0.45 | −0.03 | 0.90 | |
| Traffic Density | 0.00 | 0.01 | −0.01 | 0.00 | 0.00 | 0.15 | 0.10 | 0.10 | 0.10 | 0.09 | −0.03 | −0.02 | −0.02 | −0.02 | −0.02 | 0.01 | 0.01 | 0.00 | 0.01 | 0.01 |
Bolded correlation coefficients were statistically significant (p<0.05)
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 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.
Conflicts of Interest:
none
References
- 1.Stieb DM, et al. , Ambient air pollution, birth weight and preterm birth: a systematic review and meta-analysis. Environ Res, 2012. 117: p. 100–11. [DOI] [PubMed] [Google Scholar]
- 2.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(1): p. 67–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Schifano P, et al. , Effect of ambient temperature and air pollutants on the risk of preterm birth, Rome 2001-2010. Environ Int, 2013. 61: p. 77–87. [DOI] [PubMed] [Google Scholar]
- 4.Li X, et al. , Association between ambient fine particulate matter and preterm birth or term low birth weight: An updated systematic review and meta-analysis. Environ Pollut, 2017. 227: p. 596–605. [DOI] [PubMed] [Google Scholar]
- 5.Padula AM, et al. , Traffic-related air pollution and risk of preterm birth in the San Joaquin Valley of California. Ann Epidemiol, 2014. 24(12): p. 888–95e4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Eze IC, et al. , Association between ambient air pollution and diabetes mellitus in Europe and North America: systematic review and meta-analysis. Environ Health Perspect, 2015. 123(5): p. 381–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Rajagopalan S and Brook RD, Air pollution and type 2 diabetes: mechanistic insights. Diabetes, 2012. 61(12): p. 3037–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Park SK, Ambient Air Pollution and Type 2 Diabetes: Do the Metabolic Effects of Air Pollution Start Early in Life? Diabetes, 2017. 66(7): p. 1755–1757. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Eze IC, et al. , Association between Ambient Air Pollution and Diabetes Mellitus in Europe and North America: Systematic Review and Meta-Analysis. Environmental Health Perspectives, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Lu MC, et al. , Association of temporal distribution of fine particulate matter with glucose homeostasis during pregnancy in women of Chiayi City, Taiwan. Environ Res, 2017. 152: p. 81–87. [DOI] [PubMed] [Google Scholar]
- 11.Lu M-C, et al. , Association of temporal distribution offine particulate matter with glucose homeostasis during pregnancy in women of Chiayi City, Taiwan. Environmental research, 2017. 152(C): p. 81–87. [DOI] [PubMed] [Google Scholar]
- 12.Robledo CA, et al. , Preconception and early pregnancy air pollution exposures and risk of gestational diabetes mellitus. Environ Res, 2015. 137: p. 316–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Robledo CA, et al. , Preconception and early pregnancy air pollution exposures and risk of gestational diabetes mellitus. Environmental research, 2015. 137(c): p. 316–322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Fleisch AF, et al. , Air pollution exposure and abnormal glucose tolerance during pregnancy: the project Viva cohort. Environ Health Perspect, 2014. 122(4): p. 378–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Fleisch AF, et al. , Air pollution exposure and gestational diabetes mellitus among pregnant women in Massachusetts: a cohort study. Environ Health, 2016. 15: p. 40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lavigne E, et al. , Ambient air pollution and adverse birth outcomes: Differences by maternal comorbidities. Environ Res, 2016. 148: p. 457–466. [DOI] [PubMed] [Google Scholar]
- 17.Lavigne E, et al. , Ambient air pollution and adverse birth outcomes_ Differences by maternal comorbidities. Environmental research, 2016. 148(C): p. 457–466. [DOI] [PubMed] [Google Scholar]
- 18.Lin YT, et al. , Associations between ozone and preterm birth in women who develop gestational diabetes. Am J Epidemiol, 2015. 181(4): p. 280–7. [DOI] [PubMed] [Google Scholar]
- 19.Lin YT, et al. , Associations Between Ozone and Preterm Birth in Women Who Develop Gestational Diabetes. American journal of epidemiology, 2015. 181(4): p. 280–287. [DOI] [PubMed] [Google Scholar]
- 20.Kan H, et al. , Prospective analysis of traffic exposure as a risk factor for incident coronary heart disease: the Atherosclerosis Risk in Communities (ARIC) study. Environ Health Perspect, 2008. 116(11): p. 1463–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Padula AM, et al. , Traffic-related air pollution and risk of preterm birth in the San Joaquin Valley of California. Annals of epidemiology, 2014. 24(12): p. 888–895.e4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Baron RM and Kenny DA, The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol, 1986. 51(6): p. 1173–82. [DOI] [PubMed] [Google Scholar]
- 23.Judd CM and Kenny DA, Process Analysis: Estimating mediation in treatment evaluations. Evaluation Review, 1981. 5(5): p. 602–619. [Google Scholar]
- 24.Sobel ME, Asymptotic confidence intervals for indirect effects in structural equation models, in Sociological Methodology, Leinhardt S, Editor. 1982, American Sociological Association: Washington D.C. . p. 290–312. [Google Scholar]
- 25.Ferguson KK and Chin HB, Environmental chemicals and preterm birth: Biological mechanisms and the state of the science. Curr Epidemiol Rep, 2017. 4(1): p. 56–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Ferguson KK, O’Neill MS, and Meeker JD, Environmental contaminant exposures and preterm birth: a comprehensive review. J Toxicol Environ Health B Crit Rev, 2013. 16(2): p. 69–113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Vadillo-Ortega F, et al. , Air pollution, inflammation and preterm birth: a potential mechanistic link. Med Hypotheses, 2014. 82(2): p. 219–24. [DOI] [PMC free article] [PubMed] [Google Scholar]