Air Pollution and Term Birth Weight in New York, New York
The study population is ethnically diverse, covers a wide range of maternal age and educational levels, and includes few births of less than 2,500 g after restriction to term deliveries (Table 1). The interquartilve range for PM2.5 exposure (Figure 2) ranged from 2.5 µg/m for average exposure over the entire pregnancy to 3.3 µg/m in the first trimester, whereas the interquartile range for nitrogen dioxide ranged from 6.2 ppb for average exposure over the entire pregnancy to 8.0 ppb in both the first and second trimesters. For each pollutant, PM2.5 and nitrogen dioxide, higher exposure was correlated with slightly lower census tract–level social deprivation (Pearson's ρ was approximately −0.1 for each pollutant and each exposure window).
(Enlarge Image)
Figure 2.
Change in birth weight in grams per interquartile range (IQR) of particulate matter with aerodynamic diameter less than 2.5 μm (PM2.5) and nitrogen dioxide with 95% confidence intervals for each exposure time window based on models with the following different degrees of confounder adjustment: unadjusted (triangles), routine adjustment (circles), and fully adjusted (squares), New York, New York, 2008–2010. Corresponding numerical values are in Web Table 1, available at http://aje.oxfordjournals.org/.
Because of seasonal patterns in air pollution in relationship to the duration of pregnancy, trimester-specific exposures are correlated to varying degrees (Table 2). The seasonality of PM2.5 is bimodal and peaks both in summer and winter, and adjacent pregnancy windows were less correlated than the first and third trimesters. In contrast, the seasonality of nitrogen dioxide is monomodal and peaks in winter, which leads to higher correlation between adjacent trimesters than between distant trimesters. The correlation (Pearson's ρ) between PM2.5 and nitrogen dioxide exposure was 0.63 within the first trimester, 0.59 within the second trimester, 0.53 within the third trimester, and 0.81 for the entire pregnancy.
When we considered the covariates alone (Table 3), increasing gestational age was strongly predictive of increased birth weight, and children of black and Asian mothers, younger mothers, less educated mothers, and those who were born in later study years and in more socioeconomically deprived census tracts tended to have lower birth weights. Outdoor temperature and Medicaid status were essentially unrelated to birth weight after adjustment for the other covariates.
Figure 2 shows the estimated association between exposures to PM2.5 and nitrogen dioxide and birth weight per IQR (Web Table 1, available at http://aje.oxfordjournals.org/, provides the corresponding values). Results are expressed as grams of birth weight per 10-unit change in air pollutant exposure (10 µg/m for PM2.5 or 10 ppb for nitrogen dioxide) (Figure 3), as well as per interquartile range change in air pollutant exposure. Before covariate adjustment, pollutant exposure and birth weight were essentially unrelated for nitrogen dioxide and had weakly positive coefficients for PM2.5 across exposure windows. Adjustment had a substantial impact, primarily due to higher levels of exposure and higher birth weights among non-Hispanic whites, among older mothers, and in earlier calendar years. With or without adjustment for temperature, results indicated that higher PM2.5 and nitrogen dioxide exposures in all pregnancy windows were associated with lower birth weights. Among the trimester-specific exposure windows, for PM2.5, the strongest associations occurred in the first and third trimesters; for nitrogen dioxide, there was a less notable difference in the estimated associations across exposure windows.
(Enlarge Image)
Figure 3.
Change in birth weight in grams A) per 10 μg/m of particulate matter with aerodynamic diameter less than 2.5 μm, and B) per 10 ppb of nitrogen dioxide with 95% confidence intervals for each exposure time window based on models with the following 3 degrees of confounder adjustment: unadjusted (triangles), routine adjustment (circles), and fully adjusted (squares), New York, New York, 2008–2010. Corresponding numerical values are in Web Table 1, available at http://aje.oxfordjournals.org/.
Figure 4 compares estimates of the association between pollutant exposure and birth weight, where exposure was assigned based solely on either temporal or spatial variation to estimates from our primary analysis (Figure 3) where exposure was assigned based on both sources of variation (Web Table 2). To aid in the interpretation of health-effect estimates associated with exposures with different degrees of variability, Web Figure 1 shows box plots of the spatial and temporal components for PM2.5 and nitrogen dioxide. Whereas for nitrogen dioxide, the spatial component exhibited greater variability than the temporal component, for PM2.5 the temporal component was more variable. For PM2.5, the associations based on both sources of variation in exposure lie between the estimates based on only spatial or only temporal variation, suggesting that both sources of variability contribute to the PM2.5 associations. On the other hand, for nitrogen dioxide, the estimates based on both temporal and spatial variation are nearly identical to the estimates based on spatial variation only, providing evidence that the association between nitrogen dioxide and birth weight is driven almost completely by the variation in exposure across mothers' residences and not over time.
(Enlarge Image)
Figure 4.
Change in birth weight in grams A) per 10 μg/m of particulate matter with aerodynamic diameter less than 2.5 μm, and B) per 10 ppb of nitrogen dioxide with 95% confidence intervals for each exposure time window based on the fully adjusted model for the following 3 exposure metrics: temporal variation only (squares), spatial variation only (triangles), and combined temporal and spatial variation (circles), New York, New York, 2008–2010. Corresponding numerical values are in Web Table 1, available at http://aje.oxfordjournals.org/, (combined temporal and spatial variation) and Web Table 2 (temporal variation only and spatial variation only).
When we adjusted the pollutants for one another (2-pollutant model), higher nitrogen dioxide remained independently associated with lower birth weight, whereas PM2.5 was no longer associated with birth weight (except for during the second trimester, in which higher exposure was associated with higher birth weight) (Figure 5).
(Enlarge Image)
Figure 5.
Change in birth weight in grams A) per 10 μg/m of particulate matter with aerodynamic diameter less than 2.5 μm, and B) per 10 ppb of nitrogen dioxide with 95% confidence intervals for each exposure time window based on the 2-pollutant model, New York, New York, 2008–2010. Corresponding numerical values are in Web Table 3, available at http://aje.oxfordjournals.org/.
In sensitivity analyses allowing for a potential nonlinear exposure-response relationship, PM2.5 exhibited no evidence of a nonlinear relationship with birth weight for any of the exposure windows (Web Figure 2). For average nitrogen dioxide exposure over the study period, birth weights decreased with increasing levels of exposure until approximately 20 ppb, after which they leveled off and remained flat until approximately 35 ppb and then continued to decrease over the remaining range of the data (Web Figure 2). The form of the exposure-response function was similar for the other time windows of nitrogen dioxide exposure.
Results
Descriptive Statistics
The study population is ethnically diverse, covers a wide range of maternal age and educational levels, and includes few births of less than 2,500 g after restriction to term deliveries (Table 1). The interquartilve range for PM2.5 exposure (Figure 2) ranged from 2.5 µg/m for average exposure over the entire pregnancy to 3.3 µg/m in the first trimester, whereas the interquartile range for nitrogen dioxide ranged from 6.2 ppb for average exposure over the entire pregnancy to 8.0 ppb in both the first and second trimesters. For each pollutant, PM2.5 and nitrogen dioxide, higher exposure was correlated with slightly lower census tract–level social deprivation (Pearson's ρ was approximately −0.1 for each pollutant and each exposure window).
(Enlarge Image)
Figure 2.
Change in birth weight in grams per interquartile range (IQR) of particulate matter with aerodynamic diameter less than 2.5 μm (PM2.5) and nitrogen dioxide with 95% confidence intervals for each exposure time window based on models with the following different degrees of confounder adjustment: unadjusted (triangles), routine adjustment (circles), and fully adjusted (squares), New York, New York, 2008–2010. Corresponding numerical values are in Web Table 1, available at http://aje.oxfordjournals.org/.
Because of seasonal patterns in air pollution in relationship to the duration of pregnancy, trimester-specific exposures are correlated to varying degrees (Table 2). The seasonality of PM2.5 is bimodal and peaks both in summer and winter, and adjacent pregnancy windows were less correlated than the first and third trimesters. In contrast, the seasonality of nitrogen dioxide is monomodal and peaks in winter, which leads to higher correlation between adjacent trimesters than between distant trimesters. The correlation (Pearson's ρ) between PM2.5 and nitrogen dioxide exposure was 0.63 within the first trimester, 0.59 within the second trimester, 0.53 within the third trimester, and 0.81 for the entire pregnancy.
Statistical Model of the Associations Between PM2.5, Nitrogen Dioxide, and Birth Weight
When we considered the covariates alone (Table 3), increasing gestational age was strongly predictive of increased birth weight, and children of black and Asian mothers, younger mothers, less educated mothers, and those who were born in later study years and in more socioeconomically deprived census tracts tended to have lower birth weights. Outdoor temperature and Medicaid status were essentially unrelated to birth weight after adjustment for the other covariates.
Figure 2 shows the estimated association between exposures to PM2.5 and nitrogen dioxide and birth weight per IQR (Web Table 1, available at http://aje.oxfordjournals.org/, provides the corresponding values). Results are expressed as grams of birth weight per 10-unit change in air pollutant exposure (10 µg/m for PM2.5 or 10 ppb for nitrogen dioxide) (Figure 3), as well as per interquartile range change in air pollutant exposure. Before covariate adjustment, pollutant exposure and birth weight were essentially unrelated for nitrogen dioxide and had weakly positive coefficients for PM2.5 across exposure windows. Adjustment had a substantial impact, primarily due to higher levels of exposure and higher birth weights among non-Hispanic whites, among older mothers, and in earlier calendar years. With or without adjustment for temperature, results indicated that higher PM2.5 and nitrogen dioxide exposures in all pregnancy windows were associated with lower birth weights. Among the trimester-specific exposure windows, for PM2.5, the strongest associations occurred in the first and third trimesters; for nitrogen dioxide, there was a less notable difference in the estimated associations across exposure windows.
(Enlarge Image)
Figure 3.
Change in birth weight in grams A) per 10 μg/m of particulate matter with aerodynamic diameter less than 2.5 μm, and B) per 10 ppb of nitrogen dioxide with 95% confidence intervals for each exposure time window based on models with the following 3 degrees of confounder adjustment: unadjusted (triangles), routine adjustment (circles), and fully adjusted (squares), New York, New York, 2008–2010. Corresponding numerical values are in Web Table 1, available at http://aje.oxfordjournals.org/.
Sensitivity Analyses
Figure 4 compares estimates of the association between pollutant exposure and birth weight, where exposure was assigned based solely on either temporal or spatial variation to estimates from our primary analysis (Figure 3) where exposure was assigned based on both sources of variation (Web Table 2). To aid in the interpretation of health-effect estimates associated with exposures with different degrees of variability, Web Figure 1 shows box plots of the spatial and temporal components for PM2.5 and nitrogen dioxide. Whereas for nitrogen dioxide, the spatial component exhibited greater variability than the temporal component, for PM2.5 the temporal component was more variable. For PM2.5, the associations based on both sources of variation in exposure lie between the estimates based on only spatial or only temporal variation, suggesting that both sources of variability contribute to the PM2.5 associations. On the other hand, for nitrogen dioxide, the estimates based on both temporal and spatial variation are nearly identical to the estimates based on spatial variation only, providing evidence that the association between nitrogen dioxide and birth weight is driven almost completely by the variation in exposure across mothers' residences and not over time.
(Enlarge Image)
Figure 4.
Change in birth weight in grams A) per 10 μg/m of particulate matter with aerodynamic diameter less than 2.5 μm, and B) per 10 ppb of nitrogen dioxide with 95% confidence intervals for each exposure time window based on the fully adjusted model for the following 3 exposure metrics: temporal variation only (squares), spatial variation only (triangles), and combined temporal and spatial variation (circles), New York, New York, 2008–2010. Corresponding numerical values are in Web Table 1, available at http://aje.oxfordjournals.org/, (combined temporal and spatial variation) and Web Table 2 (temporal variation only and spatial variation only).
When we adjusted the pollutants for one another (2-pollutant model), higher nitrogen dioxide remained independently associated with lower birth weight, whereas PM2.5 was no longer associated with birth weight (except for during the second trimester, in which higher exposure was associated with higher birth weight) (Figure 5).
(Enlarge Image)
Figure 5.
Change in birth weight in grams A) per 10 μg/m of particulate matter with aerodynamic diameter less than 2.5 μm, and B) per 10 ppb of nitrogen dioxide with 95% confidence intervals for each exposure time window based on the 2-pollutant model, New York, New York, 2008–2010. Corresponding numerical values are in Web Table 3, available at http://aje.oxfordjournals.org/.
In sensitivity analyses allowing for a potential nonlinear exposure-response relationship, PM2.5 exhibited no evidence of a nonlinear relationship with birth weight for any of the exposure windows (Web Figure 2). For average nitrogen dioxide exposure over the study period, birth weights decreased with increasing levels of exposure until approximately 20 ppb, after which they leveled off and remained flat until approximately 35 ppb and then continued to decrease over the remaining range of the data (Web Figure 2). The form of the exposure-response function was similar for the other time windows of nitrogen dioxide exposure.
SHARE
