1799

Background: Prenatal exposure to ambient air pollutants has been associated with both reduced and increased birth weight (BW) in epidemiological studies. Many studies focus on the association of a single pollutant with BW rather than a mixture of pollutants.
Methods: We compared two approaches for identifying prenatal exposure to hazardous air pollutants (HAPs), (1) each HAP and (2) mixture of 15 HAPs together, as predictors to changes in BW. We evaluated 15 HAPs at the census tract level from the 2011 National Air Toxics Assessment and linked with 735,507 infant-mother pairs in North Carolina with birth years between 2006-2011. We used random intercept mixed effects regression models to estimate the change in BW in grams (g), presented as β estimates and 95% confidence intervals (CI), adjusted for maternal race/ethnicity, age, marital status, and medicaid status.
Results: Cyanide (CN), mercury, and cadmium were identified as the strongest individual predicators, with a consistent reduction in BW across all HAPs (range β: -3995.31g for CN to -1.34g for methyl-tert-butyl-ether). In the HAPs mixture, CN and xylene were the strongest predictors of increased BW (β:1908.65g CN and 213.91g for xylene) whereas toulene and benzene (β: -174.13g and -150.87g, respectively) were the strongest predictors for reduced BW. In the individual adjusted models, CN showed the strongest change in BW with a reduction of -1242.37g (-1939.27, -545.47). When analyzing HAPs as a mixture, in the adjusted models, toulene had the greatest change in BW with a reduction of -87.62g (-120.65, -54.59).
Conclusions: Comparing the two approaches, CN was the only consistent HAP identified as a predictor for change in BW. Evaluating exposure to individual pollutants in relation to BW may be neglecting the synergistic and/or antagonistic effects and underlying confounding when there are concurrent exposures to multiple pollutants. This abstract does not reflect EPA policy.

1802

Background: Exposure to environmental pollutants has been associated with both reduced birth weight (BW) through air pollutants and increased BW through endocrine disrupting chemicals. The impact of multiple environmental exposures across several domains in relation to changes in BW are not well understood.
Methods: We constructed an environmental quality index at the census tract level (trEQI) to represent environmental quality across five domains (air, water, land, sociodemographic, and built) and overall for 2006-2010 using principal components analysis. The trEQI domains were categorized into quartiles (<25^th, or best environmental quality and referent; 25^th-50^th; 50^th-75^th; >75^th, or worst environmental quality) and linked to 735,507 mother-infant pairs in North Carolina (NC) with births between 2006-2011. Random intercept mixed effects linear regression models estimated the change in BW, in grams (g), presented as β estimate and 95% confidence intervals (CI), adjusted for maternal race/ethnicity, age, marital status, and medicaid status.
Results: The worst environmental quality, or highest quartile, was associated with reduction in BW in the air [β: -10.41 g (95% CI: -17.06, -3.75)], water [β: -11.84 g (-18.82, -4.86)], and land [β: -9.17 g (-15.71, -2.65)] domains, compared to the best environmental quality. However, we observed increased BW in association with the sociodemographic [β: 34.82 g (28.44, 41.20)] domain and overall [β: 21.64 g (15.55, 27.73)] trEQI.
Conclusions: With increasing interest in how environmental mixtures can influence health, this analysis uses a more spatially resolved exposure index than previously employed (i.e., census tract vs. county) and provides a broad view of how simultaneous environmental exposures across multiple domains can result in reduced or increased BW. This abstract does not reflect EPA policy.