Date of Award

2015

Document Type

Open Access Dissertation

Department

Epidemiology and Biostatistics

Sub-Department

Epidemiology

First Advisor

Jihong Liu

Abstract

Preterm birth, generally defined as birth at <37 weeks of gestation, is an important public health issue that has multiple risk factors related to characteristics of both the mother and her environment. The purpose of this dissertation was to examine potential sources of spatially interpolated (kriged) environmental concentrations of arsenic (As) and lead (Pb) in residential soils and preterm birth in a Medicaid population of mothers giving birth in South Carolina (SC) from 1996-2001. The first objective was to investigate if a racial disparity existed for estimated soil As and Pb concentrations, after adjusting for proximal and distal sources of these metals (including distance and direction to industrial facilities) in a subset of SC Medicaid mothers living in areas of SC where soil samples were collected and analyzed for these metals. The second objective was to test the hypothesis that estimated soil As and Pb concentrations were associated with increased odds of early (<34 weeks) and late (34-36 weeks) preterm births in the same subset of SC Medicaid mothers, after adjusting for individual and neighborhood level risk factors, and examine if measure of neighborhood deprivation and racial residential segregation modified these associations. The third objective was to examine if early and all preterm births, aggregated at the county level, varied spatially and/or temporally in SC for all Medicaid mothers giving birth from 1996-2001 in Bayesian models.

For the first objective, black mothers had significantly higher estimated As and Pb soil concentrations than white mothers in the study population (adjusted betas were 0.12 and 0.22 for As and Pb, respectively; all p<0.006), and proximal sources of metals (e.g., percent of Census block group are covered by roads) were more strongly associated with estimated soil As and Pb concentrations than composite As and Pb releases from industrial facilities categorized by distance from and direction to Census block groups in which maternal residences were located.

For the second objective, estimated soil concentration of neither As nor Pb were associated with increased odds of early or late preterm birth after adjusting for maternal and neighborhood level risk factors. Only individual level covariates were associated with these birth outcomes, and associations were stronger for early as compared to late preterm births. Neighborhood deprivation and racial residential segregation were not associated with either early or late preterm birth in adjusted models, nor did they modify the main associations of interest.

Results from the final objective showed that Bayesian models including spatial and temporal parameters had a better fit, for both early and all preterm births, than those only containing temporal parameters, in the aggregate spatial-temporal analysis at the county level for all SC Medicaid mothers. Some counties in SC had significant risk of early and all preterm births over multiple years, and aggregate county measures of maternal demographics and conditions during pregnancy were associated with increased risk of early and all preterm births by both county and year.

These findings indicate the importance of proximal, historic metal sources to current residential soil concentrations, as well as the racial disparity in soil As and Pb concentrations in the studied locations of SC. They also show that accounting for the distance and direction to industrial facilities releasing these metals does not impact the racial disparity, or estimated residential soil As and Pb concentrations. In addition, odds of early and late preterm births were not associated with estimated residential soil As and Pb concentrations, and only maternal risk factors were significantly associated with these outcomes in the study population. Finally, taking into account the spatial relationship between counties in SC improved risk estimates of early and all preterm births at the county level for Medicaid mothers in SC. These finding have implications in the areas of environmental justice,

Rights

© 2015, Pamela Harley Thornton Davis

Download

Included in

Epidemiology Commons

Share

COinS