Spatiotemporal Geostatistical Methods for Exposure and Epidemiological Analyses of Groundwater Nitrate and Radon Public Deposited

Downloadable Content

Download PDF
Last Modified
  • March 19, 2019
  • Messier, Kyle
    • Affiliation: Gillings School of Global Public Health, Department of Environmental Sciences and Engineering
  • Exposure assessment and dose-response characterization are critical steps in the risk assessment of an environmental contaminant with potential human health effects. There are many established methods to conduct exposure assessments and to characterize the dose-response relationship between a contaminant of concern and a health outcome; however, many require extensive time and monetary resources that are becoming increasingly limited. Geostatistical methods are attractive approaches due to their cost-effective implementation and clear physical interpretations. Land use regression (LUR) is a type of geostatistical method that uses spatially-based explanatory variables to model outcomes using classical regression methods. Bayesian Maximum Entropy (BME) is a geostatistical framework for incorporating measurements as well as various knowledge bases in a logical and theoretically sound manner to produce estimates for variables of interest at unmonitored locations. This work advances these spatiotemporal geostatistical methods in the following three studies: 1) An exposure assessment of groundwater nitrate (NO_3^-), a biological nutrient with natural and anthropogenic sources that in excess has deleterious effects on human and ecological health; 2) An exposure assessment of groundwater radon (222Rn), a naturally occurring gas with radioactively discharged alpha particles that are known human carcinogens; and 3) An epidemiological analysis of the association between groundwater 222Rn exposure and lung and stomach cancer incidence. First, we develop a nonlinear LUR model and then integrate the model into the BME framework to produce the first space/time exposure estimates of groundwater NO_3^- concentrations across a large domain with a cross-validation r^2of 0.74. Second, an exposure model for point-level groundwater 222Rn is developed with anisotropic geological and uranium-based explanatory variables resulting in a cross-validation r^2of 0.46. Lastly, we utilize the LUR-BME exposure model for 222Rn to investigate associations with lung and stomach cancer at multiple spatial scales. It is the first epidemiological analysis of the association between groundwater 222Rn exposure and lung cancer, moreover with a significant and positive association; and the first to find a positive association between groundwater 222Rn and stomach cancer. This body of research provides advances in exposure assessment and dose-response methodology and practical real-world examples that can be used as resources for future cost-effective protection of public health.
Date of publication
Resource type
Rights statement
  • In Copyright
  • MacDonald Gibson, Jacqueline
  • Band, Lawrence
  • Characklis, Gregory
  • Serre, Marc
  • Fry, Rebecca
  • Doctor of Philosophy
Degree granting institution
  • University of North Carolina at Chapel Hill Graduate School
Graduation year
  • 2015
Place of publication
  • Chapel Hill, NC
  • There are no restrictions to this item.

This work has no parents.