Date of Award

January 2025

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)

Department

School of Public Health

First Advisor

Joshua L. Warren

Abstract

Accurately synthesizing evidence from epidemiologic studies is essential for understanding complex environmental risk relationships and establishing causal relationships. However, the majority of meta-analyses in this area ignore the possibility of variability in risk due to differences in study-specific exposure levels while typical dose-response statistical methods assume a single dose amount per study and not a distribution or range of doses (i.e., exposures). We develop a novel hierarchical Bayesian meta-regression framework that incorporates functional averaging over study-specific exposure ranges and models the underlying exposure-response relationship using a flexible Gaussian process. Using a low-rank predictive process formulation, our approach efficiently projects functional values onto a fixed grid of reference points, enabling smooth and interpretable estimates while preserving computational scalability. Through simulation, we demonstrate the limitations of existing approaches in recovering nonlinear exposure-response curves and the improved performance of our model. We apply our method to data from a published meta-analysis on long-term exposure to particulate matter (PM2.5) and all-cause mortality, identifying nonlinear patterns in the exposure-response relationship that were not captured by standard models. Comparative analyses show the proposed method provides improved model fit and more realistic uncertainty quantification.

Comments

This is an Open Access Thesis.

Download
Open Access

This Article is Open Access

Share

COinS