Estimating County-Level Opioid-Related Mortality in Georgia Using a Bayesian Conditional Autoregressive Model 公开

Abstract

Opioid-related mortality has emerged as one of the most pressing public health challenges in the United States, underscoring the need for improved surveillance and a deeper understanding of its social and structural determinants of health (SSDH). We estimated county-level opioid mortality rates in Georgia from 2020 to 2022 and examined their associations with key SSDH indicators. Using mortality data and covariates related to these determinants, including the Centers for Disease Control and Prevention’s (CDC) Social Vulnerability Index (SVI), poverty, unemployment, and distances to the nearest interstate highway and treatment center, we employed Poisson regression models with county-specific random effects and Bayesian conditional autoregressive (CAR) models to generate smoothed estimates. Most covariates were inversely associated with opioid mortality across all years, although few remained statistically significant after accounting for spatial correlation. The SVI component representing racial and ethnic minority status showed a consistently significant negative association. When spatial correlation was incorporated into the CAR models, many covariate effects became less pronounced, with estimates shifting toward the null and credible intervals becoming wider. This pattern may reflect overdispersion, weak spatial dependence, or multicollinearity among covariates. These findings highlight spatial disparities in opioid-related mortality in Georgia and provide insight to inform local prevention strategies and resource allocation. They also point to the need to further investigate the drivers of spatial heterogeneity and to incorporate spatial structure in efforts to better understand and address opioid-related health disparities.

Table of Contents

1 Introduction 1

2 Methods 3

2.1 Data 3

2.2 Poisson Regression with Random Intercepts 5

2.3 Conditional Autoregressive (CAR) Model 6

3 Results 8

3.1 2020 Results 8

3.1.1 Univariate Poisson Regression 8

3.1.2 Univariate CAR Model 9

3.1.3 Multivariate Poisson Regression 10

3.1.4 Multivariate CAR Model 13

3.2 2021 Results 16

3.2.1 Univariate Poisson Regression 16

3.2.2 Univariate CAR Model 17

3.2.3 Multivariate Poisson Regression 18

3.2.4 Multivariate CAR Model 19

3.3 2022 Results 21

3.3.1 Univariate Poisson Regression 21

3.3.2 Univariate CAR Model 22

3.3.3 Multivariate Poisson Regression 23

3.3.4 Multivariate CAR Model 24

3.4 2020–2022 Combined Results 26

3.4.1 Univariate Poisson Regression 26

3.4.2 Univariate CAR Model 27

3.4.3 Multivariate Poisson Regression 28

3.4.4 Multivariate CAR Model 29

4 Discussion 35

A Appendix 36

References 45

About this Master's Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Rollins School of Public Health
Department	Biostatistics
Degree	M.P.H.
Submission	Master's Thesis
Language	English
Research Field	Health Sciences, Public Health Health Sciences, Epidemiology Biology, Biostatistics
关键词	Social Vulnerability Disease Mapping Drug overdose Spatial Statistics CAR Model Opioid mortality
Committee Chair / Thesis Advisor	Waller, Lance, Emory University
Committee Members	Peterson, Emily, Emory University

最新修改

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	Estimating County-Level Opioid-Related Mortality in Georgia Using a Bayesian Conditional Autoregressive Model ()	2025-04-18 21:51:47 -0400	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions