The Association of Ozone and Particulate Matter Exposure to Asthma Related Hospital Visits in Mississippi Open Access

Schwartz, Gregory Joseph (2008)

Permanent URL: https://etd.library.emory.edu/concern/etds/h415p9970?locale=en
Published

Abstract

In 2006, 22.9 million Americans were estimated to have asthma. This thesis looks at the relationship between asthma related hospital visits and exposure to ambient concentrations of ozone and particulate matter < 2.5 µm (PM2.5). Exposure is measured remotely via satellite allowing the study to cover entire state of Mississippi.


A regular network of 1,318 grid cells was overlain the state of Mississippi. Mean daily Ozone and PM2.5 exposure, number of asthma related hospital visits, and demographic characteristics were determined for each cell from January 1st, 2003 to December 31st, 2005. Dual-exposure models were built using Generalized Estimating Equations (GEE) to determine the association between hospital visits and exposures controlling for demographic characteristics. The study stratified by urban or rural designation considering the entire population as well as the black portion of the
population. These models were extended with hierarchical Bayesian models to account for conditional autoregressive (CAR) spatial and non-spatial exchangeable random effects at the grid cell level.


Significant dual-exposure models were found for both urban and rural regions when considering the total population. A 2-day lag for ozone and 5-day lag for PM2.5 were used in the urban only model with relative risks of 1.003 (95% CI = (1.001, 1.006)) and 1.004 (95% CI = (1.000, 1.008)) respectively. The rural only model used a 2-day lag for ozone and 4-day lag for PM2.5 with respective relative risks of 1.002 (95% CI = (1.001, 1.005)) and 1.002 (95% CI = (1.000, 1.004)). A significant dual-exposure model was found for the black only urban area using a 4-day lag for ozone and a 2-day lag for PM2.5 with relative risks of 1.004 (95% CI = (1.001, 1.006)) and 1.003 (95% CI = (1.001, 1.006)) respectively. The Bayesian analysis found exchangeable random effects in the urban region and spatial random effects in the rural region improved the model fit allowing for risk estimates to be made at the grid cell level.

Table of Contents

Introduction......................................................................................................................... 1
Background..................................................................................................................... 1
Remote sensing of ground based pollution concentrations ............................................ 4
Models for repeated data ................................................................................................ 4
Methods............................................................................................................................... 6
Data Source .................................................................................................................... 6
Missing Data................................................................................................................... 7
Analytical Methods ......................................................................................................... 7
Descriptive Statistics................................................................................................... 7
Longitudinal Models using Generalized Estimating Equations ................................. 8
Hierarchical Bayesian Models ................................................................................. 11
Results............................................................................................................................... 16
Descriptive Statistics..................................................................................................... 16
GEE Models.................................................................................................................. 18
Single Exposure Models............................................................................................ 18
Dual-Exposure Models ............................................................................................. 19
Hierarchical Bayesian Models ..................................................................................... 20
Discussion......................................................................................................................... 23
GEE Model Discussion................................................................................................. 24
Hierarchical Bayesian Model Discussion .................................................................... 27
Strengths and Limitations ............................................................................................. 28
Conclusion .................................................................................................................... 32
Figures 1 - 8 ...................................................................................................................... 33
Tables 1 - 10...................................................................................................................... 42
Appendix........................................................................................................................... 55

About this 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
Department
Degree
Submission
Language
  • English
Research field
Keyword
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files