Identifying air pollution mixtures and investigating their associations with pediatric asthma in a time-series framework Open Access

Gass, Katherine (2014)

Permanent URL:


Every time we take a breath outside we inhale a mixture of different pollutants; however, examining the associations between these pollutant mixtures and health endpoints is challenging. This dissertation uses different methodological approaches to understand the associations between air pollution mixtures and emergency department visits for pediatric asthma. Time series with daily counts of emergency department visits for any diagnosis of asthma or wheeze among pediatric patients were obtained from hospitals in metropolitan Atlanta (1999-2010), Dallas (2006-09) and St. Louis (2001-07). Daily measurements of ambient concentrations of ozone, carbon monoxide, nitrogen dioxide, and particulate matter <2.5 microns in diameter (PM2.5) were obtained from monitors in all three metropolitan areas. In addition, daily estimates of PM2.5 source concentrations were made available for Atlanta using a Bayesian-based ensemble source apportionment technique.

A modified classification and regression tree algorithm was developed to enable the identification of multipollutant joint effects. This algorithm was then used to determine the multipollutant joint effects associated with pediatric asthma in Atlanta, as well as the common multipollutant joint effects identified in Atlanta, Dallas and St. Louis. These analyses found certain types of days, characterized by their multipollutant profiles, to be associated with a statistically significant increase in asthma emergency department visits.PM2.5 appeared to be one of the pollutants driving the formation of these harmful day types and thus further analyses were conducted to determine the associations between pediatric asthma and PM2.5 sources. A positive association was observed for the cumulative, seven-day effect a 1 microgram increase in biomass burnings (rate ratio: 1.02, 95% confidence interval: 1.01, 1.03), diesel vehicle emissions (rate ratio: 1.05, 95% confidence interval: 1.01, 1.08), and gasoline vehicle emissions (rate ratio: 1.07, 95% confidence interval: 1.03, 1.11). These confidence intervals account for uncertainties in the source apportionment estimates using multiple imputation methods.

This dissertation makes methodological contributions to the field of epidemiology with the development of a classification and regression algorithm that is well-suited for identifying joint effects of exposure mixtures. It also adds to the growing body of literature which suggests a harmful effect of multipollutant exposures on pediatric asthma.

Table of Contents

Chapter 1: Introduction 1

Chapter 2: Ambient Air Pollutants and Associated Health Effects 7

Chapter 3: Asthma and Air Pollution 25

Chapter 4: Epidemiologic Methods for Multipollutants 33

Chapter 5: Classification and Regression Trees for Epidemiologic Research (Study 1) 53

Chapter 6: A Three-City Analysis of Multipollutant Joint Effects: a comparison of classification and regression trees with conventional multipollutant models (Study 2) 79

Chapter 7: Ensemble-Based Source Apportionment of Fine Particulate Matter and Emergency Department Visits for Pediatric Asthma (Study 3) 112

Chapter 8: Conclusion 140

About this Dissertation

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.
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files