Date of Award

January 2015

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)

Department

School of Public Health

First Advisor

Susan Busch

Second Advisor

Vasilis Vasiliou

Abstract

Rationale: Fine particulate matter (<2.5 um), or PM2.5, and Ozone has been linked to a number of respiratory and cardiovascular conditions and is a known trigger for acute events. Though previous studies have addressed these correlations, few have examined PM2.5’s acute effects on inpatient admissions and health care costs, by utilizing narrower time intervals than other research projects.

Objectives: To identify whether trends in hospitalization spending in short time-intervals is associated with PM2.5 measures, and to create a predictive model for spending based on two major categories of outcomes: selected CV conditions and respiratory conditions known to be associated with PM2.5 exposure that we will identify by Medicaid charge codes. We also attempt to model inpatient admissions altogether as an alternative outcome.

Methods: We link Medicaid charge information for all procedures in Texas to daily air-quality data sourced from 63 EPA sites in the state (providing more comprehensive geographical and temporal coverage) and fit a longitudinal mixed model to extrapolate costs and risks of additional inpatient stays due to respiratory conditions from particulate matter AQI readings. Outcomes are identified by APR-DRG codes listed in the Blue Ribbon Medicaid set and exposure measurements are sourced from the EPA’s monitoring stations’ data mart. Our study covered September 2010 to August 2011. We also adjust for other potential covariates and exposures like Ozone in latter models to help explain some of the variation in outcomes.

Measurements and Main Results: We find positive association between environmental PM2.5 and healthcare spending. Our simpler multilevel linear model gave us these cost estimates: rates of Respiratory and CV charges to Medicaid increase $3.95 million dollars to each increase of ug/m^3 in PM2.5, under our simplest per capita model. After controlling for other variables in a secondary-model we find that ozone is the more significant predictor of costs/charges, and one of our models produced an annual $600,000 increase per additional 0.01 ppm of exposure over the year. We also find that ozone is a better predictor of count data and is more statistically significant when fixed time covariates are also entered into our predictive model.

Conclusion: This suggests additional savings from instituting more restrictive clear air regulations in terms of ozone and pm2.5. In addition, this creates additional burden for local medical centers. These measurements can be utilized to predict morbidities based off of air quality, as well as estimate the impact to Medicaid in terms of its financial strain. This information can be utilized in resource allocation in terms of hospital staffing and local medical needs, as well as on larger scale, aiding policy decisions. As well in a subset of our models that include other airborne contaminants and fixed time covariates, we find that ozone is more significant of a predictor.

Comments

This is an Open Access Thesis.

Share

COinS