A Pilot Study of the Feasibility of using Satellite Aerosol Optical Depth Estimates of PM2.5 as a Predictor of Acute Myocardial Infarctions and Asthma Attacks in Baltimore, Maryland during 2005
AMY HUFF (1), John T. Braggio (2), Stephanie Weber (1), Fred Dimmick (3), Rashid Malik (2), Raymond M. Hoff (4)
(1) Battelle Memorial Institute (2) Maryland Department of Health and Mental Health (3) EPA Office of Air Quality Planning and Standards (4) University of Maryland, Baltimore County
Abstract Number: 333
Preference: Poster Presentation
Last modified: November 9, 2009
Working Group: sq2
Analyses of the health effects of PM2.5 typically use data from the Environmental Protection Agency’s national PM2.5 ground-based monitor network. These monitors are located primarily in urban areas, and significant gaps in the monitor network exist nationwide. As a result, many observational studies which utilize monitor-based PM2.5 measurements and public health data are limited because of too few monitors. To address this deficiency, the authors have developed a methodology to combine PM2.5 monitor data with measurements of aerosol optical depth (AOD) from NASA’s Terra and Aqua satellites. The advantage of satellite AOD is that it provides regional information about particulate concentrations that can be used to fill in the gaps of the PM2.5 monitor network.
In this pilot study, the feasibility of using combined AOD-PM2.5 measurements as a predictor of acute MI and asthma attacks in Baltimore, Maryland during 2005 was explored using case-crossover and conditional logistic regression analyses. 2005 AOD data were converted to estimated PM2.5 concentrations using season-specific linear regression parameters for the Baltimore region. These AOD-estimated PM2.5 values were combined with PM2.5 measurements from the monitor network using a hierarchical Bayesian modeling (HBM) tool developed by EPA. The PM2.5 data were compared to daily acute MI inpatient hospitalizations and asthma Emergency Department (ED) visit information that was obtained from the 2005 Maryland Health Services Cost Review Commission (HSCRC) hospitalization and ambulatory care files. The Case-Crossover Analysis Tool (C-CAT), Beta Version 1.1 (Apex Epidemiology Research, Baltimore, MD) was used to conduct the analyses. The case-crossover analyses utilized the three available PM2.5 data sets: AOD-estimated PM2.5, PM2.5 monitor measurements, and the HBM AOD-PM2.5 surface. Each data source included 1-3 lag days, and all analyses were controlled for temperature and relative humidity.
The results show that the correlations between the PM2.5 data sets and the health data vary by race and season. Results are presented as the odds that each 1 microgram per cubic meter increase in PM2.5 resulted in a change in acute MI or ED asthma visits. For acute MI, all three PM2.5 data sources showed the same temporal trend for the second quarter, April through June, with the fourth quarter as the referent, and on lag day one. The AOD-estimated PM2.5 data showed an increase in the odds of Other Race persons of 3.4%, compared to Whites. Overall, the HBM AOD-PM2.5 surface showed a 4.8% decrease in the odds. In follow-up monthly analyses, with October as the comparison, the Odds Ratios were significant (p <0.05) and greater than 1 in January, March, June-August, and December for the HBM AOD-PM2.5 surface and AOD-estimated PM2.5 data sets. For ED asthma visits, the odds increased in the third quarter for all three data sets, but on lag day three, the odds were 2.6% lower for PM2.5 monitor measurements. Results for monthly analyses showed that in November, the odds increased for all three data sources. These results of the pilot study support the feasibility of using HBM AOD-estimated PM2.5 values as a predictor for acute MI inpatient hospitalizations and ED visits.