Estimation of historical annual PM2.5 exposures for health effects assessment

Epidemiological studies have generally found fine particle metrics such as PM2.5 (PM mass less than 2.5 mun in aerodynamic diameter) to be more strongly related to adverse health effects than PM metrics that are not size-fractionated, such as total suspended particulate matter (TSP). The latency of long-term PM exposure effects on health could potentially be investigated using the American Cancer Society Cancer Prevention Study 11 cohort and other nationwide cohorts. Unfortunately, historical PM2.5 data are not available for many past years in most of the US. With the recent introduction of a PM2.5 National Ambient Air Quality Standard (NAAQS), fine particulate data is now available through the Environmental Protection Agency's (EAP's) Aerometric Information Retrieval System (AIRS) database from 1999 onwards. Using this nationwide PM2.5 data, we have estimated ratios Of PM2.5-PM10 (PM mass less than 10 mum in aerodynamic diameter) for more than 100 Metropolitan Statistical Areas (MSAs) in the US Similarly, using TSP and PM10 data from the late 1980's, when both metric were measured, we have derived PM10/TSP ratios for hundreds of US MSAs. These MSA-specific PM ratios allow the estimation of historical annual fine particulate concentrations, for as far back as 1972, using available annual TSP or PM 10 data. We found mean ratios Of PM2.5/ PM10= 0.54+/-0.14, and PM2.5/TSP = 0.30+/-0.11. The Inhalable Particle Network (IPN), a database independent of the AIRS database, monitored TSP and PM2.5 between 1979-1982. Using a subset of MSAs common to both databases, this dataset has been used to test our hypothesis that MSA-specific mass ratios could be used to estimate PM2.5 from PM10 and TSP. Raw IPN TSP-PM2.5 concentration correlations for MSAs were non-significant (R-2 = 0.00). Using the IPN TSP and our PM2.5/TSP ratios, mean PM2.5 estimates for 26 MSA were found to correlate with the measured IPN PM2.5 at R-2 = 0.43. These results indicate that it is possible to use MSA-specific PM mass ratios to predict historical annual mean PM2.5 exposure levels from past TSP and PM10 measurements. In addition, the MSA-specific ratios were used to estimate nationwide PM2.5 concentrations for 1972-2000. These results indicate that considerable progress has been made in reducing U.S. PM2.5 levels over the 30 years. (C) 2004 Elsevier Ltd. All rights reserved.