The overall objective of this project, conducted by the University of Pittsburgh with support from Consol Energy, Ohio University, and the Allegheny County (PA) Health Department, is to develop a strategy to better define the public health implications in the Pittsburgh region of particulate matter emissions from coal-fired power plants.  Retrospective epidemiological study methods are being applied to speciated PM2.5 information, other emissions data, and other plausible explanatory factors.

In the process of achieving this overall objective, a succession of intermediate objectives will also be achieved, including:

• Assess the location, quantity, and quality of available speciated PM2.5 air monitoring data (including archived data) in the Pittsburgh region from 1999 to 2004;
•  Develop a plan including technical and economic evaluations for the construction of an air monitoring database for use in a retrospective epidemiology  study;
• Assess the quality and quantity of available mortality and morbidity data (cardiopulmonary hospital admissions, emergency room visits, physician visits, etc.) in the Pittsburgh region from 1999 to 2004;
• Develop a plan including technical and economic evaluations for the construction of a health outcomes database for use in conducting this retrospective epidemiology study;
• Investigate and/or develop methods to apportion PM2.5 constituents to sources;
• Employ the use of a geographic information system (GIS) to spatially model exposure due to traffic density and proximity to roadways;
• Explore, evaluate, compare, and validate the use of various statistical modeling techniques to help control for potential confounding of any coal-fired power plant PM2.5-related constituents and morbidity/mortality association by factors such as other sources of PM2.5 constituents, co-pollutants, weather, traffic density, and proximity to roadways;
• Determine whether a successful retrospective (1999-2004) epidemiology study modeling specific health outcomes as a function of source-related PM2.5 constituents and other confounding factors in the Pittsburgh region is feasible.

The project involves three distinct but interdependent main tasks:

• Task 1 - Inventory and assessment of air monitoring data. A plan will be developed by completing a series of subtasks during the first 6 to 7 months of the performance period.  The subtasks include inventorying existing air monitoring data, assessing the quality of these data, determining whether additional data can be acquired, and assessing the feasibility of using the data in a retrospective epidemiological study.
• Task 2 - Inventory and assessment of health impact data.  A series of subtasks will be completed during the first 6 months of the performance period in order to evaluate the feasibility of a retrospective epidemiological study and to develop a broad plan for the most efficient study design. The subtasks include inventorying existing health outcomes databases, assessing the quality of the available data, investigating whether additional data can be acquired through existing retrospective records and cohorts, and deriving an overall assessment of the utility of the data for use in a retrospective epidemiological study.
• Task 3 - Comprehensive evaluation and selection of possible methods for the statistical analysis of the data. Concurrently with Tasks 1 and 2, a plan will be developed to use GIS and exploratory data analysis techniques to provide a comprehensive overview of the region, to investigate various possible methods for dealing with multiple monitoring sites and for spatial/temporal aggregation of emission data and other explanatory factors to avoid exposure misclassification, to provide source apportionment of PM2.5 emissions, and to provide effects modeling and estimation based on available data from the Pittsburgh region and other regions.
  
  The project team also intends to investigate the use of local retrospective cohorts assembled previously for various national and local research studies such as the Cardiovascular Health Study, Health ABC, Women's Health Initiative, etc. or constructed from data gathered from health maintenance organizations (HMOs) and area physicians for the evaluation of the longer-term (months and longer vs. weeks and less) effects of air pollution on human health endpoints. This requires investigation of methods useful for constructing a plausible model for observed effects as a function of emissions having controlled for as many potential confounding factors as possible, including factors concerning the spatial locations and temporal availability of monitoring data for short term effects. Given the observational nature of retrospective investigations, it is impossible to control for all potential confounding factors but the credibility of the assessment will be in proportion to handling appropriately as many plausible confounders as possible using the existing database. The complexity of the possible methods of analysis is at least partly due to restrictions engendered by a retrospective and observational study.