A COMPUTATIONAL MODEL FOR ESTIMATING PERSONAL EXPOSURE TO AIR POLLUTANTS FOR USE IN STUDIES EXAMINING THE RELATIONSHIP BETWEEN AIR POLLUTION AND HEALTH: ISEE-495

Abstract

This paper describes a model framework for estimating the exposure to pollutants for a randomly selected member of a designated population. Such exposures can then be used in studies of the effects on health. The model makes intrinsic use of the random time-activity patterns of such individuals that have an important role in determining variations in exposures. Suitably implemented, the methodology can answer questions such as the following; (i) what fraction of the sustained ‘high’ levels of exposure, (ii) how many sustained such exposures for say 10 days in a row? (iii) Considering new air quality criteria, what impact will they have on residents under the age of 4? Over the age of 65? The framework is a foundation for the assessment of human risk of exposure to hazardous environmental contaminants. It abstracts the building blocks, along with their linkages, stochastic and structural for risk assessment. We consider the practical considerations in the implementation of the framework, including human time-activity patterns and their relationship to environmental factors that help determine such patterns. For example, in summer people may tend to stay indoors in warm weather, while in winter the reverse may be true. Thus, in summer days indoor sources of the hazardous substance may be more important in determining exposures than outdoor sources. The implementation of the above framework is referred to here as ‘pCNEM’ and can be accessed by registered users through the http://WWW.Such users are able to designate a pollutant of interest, a study area, and a study period. They can then develop a personal exposure model for that pollutant by measures such as creating local sources, setting parameters and uploading the requisite data. The model is demonstrated using by predicting the exposure to PM10 of random selected individuals from subpopulations of Greater London in 1997. These modelled exposures are compared to those which would be available using routinely monitored levels and in particular, the effect this will have on subsequent health studies is discussed. Another output of the model is an estimate of the uncertainty associated with the exposures, this measure of uncertainty is particularly useful for accounting for the variation in the pollution level, whether informally, when interpreting the regression coefficients (of the health model), or more formally via error-in-variables modelling.