Uncertainty about components of PM10 and their toxicities and its effect on estimation of health damage

To guide public policy on regulation of emissions, it is desirable to know how much damage is caused by each source of pollution and by each pollutant. Concentration-response (C-R) functions, established by epidemiological studies, play a crucial role in the estimation of the damage per kg of emitted pollutant. Of particular importance are those based on particulate matter (PM2.5 and PM10). Unfortunately there is considerable uncertainty in the application of the C-R functions for particulate matter, due to a lack of detailed information about the relative toxicity of the different components of PM in ambient air. The purpose of the present paper is to examine the effect of this uncertainty on the estimation of health damage. As starting point we assume linearity of incremental health impact with incremental concentration, in view of the lack of epidemiological evidence for no-effect thresholds or significant deviations from linearity. We write the observed C-R function as a linear combination of(unknown) C-R functions for the main components of PM. We consider four components: primary combustion particles, nitrates, sulfates, and other (mostly of soil or marine origin). We make assumptions about plausible ranges for the slopes of the component C-R functions and about the composition of ambient air in the cities where the C-R functions were obtained. Varying the unknown parameters we evaluate how much the health impacts could change with different plausible assumptions. As an example we consider the damage cost per kWh for a modern coal fired power plant.