Including uncertainty in assessments of sulfur oxide health risks.

Abstract

Methods for the incorporation of uncertainty in quantitative analysis are needed in risk assessment applications, such as the problem of estimating health risks from coal-fired power plants. Techniques including elicitation of subjective expert judgment about uncertainty and stochastic simulation modeling are combined in a demonstration analysis. Probabilistic estimates of population exposure to sulfur air pollution from a hypothetical new power plant are generated for two locations in the Ohio River Valley. Models of health responses to air pollution, obtained through elicitation of the judgment of seven leading health scientists, are applied to these exposure estimates, and uncertainty about the level of health impacts is predicted and compared. The predictions range from a significant probability of zero health effects to a small probability of effects on the order of 20 percent of the total mortality. Uncertainty about the adverse effects of sulfur air pollution on human health is far greater than the scientific uncertainty about the atmospheric processes which generate and transport it. These techniques have the potential to improve our understanding and ability to communicate about scientific uncertainty about risk, and may be useful for the analysis of the benefits of sulfur air pollution control.