A method for identifying exposure-response models for the acute respiratory effects of inhaled irritants in humans: Effects of concentration, minute ventilation, and duration

In order to conduct quantitative risk assessment with minimal uncertainty f or shortterm exposure to ozone and other respiratory irritants, one must id entify exposure-response (E-R) models that accurately predict the distribut ion of the magnitudes of response (or the proportion of individuals experie ncing a given response) across the range of ambient exposure scenarios and personal characteristics of interest. The purpose of this manuscript is to describe a method for identification of such models and to demonstrate the method and evaluate the resulting model's performance using ozone as a repr esentative respiratory irritant. This approach to model identification util izes a general knowledge of mechanistic information and E-R characteristics for the specific pollutant and traditional statistical techniques for esti mating model coefficients and prediction uncertainty. For ozone we identifi ed a two-compartment model consistent with known mechanisms and E-R charact eristics. Using data from 485 volunteers exposed for two hours to one of si x ozone concentrations at one of three activity levels, we estimated model coefficients and evaluated the predictive ability of the model for lung fun ction and symptom responses. We found that the model accurately predicted r esponses as a function of ozone concentration, minute ventilation during ex posure, and duration of exposure and concluded that this approach to model identification was successful for ozone, and we hypothesized that a similar approach would be useful for other respiratory irritants.