Concentration and Time As Components of the Exposure Metric in Occupational Epidemiology. Noah Seixas, Box 357234, University of Washington, Seattle, WA 98195-7234
Causal inference in occupational epidemiology relies on the effective definition of an exposure metric and its use in dose-response analyses. In studies of chronic disease, cumulative exposure is generally accepted as an appropriate surrogate for true dose. Cumulative exposure, however, assumes a linear effect of both time and concentration in predicting outcome which may not be an appropriate assumption for many exposure-disease relations. Several approaches to addressing this limitation have been attempted including latency or exposure window analysis, the construction and use of multiple metrics, simple and complex toxicokinetic dose modeling. Each of these approaches, and their attendant limitations are presented. A more flexible approach is presented in which the exposure and disease data are modeled, simultaneously estimating the form of the concentration and time contributions to the dose metric, and the dose-response relationship. This approach is demonstrated in a study of obstructive lung disease among coal miners. The difficulties of specifying a correct dose metric in epidemiologic studies in which the data are observational, many unmeasured risk factors may be present, and exposure data are severely limited are explored in both the coal miners study and in a study of silicosis in the diatomaceous earth industry.