Misclassification of exposure is a persistent source of bias in epidemiologic studies of occupational cancer. This occurs due to the following reasons: (a) exposure reconstructions are typically based on sparse data with significant uncertainty; (b) biologically relevant doses are not estimated, e.g., for exposures to inhaled dusts, the cumulative lung burdens of the worker do not account for retention and clearance of inhaled dusts; exposures to multiple chemicals are not accounted for. The objective of the proposed research is to develop an improved exposure and dose assessment method for epidemiologic research on occupational cancer that accounts for the uncertainties in exposure reconstruction due to sparse data, determinants of etiologically relevant dose, and exposures to multiple chemicals, using a Bayesian probabilistic framework. The methodology will be developed and demonstrated using a large occupational exposure dataset (1950-2000), from Falcon bridge Ltd., Sudbury, Ontario, which is one of the world's leading primary nickel production companies. Workers here have historically been exposed to several nickel species (oxidic, sulfitic, and soluble), diesel particulate matter, and silica - all of which are either proven or suspected human carcinogens, specifically causing lung cancer. A novel Bayesian methodology that synthesizes expert judgment and historical measurements will be developed for exposure reconstruction. The exposure reconstruction will incorporate retention and clearance models for estimating the cumulative lung dose of oxidic, sulfitic, and soluble nickel species, diesel particulate matter, and airborne silica. The exposure reconstruction will then be available for a planned epidemiologic case-control study of lung cancer in this population. However, the methods will be applicable to other industry based epidemiologic studies where the available data are sparse and exposures to chemical mixtures are the norm.
