water-consumption records have been used to estimate past exposures (Lagakos et al., 1986; Wrensch et al., 1990 a,b; Whorton et al., 1988). Mathematical modeling, toxicokinetic models, and biomarkers may improve our ability to estimate past exposures, especially where the body burden of xenobiotic chemicals is related to toxic insult.

In occupational epidemiology, it is sometimes possible to reconstruct an industrial environment for exposure estimation; exposure-monitoring records of workers may be available, and even employment records by job category may be of value. (See Applied Occupational and Environmental Hygiene, June 1991). Checkoway and colleagues (1991) have discussed methods to assess or rectify misclassification of historical exposures in occupational epidemiology (table 3-3). This work examines some indirect methods for determining possible bias in estimates of the health effects that result from nondifferential misclassification of exposure and considers misclassification of confounders. Checkoway et al. (1991) suggest that such direct approaches as simulating past exposures are often infeasible. They prefer indirect approaches for evaluating the effects of exposure and confounder misclassification.

Waste sites may expose persons to multiple chemicals (NRC, 1991a). Three important issues arise in studying exposures to complex mixtures in epidemiologic investigations: (1) quantification of exposure, (2) characterization of potential combined or interactive effects associated with exposure to multiple chemicals, and (3) identification of subpopulations that are especially sensitive to exposure from certain complex mixtures. These problems are multiplied where the mixture may vary from one site or time to another or when the mixture is not well characterized. Both problems are common in waste sites. An entire issue of the journal Toxicology (volume 105, 1995) is devoted to an examination of chemical mixtures.

Workplace studies often focus on exposure to a single chemical agent, e.g., lead in battery manufacturing or silica in a foundry. However, for the most part, epidemiologic studies of the environment must address the issue of complex mixtures. NRC (1988) has reviewed the issues and approaches to assessing the health impact of complex mixtures. One approach involves characterization of complex mixtures by toxicologic investigation. To date, toxicologic characterization of complex mixtures has received inadequate attention as a complement to environmental investigation. The advantage of toxicologic investigation is that exposure can be set by the investigator (Ozonoff and Wartenberg, 1991). Toxicologic studies are needed to characterize both mechanisms and interactive effects and to quantitate the exposure-effect relations of complex mixtures.