al., 1991). These authors conclude that improvements in exposure assessment are necessary if epidemiologic investigations are to provide reliable information on the relations between cancer incidence and pesticide exposure. These studies are noteworthy in their attempt to identify problems of reliability and validity of exposure assessments in case-control studies of cancer and pesticide exposure. For example, Brown et al. (1991) compared interview data from farmers with data from their wives or other surrogates and found excellent agreement between direct and surrogate interviews regarding the use of specific pesticides.
Blair and Zahm (1990a) concluded their study by stating, "Exposure misclassification undoubtedly occurs. Most errors from misclassification, however, are likely to be nondirectional in nature and would bias risk estimates toward the null and dilute exposure-response relationships. Methodologic investigations are needed to evaluate the reliability of current exposure assessment procedures and to develop new resources to improve assessment techniques." These recommendations are valuable, but it seems more likely that new resources are needed to enhance the best use of existing but underutilized techniques. These recommendations further illustrate the need for greater interaction between epidemiologists who recognize the importance of exposure assessment and investigators whose area of emphasis is exposure determination.
For more than 4 decades, researchers have studied the impact of air pollution, most notably respirable particles and other priority pollutants, on human health. Carefully designed studies have provided a wealth of important information about relations between ambient exposures and adverse health consequences. A brief review of some of these studies illustrates the importance of microenvironmental monitoring and indirect measures of exposure characterization.
Studies by Abbey et al. (1991), Euler et al. (1987, 1988), and Mills et al. (1991) illustrate the benefit of ambient-air monitoring in epidemiologic characterization of health-related effects of air pollution. Rather than relying on the mean concentration of ambient air pollutants only, these authors evaluated the numbers of hours that ambient air pollutants exceeded thresholds. For each participant in the studies, the levels of air pollutants at ZIP-code centroids were estimated for each month of residence (Abbey et al., 1989).
Neas et al. (1991) related annual average nitrogen dioxide concentration, measured for indoor pollution as a continuous variable and as 4 ordered categories, to respiratory symptoms and pulmonary function in children. A 15-ppb increase in the mean concentration was associated