Epidemiological assessment of hormesis in studies with low-level exposure

Despite its resurgence within toxicology and, specifically, risk assessment , the concept of hormesis remains peripheral to current epidemiological pra ctice. In this paper we examine some reasons for this, focusing on applicat ions within occupational and environmental epidemiology. Unclear in the exi sting literature is whether hormesis pertains to a single biological mechan ism or response, or the aggregate effect of all correlates of exposure. Alt hough J-shaped and U-shaped relationships between risk factors and disease endpoints have been identified epidemiologically, it is unclear whether suc h patterns reflect biological hormesis or a combination of factors resultin g in a hormetic-looking relationship. Given the potential importance of ass essing hormetic responses in epidemiological studies, we identify and discu ss key limitations of epidemiology in validly detecting and interpreting ho rmesis. For example, most observational occupational and environmental stud ies lack the ability to determine the dose received by each individual, and therefore poor surrogates of exposure are frequently used, potentially int roducing considerable systematic and random error. Further, because exposur e is not randomly assigned to humans, the potential for confounding is grea t. Finally, using a simple simulation to assess the impact of ignoring horm esis in the analysis of epidemiological data containing mild hormesis, we d emonstrate a resulting "hormetic bias," in which relative risks at exposure levels above the hormetic region are systematically overestimated.