Risk estimation for carcinogens based on epidemiological data: a structured approach, illustrated by an example on chromium.

Abstract

It is generally recognized that human, epidemiological data, if available, are preferred as the starting point for quantitative risk analysis above the use of data from animal studies. Although methods to obtain proper risk estimates from epidemiological data are available, several impediments prevent their widespread application. These impediments include unfamiliarity with epidemiological methods and the lack of a structured and transparent approach. We described a framework to conduct quantitative cancer risk assessment based on epidemiological studies in a structured, transparent, and reproducible manner. Important features of the process include a weight-of-the-evidence approach, estimation of the optimal exposure-risk function by fitting a regression model to the epidemiological data, estimation of uncertainty introduced by potential biases and missing information in the epidemiological studies, and calculation of excess lifetime risk through a life table to take into account competing risks. Sensitivity analyses are a useful tool to obtain insight into the impact of assumptions made and the variability of the underlying data. The framework is sufficiently flexible to allow many types of data, ranging from published, sometimes incomplete data to detailed individual data, while maintaining an optimal result, i.e., a state-of-the-art risk estimate with confidence intervals, based on all available evidence of sufficient quality.