Dealing with the exposure variable in occupational and environmental epidemiology.

Whereas the outcome variable in epidemiologic research is usually well-defined, the exposure is more diffuse, complex and variable, involving components of both intensity and duration. Usually, the time integral of the intensity has been used as a measure of exposure, e.g. working level months, fibre-years, etc. Toxicological models have suggested, however, that greater importance should be accorded to early exposure in cancer epidemiology, while the recent exposure may be more important as regards the development of some other types of disorders. A simplified approach in cancer epidemiology would be to consider the exposure within a time-window or otherwise allow for induction-latency time. For the practical assessment of exposure, so-called job exposure matrices might be of value, especially if recall bias is suspected, but on the other hand, there is rather little objective indication in the literature that serious recall bias is a major problem in case-referent studies.