ABSTRACT
Epidemiological methods are, by necessity, different from those adopted in the confines of the laboratory or in carefully controlled clinical trials. Epidemiological data are dirty; samples are rarely complete, independent variables are difficult to control, confounding factors may abound.
