Weak associations in environmental epidemiology.

Several suggestions have been made to enhance study designs to strengthen weak associations. Among these are reduction of non-differential misclassification, and restriction of studies to populations with low underlying risk of disease, due to low exposures to other known determinants of the health effect under investigation. To some extent, these suggestions have been put to use in environmental epidemiology, and in this paper, some prospects and problems associated with these attempts are discussed using examples from our own research.