Studies of cancer in migrant populations: methods and interpretation.

The use of descriptive epidemiological data (collected without a view to investigating any specific hypothesis) to study the risk of cancer in populations which have migrated from one country to another is reviewed. Place of birth is treated as the risk factor under study in such analyses, although they vary considerably in complexity depending on the availability of information on other explanatory or confounding variables. The underlying assumption of these studies is that migrants undergo a change in their environment (although the extent of this is rarely documented), so that differences in cancer rates confirm the importance of environmental over genetic determinants of risk. Studies which document risk according to time spent in the new country or to age at migration, or differences between migrants and their offspring, add an extra dimension, interpretable in terms of the degree of lifestyle change, or lability of the cancer to changes in exposure to the relevant determinant. Past studies have frequently used rather simplistic methodology, with insufficient attention to the presence of bias, and a reluctance to use standard epidemiological techniques to control for obvious sources of confounding. Migrant studies are divided into four broad categories, depending on the number of comparison groups (two, or more), the availability of a time dimension, and information on exposures; examples of each are described.