Ecological versus case-control studies for testing a linear-no threshold dose-response relationship.

The two basic problems with ecological studies are (A) individuals studied are not necessarily the individuals who are at risk, and (B) they are very vulnerable to confounding factors. It is shown that where the study is designed to test a linear-no threshold dose-response theory, (A) does not apply. Where the ecological study deals with the average dose and response in a large number of US counties, the available data and computer capability for reducing effects of confounders are so powerful that (B) may be no more important for the ecological than for a case-control study. The migration problem is treated and found to be relatively unimportant.