Keynote lecture. Etiologic insights from cancer mapping.

In the 1970s the epidemiology program at the U.S. National Cancer Institute made a systematic effort to identify cancer clustering by analyzing patterns of mortality at the county level, where the population is small enough to be relatively homogeneous, yet large enough to provide reliable data and stable rates. When the mortality rates for the period 1950-69 were plotted in a series of computer-generated color-coded maps, there arose a surprising number and variety of geographic patterns. This review describes how leads to the causes of several cancers have been generated and explored through a progression of descriptive and correlational studies, followed by analytical studies to determine reasons for the elevated risks in certain areas of the country. For example, the high lung cancer rates among men in some coastal areas were related mainly to asbestos exposures in shipyard work, while the elevated oral cancer rates among women in the rural south were linked to the use of smokeless tobacco (snuff). A recent update of cancer maps covering the period 1970-80 revealed geographic patterns resembling those in the earlier atlas, but with a tendency toward greater uniformity of rates around the country. Yet some new high-risk areas emerged, such as elevated rates of lung and oral cancers among women in Florida and along the Pacific coast, which seemed correlated with smoking habits, and high rates of non-Hodgkin's lymphoma among men in central areas that may be associated with agricultural exposure to herbicides. Our experience suggests that cancer mapping on a small-area scale is a useful strategy for formulating and pursuing leads to environmental and lifestyle determinants of cancer. In other countries also, the mapping approach has helped to stimulate and target research into the origins of cancer and the means of prevention.