Diet, Life Expectancy, and Chronic Disease: Studies of Seventh-Day Adventists and Other Vegetarians

Cancers that occur in different organs of the body and in different tissues are not one disease, but many different diseases. Although they share some features, they also exhibit differences. This complicates the epidemiologic picture since different cancers have different sets of causal factors, although some may be important for several cancers.

Cancers as a group are common, but cancer of a particular organ is much less so. Even the more frequent cancers, such as breast, colon, and prostate, are less common than cardiovascular disease. Thus statistical power is often a problem in prospective studies because it is difficult to find sufficient numbers of new cases of a particular cancer.

Until recently, cancer epidemiology relied mainly on retrospective casecontrol studies, which could identify large numbers of cancers at a specific site much more efficiently than prospective studies. But unfortunately, retrospective studies may not be so well suited to dietary epidemiology. The cases in a case-control study are those that already have cancer. These persons may well have altered their diet as a result of the disease and, further, may not be able to accurately recall what they ate before developing cancer. Nonetheless, given the focus of this book, it is worth mentioning a meta-analysis that combined a number of case-control studies in Italy (Tavani et al., 2000) and showed significantly increased risks in those eating more meat for cancers of the stomach, colon, rectum, pancreas, bladder, breast, uterus, and ovaries.

A person's accurate dietary recall must be the goal in this kind of preventive research, as we need to identify dietary habits that make a difference before cancer is diagnosed. The measurement error problem that pervades nutritional epidemiology is pertinent to studies of cancer and is potentially worsened when retrospective data are used. Thus, the Adventist studies described here are all prospective ones and, where possible,

-85-