Vegetarianism and Obesity,
The cohort studies of California Adventists were designed to detect new cases of cancer and coronary heart disease during the follow-up period. Thus, there was no plan for identifying new changes in body weight or new cases of hypertension, diabetes, or arthritis as study endpoints. At the beginning of the studies, however, subjects were required to answer questions about their height and weight and to say whether a doctor had ever diagnosed them as hypertensive, diabetic, or arthritic. This provided a measure of the prevalence of particular values of body mass index (BMI, a measure of obesity) and of the three other listed medical conditions at the beginning of the Adventist Mortality Study (AMS) in 1960, and prior to the Adventist Health Study (AHS) in 1976.
Epidemiologists in recent years, when searching for causal variables, have quite rightly tended not to favor studies reporting associations between disease prevalence and risk factors. The important difference between prevalence and incidence (prospective) studies is that the disease of interest is already present at the beginning of a prevalence study. By contrast, at the beginning of an incidence study the disease of interest is not present and new cases are detected during a follow-up period. The reasons for preferring incidence studies are that (1) prevalence studies tend to include a smaller proportion of subjects with the more severe and perhaps fatal forms of the disease than actually occurs in a population during the follow-up—this is known as length bias (Jekel et al., 1996); and (2) the possible risk factor being investigated (e.g., diet) and the presence of disease are both measured at the same time. Thus, it is possible that the presence of the disease had led subjects to change either their dietary habits or their perception of their previous diet. If so, the reported diet would not