Stories from the evolution of guidelines for causal inference in epidemiologic associations: 1953-1965

Abstract

Guidelines for causal inference in epidemiologic associations were a major contribution to modern epidemiologic analysis in the 1960s. This story recounts dramatic elements in a series of exchanges leading to their formulation and effective use in the 1964 Report of the Advisory Committee to the US Surgeon General on Smoking and Health, the landmark report which concluded that cigarette smoking caused lung cancer. The opening salvo was precipitated by Ancel Keys' presentation of an ecologic correlation between diet and cardiac death, which was vigorously criticized in an article by Jacob Yerushalmy calling for "proper handling" of bias and confounding in observational evidence. The dispute demonstrated a need for guidelines for causal inference and set off their serial refinement among US thinkers. Less well documented parallel efforts went on in the United Kingdom, leading to the criteria that Bradford Hill presented in his 1965 President's Address to the Royal Society of Medicine. Here the authors recount experiences with some of the principals involved in development of these criteria and note the omission from both classic reports of proper attribution to those who helped create the guidelines. They also present direct, if unsatisfying, evidence about those particular lapses.

Figure 1.

Mortality from degenerative heart disease according to dietary fat (fat calories as a percentage of total calories) among men aged 45–49 years (lower curve) and men aged 55–59 years (upper curve), 1948–1949. Results were calculated from national food balance data for 1949 supplied by the Nutrition Division of the Food and Agriculture Organization of the United Nations. (Reproduced with permission from the Journal of Mount Sinai Hospital (8).)