Diet-Heart Hypothesis

August 24, 1957

DIET AND THE EPIDEMIOLOGY OF CORONARY HEART DISEASE

Ancel Keys, Ph.D.

Author Affiliations

JAMA. 1957;164(17):1912-1919. doi:10.1001/jama.1957.62980170024007e

FullText

Abstract

Researches ranging from laboratory experiments to epidemiology on a world-wide basis have led to the hypothesis that the fat content of the habitual diets of populations has an important effect on the frequency of coronary heart disease in those populations. The general nature and some of the evidence for this hypothesis have been summarized several times in recent years,1 but progress in this field manages to outdate each new résumé by the time it reaches print. Not that the previously cited facts or the hypothesis itself have been found to be erroneous; on the contrary, each new research adds detail, reduces areas of uncertainty, and, so far, provides further reason to believe that along the line of this hypothesis we may hope to find effective measures to reduce the incidence of coronary heart disease.

A third famous clinical trial that is cited again and again is the Finnish Mental Hospital study. I first heard about this study from a top nutrition expert who assured me that it was really “the best possible proof” that saturated fat is unhealthy.

In 1958, researchers seeking to compare a traditional diet high in animal fats to a new one high in polyunsaturated fats selected two mental hospitals near Helsinki. One they called Hospital K and the other, Hospital N. For the first six years of the trial, inmates at Hospital N were fed a diet very high in vegetable fat. Ordinary milk was replaced with an emulsion of soybean oil in skim milk, and butter was replaced by a special margarine high in polyunsaturated fats. The vegetable oil content of the special diet was six times higher than in a normal diet. Meanwhile, inmates of Hospital K ate their regular fare. Then the hospitals swapped, and for the next six years, Hospital K inmates got the special diet while Hospital N returned to their normal one.

In the special-diet group, serum cholesterol went down by 12 percent to 18 percent, and “heart disease was halved.” This is how the study is remembered and is the conclusion that the study directors, Matti Miettinen and Osmo Turpeinen, themselves drew. In a population of middle-aged men, they said, a diet low in saturated fats “exerted a substantial preventive effect upon coronary heart disease.”

But a closer look reveals a different picture. Heart disease incidence (which the investigators defined as deaths plus heart attacks) did go down dramatically for the men at Hospital N: there were sixteen such cases among men on the normal diet compared to only four on the special diet. But the difference found in Hospital K was not significant. Nor was any difference observed among the women. The biggest problem with the study, however, was that, like the subjects in the LA Veterans Trial, its population was a moving target. With admissions and discharges over the years, the composition of the groups changed by half. A shifting population means that an inmate in the group who died of a heart attack might have been admitted three days earlier and the death would have had nothing to do with his diet; and, vice versa, a patient who was released might have died soon thereafter but would not have been recorded in the study.

This and other design problems were so great that two high-level NIH officials together with a professor at George Washington University felt moved to criticize the study in a letter to The Lancet asserting that the authors’ conclusions were too statistically weak to be used as any kind of evidence for the diet-heart hypothesis. Miettinen and Turpeinen acknowledged that their study design was “not ideal,” including the fact that the study population was far from stable, but asserted in their defense that a perfect trial would be “so elaborate and costly . . . [that it] may perhaps never be performed.” Their imperfect trial, meanwhile, would have to stand: “we do not see any reason to change or modify our conclusions,” they wrote. The research community accepted this “good-enough” reasoning, and the Finnish Mental Hospital study earned a spot as one of the linchpins of evidence for the diet-heart hypothesis.

Nina Teicholz - Page 77

 On behalf of the skeptics, nonbelievers in a high longevity among the pre-statistical Eskimos, I quote Dr. Ancel Keys, director of the Laboratory of Physiological Hygiene of the University of Minnesota. In a newspaper exchange with Mr. C. N. Pearson, Dr. Keys wrote on December 30, 1958:

“May I correct some errors in your letter to the Minneapolis Star dated November 23rd?

“First, you should know that extremely little is known about the health of primitive peoples, including the tiny remnant of primitive Eskimos. It is known, however, that their life expectancy is very short and that a primitive Eskimo above the age of 50 is a great rarity.

“Second, primitive Eskimos eat no beef, pork, lamb, or chicken and never have butter, milk, ice cream or cheese.”

The frontier doctors, quoted in this book as believing that the natives formerly had good health and kept it so long as they remained primitive, would surely agree with Dr. Keys that an Eskimo can no longer be called primitive if he habitually eats the foods here listed. Some Eskimos have been eating most of them for decades, if chiefly in canned form. To find Eskimos who had not been contaminated by civilized diet, we must go to one of two sources, the records of the Moravian Church in Labrador or the Russian Church in Alaska. I have consulted both, seeking statistically significant material bearing upon the divergent views of Dr. Greist and Dr. Keys.

In 1959, they jointly reported “to the nation” on “A Decade of Progress against Cardiovascular Disease.”

​

Dr Wilkins, Dr. Sprague, friends, since 1950, a number of new drugs have come lnto use for the treatment of high blood pressure and, considered together, they represent one of the great advances of medical science durIng the perlod of our report. These drugs have changed not only our treatment of hypertension, but also our concepts of the nature of this disease. Except In a few rarer forms. the cause of hlgh blood pressure is unknown. None of the new drugs used ln treatment, therefore, was designed to get at the cause. Rather, they were designed primarily to relieve the result; namely, the elevation of blood pressure, whatever its cause might be. However, the actlon of these drugs in lowering blood pressure has provlded new information on the nature of hypertension and has brought us closer to a true concept of the cause or causes of this condttion.

​

​

"Despite the limitations of the tools and procedures used in this study some provocative findings have been produced. There appears to be a real difference in risk of developing severe manifestations of CHD for farmers as contrasted with other occupations as a group. The explanation for this difference deserves more intensive study. Some comments should be made on the lack of apparent differences in recent dietary patterns of coronary cases in comparison with controls. This does not necessarily mean that dietary factors may not be important in the development of coronary heart disease. Mean dietary intake will have to be assessed in relation to height-weight-activity characteristics. Even this may not reveal differences between coronary cases and controls since the fat consumption in the population studied was surprisingly uniform. As is shown in Table 6, the calories from fat ranged only between 40 and 50 per cent in two-thirds of the men studied. Under such conditions, considering the potential inherent error in dietary interview procedures, it seems unlikely that any relationship between diet and CHD can be established. These dietary findings suggest the probable importance of factors other than diet in determining why, in populations on relatively high fat diets, some males develop CHD and others do not."