Diet-heart disease hypothesis is wishful thinkingBMJ 2002; 324 doi: https://doi.org/10.1136/bmj.324.7331.238 (Published 26 January 2002) Cite this as: BMJ 2002;324:238
EDITOR—Mann et al and Hu et al list several shortcomings in the trials reviewed by Hooper et al that explain why dietary treatments for patients with coronary heart disease were ineffective.1–3 Inferior studies with negative results are prevalent, but where is the positive evidence that justifies the dietary recommendations?
Mann et al point to the improved outcome in the subgroup analysis of the five prolonged trials. But in that analysis Hooper et al excluded the Sydney diet-heart study, where total mortality was significantly increased, and included the Veterans Administration Trial, which was biased by a significant higher number of heavy smokers in the control group.
Mann et al also say that there is an enormous body of descriptive epidemiology that supports the link between dietary fat, cholesterol concentrations, and coronary heart disease. The accumulated epidemiology actually strongly contradicts such a link, as illustrated by a systematic review.4 In a study of Japanese migrants in the United States the cultural upbringing was the strongest predictor of coronary heart disease. Those who were brought up in a non-Japanese fashion but preferred the lean Japanese food had a heart attack almost twice as often as those who were brought up in the Japanese way but preferred fatty American food.4
Masai people probably have the highest intake of animal fat in the world, but abnormalities on electrocardiography were far less frequent than in Americans and raised atherosclerotic lesions were rare.4 Mortality from coronary heart disease in southern India was seven times higher than in the north and the age at death 44 years compared with 52, although people in the north ate 19 times more fat, mostly animal fat, and also smoked much more.4
In 30 of 103 time periods in 33 countries fat intake increased along with coronary mortality, but in 33 periods where the intake also increased, coronary mortality was unchanged in 10 and decreased in 23.4 In six case-control studies the diet of the coronary patients did not differ appreciably from that of the controls.4 In 21 cohort studies including more than 150 000 participants with and without coronary heart disease no study found an eating pattern in accordance with the current view on the influence of dietary fat.4 Hu et al have published several studies with similar findings.
But just as is the case with the analysis by Hooper et al, these findings and many other contradictions are always explained away with more or less valid arguments. As Karl Popper would have said: the diet-heart disease hypothesis is unfalsifiable and should therefore be classified as non-science.