Diet-Heart Debate Turned on Its Head: New Interpretation of Old Data Questions the Harm of Saturated Fat

Addressing a debate that has been dragging on for more than half a century, a new interpretation of old data could potentially swing the pendulum away from the commonly held belief that replacing saturated fats with polyunsaturated fats improves heart disease outcomes.

Termed by researchers in the 1950s as the “diet-heart hypothesis,” several observational studies supported the idea that not only do various fats affect the body differently but also that diets low in fat could impart greater heart health. Healthcare professionals and policy makers have grappled since then with the existing data—published studies showing a positive link were observational and incomplete in nature—and, last year, removed dietary cholesterol and total fat as important risk factors for heart disease from US dietary guidelines.

The Take Home. Diet-Heart Debate Turned on Its Head: New Interpretation of Old Data Questions the Harm of Saturated Fat

Now, researchers led by Christopher Ramsden, MD (National Institutes of Health, Bethesda, MD), have recovered previously unpublished data from the Minnesota Coronary Experiment, a randomized trial of 2,355 subjects designed to look at the effects of replacing saturated fat with vegetable oil rich in linoleic acid. In a study published this week in the BMJ, they show that replacing saturated fats with corn oil and corn-based margarine for at least 1 year does reduce serum cholesterol (mean change from baseline -13.8% vs -1.0%; P < 0.001).

But the investigators also found that this diet change does not lower the risks of coronary atherosclerosis or MI. Additionally, those on the study diet had no better survival than controls, and in fact, patients older than 65 actually had a 22% higher chance of dying on the study diet for every 30 mg/dL reduction in serum cholesterol (P < 0.001).

In an additional meta-analysis of 10,808 patients from five randomized trials that compared the same fats, again mean serum cholesterol was lowered by replacing saturated fats with vegetable oils, but there was still no beneficial effect on cardiac (HR 1.13; 95% CI 0.83-1.54) or all-cause mortality (HR 1.07; 95% CI 0.90-1.27).

An explanation for the “seemingly paradoxical results” may lie in the corn oil used in the study diet, the authors suggest. “As the major vehicle for delivery of cholesterol to vascular tissues, low density lipoprotein is often considered a causal mediator for coronary heart disease,” they write, adding that replacing saturated fat with linoleic acid decreases the level of low density lipoproteins, which in turn would be thought to lower the risk of coronary heart disease.

However, the authors continue, consuming linoleic acids “produces a wide range of biochemical consequences . . . that could plausibly increase the risk of coronary heart disease.” Understanding how the benefits of each type of fat balance out could “help to explain why some agents that decrease low density lipoprotein have been shown to reduce the risk of coronary heart disease, while others have no clear effect, and still others might actually increase risk,” they say.

Collectively, Ramsden et al interpret the data to show “no clear benefit” of lowering serum cholesterol by replacing saturated fat with corn oils.

Times Have Changed

The trial was conducted between 1968 and 1973 in a nursing home and six state mental hospitals in Minnesota, but was not published until 1989 in Arteriosclerosis, Thrombosis, and Vascular Biology. The original researchers also found no differences in cardiovascular events and mortality, but focused on the presumed benefit of the vegetable-based diet for younger patients.

When the study was completed, the crude results were “clearly at odds with prevailing beliefs,” the authors of the new paper write. “One can speculate that investigators and sponsors would have wanted to distinguish between a failed theory and a failed trial before publication.” Those who initially did the Minnesota study might have also been concerned of heavy censoring by journals or the fact that there was “little or no scientific or clinical trial literature at the time” to support their findings, they add.

But whatever the reason the original investigators had for omitting “key results,” Ramsden and colleagues write that “there is growing recognition that incomplete publication of negative or inconclusive results can contribute to skewed research priorities and public health initiatives.”

Now, this information “adds to the literature in terms of reinforcing the fact that this isn’t a new debate,” Todd Olszewski, PhD (Providence College, RI), who was not part of either the Minnesota study or the current analysis, told TCTMD. Early challenges to the body of evidence supporting the diet-heart link “focused on how to distinguish between statistical association and causation. . . . This recent study shows this attempt to review preexisting data and to formulate new interpretations based on the data reviewed.”

Fifty years ago, both supporters and critics of the diet-heart hypotheses “challenged their respective study designs and techniques in part because many of the early diet-heart researchers came into cardiovascular epidemiology from different disciplines,” he explained. “By and large, uncertainty was always present and continues to be present when we question how best to convert large population-based studies into heart healthy recommendations for individuals.”

Because of all of the back and forth over the years, Olszewski said he is not sure if the new paper will have an “immediate impact on how doctors make dietary recommendations to their patients. . . . It may very well be an impossibility to be able to finally obtain definitive conclusive evidence with respect to the diet-heart hypothesis because of the complicated nature of the disease, the fact that we’re talking about a chronic disease that develops over years or decades, [and] that the American public sometimes has conflated association with causation.”

Future researchers and policy makers should “be able to clarify and explain the science supporting nutrition guidelines to the American public regardless of what those nutrition guidelines might say about the limiting of any form of dietary fat,” he added.

Clinical Outcomes, Not Surrogates

In an accompanying editorial, J. Lennert Veerman, MD, PhD (University of Queensland, Herston, Australia), writes that “ideally, recommendations should be based on clinical outcomes, not surrogates such as cholesterol concentration.” The caveat to that, however, is that “clinical outcomes do not point uniformly in the same direction in all studies,” he adds.

Additionally, while the randomized Minnesota trial is stronger, by default, than observational data, it cannot necessarily be generalized outside of a hospital or nursing home setting where meals are modulated and monitored.

With regard to the meta-analysis, Veerman says, “it is tempting to speculate that the strong belief that polyunsaturated fats are good for health might have led to a self-fulfilling prophesy, whereby more health conscious participants consume more polyunsaturated fats and residual confounding by other healthy lifestyle choices produces better health and longer survival.”

Regardless of whether a definitive answer will ever be reached with regard to the diet-heart debate, he concludes that “we should continue to eat (and advise others to eat) more fish, fruits, vegetables, and whole grains. We should avoid salt, sugar, industrial trans fats, and avoid over eating.”

  • Ramsden CE, Zamora D, Majchrzak-Hong S, et al. Re-evaluation of the traditional diet-heart hypothesis: analysis of recovered data from Minnesota Coronary Experiment (1968-73). BMJ. 2016;353:i1246.
  • Veerman JL. Dietary fats: a new look at old data challenges established wisdom. BMJ. 2016;353:i1512.

Related Stories:


  • Ramsden and Olszewski report no relevant conflicts of interest.
  • Veerman reports receiving research funding from Pfizer and the National Health and Medical Research Council.