The Minnesota Coronary Experiment

Key Takeaway

The Minnesota Coronary Experiment showed that replacing much of the saturated fat in institutional diets with corn oil lowered total serum cholesterol. It did not show a survival benefit. That important result challenges a simple “lower cholesterol by any dietary substitution and lives will be saved” story, but it does not prove that saturated fat is protective or that every cholesterol-lowering strategy is harmful.

Why This 1960s Trial Still Matters

The Minnesota Coronary Experiment (MCE) has become a touchstone in debates about saturated fat, vegetable oils, and heart disease. The reason for that is obvious: it was a large, tightly controlled feeding trial, and the 2016 recovery of previously unpublished analyses revealed a result that was less tidy than the nutrition message many people had absorbed—pun intended. The intervention lowered serum cholesterol, yet the recovered survival analyses did not show that the intervention group lived longer [1].

The lesson is not that one historical study settles the fat debate. It is that nutrition claims should be judged by the outcomes people actually care about—heart attacks, disability, and survival—not only by a laboratory marker. It is also a case study in why complete reporting matters. When key analyses are missing, delayed, or difficult to retrieve, the scientific record can become more confident than the underlying evidence warrants.

What Did The MCE Ask?

By the 1960s, the “diet-heart hypothesis” had a clear logic: saturated fat raises blood cholesterol; higher cholesterol is associated with coronary heart disease; therefore, replacing saturated fat with polyunsaturated vegetable oils should lower cholesterol and reduce cardiovascular deaths. Linoleic acid—an omega-6 polyunsaturated fatty acid abundant in corn, safflower, and sunflower oils—was central to this approach. The MCE was designed to test the full chain of reasoning, not just the first link [1].

That distinction matters. A surrogate outcome is a measurable stand-in for a clinical outcome. In the MCE, serum cholesterol was the surrogate; death, heart attack, and atherosclerosis were the clinical outcomes. Surrogates can be useful, but they are not interchangeable with the outcomes they are meant to predict. A treatment can improve a number on a lab report without necessarily improving long-term health in every population, dose, or dietary context.

Ancel Keys and the Diet-Heart Era

Figure 1. Ancel Keys

University of Minnesota Archives

Ancel Keys was a physiologist and one of the most influential nutrition researchers of the twentieth century. He was a central figure in the MCE: Ivan Frantz was the principal investigator and Keys was the co-principal investigator [1]. Keys also led the Seven Countries Study, a long-running observational study that compared diet, lifestyle, and cardiovascular outcomes across populations. Its early findings helped focus attention on the relationships among dietary saturated fat, serum cholesterol, and coronary heart disease [2].

Keys’s work shaped the intellectual climate in which dietary policy developed, but he did not single-handedly create U.S. nutrition guidance. The 1977 Senate “Dietary Goals” helped popularize a public-health case for reducing fat, saturated fat, and cholesterol; the first formal Dietary Guidelines for Americans were issued by USDA and HHS in 1980 [3]. These recommendations reflected a broader body of epidemiology, mechanistic research, clinical studies, and public-health judgment. The MCE matters because it was a rare attempt to test the causal step that observational evidence alone cannot settle.

How the MCE Worked

The MCE was a double-blind, parallel-group randomized feeding trial conducted from 1968 to 1973 in six Minnesota state mental hospitals and one nursing home. Participants were adults aged 20 to 97. The historical record contains some inconsistent totals, but Ramsden and colleagues report that 9,570 people were randomized, and the researchers recovered completed survival analyses for a cohort of 9,423 [1]. The institutional setting was a blessing and curse: it gave investigators unusual control over meals, but it also resulted in a population unlike most people making food choices at home. More on this below.

The intervention replaced a large share of saturated fat with corn oil and corn-oil margarine. Saturated fat fell from about 18.5% to 9.2% of calories, while linoleic acid rose from roughly 3.4% to 13.2% of calories. The control diet retained a high saturated-fat content, but it was not a modern “usual diet”: it also contained more linoleic acid than the baseline diet and substantial industrial trans fat from the margarines and shortenings common at the time [1]. This means the trial tested one specific food substitution in one historical setting, not every possible version of “saturated fat versus vegetable oil.”

Figure 2. Redrawn from MCE dietary-composition data. The intervention sharply lowered saturated fat and raised linoleic acid; the control diet also raised linoleic acid modestly. Data: Ramsden et al., 2016 [1].

What the MCE Trial Found

On its biochemical target, the intervention worked. Among 2,355 participants exposed to the study diets for at least one year and with repeated cholesterol measurements, mean serum cholesterol fell 13.8% from baseline in the intervention group, compared with 1% in the control group [1]. In other words, the dietary contrast was large enough to move the marker it was intended to move.

Figure 3. Mean change in total serum cholesterol among participants with at least one year of diet exposure and longitudinal measurements (n=2,355). Data: Ramsden et al., 2016 [1].

The clinical-outcome story was different. Kaplan-Meier survival curves reproduced from a 1981 master’s thesis found no mortality benefit for the intervention in the full randomized cohort or in the prespecified sex and age groups. The curves suggested a possible unfavorable pattern among participants aged 65 or older, but the individual-level mortality data needed to repeat that comparison were not recovered, so its statistical significance could not be reassessed [1]. The 1989 publication likewise reported no mortality benefit in the full population.

The recovered autopsy analysis adds another caution, not a final verdict. Only 149 of 295 completed autopsy files were recovered. In this partial sample, at least one myocardial infarction was found in 31 of 76 intervention participants (41%) and 16 of 73 control participants (22%), an incidence-rate ratio of 1.90. There was no evidence that the intervention reduced coronary or aortic atherosclerosis. Because roughly half of the completed autopsy files were unavailable, the authors explicitly described these findings as provisional [1].

Limits in Generalizing the MCE

The MCE was unusually ambitious, but it was not a perfect modern trial. The original documents did not specify one primary outcome, which makes it harder to know which result should carry the greatest statistical weight. Individual exposure was also often shorter than the trial's calendar span. Although the diet phase lasted 41 to 56 months across hospitals, the cholesterol-analysis cohort had a mean follow-up of 2.9 years because participants were followed only while they remained in the institution [1].

Its setting is both a strength and a limitation. Centrally prepared meals and masked assignment reduced the dietary uncertainty that plagues free-living nutrition studies. At the same time, residents of state mental hospitals and a nursing home may have differed from community-dwelling adults in illness burden, medication use, smoking, mobility, and reasons for hospitalization. The results therefore cannot be assumed to apply directly to younger, healthier people choosing among contemporary foods.

The dietary comparison also matters. The intervention did not simply swap butter for olive oil; it substantially increased corn oil and linoleic acid across many institutional foods. The control diet contained trans fat that would now be recognized as harmful, and it was not metabolically neutral. Finally, the 2016 investigators recovered only selected datasets: not the complete individual-level mortality file, not all autopsies, and not cholesterol measurements for participants exposed for less than one year. These limitations do not nullify the findings, they define the boundaries of what the trial can show.

The Result that Needs the Most Careful Reading

The most provocative MCE result is often summarized as “lower cholesterol, increased mortality.” That is too strong. In the one-year-plus cohort, greater within-person reductions in serum cholesterol were associated with a higher risk of death: each 30 mg/dL decrease was associated with a 22% higher adjusted hazard of death (hazard ratio 1.22, 95% CI, 1.14 to 1.32). The association was stronger in people aged 65 or older and absent in those younger than 65 [1].

But this was not the same as the randomized intervention comparison. It was an observational analysis within the trial: people whose cholesterol fell more may also have differed in illness burden, frailty, weight change, or other factors linked to mortality. Ramsden and colleagues adjusted for several measured variables, including age, body mass index, blood pressure, and diet adherence, and ran sensitivity analyses using changes in weight and blood pressure. Those steps strengthen the analysis, but they cannot remove every form of reverse causation or confounding factors. The responsible interpretation is narrower: in this population and dataset, a larger fall in serum cholesterol did not behave as a straightforward signal of lower mortality risk.

Why the Reporting History Matters

The MCE did not vanish completely. A 1989 paper reported some findings, and an unpublished 1981 thesis contained survival analyses. Yet several analyses pre-specified in the original study documents - including subgroup survival results, the cholesterol-change and death analysis, and autopsy endpoints - had not entered the peer-reviewed evidence base until the 2016 re-analysis. “Incomplete publication” is more precise than claiming proven suppression, but it is still consequential. Scientific syntheses can only be as complete as the results available to them.

This is one reason the MCE is larger than a debate about corn oil. It illustrates the importance of preregistered outcomes, public protocols, data preservation, and reporting results that are inconvenient as well as results that are exciting. In nutrition, where interventions are difficult to blind and effects can take years to emerge, transparent outcome reporting is especially important.

What Later Evidence Adds

Ramsden and colleagues paired the MCE with a meta-analysis of five trials that specifically replaced saturated fat with linoleic-acid-rich oils. Across 10,808 participants, they found no clear benefit for coronary-heart-disease mortality or all-cause mortality [1]. That analysis asks a narrow question: what happens when saturated fat is replaced with high-linoleic-acid vegetable oils in the kinds of interventions available at the time?

Broader and more recent evidence asks a different question. A 2020 Cochrane review of long-term randomized trials of saturated-fat reduction found a probable reduction in combined cardiovascular events (risk ratio 0.79, 95% CI, 0.66 to 0.94) but little or no effect on all-cause mortality (0.96, 0.90 to 1.03) or cardiovascular mortality (0.95, 0.80 to 1.12) [4]. The difference is not a contradiction to ignore; it is a reminder that the replacement nutrient, the food source, baseline risk, trial duration, and outcome chosen all matter.

Current international guidance has not abandoned the emphasis on fat quality. The World Health Organization’s 2023 guideline recommends that most dietary fat be unsaturated, limits saturated fat to no more than 10% of energy intake, and emphasizes minimizing trans fat [5]. Still, the practical implication is more nuanced than “replace every saturated-fat food with any vegetable oil.” Food patterns matter: replacing butter with an oil-rich, minimally processed meal is not nutritionally equivalent to replacing it with a refined snack food, and no individual dietary decision should rest on one trial from a very different era.

What Health-Conscious Readers Should Take from the MCE

The MCE can support this conclusion	The MCE cannot support this conclusion
A major corn-oil substitution lowered total serum cholesterol in a controlled institutional setting.	Saturated fat is harmless, or any reduction in cholesterol is harmful.
Lowering a surrogate marker did not guarantee a mortality benefit in this trial.	All modern recommendations to reduce saturated fat are disproven.
Incomplete reporting can distort how confidently a field interprets its evidence.	The partial recovered autopsy sample proves the intervention caused heart attacks.

The MCE is best read as a corrective against nutritional overconfidence. It did not validate the simple prediction that replacing saturated fat with large amounts of corn oil would lower total cholesterol and, therefore, lower mortality. It also cannot tell us that every source of saturated fat has the same effect, that linoleic acid is intrinsically dangerous, or that present-day dietary guidelines should be discarded. Its enduring contribution is methodological: clinical outcomes matter, substitution matters, and all planned analyses deserve to be seen.

References

1. 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. doi:10.1136/bmj.i1246

2. Keys A, Menotti A, Aravanis C, et al. The seven countries study: 2,289 deaths in 15 years. Prev Med. 1984;13(2):141-154. doi:10.1016/0091-7435(84)90047-1

3. 1980 Dietary Guidelines for Americans | Dietary Guidelines for Americans. Accessed June 23, 2026. https://www.dietaryguidelines.gov/about-dietary-guidelines/previous-editions/1980-dietary-guidelines-americans

4. Hooper L, Martin N, Jimoh OF, Kirk C, Foster E, Abdelhamid AS. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database Syst Rev. 2020;2020(5):CD011737. doi:10.1002/14651858.CD011737.pub2

5. Saturated Fatty Acid and Trans-Fatty Acid Intake for Adults and Children: WHO Guideline. World Health Organization; 2023. Accessed June 23, 2026. http://www.ncbi.nlm.nih.gov/books/NBK593397/