Article Text

Download PDFPDF

Randomized controlled trial
Diet–heart disease hypothesis is unaffected by results of analysis of recovered data from Minnesota Coronary Experiment
  1. C Murray Skeaff,
  2. Jim I Mann
  1. Department of Human Nutrition, University of Otago, Dunedin, New Zealand
  1. Correspondence to: Professor Murray Skeaff, Department of Human Nutrition, University of Otago, P.O. Box 56, Dunedin 9054, New Zealand; murray.skeaff{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Commentary on: OpenUrlAbstract/FREE Full Text.


Replacement of a proportion of saturated fat with polyunsaturated fats is a cornerstone of dietary recommendations for the prevention of coronary heart disease.1 Ramsden and colleagues have hypothesised that n-6 polyunsaturates may increase the risk of heart disease.2

The Minnesota Coronary Experiment (MCE) was one of a small number of randomised controlled trials conducted and published between the mid-1960s and early 1990s to test the effect of replacing dietary saturated fat with polyunsaturated fat on coronary heart disease.3 The relative risk of coronary events in the treatment group was 1.08 (95% CI 0.84 to 1.37) and this result, reported in the trial's original paper, has been included in recent systematic reviews and meta-analyses on the topic of dietary fat and cardiovascular disease4 but not in the most recent Cochrane review because the duration of the intervention was too short.5 The title of the paper by Ramsden and colleagues is misleading because there is no result from the analysis of the recovered data that draws into question the original results of the MCE.


In the original trial, 9057 patients at six Minnesota state psychiatric hospitals and one nursing home were randomised to a control diet specified to be high in saturated fat (19%kJ) or an intervention diet specified to be high in polyunsaturated fat (20%kJ). Study meals were prepared and available to all participants when resident in an institution. Not all patients were in hospital continuously for the duration of the trial; many patients had more than one hospital stay. This is important because participants in the intervention group would likely have consumed meals more similar to the control diet when in the community. Fasting blood samples were collected and serum stored for the analysis of total cholesterol at a later date. Frantz et al3 explain: ‘when a subject left the hospital and then returned, blood for the cholesterol determination was drawn after he had been back on the diet for 3 weeks’.

Ramsden and colleagues did not re-analyse the results of the randomised controlled trial. Their original contribution is to report on the prospective association between change in serum cholesterol concentration and risk of death. This analysis of the recovered data is not ‘intention-to-treat’ but is limited to the 2355 participants who were exposed to study diets for more than 1 year and had cholesterol measurements at baseline and follow-up.


Prospective analysis showed that a decrease in serum cholesterol (0.78 mmol/L) from baseline was associated with an increased risk of death from all causes (1.22; 95% CI 1.14 to 1.32), irrespective of which treatment group participants were assigned to. The association was confined to participants who were aged at least 65 years.

The meta-analysis of randomised controlled trials of saturated fat replacement with polyunsaturated fat, which Ramsden et al3 include in their paper, uses results for the MCE extracted from the original paper so provides no new information to test the authors’ assertion that ‘…incomplete publication has contributed to overestimation of the benefits of replacing saturated fat with vegetable oils rich in linoleic acid’.


The non-continuous stay of participants in hospital would tend to bias the effect of the original intervention towards the null, furthermore, the delay in blood sampling of readmitted participants would tend to exaggerate the extent of cholesterol-lowering in the intervention group. The finding by Ramsden et al that a drop in cholesterol in those aged over 65 years increases death from all causes, regardless of dietary treatment allocation, may simply indicate that recent low cholesterol levels are a reflection of ill health. The importance of nutrient–disease relationships and recommendations that follow from them are based on the totality of evidence, not on results from single studies or even on a single research approach. The benefits of replacing saturated with polyunsaturated fats is supported by ecological data, cohort studies,6 carefully controlled experiments examining the effect of altering fat source on recognised biomarkers for coronary heart disease and randomised controlled trials.5

Implications for practice

In our view, this paper provides no new evidence with which to re-evaluate the diet–heart disease hypothesis or that the guidelines regarding the nature of dietary fat should be altered.



  • Competing interests None declared.

  • Provenance and peer review Commissioned; internally peer reviewed.